# File ext/json/lib/json/add/string.rb, line 11
def self.json_create(object)
  object["raw"].pack("C*")
end

Raw Strings are JSON Objects (the raw bytes are stored in an array for the key “raw”). The Ruby String can be created by this class method.

static VALUE
rb_str_init(int argc, VALUE *argv, VALUE str)
{
    static ID keyword_ids[2];
    VALUE orig, opt, venc, vcapa;
    VALUE kwargs[2];
    rb_encoding *enc = 0;
    int n;

    if (!keyword_ids[0]) {
        keyword_ids[0] = rb_id_encoding();
        CONST_ID(keyword_ids[1], "capacity");
    }

    n = rb_scan_args(argc, argv, "01:", &orig, &opt);
    if (!NIL_P(opt)) {
        rb_get_kwargs(opt, keyword_ids, 0, 2, kwargs);
        venc = kwargs[0];
        vcapa = kwargs[1];
        if (!UNDEF_P(venc) && !NIL_P(venc)) {
            enc = rb_to_encoding(venc);
        }
        if (!UNDEF_P(vcapa) && !NIL_P(vcapa)) {
            long capa = NUM2LONG(vcapa);
            long len = 0;
            int termlen = enc ? rb_enc_mbminlen(enc) : 1;

            if (capa < STR_BUF_MIN_SIZE) {
                capa = STR_BUF_MIN_SIZE;
            }
            if (n == 1) {
                StringValue(orig);
                len = RSTRING_LEN(orig);
                if (capa < len) {
                    capa = len;
                }
                if (orig == str) n = 0;
            }
            str_modifiable(str);
            if (STR_EMBED_P(str) || FL_TEST(str, STR_SHARED|STR_NOFREE)) {
                /* make noembed always */
                const size_t size = (size_t)capa + termlen;
                const char *const old_ptr = RSTRING_PTR(str);
                const size_t osize = RSTRING_LEN(str) + TERM_LEN(str);
                char *new_ptr = ALLOC_N(char, size);
                if (STR_EMBED_P(str)) RUBY_ASSERT((long)osize <= str_embed_capa(str));
                memcpy(new_ptr, old_ptr, osize < size ? osize : size);
                FL_UNSET_RAW(str, STR_SHARED|STR_NOFREE);
                RSTRING(str)->as.heap.ptr = new_ptr;
            }
            else if (STR_HEAP_SIZE(str) != (size_t)capa + termlen) {
                SIZED_REALLOC_N(RSTRING(str)->as.heap.ptr, char,
                        (size_t)capa + termlen, STR_HEAP_SIZE(str));
            }
            STR_SET_LEN(str, len);
            TERM_FILL(&RSTRING(str)->as.heap.ptr[len], termlen);
            if (n == 1) {
                memcpy(RSTRING(str)->as.heap.ptr, RSTRING_PTR(orig), len);
                rb_enc_cr_str_exact_copy(str, orig);
            }
            FL_SET(str, STR_NOEMBED);
            RSTRING(str)->as.heap.aux.capa = capa;
        }
        else if (n == 1) {
            rb_str_replace(str, orig);
        }
        if (enc) {
            rb_enc_associate(str, enc);
            ENC_CODERANGE_CLEAR(str);
        }
    }
    else if (n == 1) {
        rb_str_replace(str, orig);
    }
    return str;
}

Returns a new String object containing the given string.

The options are optional keyword options (see below).

With no argument given and keyword encoding also not given, returns an empty string with the Encoding ASCII-8BIT:

s = String.new # => ""
s.encoding     # => #<Encoding:ASCII-8BIT>

With argument string given and keyword option encoding not given, returns a new string with the same encoding as string:

s0 = 'foo'.encode(Encoding::UTF_16)
s1 = String.new(s0)
s1.encoding # => #<Encoding:UTF-16 (dummy)>

(Unlike String.new, a string literal like '' or a here document literal always has script encoding.)

With keyword option encoding given, returns a string with the specified encoding; the encoding may be an Encoding object, an encoding name, or an encoding name alias:

String.new(encoding: Encoding::US_ASCII).encoding        # => #<Encoding:US-ASCII>
String.new('', encoding: Encoding::US_ASCII).encoding    # => #<Encoding:US-ASCII>
String.new('foo', encoding: Encoding::US_ASCII).encoding # => #<Encoding:US-ASCII>
String.new('foo', encoding: 'US-ASCII').encoding         # => #<Encoding:US-ASCII>
String.new('foo', encoding: 'ASCII').encoding            # => #<Encoding:US-ASCII>

The given encoding need not be valid for the string’s content, and its validity is not checked:

s = String.new('こんにちは', encoding: 'ascii')
s.valid_encoding? # => false

But the given encoding itself is checked:

String.new('foo', encoding: 'bar') # Raises ArgumentError.

With keyword option capacity given, the given value is advisory only, and may or may not set the size of the internal buffer, which may in turn affect performance:

String.new('foo', capacity: 1)    # Buffer size is at least 4 (includes terminal null byte).
String.new('foo', capacity: 4096) # Buffer size is at least 4;
                                  # may be equal to, greater than, or less than 4096.

static VALUE
rb_str_s_try_convert(VALUE dummy, VALUE str)
{
    return rb_check_string_type(str);
}

Attempts to convert the given object to a string.

If object is already a string, returns object, unmodified.

Otherwise if object responds to :to_str, calls object.to_str and returns the result.

Returns nil if object does not respond to :to_str.

Raises an exception unless object.to_str returns a string.

static VALUE
rb_str_format_m(VALUE str, VALUE arg)
{
    VALUE tmp = rb_check_array_type(arg);

    if (!NIL_P(tmp)) {
        return rb_str_format(RARRAY_LENINT(tmp), RARRAY_CONST_PTR(tmp), str);
    }
    return rb_str_format(1, &arg, str);
}

Returns the result of formatting object into the format specifications contained in self (see Format Specifications):

'%05d' % 123 # => "00123"

If self contains multiple format specifications, object must be an array or hash containing the objects to be formatted:

'%-5s: %016x' % [ 'ID', self.object_id ]                # => "ID   : 00002b054ec93168"
'foo = %{foo}' % {foo: 'bar'}                           # => "foo = bar"
'foo = %{foo}, baz = %{baz}' % {foo: 'bar', baz: 'bat'} # => "foo = bar, baz = bat"

Related: see Converting to New String.

VALUE
rb_str_times(VALUE str, VALUE times)
{
    VALUE str2;
    long n, len;
    char *ptr2;
    int termlen;

    if (times == INT2FIX(1)) {
        return str_duplicate(rb_cString, str);
    }
    if (times == INT2FIX(0)) {
        str2 = str_alloc_embed(rb_cString, 0);
        rb_enc_copy(str2, str);
        return str2;
    }
    len = NUM2LONG(times);
    if (len < 0) {
        rb_raise(rb_eArgError, "negative argument");
    }
    if (RSTRING_LEN(str) == 1 && RSTRING_PTR(str)[0] == 0) {
        if (STR_EMBEDDABLE_P(len, 1)) {
            str2 = str_alloc_embed(rb_cString, len + 1);
            memset(RSTRING_PTR(str2), 0, len + 1);
        }
        else {
            str2 = str_alloc_heap(rb_cString);
            RSTRING(str2)->as.heap.aux.capa = len;
            RSTRING(str2)->as.heap.ptr = ZALLOC_N(char, (size_t)len + 1);
        }
        STR_SET_LEN(str2, len);
        rb_enc_copy(str2, str);
        return str2;
    }
    if (len && LONG_MAX/len <  RSTRING_LEN(str)) {
        rb_raise(rb_eArgError, "argument too big");
    }

    len *= RSTRING_LEN(str);
    termlen = TERM_LEN(str);
    str2 = str_enc_new(rb_cString, 0, len, STR_ENC_GET(str));
    ptr2 = RSTRING_PTR(str2);
    if (len) {
        n = RSTRING_LEN(str);
        memcpy(ptr2, RSTRING_PTR(str), n);
        while (n <= len/2) {
            memcpy(ptr2 + n, ptr2, n);
            n *= 2;
        }
        memcpy(ptr2 + n, ptr2, len-n);
    }
    STR_SET_LEN(str2, len);
    TERM_FILL(&ptr2[len], termlen);
    rb_enc_cr_str_copy_for_substr(str2, str);

    return str2;
}

Returns a new string containing n copies of self:

'Ho!' * 3 # => "Ho!Ho!Ho!"
'No!' * 0 # => ""

Related: see Converting to New String.

VALUE
rb_str_plus(VALUE str1, VALUE str2)
{
    VALUE str3;
    rb_encoding *enc;
    char *ptr1, *ptr2, *ptr3;
    long len1, len2;
    int termlen;

    StringValue(str2);
    enc = rb_enc_check_str(str1, str2);
    RSTRING_GETMEM(str1, ptr1, len1);
    RSTRING_GETMEM(str2, ptr2, len2);
    termlen = rb_enc_mbminlen(enc);
    if (len1 > LONG_MAX - len2) {
        rb_raise(rb_eArgError, "string size too big");
    }
    str3 = str_enc_new(rb_cString, 0, len1+len2, enc);
    ptr3 = RSTRING_PTR(str3);
    memcpy(ptr3, ptr1, len1);
    memcpy(ptr3+len1, ptr2, len2);
    TERM_FILL(&ptr3[len1+len2], termlen);

    ENCODING_CODERANGE_SET(str3, rb_enc_to_index(enc),
                           ENC_CODERANGE_AND(ENC_CODERANGE(str1), ENC_CODERANGE(str2)));
    RB_GC_GUARD(str1);
    RB_GC_GUARD(str2);
    return str3;
}

Returns a new string containing other_string concatenated to self:

'Hello from ' + self.to_s # => "Hello from main"

Related: see Converting to New String.

static VALUE
str_uplus(VALUE str)
{
    if (OBJ_FROZEN(str) || CHILLED_STRING_P(str)) {
        return rb_str_dup(str);
    }
    else {
        return str;
    }
}

Returns self if self is not frozen and can be mutated without warning issuance.

Otherwise returns self.dup, which is not frozen.

Related: see Freezing/Unfreezing.

static VALUE
str_uminus(VALUE str)
{
    if (!BARE_STRING_P(str) && !rb_obj_frozen_p(str)) {
        str = rb_str_dup(str);
    }
    return rb_fstring(str);
}

Returns a frozen string equal to self.

The returned string is self if and only if all of the following are true:

• self is already frozen.

• self is an instance of String (rather than of a subclass of String)

• self has no instance variables set on it.

Otherwise, the returned string is a frozen copy of self.

Returning self, when possible, saves duplicating self; see Data deduplication.

It may also save duplicating other, already-existing, strings:

s0 = 'foo'
s1 = 'foo'
s0.object_id == s1.object_id       # => false
(-s0).object_id == (-s1).object_id # => true

Note that method -@ is convenient for defining a constant:

FileName = -'config/database.yml'

While its alias dedup is better suited for chaining:

'foo'.dedup.gsub!('o')

Related: see Freezing/Unfreezing.

VALUE
rb_str_concat(VALUE str1, VALUE str2)
{
    unsigned int code;
    rb_encoding *enc = STR_ENC_GET(str1);
    int encidx;

    if (RB_INTEGER_TYPE_P(str2)) {
        if (rb_num_to_uint(str2, &code) == 0) {
        }
        else if (FIXNUM_P(str2)) {
            rb_raise(rb_eRangeError, "%ld out of char range", FIX2LONG(str2));
        }
        else {
            rb_raise(rb_eRangeError, "bignum out of char range");
        }
    }
    else {
        return rb_str_append(str1, str2);
    }

    encidx = rb_ascii8bit_appendable_encoding_index(enc, code);

    if (encidx >= 0) {
        rb_str_buf_cat_byte(str1, (unsigned char)code);
    }
    else {
        long pos = RSTRING_LEN(str1);
        int cr = ENC_CODERANGE(str1);
        int len;
        char *buf;

        switch (len = rb_enc_codelen(code, enc)) {
          case ONIGERR_INVALID_CODE_POINT_VALUE:
            rb_raise(rb_eRangeError, "invalid codepoint 0x%X in %s", code, rb_enc_name(enc));
            break;
          case ONIGERR_TOO_BIG_WIDE_CHAR_VALUE:
          case 0:
            rb_raise(rb_eRangeError, "%u out of char range", code);
            break;
        }
        buf = ALLOCA_N(char, len + 1);
        rb_enc_mbcput(code, buf, enc);
        if (rb_enc_precise_mbclen(buf, buf + len + 1, enc) != len) {
            rb_raise(rb_eRangeError, "invalid codepoint 0x%X in %s", code, rb_enc_name(enc));
        }
        rb_str_resize(str1, pos+len);
        memcpy(RSTRING_PTR(str1) + pos, buf, len);
        if (cr == ENC_CODERANGE_7BIT && code > 127) {
            cr = ENC_CODERANGE_VALID;
        }
        else if (cr == ENC_CODERANGE_BROKEN) {
            cr = ENC_CODERANGE_UNKNOWN;
        }
        ENC_CODERANGE_SET(str1, cr);
    }
    return str1;
}

Appends a string representation of object to self; returns self.

If object is a string, appends it to self:

s = 'foo'
s << 'bar' # => "foobar"
s          # => "foobar"

If object is an integer, its value is considered a codepoint; converts the value to a character before concatenating:

s = 'foo'
s << 33 # => "foo!"

Additionally, if the codepoint is in range 0..0xff and the encoding of self is Encoding::US_ASCII, changes the encoding to Encoding::ASCII_8BIT:

s = 'foo'.encode(Encoding::US_ASCII)
s.encoding # => #<Encoding:US-ASCII>
s << 0xff  # => "foo\xFF"
s.encoding # => #<Encoding:BINARY (ASCII-8BIT)>

Raises RangeError if that codepoint is not representable in the encoding of self:

s = 'foo'
s.encoding              # => <Encoding:UTF-8>
s << 0x00110000         # 1114112 out of char range (RangeError)
s = 'foo'.encode(Encoding::EUC_JP)
s << 0x00800080         # invalid codepoint 0x800080 in EUC-JP (RangeError)

Related: see Modifying.

static VALUE
rb_str_cmp_m(VALUE str1, VALUE str2)
{
    int result;
    VALUE s = rb_check_string_type(str2);
    if (NIL_P(s)) {
        return rb_invcmp(str1, str2);
    }
    result = rb_str_cmp(str1, s);
    return INT2FIX(result);
}

Compares self and other, evaluating their contents, not their lengths.

Returns:

• -1, if self is smaller.

• 0, if the two are equal.

• 1, if self is larger.

• nil, if the two are incomparable.

Examples:

'a'  <=> 'b'  # => -1
'a'  <=> 'ab' # => -1
'a'  <=> 'a'  # => 0
'b'  <=> 'a'  # => 1
'ab' <=> 'a'  # => 1
'a'  <=> :a   # => nil

Class String includes module Comparable, each of whose methods uses String#<=> for comparison.

Related: see Comparing.

VALUE
rb_str_equal(VALUE str1, VALUE str2)
{
    if (str1 == str2) return Qtrue;
    if (!RB_TYPE_P(str2, T_STRING)) {
        if (!rb_respond_to(str2, idTo_str)) {
            return Qfalse;
        }
        return rb_equal(str2, str1);
    }
    return rb_str_eql_internal(str1, str2);
}

Returns whether object is equal to self.

When object is a string, returns whether object has the same length and content as self:

s = 'foo'
s == 'foo'  # => true
s == 'food' # => false
s == 'FOO'  # => false

Returns false if the two strings’ encodings are not compatible:

"\u{e4 f6 fc}".encode(Encoding::ISO_8859_1) == ("\u{c4 d6 dc}") # => false

When object is not a string:

• If object responds to method to_str, object == self is called and its return value is returned.

• If object does not respond to to_str, false is returned.

Related: Comparing.

static VALUE
rb_str_match(VALUE x, VALUE y)
{
    switch (OBJ_BUILTIN_TYPE(y)) {
      case T_STRING:
        rb_raise(rb_eTypeError, "type mismatch: String given");

      case T_REGEXP:
        return rb_reg_match(y, x);

      default:
        return rb_funcall(y, idEqTilde, 1, x);
    }
}

When object is a Regexp, returns the index of the first substring in self matched by object, or nil if no match is found; updates Regexp-related global variables:

'foo' =~ /f/ # => 0
$~           # => #<MatchData "f">
'foo' =~ /o/ # => 1
$~           # => #<MatchData "o">
'foo' =~ /x/ # => nil
$~           # => nil

Note that string =~ regexp is different from regexp =~ string (see Regexp#=~):

number = nil
'no. 9' =~ /(?<number>\d+)/ # => 4
number                      # => nil # Not assigned.
/(?<number>\d+)/ =~ 'no. 9' # => 4
number                      # => "9" # Assigned.

If object is not a Regexp, returns the value returned by object =~ self.

Related: see Querying.

static VALUE
rb_str_aref_m(int argc, VALUE *argv, VALUE str)
{
    if (argc == 2) {
        if (RB_TYPE_P(argv[0], T_REGEXP)) {
            return rb_str_subpat(str, argv[0], argv[1]);
        }
        else {
            return rb_str_substr_two_fixnums(str, argv[0], argv[1], TRUE);
        }
    }
    rb_check_arity(argc, 1, 2);
    return rb_str_aref(str, argv[0]);
}

Returns the substring of self specified by the arguments.

Form self[index]

With non-negative integer argument index given, returns the 1-character substring found in self at character offset index:

'hello'[0]    # => "h"
'hello'[4]    # => "o"
'hello'[5]    # => nil
'Привет'[2]   # => "и"
'こんにちは'[4] # => "は"

With negative integer argument index given, counts backward from the end of self:

'hello'[-1] # => "o"
'hello'[-5] # => "h"
'hello'[-6] # => nil

Form self[start, length]

With integer arguments start and length given, returns a substring of size length characters (as available) beginning at character offset specified by start.

If argument start is non-negative, the offset is start:

'hello'[0, 1]  # => "h"
'hello'[0, 5]  # => "hello"
'hello'[0, 6]  # => "hello"
'hello'[2, 3]  # => "llo"
'hello'[2, 0]  # => ""
'hello'[2, -1] # => nil

If argument start is negative, counts backward from the end of self:

'hello'[-1, 1] # => "o"
'hello'[-5, 5] # => "hello"
'hello'[-1, 0] # => ""
'hello'[-6, 5] # => nil

Special case: if start equals the length of self, returns a new empty string:

'hello'[5, 3]  # => ""

Form self[range]

With Range argument range given, forms substring self[range.start, range.size]:

'hello'[0..2]  # => "hel"
'hello'[0, 3]  # => "hel"

'hello'[0...2] # => "he"
'hello'[0, 2]  # => "he"

'hello'[0, 0]  # => ""
'hello'[0...0] # => ""

Form self[regexp, capture = 0]

With Regexp argument regexp given and capture as zero, searches for a matching substring in self; updates Regexp-related global variables:

'hello'[/ell/]     # => "ell"
'hello'[/l+/]      # => "ll"
'hello'[//]        # => ""
'hello'[/nosuch/]  # => nil

With capture as a positive integer n, returns the +n+th matched group:

'hello'[/(h)(e)(l+)(o)/]    # => "hello"
'hello'[/(h)(e)(l+)(o)/, 1] # => "h"
$1                          # => "h"
'hello'[/(h)(e)(l+)(o)/, 2] # => "e"
$2                          # => "e"
'hello'[/(h)(e)(l+)(o)/, 3] # => "ll"
'hello'[/(h)(e)(l+)(o)/, 4] # => "o"
'hello'[/(h)(e)(l+)(o)/, 5] # => nil

Form self[substring]

With string argument substring given, returns the matching substring of self, if found:

'hello'['ell']      # => "ell"
'hello'['']         # => ""
'hello'['nosuch']   # => nil
'Привет'['ив']      # => "ив"
'こんにちは'['んにち'] # => "んにち"

Related: see Converting to New String.

static VALUE
rb_str_aset_m(int argc, VALUE *argv, VALUE str)
{
    if (argc == 3) {
        if (RB_TYPE_P(argv[0], T_REGEXP)) {
            rb_str_subpat_set(str, argv[0], argv[1], argv[2]);
        }
        else {
            rb_str_update(str, NUM2LONG(argv[0]), NUM2LONG(argv[1]), argv[2]);
        }
        return argv[2];
    }
    rb_check_arity(argc, 2, 3);
    return rb_str_aset(str, argv[0], argv[1]);
}

Returns self with all, a substring, or none of its contents replaced; returns the argument other_string.

Form self[index] = other_string

With non-negative integer argument index given, searches for the 1-character substring found in self at character offset index:

s = 'hello'
s[0] = 'foo' # => "foo"
s            # => "fooello"

s = 'hello'
s[4] = 'foo' # => "foo"
s            # => "hellfoo"

s = 'hello'
s[5] = 'foo' # => "foo"
s            # => "hellofoo"

s = 'hello'
s[6] = 'foo' # Raises IndexError: index 6 out of string.

With negative integer argument index given, counts backward from the end of self:

s = 'hello'
s[-1] = 'foo'  # => "foo"
s              # => "hellfoo"

s = 'hello'
s[-5] = 'foo'  # => "foo"
s              # => "fooello"

s = 'hello'
s[-6] = 'foo'  # Raises IndexError: index -6 out of string.

Form self[start, length] = other_string

With integer arguments start and length given, searches for a substring of size length characters (as available) beginning at character offset specified by start.

If argument start is non-negative, the offset is start:

s = 'hello'
s[0, 1] = 'foo'  # => "foo"
s                # => "fooello"

s = 'hello'
s[0, 5] = 'foo'  # => "foo"
s                # => "foo"

s = 'hello'
s[0, 9] = 'foo'  # => "foo"
s                # => "foo"

s = 'hello'
s[2, 0] = 'foo'  # => "foo"
s                # => "hefoollo"

s = 'hello'
s[2, -1] = 'foo' # Raises IndexError: negative length -1.

If argument start is negative, counts backward from the end of self:

s = 'hello'
s[-1, 1] = 'foo' # => "foo"
s                # => "hellfoo"

s = 'hello'
s[-1, 9] = 'foo' # => "foo"
s                # => "hellfoo"

s = 'hello'
s[-5, 2] = 'foo' # => "foo"
s                # => "foollo"

s = 'hello'
s[-3, 0] = 'foo' # => "foo"
s                # => "hefoollo"

s = 'hello'
s[-6, 2] = 'foo' # Raises IndexError: index -6 out of string.

Special case: if start equals the length of self, the argument is appended to self:

s = 'hello'
s[5, 3] = 'foo' # => "foo"
s               # => "hellofoo"

Form self[range] = other_string

With Range argument range given, equivalent to self[range.start, range.size] = other_string:

s0 = 'hello'
s1 = 'hello'
s0[0..2] = 'foo' # => "foo"
s1[0, 3] = 'foo' # => "foo"
s0               # => "foolo"
s1               # => "foolo"

s = 'hello'
s[0...2] = 'foo' # => "foo"
s                # => "foollo"

s = 'hello'
s[0...0] = 'foo' # => "foo"
s                # => "foohello"

s = 'hello'
s[9..10] = 'foo' # Raises RangeError: 9..10 out of range

Form self[regexp, capture = 0] = other_string

With Regexp argument regexp given and capture as zero, searches for a matching substring in self; updates Regexp-related global variables:

s = 'hello'
s[/l/] = 'L'       # => "L"
[$`, $&, $']       # => ["he", "l", "lo"]
s[/eLlo/] = 'owdy' # => "owdy"
[$`, $&, $']       # => ["h", "eLlo", ""]
s[/eLlo/] = 'owdy' # Raises IndexError: regexp not matched.
[$`, $&, $']       # => [nil, nil, nil]

With capture as a positive integer n, searches for the +n+th matched group:

s = 'hello'
s[/(h)(e)(l+)(o)/] = 'foo'    # => "foo"
[$`, $&, $']                  # => ["", "hello", ""]

s = 'hello'
s[/(h)(e)(l+)(o)/, 1] = 'foo' # => "foo"
s                             # => "fooello"
[$`, $&, $']                  # => ["", "hello", ""]

s = 'hello'
s[/(h)(e)(l+)(o)/, 2] = 'foo' # => "foo"
s                             # => "hfoollo"
[$`, $&, $']                  # => ["", "hello", ""]

s = 'hello'
s[/(h)(e)(l+)(o)/, 4] = 'foo' # => "foo"
s                             # => "hellfoo"
[$`, $&, $']                  # => ["", "hello", ""]

s = 'hello'
# => "hello"
s[/(h)(e)(l+)(o)/, 5] = 'foo  # Raises IndexError: index 5 out of regexp.

s = 'hello'
s[/nosuch/] = 'foo'           # Raises IndexError: regexp not matched.

Form self[substring] = other_string

With string argument substring given:

s = 'hello'
s['l'] = 'foo'  # => "foo"
s  # => "hefoolo"

s = 'hello'
s['ll'] = 'foo'  # => "foo"
s  # => "hefooo"

s = 'Привет'
s['ив'] = 'foo'  # => "foo"
s  # => "Прfooет"

s = 'こんにちは'
s['んにち'] = 'foo'  # => "foo"
s  # => "こfooは"

s['nosuch'] = 'foo' # Raises IndexError: string not matched.

Related: see Modifying.

VALUE
rb_str_append_as_bytes(int argc, VALUE *argv, VALUE str)
{
    long needed_capacity = 0;
    volatile VALUE t0;
    enum ruby_value_type *types = ALLOCV_N(enum ruby_value_type, t0, argc);

    for (int index = 0; index < argc; index++) {
        VALUE obj = argv[index];
        enum ruby_value_type type = types[index] = rb_type(obj);
        switch (type) {
          case T_FIXNUM:
          case T_BIGNUM:
            needed_capacity++;
            break;
          case T_STRING:
            needed_capacity += RSTRING_LEN(obj);
            break;
          default:
            rb_raise(
                rb_eTypeError,
                "wrong argument type %"PRIsVALUE" (expected String or Integer)",
                rb_obj_class(obj)
            );
            break;
        }
    }

    str_ensure_available_capa(str, needed_capacity);
    char *sptr = RSTRING_END(str);

    for (int index = 0; index < argc; index++) {
        VALUE obj = argv[index];
        enum ruby_value_type type = types[index];
        switch (type) {
          case T_FIXNUM:
          case T_BIGNUM: {
            argv[index] = obj = rb_int_and(obj, INT2FIX(0xff));
            char byte = (char)(NUM2INT(obj) & 0xFF);
            *sptr = byte;
            sptr++;
            break;
          }
          case T_STRING: {
            const char *ptr;
            long len;
            RSTRING_GETMEM(obj, ptr, len);
            memcpy(sptr, ptr, len);
            sptr += len;
            break;
          }
          default:
            rb_bug("append_as_bytes arguments should have been validated");
        }
    }

    STR_SET_LEN(str, RSTRING_LEN(str) + needed_capacity);
    TERM_FILL(sptr, TERM_LEN(str)); /* sentinel */

    int cr = ENC_CODERANGE(str);
    switch (cr) {
      case ENC_CODERANGE_7BIT: {
        for (int index = 0; index < argc; index++) {
            VALUE obj = argv[index];
            enum ruby_value_type type = types[index];
            switch (type) {
              case T_FIXNUM:
              case T_BIGNUM: {
                if (!ISASCII(NUM2INT(obj))) {
                    goto clear_cr;
                }
                break;
              }
              case T_STRING: {
                if (ENC_CODERANGE(obj) != ENC_CODERANGE_7BIT) {
                    goto clear_cr;
                }
                break;
              }
              default:
                rb_bug("append_as_bytes arguments should have been validated");
            }
        }
        break;
      }
      case ENC_CODERANGE_VALID:
        if (ENCODING_GET_INLINED(str) == ENCINDEX_ASCII_8BIT) {
            goto keep_cr;
        }
        else {
            goto clear_cr;
        }
        break;
      default:
        goto clear_cr;
        break;
    }

    RB_GC_GUARD(t0);

  clear_cr:
    // If no fast path was hit, we clear the coderange.
    // append_as_bytes is predominantly meant to be used in
    // buffering situation, hence it's likely the coderange
    // will never be scanned, so it's not worth spending time
    // precomputing the coderange except for simple and common
    // situations.
    ENC_CODERANGE_CLEAR(str);
  keep_cr:
    return str;
}

Concatenates each object in objects into self; returns self; performs no encoding validation or conversion:

s = 'foo'
s.append_as_bytes(" \xE2\x82") # => "foo \xE2\x82"
s.valid_encoding?              # => false
s.append_as_bytes("\xAC 12")
s.valid_encoding?              # => true

When a given object is an integer, the value is considered an 8-bit byte; if the integer occupies more than one byte (i.e,. is greater than 255), appends only the low-order byte (similar to String#setbyte):

s = ""
s.append_as_bytes(0, 257) # => "\u0000\u0001"
s.bytesize                # => 2

Related: see Modifying.

static VALUE
rb_str_is_ascii_only_p(VALUE str)
{
    int cr = rb_enc_str_coderange(str);

    return RBOOL(cr == ENC_CODERANGE_7BIT);
}

Returns whether self contains only ASCII characters:

'abc'.ascii_only?         # => true
"abc\u{6666}".ascii_only? # => false

Related: see Querying.

static VALUE
rb_str_b(VALUE str)
{
    VALUE str2;
    if (STR_EMBED_P(str)) {
        str2 = str_alloc_embed(rb_cString, RSTRING_LEN(str) + TERM_LEN(str));
    }
    else {
        str2 = str_alloc_heap(rb_cString);
    }
    str_replace_shared_without_enc(str2, str);

    if (rb_enc_asciicompat(STR_ENC_GET(str))) {
        // BINARY strings can never be broken; they're either 7-bit ASCII or VALID.
        // If we know the receiver's code range then we know the result's code range.
        int cr = ENC_CODERANGE(str);
        switch (cr) {
          case ENC_CODERANGE_7BIT:
            ENC_CODERANGE_SET(str2, ENC_CODERANGE_7BIT);
            break;
          case ENC_CODERANGE_BROKEN:
          case ENC_CODERANGE_VALID:
            ENC_CODERANGE_SET(str2, ENC_CODERANGE_VALID);
            break;
          default:
            ENC_CODERANGE_CLEAR(str2);
            break;
        }
    }

    return str2;
}

Returns a copy of self that has ASCII-8BIT encoding; the underlying bytes are not modified:

s = "\x99"
s.encoding   # => #<Encoding:UTF-8>
t = s.b      # => "\x99"
t.encoding   # => #<Encoding:ASCII-8BIT>

s = "\u4095" # => "䂕"
s.encoding   # => #<Encoding:UTF-8>
s.bytes      # => [228, 130, 149]
t = s.b      # => "\xE4\x82\x95"
t.encoding   # => #<Encoding:ASCII-8BIT>
t.bytes      # => [228, 130, 149]

Related: see Converting to New String.

static VALUE
rb_str_byteindex_m(int argc, VALUE *argv, VALUE str)
{
    VALUE sub;
    VALUE initpos;
    long pos;

    if (rb_scan_args(argc, argv, "11", &sub, &initpos) == 2) {
        long slen = RSTRING_LEN(str);
        pos = NUM2LONG(initpos);
        if (pos < 0 ? (pos += slen) < 0 : pos > slen) {
            if (RB_TYPE_P(sub, T_REGEXP)) {
                rb_backref_set(Qnil);
            }
            return Qnil;
        }
    }
    else {
        pos = 0;
    }

    str_ensure_byte_pos(str, pos);

    if (RB_TYPE_P(sub, T_REGEXP)) {
        if (rb_reg_search(sub, str, pos, 0) >= 0) {
            VALUE match = rb_backref_get();
            struct re_registers *regs = RMATCH_REGS(match);
            pos = BEG(0);
            return LONG2NUM(pos);
        }
    }
    else {
        StringValue(sub);
        pos = rb_str_byteindex(str, sub, pos);
        if (pos >= 0) return LONG2NUM(pos);
    }
    return Qnil;
}

Returns the 0-based integer index of a substring of self specified by object (a string or Regexp) and offset, or nil if there is no such substring; the returned index is the count of bytes (not characters).

When object is a string, returns the index of the first found substring equal to object:

s = 'foo'          # => "foo"
s.size             # => 3 # Three 1-byte characters.
s.bytesize         # => 3 # Three bytes.
s.byteindex('f')   # => 0
s.byteindex('o')   # => 1
s.byteindex('oo')  # => 1
s.byteindex('ooo') # => nil

When object is a Regexp, returns the index of the first found substring matching object; updates Regexp-related global variables:

s = 'foo'
s.byteindex(/f/)   # => 0
$~                 # => #<MatchData "f">
s.byteindex(/o/)   # => 1
s.byteindex(/oo/)  # => 1
s.byteindex(/ooo/) # => nil
$~                 # => nil

Integer argument offset, if given, specifies the 0-based index of the byte where searching is to begin.

When offset is non-negative, searching begins at byte position offset:

s = 'foo'
s.byteindex('o', 1) # => 1
s.byteindex('o', 2) # => 2
s.byteindex('o', 3) # => nil

When offset is negative, counts backward from the end of self:

s = 'foo'
s.byteindex('o', -1) # => 2
s.byteindex('o', -2) # => 1
s.byteindex('o', -3) # => 1
s.byteindex('o', -4) # => nil

Raises IndexError if the byte at offset is not the first byte of a character:

s = "\uFFFF\uFFFF"       # => "\uFFFF\uFFFF"
s.size                   # => 2 # Two 3-byte characters.
s.bytesize               # => 6 # Six bytes.
s.byteindex("\uFFFF")    # => 0
s.byteindex("\uFFFF", 1) # Raises IndexError
s.byteindex("\uFFFF", 2) # Raises IndexError
s.byteindex("\uFFFF", 3) # => 3
s.byteindex("\uFFFF", 4) # Raises IndexError
s.byteindex("\uFFFF", 5) # Raises IndexError
s.byteindex("\uFFFF", 6) # => nil

Related: see Querying.

static VALUE
rb_str_byterindex_m(int argc, VALUE *argv, VALUE str)
{
    VALUE sub;
    VALUE initpos;
    long pos, len = RSTRING_LEN(str);

    if (rb_scan_args(argc, argv, "11", &sub, &initpos) == 2) {
        pos = NUM2LONG(initpos);
        if (pos < 0 && (pos += len) < 0) {
            if (RB_TYPE_P(sub, T_REGEXP)) {
                rb_backref_set(Qnil);
            }
            return Qnil;
        }
        if (pos > len) pos = len;
    }
    else {
        pos = len;
    }

    str_ensure_byte_pos(str, pos);

    if (RB_TYPE_P(sub, T_REGEXP)) {
        if (rb_reg_search(sub, str, pos, 1) >= 0) {
            VALUE match = rb_backref_get();
            struct re_registers *regs = RMATCH_REGS(match);
            pos = BEG(0);
            return LONG2NUM(pos);
        }
    }
    else {
        StringValue(sub);
        pos = rb_str_byterindex(str, sub, pos);
        if (pos >= 0) return LONG2NUM(pos);
    }
    return Qnil;
}

Returns the 0-based integer index of a substring of self that is the last match for the given object (a string or Regexp) and offset, or nil if there is no such substring; the returned index is the count of bytes (not characters).

When object is a string, returns the index of the last found substring equal to object:

s = 'foo'           # => "foo"
s.size              # => 3 # Three 1-byte characters.
s.bytesize          # => 3 # Three bytes.
s.byterindex('f')   # => 0
s.byterindex('o')   # => 2
s.byterindex('oo')  # => 1
s.byterindex('ooo') # => nil

When object is a Regexp, returns the index of the last found substring matching object; updates Regexp-related global variables:

s = 'foo'
s.byterindex(/f/)   # => 0
$~                  # => #<MatchData "f">
s.byterindex(/o/)   # => 2
s.byterindex(/oo/)  # => 1
s.byterindex(/ooo/) # => nil
$~                  # => nil

The last match means starting at the possible last position, not the last of the longest matches:

s = 'foo'
s.byterindex(/o+/) # => 2
$~                 #=> #<MatchData "o">

To get the last longest match, use a negative lookbehind:

s = 'foo'
s.byterindex(/(?<!o)o+/) # => 1
$~                       # => #<MatchData "oo">

Or use method byteindex with negative lookahead:

s = 'foo'
s.byteindex(/o+(?!.*o)/) # => 1
$~                       #=> #<MatchData "oo">

Integer argument offset, if given, specifies the 0-based index of the byte where searching is to end.

When offset is non-negative, searching ends at byte position offset:

s = 'foo'
s.byterindex('o', 0) # => nil
s.byterindex('o', 1) # => 1
s.byterindex('o', 2) # => 2
s.byterindex('o', 3) # => 2

When offset is negative, counts backward from the end of self:

s = 'foo'
s.byterindex('o', -1) # => 2
s.byterindex('o', -2) # => 1
s.byterindex('o', -3) # => nil

Raises IndexError if the byte at offset is not the first byte of a character:

s = "\uFFFF\uFFFF"        # => "\uFFFF\uFFFF"
s.size                    # => 2 # Two 3-byte characters.
s.bytesize                # => 6 # Six bytes.
s.byterindex("\uFFFF")    # => 3
s.byterindex("\uFFFF", 1) # Raises IndexError
s.byterindex("\uFFFF", 2) # Raises IndexError
s.byterindex("\uFFFF", 3) # => 3
s.byterindex("\uFFFF", 4) # Raises IndexError
s.byterindex("\uFFFF", 5) # Raises IndexError
s.byterindex("\uFFFF", 6) # => nil

Related: see Querying.

static VALUE
rb_str_bytes(VALUE str)
{
    VALUE ary = WANTARRAY("bytes", RSTRING_LEN(str));
    return rb_str_enumerate_bytes(str, ary);
}

Returns an array of the bytes in self:

'hello'.bytes  # => [104, 101, 108, 108, 111]
'Привет'.bytes # => [208, 159, 209, 128, 208, 184, 208, 178, 208, 181, 209, 130]
'こんにちは'.bytes
# => [227, 129, 147, 227, 130, 147, 227, 129, 171, 227, 129, 161, 227, 129, 175]

Related: see Converting to Non-String.

VALUE
rb_str_bytesize(VALUE str)
{
    return LONG2NUM(RSTRING_LEN(str));
}

Returns the count of bytes in self.

Note that the byte count may be different from the character count (returned by size):

s = 'foo'
s.bytesize # => 3
s.size     # => 3
s = 'Привет'
s.bytesize # => 12
s.size     # => 6
s = 'こんにちは'
s.bytesize # => 15
s.size     # => 5

Related: see Querying.

static VALUE
rb_str_byteslice(int argc, VALUE *argv, VALUE str)
{
    if (argc == 2) {
        long beg = NUM2LONG(argv[0]);
        long len = NUM2LONG(argv[1]);
        return str_byte_substr(str, beg, len, TRUE);
    }
    rb_check_arity(argc, 1, 2);
    return str_byte_aref(str, argv[0]);
}

Returns a substring of self, or nil if the substring cannot be constructed.

With integer arguments offset and length given, returns the substring beginning at the given offset and of the given length (as available):

s = '0123456789'   # => "0123456789"
s.byteslice(2)     # => "2"
s.byteslice(200)   # => nil
s.byteslice(4, 3)  # => "456"
s.byteslice(4, 30) # => "456789"

Returns nil if length is negative or offset falls outside of self:

s.byteslice(4, -1) # => nil
s.byteslice(40, 2) # => nil

Counts backwards from the end of self if offset is negative:

s = '0123456789'   # => "0123456789"
s.byteslice(-4)    # => "6"
s.byteslice(-4, 3) # => "678"

With Range argument range given, returns byteslice(range.begin, range.size):

s = '0123456789'    # => "0123456789"
s.byteslice(4..6)   # => "456"
s.byteslice(-6..-4) # => "456"
s.byteslice(5..2)   # => "" # range.size is zero.
s.byteslice(40..42) # => nil

The starting and ending offsets need not be on character boundaries:

s = 'こんにちは'
s.byteslice(0, 3) # => "こ"
s.byteslice(1, 3) # => "\x81\x93\xE3"

The encodings of self and the returned substring are always the same:

s.encoding                 # => #<Encoding:UTF-8>
s.byteslice(0, 3).encoding # => #<Encoding:UTF-8>
s.byteslice(1, 3).encoding # => #<Encoding:UTF-8>

But, depending on the character boundaries, the encoding of the returned substring may not be valid:

s.valid_encoding?                 # => true
s.byteslice(0, 3).valid_encoding? # => true
s.byteslice(1, 3).valid_encoding? # => false

Related: see Converting to New String.

static VALUE
rb_str_bytesplice(int argc, VALUE *argv, VALUE str)
{
    long beg, len, vbeg, vlen;
    VALUE val;
    int cr;

    rb_check_arity(argc, 2, 5);
    if (!(argc == 2 || argc == 3 || argc == 5)) {
        rb_raise(rb_eArgError, "wrong number of arguments (given %d, expected 2, 3, or 5)", argc);
    }
    if (argc == 2 || (argc == 3 && !RB_INTEGER_TYPE_P(argv[0]))) {
        if (!rb_range_beg_len(argv[0], &beg, &len, RSTRING_LEN(str), 2)) {
            rb_raise(rb_eTypeError, "wrong argument type %s (expected Range)",
                     rb_builtin_class_name(argv[0]));
        }
        val = argv[1];
        StringValue(val);
        if (argc == 2) {
            /* bytesplice(range, str) */
            vbeg = 0;
            vlen = RSTRING_LEN(val);
        }
        else {
            /* bytesplice(range, str, str_range) */
            if (!rb_range_beg_len(argv[2], &vbeg, &vlen, RSTRING_LEN(val), 2)) {
                rb_raise(rb_eTypeError, "wrong argument type %s (expected Range)",
                         rb_builtin_class_name(argv[2]));
            }
        }
    }
    else {
        beg = NUM2LONG(argv[0]);
        len = NUM2LONG(argv[1]);
        val = argv[2];
        StringValue(val);
        if (argc == 3) {
            /* bytesplice(index, length, str) */
            vbeg = 0;
            vlen = RSTRING_LEN(val);
        }
        else {
            /* bytesplice(index, length, str, str_index, str_length) */
            vbeg = NUM2LONG(argv[3]);
            vlen = NUM2LONG(argv[4]);
        }
    }
    str_check_beg_len(str, &beg, &len);
    str_check_beg_len(val, &vbeg, &vlen);
    str_modify_keep_cr(str);

    if (RB_UNLIKELY(ENCODING_GET_INLINED(str) != ENCODING_GET_INLINED(val))) {
        rb_enc_associate(str, rb_enc_check(str, val));
    }

    rb_str_update_1(str, beg, len, val, vbeg, vlen);
    cr = ENC_CODERANGE_AND(ENC_CODERANGE(str), ENC_CODERANGE(val));
    if (cr != ENC_CODERANGE_BROKEN)
        ENC_CODERANGE_SET(str, cr);
    return str;
}

Replaces target bytes in self with source bytes from the given string str; returns self.

In the first form, arguments offset and length determine the target bytes, and the source bytes are all of the given str:

'0123456789'.bytesplice(0, 3, 'abc')  # => "abc3456789"
'0123456789'.bytesplice(3, 3, 'abc')  # => "012abc6789"
'0123456789'.bytesplice(0, 50, 'abc') # => "abc"
'0123456789'.bytesplice(50, 3, 'abc') # Raises IndexError.

The counts of the target bytes and source source bytes may be different:

'0123456789'.bytesplice(0, 6, 'abc') # => "abc6789"      # Shorter source.
'0123456789'.bytesplice(0, 1, 'abc') # => "abc123456789" # Shorter target.

And either count may be zero (i.e., specifying an empty string):

'0123456789'.bytesplice(0, 3, '')    # => "3456789"       # Empty source.
'0123456789'.bytesplice(0, 0, 'abc') # => "abc0123456789" # Empty target.

In the second form, just as in the first, arugments offset and length determine the target bytes; argument str contains the source bytes, and the additional arguments str_offset and str_length determine the actual source bytes:

'0123456789'.bytesplice(0, 3, 'abc', 0, 3) # => "abc3456789"
'0123456789'.bytesplice(0, 3, 'abc', 1, 1) # => "b3456789"      # Shorter source.
'0123456789'.bytesplice(0, 1, 'abc', 0, 3) # => "abc123456789"  # Shorter target.
'0123456789'.bytesplice(0, 3, 'abc', 1, 0) # => "3456789"       # Empty source.
'0123456789'.bytesplice(0, 0, 'abc', 0, 3) # => "abc0123456789" # Empty target.

In the third form, argument range determines the target bytes and the source bytes are all of the given str:

'0123456789'.bytesplice(0..2, 'abc')  # => "abc3456789"
'0123456789'.bytesplice(3..5, 'abc')  # => "012abc6789"
'0123456789'.bytesplice(0..5, 'abc')  # => "abc6789"       # Shorter source.
'0123456789'.bytesplice(0..0, 'abc')  # => "abc123456789"  # Shorter target.
'0123456789'.bytesplice(0..2, '')     # => "3456789"       # Empty source.
'0123456789'.bytesplice(0...0, 'abc') # => "abc0123456789" # Empty target.

In the fourth form, just as in the third, arugment range determines the target bytes; argument str contains the source bytes, and the additional argument str_range determines the actual source bytes:

'0123456789'.bytesplice(0..2, 'abc', 0..2)  # => "abc3456789"
'0123456789'.bytesplice(3..5, 'abc', 0..2)  # => "012abc6789"
'0123456789'.bytesplice(0..2, 'abc', 0..1)  # => "ab3456789"     # Shorter source.
'0123456789'.bytesplice(0..1, 'abc', 0..2)  # => "abc23456789"   # Shorter target.
'0123456789'.bytesplice(0..2, 'abc', 0...0) # => "3456789"       # Empty source.
'0123456789'.bytesplice(0...0, 'abc', 0..2) # => "abc0123456789" # Empty target.

In any of the forms, the beginnings and endings of both source and target must be on character boundaries.

In these examples, self has five 3-byte characters, and so has character boundaries at offsets 0, 3, 6, 9, 12, and 15.

'こんにちは'.bytesplice(0, 3, 'abc') # => "abcんにちは"
'こんにちは'.bytesplice(1, 3, 'abc') # Raises IndexError.
'こんにちは'.bytesplice(0, 2, 'abc') # Raises IndexError.

static VALUE
rb_str_capitalize(int argc, VALUE *argv, VALUE str)
{
    rb_encoding *enc;
    OnigCaseFoldType flags = ONIGENC_CASE_UPCASE | ONIGENC_CASE_TITLECASE;
    VALUE ret;

    flags = check_case_options(argc, argv, flags);
    enc = str_true_enc(str);
    if (RSTRING_LEN(str) == 0 || !RSTRING_PTR(str)) return str;
    if (flags&ONIGENC_CASE_ASCII_ONLY) {
        ret = rb_str_new(0, RSTRING_LEN(str));
        rb_str_ascii_casemap(str, ret, &flags, enc);
    }
    else {
        ret = rb_str_casemap(str, &flags, enc);
    }
    return ret;
}

Returns a string containing the characters in self, each with possibly changed case:

• The first character made uppercase.

• All other characters are made lowercase.

Examples:

'hello'.capitalize  # => "Hello"
'HELLO'.capitalize  # => "Hello"
'straße'.capitalize # => "Straße"  # Lowercase 'ß' not changed.
'STRAẞE'.capitalize # => "Straße"  # Uppercase 'ẞ' downcased to 'ß'.
'привет'.capitalize # => "Привет"
'ПРИВЕТ'.capitalize # => "Привет"

Some characters (and some character sets) do not have upcase and downcase versions; see Case Mapping:

s = '1, 2, 3, ...'
s.capitalize == s # => true
s = 'こんにちは'
s.capitalize == s # => true

The casing is affected by the given mapping, which may be :ascii, :fold, or :turkic; see Case Mappings.

Related: see Converting to New String.

static VALUE
rb_str_capitalize_bang(int argc, VALUE *argv, VALUE str)
{
    rb_encoding *enc;
    OnigCaseFoldType flags = ONIGENC_CASE_UPCASE | ONIGENC_CASE_TITLECASE;

    flags = check_case_options(argc, argv, flags);
    str_modify_keep_cr(str);
    enc = str_true_enc(str);
    if (RSTRING_LEN(str) == 0 || !RSTRING_PTR(str)) return Qnil;
    if (flags&ONIGENC_CASE_ASCII_ONLY)
        rb_str_ascii_casemap(str, str, &flags, enc);
    else
        str_shared_replace(str, rb_str_casemap(str, &flags, enc));

    if (ONIGENC_CASE_MODIFIED&flags) return str;
    return Qnil;
}

Like String#capitalize, except that:

• Changes character casings in self (not in a copy of self).

• Returns self if any changes are made, nil otherwise.

Related: See Modifying.

static VALUE
rb_str_casecmp(VALUE str1, VALUE str2)
{
    VALUE s = rb_check_string_type(str2);
    if (NIL_P(s)) {
        return Qnil;
    }
    return str_casecmp(str1, s);
}

Ignoring case, compares self and other_string; returns:

• -1 if self.downcase is smaller than other_string.downcase.

• 0 if the two are equal.

• 1 if self.downcase is larger than other_string.downcase.

• nil if the two are incomparable.

See Case Mapping.

Examples:

'foo'.casecmp('goo')  # => -1
'goo'.casecmp('foo')  # => 1
'foo'.casecmp('food') # => -1
'food'.casecmp('foo') # => 1
'FOO'.casecmp('foo')  # => 0
'foo'.casecmp('FOO')  # => 0
'foo'.casecmp(1)      # => nil

Related: see Comparing.

static VALUE
rb_str_casecmp_p(VALUE str1, VALUE str2)
{
    VALUE s = rb_check_string_type(str2);
    if (NIL_P(s)) {
        return Qnil;
    }
    return str_casecmp_p(str1, s);
}

Returns true if self and other_string are equal after Unicode case folding, false if unequal, nil if incomparable.

See Case Mapping.

Examples:

'foo'.casecmp?('goo')  # => false
'goo'.casecmp?('foo')  # => false
'foo'.casecmp?('food') # => false
'food'.casecmp?('foo') # => false
'FOO'.casecmp?('foo')  # => true
'foo'.casecmp?('FOO')  # => true
'foo'.casecmp?(1)      # => nil

Related: see Comparing.

static VALUE
rb_str_center(int argc, VALUE *argv, VALUE str)
{
    return rb_str_justify(argc, argv, str, 'c');
}

Returns a centered copy of self.

If integer argument size is greater than the size (in characters) of self, returns a new string of length size that is a copy of self, centered and padded on one or both ends with pad_string:

'hello'.center(6)             # => "hello "               # Padded on one end.
'hello'.center(10)            # => "  hello   "           # Padded on both ends.
'hello'.center(20, '-|')      # => "-|-|-|-hello-|-|-|-|" # Some padding repeated.
'hello'.center(10, 'abcdefg') # => "abhelloabc"           # Some padding not used.
'  hello  '.center(13)        # => "    hello    "
'Привет'.center(10)           # => "  Привет  "
'こんにちは'.center(10)         # => "  こんにちは   "      # Multi-byte characters.

If size is less than or equal to the size of self, returns an unpadded copy of self:

'hello'.center(5)   # => "hello"
'hello'.center(-10) # => "hello"

Related: see Converting to New String.

static VALUE
rb_str_chars(VALUE str)
{
    VALUE ary = WANTARRAY("chars", rb_str_strlen(str));
    return rb_str_enumerate_chars(str, ary);
}

Returns an array of the characters in self:

'hello'.chars     # => ["h", "e", "l", "l", "o"]
'Привет'.chars    # => ["П", "р", "и", "в", "е", "т"]
'こんにちは'.chars # => ["こ", "ん", "に", "ち", "は"]
''.chars          # => []

Related: see Converting to Non-String.

static VALUE
rb_str_chomp(int argc, VALUE *argv, VALUE str)
{
    VALUE rs = chomp_rs(argc, argv);
    if (NIL_P(rs)) return str_duplicate(rb_cString, str);
    return rb_str_subseq(str, 0, chompped_length(str, rs));
}

Returns a new string copied from self, with trailing characters possibly removed:

When line_sep is "\n", removes the last one or two characters if they are "\r", "\n", or "\r\n" (but not "\n\r"):

$/                    # => "\n"
"abc\r".chomp         # => "abc"
"abc\n".chomp         # => "abc"
"abc\r\n".chomp       # => "abc"
"abc\n\r".chomp       # => "abc\n"
"тест\r\n".chomp      # => "тест"
"こんにちは\r\n".chomp  # => "こんにちは"

When line_sep is '' (an empty string), removes multiple trailing occurrences of "\n" or "\r\n" (but not "\r" or "\n\r"):

"abc\n\n\n".chomp('')           # => "abc"
"abc\r\n\r\n\r\n".chomp('')     # => "abc"
"abc\n\n\r\n\r\n\n\n".chomp('') # => "abc"
"abc\n\r\n\r\n\r".chomp('')     # => "abc\n\r\n\r\n\r"
"abc\r\r\r".chomp('')           # => "abc\r\r\r"

When line_sep is neither "\n" nor '', removes a single trailing line separator if there is one:

'abcd'.chomp('cd')   # => "ab"
'abcdcd'.chomp('cd') # => "abcd"
'abcd'.chomp('xx')   # => "abcd"

Related: see Converting to New String.

static VALUE
rb_str_chomp_bang(int argc, VALUE *argv, VALUE str)
{
    VALUE rs;
    str_modifiable(str);
    if (RSTRING_LEN(str) == 0 && argc < 2) return Qnil;
    rs = chomp_rs(argc, argv);
    if (NIL_P(rs)) return Qnil;
    return rb_str_chomp_string(str, rs);
}

Like String#chomp, except that:

• Removes trailing characters from self (not from a copy of self).

• Returns self if any characters are removed, nil otherwise.

Related: see Modifying.

static VALUE
rb_str_chop(VALUE str)
{
    return rb_str_subseq(str, 0, chopped_length(str));
}

Returns a new string copied from self, with trailing characters possibly removed.

Removes "\r\n" if those are the last two characters.

"abc\r\n".chop      # => "abc"
"тест\r\n".chop     # => "тест"
"こんにちは\r\n".chop # => "こんにちは"

Otherwise removes the last character if it exists.

'abcd'.chop     # => "abc"
'тест'.chop     # => "тес"
'こんにちは'.chop # => "こんにち"
''.chop         # => ""

If you only need to remove the newline separator at the end of the string, String#chomp is a better alternative.

Related: see Converting to New String.

static VALUE
rb_str_chop_bang(VALUE str)
{
    str_modify_keep_cr(str);
    if (RSTRING_LEN(str) > 0) {
        long len;
        len = chopped_length(str);
        STR_SET_LEN(str, len);
        TERM_FILL(&RSTRING_PTR(str)[len], TERM_LEN(str));
        if (ENC_CODERANGE(str) != ENC_CODERANGE_7BIT) {
            ENC_CODERANGE_CLEAR(str);
        }
        return str;
    }
    return Qnil;
}

Like String#chop, except that:

• Removes trailing characters from self (not from a copy of self).

• Returns self if any characters are removed, nil otherwise.

Related: see Modifying.

static VALUE
rb_str_chr(VALUE str)
{
    return rb_str_substr(str, 0, 1);
}

Returns a string containing the first character of self:

'hello'.chr     # => "h"
'тест'.chr      # => "т"
'こんにちは'.chr # => "こ"
''.chr          # => ""

Related: see Converting to New String.

static VALUE
rb_str_clear(VALUE str)
{
    str_discard(str);
    STR_SET_EMBED(str);
    STR_SET_LEN(str, 0);
    RSTRING_PTR(str)[0] = 0;
    if (rb_enc_asciicompat(STR_ENC_GET(str)))
        ENC_CODERANGE_SET(str, ENC_CODERANGE_7BIT);
    else
        ENC_CODERANGE_SET(str, ENC_CODERANGE_VALID);
    return str;
}

Removes the contents of self:

s = 'foo'
s.clear # => ""
s       # => ""

Related: see Modifying.

static VALUE
rb_str_codepoints(VALUE str)
{
    VALUE ary = WANTARRAY("codepoints", rb_str_strlen(str));
    return rb_str_enumerate_codepoints(str, ary);
}

Returns an array of the codepoints in self; each codepoint is the integer value for a character:

'hello'.codepoints     # => [104, 101, 108, 108, 111]
'тест'.codepoints      # => [1090, 1077, 1089, 1090]
'こんにちは'.codepoints # => [12371, 12435, 12395, 12385, 12399]
''.codepoints          # => []

Related: see Converting to Non-String.

static VALUE
rb_str_concat_multi(int argc, VALUE *argv, VALUE str)
{
    str_modifiable(str);

    if (argc == 1) {
        return rb_str_concat(str, argv[0]);
    }
    else if (argc > 1) {
        int i;
        VALUE arg_str = rb_str_tmp_new(0);
        rb_enc_copy(arg_str, str);
        for (i = 0; i < argc; i++) {
            rb_str_concat(arg_str, argv[i]);
        }
        rb_str_buf_append(str, arg_str);
    }

    return str;
}

Concatenates each object in objects to self; returns self:

'foo'.concat('bar', 'baz') # => "foobarbaz"

For each given object object that is an integer, the value is considered a codepoint and converted to a character before concatenation:

'foo'.concat(32, 'bar', 32, 'baz') # => "foo bar baz" # Embeds spaces.
'те'.concat(1089, 1090)            # => "тест"
'こん'.concat(12395, 12385, 12399)  # => "こんにちは"

Related: see Converting to New String.

static VALUE
rb_str_count(int argc, VALUE *argv, VALUE str)
{
    char table[TR_TABLE_SIZE];
    rb_encoding *enc = 0;
    VALUE del = 0, nodel = 0, tstr;
    char *s, *send;
    int i;
    int ascompat;
    size_t n = 0;

    rb_check_arity(argc, 1, UNLIMITED_ARGUMENTS);

    tstr = argv[0];
    StringValue(tstr);
    enc = rb_enc_check(str, tstr);
    if (argc == 1) {
        const char *ptstr;
        if (RSTRING_LEN(tstr) == 1 && rb_enc_asciicompat(enc) &&
            (ptstr = RSTRING_PTR(tstr),
             ONIGENC_IS_ALLOWED_REVERSE_MATCH(enc, (const unsigned char *)ptstr, (const unsigned char *)ptstr+1)) &&
            !is_broken_string(str)) {
            int clen;
            unsigned char c = rb_enc_codepoint_len(ptstr, ptstr+1, &clen, enc);

            s = RSTRING_PTR(str);
            if (!s || RSTRING_LEN(str) == 0) return INT2FIX(0);
            send = RSTRING_END(str);
            while (s < send) {
                if (*(unsigned char*)s++ == c) n++;
            }
            return SIZET2NUM(n);
        }
    }

    tr_setup_table(tstr, table, TRUE, &del, &nodel, enc);
    for (i=1; i<argc; i++) {
        tstr = argv[i];
        StringValue(tstr);
        enc = rb_enc_check(str, tstr);
        tr_setup_table(tstr, table, FALSE, &del, &nodel, enc);
    }

    s = RSTRING_PTR(str);
    if (!s || RSTRING_LEN(str) == 0) return INT2FIX(0);
    send = RSTRING_END(str);
    ascompat = rb_enc_asciicompat(enc);
    while (s < send) {
        unsigned int c;

        if (ascompat && (c = *(unsigned char*)s) < 0x80) {
            if (table[c]) {
                n++;
            }
            s++;
        }
        else {
            int clen;
            c = rb_enc_codepoint_len(s, send, &clen, enc);
            if (tr_find(c, table, del, nodel)) {
                n++;
            }
            s += clen;
        }
    }

    return SIZET2NUM(n);
}

Returns the total number of characters in self that are specified by the given selectors.

For one 1-character selector, returns the count of instances of that character:

s = 'abracadabra'
s.count('a') # => 5
s.count('b') # => 2
s.count('x') # => 0
s.count('')  # => 0

s = 'тест'
s.count('т')  # => 2
s.count('е')  # => 1

s = 'よろしくお願いします'
s.count('よ')  # => 1
s.count('し')  # => 2

For one multi-character selector, returns the count of instances for all specified characters:

s = 'abracadabra'
s.count('ab')     # => 7
s.count('abc')    # => 8
s.count('abcd')   # => 9
s.count('abcdr')  # => 11
s.count('abcdrx') # => 11

Order and repetition do not matter:

s.count('ba')   == s.count('ab') # => true
s.count('baab') == s.count('ab') # => true

For multiple selectors, forms a single selector that is the intersection of characters in all selectors and returns the count of instances for that selector:

s = 'abcdefg'
s.count('abcde', 'dcbfg') == s.count('bcd') # => true
s.count('abc', 'def')     == s.count('')    # => true

In a character selector, three characters get special treatment:

• A caret ('^') functions as a negation operator for the immediately following characters:

  s = 'abracadabra'
  s.count('^bc') # => 8  # Count of all except 'b' and 'c'.
  

• A hyphen ('-') between two other characters defines a range of characters:

  s = 'abracadabra'
  s.count('a-c') # => 8  # Count of all 'a', 'b', and 'c'.
  

• A backslash ('\') acts as an escape for a caret, a hyphen, or another backslash:

  s = 'abracadabra'
  s.count('\^bc')           # => 3  # Count of '^', 'b', and 'c'.
  s.count('a\-c')           # => 6  # Count of 'a', '-', and 'c'.
  'foo\bar\baz'.count('\\') # => 2  # Count of '\'.
  

These usages may be mixed:

s = 'abracadabra'
s.count('a-cq-t') # => 10  # Multiple ranges.
s.count('ac-d')   # => 7   # Range mixed with plain characters.
s.count('^a-c')   # => 3   # Range mixed with negation.

For multiple selectors, all forms may be used, including negations, ranges, and escapes.

s = 'abracadabra'
s.count('^abc', '^def') == s.count('^abcdef') # => true
s.count('a-e', 'c-g')   == s.count('cde')     # => true
s.count('^abc', 'c-g')  == s.count('defg')    # => true

Related: see Querying.

static VALUE
rb_str_crypt(VALUE str, VALUE salt)
{
#ifdef HAVE_CRYPT_R
    VALUE databuf;
    struct crypt_data *data;
#   define CRYPT_END() ALLOCV_END(databuf)
#else
    char *tmp_buf;
    extern char *crypt(const char *, const char *);
#   define CRYPT_END() rb_nativethread_lock_unlock(&crypt_mutex.lock)
#endif
    VALUE result;
    const char *s, *saltp;
    char *res;
#ifdef BROKEN_CRYPT
    char salt_8bit_clean[3];
#endif

    StringValue(salt);
    mustnot_wchar(str);
    mustnot_wchar(salt);
    s = StringValueCStr(str);
    saltp = RSTRING_PTR(salt);
    if (RSTRING_LEN(salt) < 2 || !saltp[0] || !saltp[1]) {
        rb_raise(rb_eArgError, "salt too short (need >=2 bytes)");
    }

#ifdef BROKEN_CRYPT
    if (!ISASCII((unsigned char)saltp[0]) || !ISASCII((unsigned char)saltp[1])) {
        salt_8bit_clean[0] = saltp[0] & 0x7f;
        salt_8bit_clean[1] = saltp[1] & 0x7f;
        salt_8bit_clean[2] = '\0';
        saltp = salt_8bit_clean;
    }
#endif
#ifdef HAVE_CRYPT_R
    data = ALLOCV(databuf, sizeof(struct crypt_data));
# ifdef HAVE_STRUCT_CRYPT_DATA_INITIALIZED
    data->initialized = 0;
# endif
    res = crypt_r(s, saltp, data);
#else
    rb_nativethread_lock_lock(&crypt_mutex.lock);
    res = crypt(s, saltp);
#endif
    if (!res) {
        int err = errno;
        CRYPT_END();
        rb_syserr_fail(err, "crypt");
    }
#ifdef HAVE_CRYPT_R
    result = rb_str_new_cstr(res);
    CRYPT_END();
#else
    // We need to copy this buffer because it's static and we need to unlock the mutex
    // before allocating a new object (the string to be returned). If we allocate while
    // holding the lock, we could run GC which fires the VM barrier and causes a deadlock
    // if other ractors are waiting on this lock.
    size_t res_size = strlen(res)+1;
    tmp_buf = ALLOCA_N(char, res_size); // should be small enough to alloca
    memcpy(tmp_buf, res, res_size);
    res = tmp_buf;
    CRYPT_END();
    result = rb_str_new_cstr(res);
#endif
    return result;
}

Returns the string generated by calling crypt(3) standard library function with str and salt_str, in this order, as its arguments. Please do not use this method any longer. It is legacy; provided only for backward compatibility with ruby scripts in earlier days. It is bad to use in contemporary programs for several reasons:

• Behaviour of C’s crypt(3) depends on the OS it is run. The generated string lacks data portability.

• On some OSes such as Mac OS, crypt(3) never fails (i.e. silently ends up in unexpected results).

• On some OSes such as Mac OS, crypt(3) is not thread safe.

• So-called “traditional” usage of crypt(3) is very very very weak. According to its manpage, Linux’s traditional crypt(3) output has only 2**56 variations; too easy to brute force today. And this is the default behaviour.

• In order to make things robust some OSes implement so-called “modular” usage. To go through, you have to do a complex build-up of the salt_str parameter, by hand. Failure in generation of a proper salt string tends not to yield any errors; typos in parameters are normally not detectable.

  • For instance, in the following example, the second invocation of String#crypt is wrong; it has a typo in “round=” (lacks “s”). However the call does not fail and something unexpected is generated.

    "foo".crypt("$5$rounds=1000$salt$") # OK, proper usage
    "foo".crypt("$5$round=1000$salt$")  # Typo not detected
    

• Even in the “modular” mode, some hash functions are considered archaic and no longer recommended at all; for instance module $1$ is officially abandoned by its author: see phk.freebsd.dk/sagas/md5crypt_eol/ . For another instance module $3$ is considered completely broken: see the manpage of FreeBSD.

• On some OS such as Mac OS, there is no modular mode. Yet, as written above, crypt(3) on Mac OS never fails. This means even if you build up a proper salt string it generates a traditional DES hash anyways, and there is no way for you to be aware of.

  "foo".crypt("$5$rounds=1000$salt$") # => "$5fNPQMxC5j6."
  

If for some reason you cannot migrate to other secure contemporary password hashing algorithms, install the string-crypt gem and require 'string/crypt' to continue using it.

static VALUE
rb_str_delete(int argc, VALUE *argv, VALUE str)
{
    str = str_duplicate(rb_cString, str);
    rb_str_delete_bang(argc, argv, str);
    return str;
}

Returns a new string that is a copy of self with certain characters removed; the removed characters are all instances of those specified by the given string selectors.

For one 1-character selector, removes all instances of that character:

s = 'abracadabra'
s.delete('a') # => "brcdbr"
s.delete('b') # => "aracadara"
s.delete('x') # => "abracadabra"
s.delete('')  # => "abracadabra"

s = 'тест'
s.delete('т') # => "ес"
s.delete('е') # => "тст"

s = 'よろしくお願いします'
s.delete('よ') # => "ろしくお願いします"
s.delete('し') # => "よろくお願います"

For one multi-character selector, removes all instances of the specified characters:

s = 'abracadabra'
s.delete('ab')     # => "rcdr"
s.delete('abc')    # => "rdr"
s.delete('abcd')   # => "rr"
s.delete('abcdr')  # => ""
s.delete('abcdrx') # => ""

Order and repetition do not matter:

s.delete('ba')   == s.delete('ab') # => true
s.delete('baab') == s.delete('ab') # => true

For multiple selectors, forms a single selector that is the intersection of characters in all selectors and removes all instances of characters specified by that selector:

s = 'abcdefg'
s.delete('abcde', 'dcbfg') == s.delete('bcd') # => true
s.delete('abc', 'def')     == s.delete('')    # => true

In a character selector, three characters get special treatment:

• A caret ('^') functions as a negation operator for the immediately following characters:

  s = 'abracadabra'
  s.delete('^bc') # => "bcb"  # Deletes all except 'b' and 'c'.
  

• A hyphen ('-') between two other characters defines a range of characters:

  s = 'abracadabra'
  s.delete('a-c') # => "rdr"  # Deletes all 'a', 'b', and 'c'.
  

• A backslash ('\') acts as an escape for a caret, a hyphen, or another backslash:

  s = 'abracadabra'
  s.delete('\^bc')           # => "araadara"   # Deletes all '^', 'b', and 'c'.
  s.delete('a\-c')           # => "brdbr"      # Deletes all 'a', '-', and 'c'.
  'foo\bar\baz'.delete('\\') # => "foobarbaz"  # Deletes all '\'.
  

These usages may be mixed:

s = 'abracadabra'
s.delete('a-cq-t') # => "d"         # Multiple ranges.
s.delete('ac-d')   # => "brbr"      # Range mixed with plain characters.
s.delete('^a-c')   # => "abacaaba"  # Range mixed with negation.

For multiple selectors, all forms may be used, including negations, ranges, and escapes.

s = 'abracadabra'
s.delete('^abc', '^def') == s.delete('^abcdef') # => true
s.delete('a-e', 'c-g')   == s.delete('cde')     # => true
s.delete('^abc', 'c-g')  == s.delete('defg')    # => true

Related: see Converting to New String.

static VALUE
rb_str_delete_bang(int argc, VALUE *argv, VALUE str)
{
    char squeez[TR_TABLE_SIZE];
    rb_encoding *enc = 0;
    char *s, *send, *t;
    VALUE del = 0, nodel = 0;
    int modify = 0;
    int i, ascompat, cr;

    if (RSTRING_LEN(str) == 0 || !RSTRING_PTR(str)) return Qnil;
    rb_check_arity(argc, 1, UNLIMITED_ARGUMENTS);
    for (i=0; i<argc; i++) {
        VALUE s = argv[i];

        StringValue(s);
        enc = rb_enc_check(str, s);
        tr_setup_table(s, squeez, i==0, &del, &nodel, enc);
    }

    str_modify_keep_cr(str);
    ascompat = rb_enc_asciicompat(enc);
    s = t = RSTRING_PTR(str);
    send = RSTRING_END(str);
    cr = ascompat ? ENC_CODERANGE_7BIT : ENC_CODERANGE_VALID;
    while (s < send) {
        unsigned int c;
        int clen;

        if (ascompat && (c = *(unsigned char*)s) < 0x80) {
            if (squeez[c]) {
                modify = 1;
            }
            else {
                if (t != s) *t = c;
                t++;
            }
            s++;
        }
        else {
            c = rb_enc_codepoint_len(s, send, &clen, enc);

            if (tr_find(c, squeez, del, nodel)) {
                modify = 1;
            }
            else {
                if (t != s) rb_enc_mbcput(c, t, enc);
                t += clen;
                if (cr == ENC_CODERANGE_7BIT) cr = ENC_CODERANGE_VALID;
            }
            s += clen;
        }
    }
    TERM_FILL(t, TERM_LEN(str));
    STR_SET_LEN(str, t - RSTRING_PTR(str));
    ENC_CODERANGE_SET(str, cr);

    if (modify) return str;
    return Qnil;
}

Like String#delete, but modifies self in place; returns self if any characters were deleted, nil otherwise.

Related: see Modifying.

static VALUE
rb_str_delete_prefix(VALUE str, VALUE prefix)
{
    long prefixlen;

    prefixlen = deleted_prefix_length(str, prefix);
    if (prefixlen <= 0) return str_duplicate(rb_cString, str);

    return rb_str_subseq(str, prefixlen, RSTRING_LEN(str) - prefixlen);
}

Returns a copy of self with leading substring prefix removed:

'oof'.delete_prefix('o')          # => "of"
'oof'.delete_prefix('oo')         # => "f"
'oof'.delete_prefix('oof')        # => ""
'oof'.delete_prefix('x')          # => "oof"
'тест'.delete_prefix('те')        # => "ст"
'こんにちは'.delete_prefix('こん')  # => "にちは"

Related: see Converting to New String.

static VALUE
rb_str_delete_prefix_bang(VALUE str, VALUE prefix)
{
    long prefixlen;
    str_modify_keep_cr(str);

    prefixlen = deleted_prefix_length(str, prefix);
    if (prefixlen <= 0) return Qnil;

    return rb_str_drop_bytes(str, prefixlen);
}

Like String#delete_prefix, except that self is modified in place; returns self if the prefix is removed, nil otherwise.

Related: see Modifying.

static VALUE
rb_str_delete_suffix(VALUE str, VALUE suffix)
{
    long suffixlen;

    suffixlen = deleted_suffix_length(str, suffix);
    if (suffixlen <= 0) return str_duplicate(rb_cString, str);

    return rb_str_subseq(str, 0, RSTRING_LEN(str) - suffixlen);
}

Returns a copy of self with trailing substring suffix removed:

'foo'.delete_suffix('o')           # => "fo"
'foo'.delete_suffix('oo')          # => "f"
'foo'.delete_suffix('foo')         # => ""
'foo'.delete_suffix('f')           # => "foo"
'foo'.delete_suffix('x')           # => "foo"
'тест'.delete_suffix('ст')         # => "те"
'こんにちは'.delete_suffix('ちは')  # => "こんに"

Related: see Converting to New String.

static VALUE
rb_str_delete_suffix_bang(VALUE str, VALUE suffix)
{
    long olen, suffixlen, len;
    str_modifiable(str);

    suffixlen = deleted_suffix_length(str, suffix);
    if (suffixlen <= 0) return Qnil;

    olen = RSTRING_LEN(str);
    str_modify_keep_cr(str);
    len = olen - suffixlen;
    STR_SET_LEN(str, len);
    TERM_FILL(&RSTRING_PTR(str)[len], TERM_LEN(str));
    if (ENC_CODERANGE(str) != ENC_CODERANGE_7BIT) {
        ENC_CODERANGE_CLEAR(str);
    }
    return str;
}

Like String#delete_suffix, except that self is modified in place; returns self if the suffix is removed, nil otherwise.

Related: see Modifying.

static VALUE
rb_str_downcase(int argc, VALUE *argv, VALUE str)
{
    rb_encoding *enc;
    OnigCaseFoldType flags = ONIGENC_CASE_DOWNCASE;
    VALUE ret;

    flags = check_case_options(argc, argv, flags);
    enc = str_true_enc(str);
    if (case_option_single_p(flags, enc, str)) {
        ret = rb_str_new(RSTRING_PTR(str), RSTRING_LEN(str));
        str_enc_copy_direct(ret, str);
        downcase_single(ret);
    }
    else if (flags&ONIGENC_CASE_ASCII_ONLY) {
        ret = rb_str_new(0, RSTRING_LEN(str));
        rb_str_ascii_casemap(str, ret, &flags, enc);
    }
    else {
        ret = rb_str_casemap(str, &flags, enc);
    }

    return ret;
}

Returns a new string containing the downcased characters in self:

'HELLO'.downcase        # => "hello"
'STRAẞE'.downcase       # => "straße"
'ПРИВЕТ'.downcase       # => "привет"
'RubyGems.org'.downcase # => "rubygems.org"

Some characters (and some character sets) do not have upcase and downcase versions; see Case Mapping:

s = '1, 2, 3, ...'
s.downcase == s # => true
s = 'こんにちは'
s.downcase == s # => true

The casing is affected by the given mapping, which may be :ascii, :fold, or :turkic; see Case Mappings.

Related: see Converting to New String.

static VALUE
rb_str_downcase_bang(int argc, VALUE *argv, VALUE str)
{
    rb_encoding *enc;
    OnigCaseFoldType flags = ONIGENC_CASE_DOWNCASE;

    flags = check_case_options(argc, argv, flags);
    str_modify_keep_cr(str);
    enc = str_true_enc(str);
    if (case_option_single_p(flags, enc, str)) {
        if (downcase_single(str))
            flags |= ONIGENC_CASE_MODIFIED;
    }
    else if (flags&ONIGENC_CASE_ASCII_ONLY)
        rb_str_ascii_casemap(str, str, &flags, enc);
    else
        str_shared_replace(str, rb_str_casemap(str, &flags, enc));

    if (ONIGENC_CASE_MODIFIED&flags) return str;
    return Qnil;
}

Like String#downcase, except that:

• Changes character casings in self (not in a copy of self).

• Returns self if any changes are made, nil otherwise.

Related: See Modifying.

VALUE
rb_str_dump(VALUE str)
{
    int encidx = rb_enc_get_index(str);
    rb_encoding *enc = rb_enc_from_index(encidx);
    long len;
    const char *p, *pend;
    char *q, *qend;
    VALUE result;
    int u8 = (encidx == rb_utf8_encindex());
    static const char nonascii_suffix[] = ".dup.force_encoding(\"%s\")";

    len = 2;                    /* "" */
    if (!rb_enc_asciicompat(enc)) {
        len += strlen(nonascii_suffix) - rb_strlen_lit("%s");
        len += strlen(enc->name);
    }

    p = RSTRING_PTR(str); pend = p + RSTRING_LEN(str);
    while (p < pend) {
        int clen;
        unsigned char c = *p++;

        switch (c) {
          case '"':  case '\\':
          case '\n': case '\r':
          case '\t': case '\f':
          case '\013': case '\010': case '\007': case '\033':
            clen = 2;
            break;

          case '#':
            clen = IS_EVSTR(p, pend) ? 2 : 1;
            break;

          default:
            if (ISPRINT(c)) {
                clen = 1;
            }
            else {
                if (u8 && c > 0x7F) {   /* \u notation */
                    int n = rb_enc_precise_mbclen(p-1, pend, enc);
                    if (MBCLEN_CHARFOUND_P(n)) {
                        unsigned int cc = rb_enc_mbc_to_codepoint(p-1, pend, enc);
                        if (cc <= 0xFFFF)
                            clen = 6;  /* \uXXXX */
                        else if (cc <= 0xFFFFF)
                            clen = 9;  /* \u{XXXXX} */
                        else
                            clen = 10; /* \u{XXXXXX} */
                        p += MBCLEN_CHARFOUND_LEN(n)-1;
                        break;
                    }
                }
                clen = 4;       /* \xNN */
            }
            break;
        }

        if (clen > LONG_MAX - len) {
            rb_raise(rb_eRuntimeError, "string size too big");
        }
        len += clen;
    }

    result = rb_str_new(0, len);
    p = RSTRING_PTR(str); pend = p + RSTRING_LEN(str);
    q = RSTRING_PTR(result); qend = q + len + 1;

    *q++ = '"';
    while (p < pend) {
        unsigned char c = *p++;

        if (c == '"' || c == '\\') {
            *q++ = '\\';
            *q++ = c;
        }
        else if (c == '#') {
            if (IS_EVSTR(p, pend)) *q++ = '\\';
            *q++ = '#';
        }
        else if (c == '\n') {
            *q++ = '\\';
            *q++ = 'n';
        }
        else if (c == '\r') {
            *q++ = '\\';
            *q++ = 'r';
        }
        else if (c == '\t') {
            *q++ = '\\';
            *q++ = 't';
        }
        else if (c == '\f') {
            *q++ = '\\';
            *q++ = 'f';
        }
        else if (c == '\013') {
            *q++ = '\\';
            *q++ = 'v';
        }
        else if (c == '\010') {
            *q++ = '\\';
            *q++ = 'b';
        }
        else if (c == '\007') {
            *q++ = '\\';
            *q++ = 'a';
        }
        else if (c == '\033') {
            *q++ = '\\';
            *q++ = 'e';
        }
        else if (ISPRINT(c)) {
            *q++ = c;
        }
        else {
            *q++ = '\\';
            if (u8) {
                int n = rb_enc_precise_mbclen(p-1, pend, enc) - 1;
                if (MBCLEN_CHARFOUND_P(n)) {
                    int cc = rb_enc_mbc_to_codepoint(p-1, pend, enc);
                    p += n;
                    if (cc <= 0xFFFF)
                        snprintf(q, qend-q, "u%04X", cc);    /* \uXXXX */
                    else
                        snprintf(q, qend-q, "u{%X}", cc);  /* \u{XXXXX} or \u{XXXXXX} */
                    q += strlen(q);
                    continue;
                }
            }
            snprintf(q, qend-q, "x%02X", c);
            q += 3;
        }
    }
    *q++ = '"';
    *q = '\0';
    if (!rb_enc_asciicompat(enc)) {
        snprintf(q, qend-q, nonascii_suffix, enc->name);
        encidx = rb_ascii8bit_encindex();
    }
    /* result from dump is ASCII */
    rb_enc_associate_index(result, encidx);
    ENC_CODERANGE_SET(result, ENC_CODERANGE_7BIT);
    return result;
}

For an ordinary string, this method, +String#dump+, returns a printable ASCII-only version of self, enclosed in double-quotes.

For a dumped string, method String#undump is the inverse of +String#dump+; it returns a “restored” version of self, where all the dumping changes have been undone.

In the simplest case, the dumped string contains the original string, enclosed in double-quotes; this example is done in irb (interactive Ruby), which uses method ‘inspect` to render the results:

s = 'hello'   # => "hello"
s.dump        # => "\"hello\""
s.dump.undump # => "hello"

Keep in mind that in the second line above:

• The outer double-quotes are put on by inspect, and are not part of the output of dump.

• The inner double-quotes are part of the output of dump, and are escaped by inspect because they are within the outer double-quotes.

To avoid confusion, we’ll use this helper method to omit the outer double-quotes:

def dump(s)
  print "String:   ", s, "\n"
  print "Dumped:   ", s.dump, "\n"
  print "Undumped: ", s.dump.undump, "\n"
end

So that for string 'hello', we’ll see:

String:    hello
Dumped:    "hello"
Undumped:  hello

In a dump, certain special characters are escaped:

String:    "
Dumped:    "\""
Undumped:  "

String:    \
Dumped:    "\\"
Undumped:  \

In a dump, unprintable characters are replaced by printable ones; the unprintable characters are the whitespace characters (other than space itself); here we see the ordinals for those characers, together with explanatory text:

h = {
   7 => 'Alert (BEL)',
   8 => 'Backspace (BS)',
   9 => 'Horizontal tab (HT)',
  10 => 'Linefeed (LF)',
  11 => 'Vertical tab (VT)',
  12 => 'Formfeed (FF)',
  13 => 'Carriage return (CR)'
}

In this example, the dumped output is printed by method inspect, and so contains both outer double-quotes and escaped inner double-quotes:

s = ''
h.keys.each {|i| s << i } # => [7, 8, 9, 10, 11, 12, 13]
s                         # => "\a\b\t\n\v\f\r"
s.dump                    # => "\"\\a\\b\\t\\n\\v\\f\\r\""

If self is encoded in UTF-8 and contains Unicode characters, each Unicode character is dumped as a Unicode escape sequence:

String:    тест
Dumped:    "\u0442\u0435\u0441\u0442"
Undumped:  тест

String:    こんにちは
Dumped:    "\u3053\u3093\u306B\u3061\u306F"
Undumped:  こんにちは

If the encoding of self is not ASCII-compatible (i.e., if self.encoding.ascii_compatible? returns false), each ASCII-compatible byte is dumped as an ASCII character, and all other bytes are dumped as hexadecimal; also appends .dup.force_encoding(\"encoding\"), where <encoding> is self.encoding.name:

String:    hello
Dumped:    "\xFE\xFF\x00h\x00e\x00l\x00l\x00o".dup.force_encoding("UTF-16")
Undumped:  hello

String:    тест
Dumped:    "\xFE\xFF\x04B\x045\x04A\x04B".dup.force_encoding("UTF-16")
Undumped:  тест

String:    こんにちは
Dumped:    "\xFE\xFF0S0\x930k0a0o".dup.force_encoding("UTF-16")
Undumped:  こんにちは

static VALUE
rb_str_each_byte(VALUE str)
{
    RETURN_SIZED_ENUMERATOR(str, 0, 0, rb_str_each_byte_size);
    return rb_str_enumerate_bytes(str, 0);
}

With a block given, calls the block with each successive byte from self; returns self:

a = []
'hello'.each_byte {|byte| a.push(byte) }     # Five 1-byte characters.
a # => [104, 101, 108, 108, 111]
a = []
'тест'.each_byte {|byte| a.push(byte) }      # Four 2-byte characters.
a # => [209, 130, 208, 181, 209, 129, 209, 130]
a = []
'こんにちは'.each_byte {|byte| a.push(byte) }  # Five 3-byte characters.
a # => [227, 129, 147, 227, 130, 147, 227, 129, 171, 227, 129, 161, 227, 129, 175]

With no block given, returns an enumerator.

Related: see Iterating.

static VALUE
rb_str_each_char(VALUE str)
{
    RETURN_SIZED_ENUMERATOR(str, 0, 0, rb_str_each_char_size);
    return rb_str_enumerate_chars(str, 0);
}

With a block given, calls the block with each successive character from self; returns self:

a = []
'hello'.each_char do |char|
  a.push(char)
end
a # => ["h", "e", "l", "l", "o"]
a = []
'тест'.each_char do |char|
  a.push(char)
end
a # => ["т", "е", "с", "т"]
a = []
'こんにちは'.each_char do |char|
  a.push(char)
end
a # => ["こ", "ん", "に", "ち", "は"]

With no block given, returns an enumerator.

Related: see Iterating.

static VALUE
rb_str_each_codepoint(VALUE str)
{
    RETURN_SIZED_ENUMERATOR(str, 0, 0, rb_str_each_char_size);
    return rb_str_enumerate_codepoints(str, 0);
}

With a block given, calls the block with each successive codepoint from self; each codepoint is the integer value for a character; returns self:

a = []
'hello'.each_codepoint do |codepoint|
  a.push(codepoint)
end
a # => [104, 101, 108, 108, 111]
a = []
'тест'.each_codepoint do |codepoint|
  a.push(codepoint)
end
a # => [1090, 1077, 1089, 1090]
a = []
'こんにちは'.each_codepoint do |codepoint|
  a.push(codepoint)
end
a # => [12371, 12435, 12395, 12385, 12399]

With no block given, returns an enumerator.

Related: see Iterating.

static VALUE
rb_str_each_grapheme_cluster(VALUE str)
{
    RETURN_SIZED_ENUMERATOR(str, 0, 0, rb_str_each_grapheme_cluster_size);
    return rb_str_enumerate_grapheme_clusters(str, 0);
}

With a block given, calls the given block with each successive grapheme cluster from self (see Unicode Grapheme Cluster Boundaries); returns self:

a = []
'hello'.each_grapheme_cluster do |grapheme_cluster|
  a.push(grapheme_cluster)
end
a  # => ["h", "e", "l", "l", "o"]

a = []
'тест'.each_grapheme_cluster do |grapheme_cluster|
  a.push(grapheme_cluster)
end
a # => ["т", "е", "с", "т"]

a = []
'こんにちは'.each_grapheme_cluster do |grapheme_cluster|
  a.push(grapheme_cluster)
end
a # => ["こ", "ん", "に", "ち", "は"]

With no block given, returns an enumerator.

Related: see Iterating.

static VALUE
rb_str_each_line(int argc, VALUE *argv, VALUE str)
{
    RETURN_SIZED_ENUMERATOR(str, argc, argv, 0);
    return rb_str_enumerate_lines(argc, argv, str, 0);
}

With a block given, forms the substrings (lines) that are the result of splitting self at each occurrence of the given record_separator; passes each line to the block; returns self.

With the default record_separator:

$/ # => "\n"
s = <<~EOT
This is the first line.
This is line two.

This is line four.
This is line five.
EOT
s.each_line {|line| p line }

Output:

"This is the first line.\n"
"This is line two.\n"
"\n"
"This is line four.\n"
"This is line five.\n"

With a different record_separator:

record_separator = ' is '
s.each_line(record_separator) {|line| p line }

Output:

"This is "
"the first line.\nThis is "
"line two.\n\nThis is "
"line four.\nThis is "
"line five.\n"

With chomp as true, removes the trailing record_separator from each line:

s.each_line(chomp: true) {|line| p line }

Output:

"This is the first line."
"This is line two."
""
"This is line four."
"This is line five."

With an empty string as record_separator, forms and passes “paragraphs” by splitting at each occurrence of two or more newlines:

record_separator = ''
s.each_line(record_separator) {|line| p line }

Output:

"This is the first line.\nThis is line two.\n\n"
"This is line four.\nThis is line five.\n"

With no block given, returns an enumerator.

Related: see Iterating.

static VALUE
rb_str_empty(VALUE str)
{
    return RBOOL(RSTRING_LEN(str) == 0);
}

Returns whether the length of self is zero:

'hello'.empty? # => false
' '.empty? # => false
''.empty? # => true

Related: see Querying.

static VALUE
str_encode(int argc, VALUE *argv, VALUE str)
{
    VALUE newstr = str;
    int encidx = str_transcode(argc, argv, &newstr);
    return encoded_dup(newstr, str, encidx);
}

Returns a copy of self transcoded as determined by dst_encoding; see Encodings.

By default, raises an exception if self contains an invalid byte or a character not defined in dst_encoding; that behavior may be modified by encoding options; see below.

With no arguments:

• Uses the same encoding if Encoding.default_internal is nil (the default):

  Encoding.default_internal # => nil
  s = "Ruby\x99".force_encoding('Windows-1252')
  s.encoding                # => #<Encoding:Windows-1252>
  s.bytes                   # => [82, 117, 98, 121, 153]
  t = s.encode              # => "Ruby\x99"
  t.encoding                # => #<Encoding:Windows-1252>
  t.bytes                   # => [82, 117, 98, 121, 226, 132, 162]
  

• Otherwise, uses the encoding Encoding.default_internal:

  Encoding.default_internal = 'UTF-8'
  t = s.encode              # => "Ruby™"
  t.encoding                # => #<Encoding:UTF-8>
  

With only argument dst_encoding given, uses that encoding:

s = "Ruby\x99".force_encoding('Windows-1252')
s.encoding            # => #<Encoding:Windows-1252>
t = s.encode('UTF-8') # => "Ruby™"
t.encoding            # => #<Encoding:UTF-8>

With arguments dst_encoding and src_encoding given, interprets self using src_encoding, encodes the new string using dst_encoding:

s = "Ruby\x99"
t = s.encode('UTF-8', 'Windows-1252') # => "Ruby™"
t.encoding                            # => #<Encoding:UTF-8>

Optional keyword arguments enc_opts specify encoding options; see Encoding Options.

Please note that, unless invalid: :replace option is given, conversion from an encoding enc to the same encoding enc (independent of whether enc is given explicitly or implicitly) is a no-op, i.e. the string is simply copied without any changes, and no exceptions are raised, even if there are invalid bytes.

Related: see Converting to New String.

static VALUE
str_encode_bang(int argc, VALUE *argv, VALUE str)
{
    VALUE newstr;
    int encidx;

    rb_check_frozen(str);

    newstr = str;
    encidx = str_transcode(argc, argv, &newstr);

    if (encidx < 0) return str;
    if (newstr == str) {
        rb_enc_associate_index(str, encidx);
        return str;
    }
    rb_str_shared_replace(str, newstr);
    return str_encode_associate(str, encidx);
}

Like encode, but applies encoding changes to self; returns self.

Related: see Modifying.

VALUE
rb_obj_encoding(VALUE obj)
{
    int idx = rb_enc_get_index(obj);
    if (idx < 0) {
        rb_raise(rb_eTypeError, "unknown encoding");
    }
    return rb_enc_from_encoding_index(idx & ENC_INDEX_MASK);
}

Returns an Encoding object that represents the encoding of self; see Encodings.

Related: see Querying.

static VALUE
rb_str_end_with(int argc, VALUE *argv, VALUE str)
{
    int i;

    for (i=0; i<argc; i++) {
        VALUE tmp = argv[i];
        const char *p, *s, *e;
        long slen, tlen;
        rb_encoding *enc;

        StringValue(tmp);
        enc = rb_enc_check(str, tmp);
        if ((tlen = RSTRING_LEN(tmp)) == 0) return Qtrue;
        if ((slen = RSTRING_LEN(str)) < tlen) continue;
        p = RSTRING_PTR(str);
        e = p + slen;
        s = e - tlen;
        if (!at_char_boundary(p, s, e, enc))
            continue;
        if (memcmp(s, RSTRING_PTR(tmp), tlen) == 0)
            return Qtrue;
    }
    return Qfalse;
}

Returns whether self ends with any of the given strings:

'foo'.end_with?('oo')         # => true
'foo'.end_with?('bar', 'oo')  # => true
'foo'.end_with?('bar', 'baz') # => false
'foo'.end_with?('')           # => true
'тест'.end_with?('т')         # => true
'こんにちは'.end_with?('は')   # => true

Related: see Querying.

VALUE
rb_str_eql(VALUE str1, VALUE str2)
{
    if (str1 == str2) return Qtrue;
    if (!RB_TYPE_P(str2, T_STRING)) return Qfalse;
    return rb_str_eql_internal(str1, str2);
}

Returns whether self and object have the same length and content:

s = 'foo'
s.eql?('foo')  # => true
s.eql?('food') # => false
s.eql?('FOO')  # => false

Returns false if the two strings’ encodings are not compatible:

s0 = "äöü"                           # => "äöü"
s1 = s0.encode(Encoding::ISO_8859_1) # => "\xE4\xF6\xFC"
s0.encoding                          # => #<Encoding:UTF-8>
s1.encoding                          # => #<Encoding:ISO-8859-1>
s0.eql?(s1)                          # => false

See Encodings.

Related: see Querying.

static VALUE
rb_str_force_encoding(VALUE str, VALUE enc)
{
    str_modifiable(str);

    rb_encoding *encoding = rb_to_encoding(enc);
    int idx = rb_enc_to_index(encoding);

    // If the encoding is unchanged, we do nothing.
    if (ENCODING_GET(str) == idx) {
        return str;
    }

    rb_enc_associate_index(str, idx);

    // If the coderange was 7bit and the new encoding is ASCII-compatible
    // we can keep the coderange.
    if (ENC_CODERANGE(str) == ENC_CODERANGE_7BIT && encoding && rb_enc_asciicompat(encoding)) {
        return str;
    }

    ENC_CODERANGE_CLEAR(str);
    return str;
}

Changes the encoding of self to the given encoding, which may be a string encoding name or an Encoding object; does not change the underlying bytes; returns self:

s = 'łał'
s.bytes                   # => [197, 130, 97, 197, 130]
s.encoding                # => #<Encoding:UTF-8>
s.force_encoding('ascii') # => "\xC5\x82a\xC5\x82"
s.encoding                # => #<Encoding:US-ASCII>
s.valid_encoding?         # => true
s.bytes                   # => [197, 130, 97, 197, 130]

Makes the change even if the given encoding is invalid for self (as is the change above):

s.valid_encoding?         # => false

See Encodings.

Related: see Modifying.

VALUE
rb_str_getbyte(VALUE str, VALUE index)
{
    long pos = NUM2LONG(index);

    if (pos < 0)
        pos += RSTRING_LEN(str);
    if (pos < 0 ||  RSTRING_LEN(str) <= pos)
        return Qnil;

    return INT2FIX((unsigned char)RSTRING_PTR(str)[pos]);
}

Returns the byte at zero-based index as an integer:

s = 'foo'
s.getbyte(0)    # => 102
s.getbyte(1)    # => 111
s.getbyte(2)    # => 111

Counts backward from the end if index is negative:

s.getbyte(-3) # => 102

Returns nil if index is out of range:

s.getbyte(3)  # => nil
s.getbyte(-4) # => nil

More examples:

s = 'тест'
s.bytes      # => [209, 130, 208, 181, 209, 129, 209, 130]
s.getbyte(2) # => 208
s = 'こんにちは'
s.bytes      # => [227, 129, 147, 227, 130, 147, 227, 129, 171, 227, 129, 161, 227, 129, 175]
s.getbyte(2) # => 147

Related: see Converting to Non-String.

static VALUE
rb_str_grapheme_clusters(VALUE str)
{
    VALUE ary = WANTARRAY("grapheme_clusters", rb_str_strlen(str));
    return rb_str_enumerate_grapheme_clusters(str, ary);
}

Returns an array of the grapheme clusters in self (see Unicode Grapheme Cluster Boundaries):

s = "ä-pqr-b̈-xyz-c̈"
s.size                   # => 16
s.bytesize               # => 19
s.grapheme_clusters.size # => 13
s.grapheme_clusters
# => ["ä", "-", "p", "q", "r", "-", "b̈", "-", "x", "y", "z", "-", "c̈"]

Details:

s = "ä"
s.grapheme_clusters             # => ["ä"]           # One grapheme cluster.
s.bytes                         # => [97, 204, 136]  # Three bytes.
s.chars                         # => ["a", "̈"]       # Two characters.
s.chars.map {|char| char.ord }  # => [97, 776]       # Their values.

Related: see Converting to Non-String.

static VALUE
rb_str_gsub(int argc, VALUE *argv, VALUE str)
{
    return str_gsub(argc, argv, str, 0);
}

Returns a copy of self with zero or more substrings replaced.

Argument pattern may be a string or a Regexp; argument replacement may be a string or a Hash. Varying types for the argument values makes this method very versatile.

Below are some simple examples; for many more examples, see Substitution Methods.

With arguments pattern and string replacement given, replaces each matching substring with the given replacement string:

s = 'abracadabra'
s.gsub('ab', 'AB')   # => "ABracadABra"
s.gsub(/[a-c]/, 'X') # => "XXrXXXdXXrX"

With arguments pattern and hash replacement given, replaces each matching substring with a value from the given replacement hash, or removes it:

h = {'a' => 'A', 'b' => 'B', 'c' => 'C'}
s.gsub(/[a-c]/, h) # => "ABrACAdABrA"  # 'a', 'b', 'c' replaced.
s.gsub(/[a-d]/, h) # => "ABrACAABrA"   # 'd' removed.

With argument pattern and a block given, calls the block with each matching substring; replaces that substring with the block’s return value:

s.gsub(/[a-d]/) {|substring| substring.upcase }
# => "ABrACADABrA"

With argument pattern and no block given, returns a new Enumerator.

Related: see Converting to New String.

static VALUE
rb_str_gsub_bang(int argc, VALUE *argv, VALUE str)
{
    str_modify_keep_cr(str);
    return str_gsub(argc, argv, str, 1);
}

Like String#gsub, except that:

• Performs substitutions in self (not in a copy of self).

• Returns self if any characters are removed, nil otherwise.

Related: see Modifying.

static VALUE
rb_str_hash_m(VALUE str)
{
    st_index_t hval = rb_str_hash(str);
    return ST2FIX(hval);
}

Returns the integer hash value for self.

Two String objects that have identical content and compatible encodings also have the same hash value; see Object#hash and Encodings:

s = 'foo'
h = s.hash       # => -569050784
h == 'foo'.hash  # => true
h == 'food'.hash # => false
h == 'FOO'.hash  # => false

s0 = "äöü"
s1 = s0.encode(Encoding::ISO_8859_1)
s0.encoding        # => #<Encoding:UTF-8>
s1.encoding        # => #<Encoding:ISO-8859-1>
s0.hash == s1.hash # => false

Related: see Querying.

static VALUE
rb_str_hex(VALUE str)
{
    return rb_str_to_inum(str, 16, FALSE);
}

Interprets the leading substring of self as hexadecimal, possibly signed; returns its value as an integer.

The leading substring is interpreted as hexadecimal when it begins with:

• One or more character representing hexadecimal digits (each in one of the ranges '0'..'9', 'a'..'f', or 'A'..'F'); the string to be interpreted ends at the first character that does not represent a hexadecimal digit:

  'f'.hex        # => 15
  '11'.hex       # => 17
  'FFF'.hex      # => 4095
  'fffg'.hex     # => 4095
  'foo'.hex      # => 15   # 'f' hexadecimal, 'oo' not.
  'bar'.hex      # => 186  # 'ba' hexadecimal, 'r' not.
  'deadbeef'.hex # => 3735928559
  

• '0x' or '0X', followed by one or more hexadecimal digits:

  '0xfff'.hex    # => 4095
  '0xfffg'.hex   # => 4095
  

Any of the above may prefixed with '-', which negates the interpreted value:

'-fff'.hex      # => -4095
'-0xFFF'.hex    # => -4095

For any substring not described above, returns zero:

'xxx'.hex     # => 0
''.hex        # => 0

Note that, unlike oct, this method interprets only hexadecimal, and not binary, octal, or decimal notations:

'0b111'.hex   # => 45329
'0o777'.hex   # => 0
'0d999'.hex   # => 55705

Related: See Converting to Non-String.

VALUE
rb_str_include(VALUE str, VALUE arg)
{
    long i;

    StringValue(arg);
    i = rb_str_index(str, arg, 0);

    return RBOOL(i != -1);
}

Returns whether self contains other_string:

s = 'bar'
s.include?('ba')  # => true
s.include?('ar')  # => true
s.include?('bar') # => true
s.include?('a')   # => true
s.include?('')    # => true
s.include?('foo') # => false

Related: see Querying.

static VALUE
rb_str_index_m(int argc, VALUE *argv, VALUE str)
{
    VALUE sub;
    VALUE initpos;
    rb_encoding *enc = STR_ENC_GET(str);
    long pos;

    if (rb_scan_args(argc, argv, "11", &sub, &initpos) == 2) {
        long slen = str_strlen(str, enc); /* str's enc */
        pos = NUM2LONG(initpos);
        if (pos < 0 ? (pos += slen) < 0 : pos > slen) {
            if (RB_TYPE_P(sub, T_REGEXP)) {
                rb_backref_set(Qnil);
            }
            return Qnil;
        }
    }
    else {
        pos = 0;
    }

    if (RB_TYPE_P(sub, T_REGEXP)) {
        pos = str_offset(RSTRING_PTR(str), RSTRING_END(str), pos,
                         enc, single_byte_optimizable(str));

        if (rb_reg_search(sub, str, pos, 0) >= 0) {
            VALUE match = rb_backref_get();
            struct re_registers *regs = RMATCH_REGS(match);
            pos = rb_str_sublen(str, BEG(0));
            return LONG2NUM(pos);
        }
    }
    else {
        StringValue(sub);
        pos = rb_str_index(str, sub, pos);
        if (pos >= 0) {
            pos = rb_str_sublen(str, pos);
            return LONG2NUM(pos);
        }
    }
    return Qnil;
}

Returns the integer position of the first substring that matches the given argument pattern, or nil if none found.

When pattern is a string, returns the index of the first matching substring in self:

'foo'.index('f')         # => 0
'foo'.index('o')         # => 1
'foo'.index('oo')        # => 1
'foo'.index('ooo')       # => nil
'тест'.index('с')        # => 2  # Characters, not bytes.
'こんにちは'.index('ち')  # => 3

When pattern is a Regexp, returns the index of the first match in self:

'foo'.index(/o./) # => 1
'foo'.index(/.o/) # => 0

When offset is non-negative, begins the search at position offset; the returned index is relative to the beginning of self:

'bar'.index('r', 0)        # => 2
'bar'.index('r', 1)        # => 2
'bar'.index('r', 2)        # => 2
'bar'.index('r', 3)        # => nil
'bar'.index(/[r-z]/, 0)    # => 2
'тест'.index('с', 1)       # => 2
'тест'.index('с', 2)       # => 2
'тест'.index('с', 3)       # => nil  # Offset in characters, not bytes.
'こんにちは'.index('ち', 2) # => 3

With negative integer argument offset, selects the search position by counting backward from the end of self:

'foo'.index('o', -1)  # => 2
'foo'.index('o', -2)  # => 1
'foo'.index('o', -3)  # => 1
'foo'.index('o', -4)  # => nil
'foo'.index(/o./, -2) # => 1
'foo'.index(/.o/, -2) # => 1

Related: see Querying.

static VALUE
rb_str_insert(VALUE str, VALUE idx, VALUE str2)
{
    long pos = NUM2LONG(idx);

    if (pos == -1) {
        return rb_str_append(str, str2);
    }
    else if (pos < 0) {
        pos++;
    }
    rb_str_update(str, pos, 0, str2);
    return str;
}

Inserts the given other_string into self; returns self.

If the given index is non-negative, inserts other_string at offset index:

'foo'.insert(0, 'bar')       # => "barfoo"
'foo'.insert(1, 'bar')       # => "fbaroo"
'foo'.insert(3, 'bar')       # => "foobar"
'тест'.insert(2, 'bar')      # => "теbarст"  # Characters, not bytes.
'こんにちは'.insert(2, 'bar') # => "こんbarにちは"

If the index is negative, counts backward from the end of self and inserts other_string after the offset:

'foo'.insert(-2, 'bar') # => "fobaro"

Related: see Modifying.

VALUE
rb_str_inspect(VALUE str)
{
    int encidx = ENCODING_GET(str);
    rb_encoding *enc = rb_enc_from_index(encidx);
    const char *p, *pend, *prev;
    char buf[CHAR_ESC_LEN + 1];
    VALUE result = rb_str_buf_new(0);
    rb_encoding *resenc = rb_default_internal_encoding();
    int unicode_p = rb_enc_unicode_p(enc);
    int asciicompat = rb_enc_asciicompat(enc);

    if (resenc == NULL) resenc = rb_default_external_encoding();
    if (!rb_enc_asciicompat(resenc)) resenc = rb_usascii_encoding();
    rb_enc_associate(result, resenc);
    str_buf_cat2(result, "\"");

    p = RSTRING_PTR(str); pend = RSTRING_END(str);
    prev = p;
    while (p < pend) {
        unsigned int c, cc;
        int n;

        n = rb_enc_precise_mbclen(p, pend, enc);
        if (!MBCLEN_CHARFOUND_P(n)) {
            if (p > prev) str_buf_cat(result, prev, p - prev);
            n = rb_enc_mbminlen(enc);
            if (pend < p + n)
                n = (int)(pend - p);
            while (n--) {
                snprintf(buf, CHAR_ESC_LEN, "\\x%02X", *p & 0377);
                str_buf_cat(result, buf, strlen(buf));
                prev = ++p;
            }
            continue;
        }
        n = MBCLEN_CHARFOUND_LEN(n);
        c = rb_enc_mbc_to_codepoint(p, pend, enc);
        p += n;
        if ((asciicompat || unicode_p) &&
          (c == '"'|| c == '\\' ||
            (c == '#' &&
             p < pend &&
             MBCLEN_CHARFOUND_P(rb_enc_precise_mbclen(p,pend,enc)) &&
             (cc = rb_enc_codepoint(p,pend,enc),
              (cc == '$' || cc == '@' || cc == '{'))))) {
            if (p - n > prev) str_buf_cat(result, prev, p - n - prev);
            str_buf_cat2(result, "\\");
            if (asciicompat || enc == resenc) {
                prev = p - n;
                continue;
            }
        }
        switch (c) {
          case '\n': cc = 'n'; break;
          case '\r': cc = 'r'; break;
          case '\t': cc = 't'; break;
          case '\f': cc = 'f'; break;
          case '\013': cc = 'v'; break;
          case '\010': cc = 'b'; break;
          case '\007': cc = 'a'; break;
          case 033: cc = 'e'; break;
          default: cc = 0; break;
        }
        if (cc) {
            if (p - n > prev) str_buf_cat(result, prev, p - n - prev);
            buf[0] = '\\';
            buf[1] = (char)cc;
            str_buf_cat(result, buf, 2);
            prev = p;
            continue;
        }
        /* The special casing of 0x85 (NEXT_LINE) here is because
         * Oniguruma historically treats it as printable, but it
         * doesn't match the print POSIX bracket class or character
         * property in regexps.
         *
         * See Ruby Bug #16842 for details:
         * https://bugs.ruby-lang.org/issues/16842
         */
        if ((enc == resenc && rb_enc_isprint(c, enc) && c != 0x85) ||
            (asciicompat && rb_enc_isascii(c, enc) && ISPRINT(c))) {
            continue;
        }
        else {
            if (p - n > prev) str_buf_cat(result, prev, p - n - prev);
            rb_str_buf_cat_escaped_char(result, c, unicode_p);
            prev = p;
            continue;
        }
    }
    if (p > prev) str_buf_cat(result, prev, p - prev);
    str_buf_cat2(result, "\"");

    return result;
}

Returns a printable version of self, enclosed in double-quotes.

Most printable characters are rendered simply as themselves:

'abc'.inspect        # => "\"abc\""
'012'.inspect        # => "\"012\""
''.inspect           # => "\"\""
"\u000012".inspect   # => "\"\\u000012\""
'тест'.inspect       # => "\"тест\""
'こんにちは'.inspect  # => "\"こんにちは\""

But printable characters double-quote ('"') and backslash and ('\') are escaped:

'"'.inspect  # => "\"\\\"\""
'\\'.inspect # => "\"\\\\\""

Unprintable characters are the ASCII characters whose values are in range 0..31, along with the character whose value is 127.

Most of these characters are rendered thus:

0.chr.inspect # => "\"\\x00\""
1.chr.inspect # => "\"\\x01\""
2.chr.inspect # => "\"\\x02\""
# ...

A few, however, have special renderings:

7.chr.inspect  # => "\"\\a\""  # BEL
8.chr.inspect  # => "\"\\b\""  # BS
9.chr.inspect  # => "\"\\t\""  # TAB
10.chr.inspect # => "\"\\n\""  # LF
11.chr.inspect # => "\"\\v\""  # VT
12.chr.inspect # => "\"\\f\""  # FF
13.chr.inspect # => "\"\\r\""  # CR
27.chr.inspect # => "\"\\e\""  # ESC

Related: see Converting to Non-String.

VALUE
rb_str_intern(VALUE str)
{
    return sym_find_or_insert_dynamic_symbol(&ruby_global_symbols, str);
}

Returns the Symbol object derived from self, creating it if it did not already exist:

'foo'.intern       # => :foo
'тест'.intern      # => :тест
'こんにちは'.intern # => :こんにちは

Related: see Converting to Non-String.

VALUE
rb_str_length(VALUE str)
{
    return LONG2NUM(str_strlen(str, NULL));
}

Returns the count of characters (not bytes) in self:

'foo'.length        # => 3
'тест'.length       # => 4
'こんにちは'.length  # => 5

Contrast with String#bytesize:

'foo'.bytesize        # => 3
'тест'.bytesize       # => 8
'こんにちは'.bytesize  # => 15

Related: see Querying.

static VALUE
rb_str_lines(int argc, VALUE *argv, VALUE str)
{
    VALUE ary = WANTARRAY("lines", 0);
    return rb_str_enumerate_lines(argc, argv, str, ary);
}

Returns substrings (“lines”) of self according to the given arguments:

s = <<~EOT
This is the first line.
This is line two.

This is line four.
This is line five.
EOT

With the default argument values:

$/ # => "\n"
s.lines
# =>
["This is the first line.\n",
 "This is line two.\n",
 "\n",
 "This is line four.\n",
 "This is line five.\n"]

With a different record_separator:

record_separator = ' is '
s.lines(record_separator)
# =>
["This is ",
 "the first line.\nThis is ",
 "line two.\n\nThis is ",
 "line four.\nThis is ",
 "line five.\n"]

With keyword argument chomp as true, removes the trailing newline from each line:

s.lines(chomp: true)
# =>
["This is the first line.",
 "This is line two.",
 "",
 "This is line four.",
 "This is line five."]

Related: see Converting to Non-String.

static VALUE
rb_str_ljust(int argc, VALUE *argv, VALUE str)
{
    return rb_str_justify(argc, argv, str, 'l');
}

Returns a copy of self, left-justified and, if necessary, right-padded with the pad_string:

'hello'.ljust(10)       # => "hello     "
'  hello'.ljust(10)     # => "  hello   "
'hello'.ljust(10, 'ab') # => "helloababa"
'тест'.ljust(10)        # => "тест      "
'こんにちは'.ljust(10)   # => "こんにちは     "

If width <= self.length, returns a copy of self:

'hello'.ljust(5)  # => "hello"
'hello'.ljust(1)  # => "hello"  # Does not truncate to width.

Related: see Converting to New String.

static VALUE
rb_str_lstrip(int argc, VALUE *argv, VALUE str)
{
    char *start;
    long len, loffset;

    RSTRING_GETMEM(str, start, len);
    if (argc > 0) {
        char table[TR_TABLE_SIZE];
        VALUE del = 0, nodel = 0;

        tr_setup_table_multi(table, &del, &nodel, str, argc, argv);
        loffset = lstrip_offset_table(str, start, start+len, STR_ENC_GET(str), table, del, nodel);
    }
    else {
        loffset = lstrip_offset(str, start, start+len, STR_ENC_GET(str));
    }
    if (loffset <= 0) return str_duplicate(rb_cString, str);
    return rb_str_subseq(str, loffset, len - loffset);
}

Returns a copy of self with leading whitespace removed; see Whitespace in Strings:

whitespace = "\x00\t\n\v\f\r "
s = whitespace + 'abc' + whitespace
# => "\u0000\t\n\v\f\r abc\u0000\t\n\v\f\r "
s.lstrip
# => "abc\u0000\t\n\v\f\r "

If selectors are given, removes characters of selectors from the beginning of self:

s = "---abc+++"
s.lstrip("-") # => "abc+++"

selectors must be valid character selectors (see Character Selectors), and may use any of its valid forms, including negation, ranges, and escapes:

"01234abc56789".lstrip("0-9") # "abc56789"
"01234abc56789".lstrip("0-9", "^4-6") # "4abc56789"

Related: see Converting to New String.

static VALUE
rb_str_lstrip_bang(int argc, VALUE *argv, VALUE str)
{
    rb_encoding *enc;
    char *start, *s;
    long olen, loffset;

    str_modify_keep_cr(str);
    enc = STR_ENC_GET(str);
    RSTRING_GETMEM(str, start, olen);
    if (argc > 0) {
        char table[TR_TABLE_SIZE];
        VALUE del = 0, nodel = 0;

        tr_setup_table_multi(table, &del, &nodel, str, argc, argv);
        loffset = lstrip_offset_table(str, start, start+olen, enc, table, del, nodel);
    }
    else {
        loffset = lstrip_offset(str, start, start+olen, enc);
    }

    if (loffset > 0) {
        long len = olen-loffset;
        s = start + loffset;
        memmove(start, s, len);
        STR_SET_LEN(str, len);
        TERM_FILL(start+len, rb_enc_mbminlen(enc));
        return str;
    }
    return Qnil;
}

Like String#lstrip, except that:

• Performs stripping in self (not in a copy of self).

• Returns self if any characters are stripped, nil otherwise.

Related: see Modifying.

static VALUE
rb_str_match_m(int argc, VALUE *argv, VALUE str)
{
    VALUE re, result;
    if (argc < 1)
        rb_check_arity(argc, 1, 2);
    re = argv[0];
    argv[0] = str;
    result = rb_funcallv(get_pat(re), rb_intern("match"), argc, argv);
    if (!NIL_P(result) && rb_block_given_p()) {
        return rb_yield(result);
    }
    return result;
}

Creates a MatchData object based on self and the given arguments; updates Regexp Global Variables.

• Computes regexp by converting pattern (if not already a Regexp).

  regexp = Regexp.new(pattern)
  

• Computes matchdata, which will be either a MatchData object or nil (see Regexp#match):

  matchdata = regexp.match(self[offset..])
  

With no block given, returns the computed matchdata or nil:

'foo'.match('f')    # => #<MatchData "f">
'foo'.match('o')    # => #<MatchData "o">
'foo'.match('x')    # => nil
'foo'.match('f', 1) # => nil
'foo'.match('o', 1) # => #<MatchData "o">

With a block given and computed matchdata non-nil, calls the block with matchdata; returns the block’s return value:

'foo'.match(/o/) {|matchdata| matchdata } # => #<MatchData "o">

With a block given and nil matchdata, does not call the block:

'foo'.match(/x/) {|matchdata| fail 'Cannot happen' } # => nil

Related: see Querying.

static VALUE
rb_str_match_m_p(int argc, VALUE *argv, VALUE str)
{
    VALUE re;
    rb_check_arity(argc, 1, 2);
    re = get_pat(argv[0]);
    return rb_reg_match_p(re, str, argc > 1 ? NUM2LONG(argv[1]) : 0);
}

Returns whether a match is found for self and the given arguments; does not update Regexp Global Variables.

Computes regexp by converting pattern (if not already a Regexp):

regexp = Regexp.new(pattern)

Returns true if self[offset..].match(regexp) returns a MatchData object, false otherwise:

'foo'.match?(/o/) # => true
'foo'.match?('o') # => true
'foo'.match?(/x/) # => false
'foo'.match?('f', 1) # => false
'foo'.match?('o', 1) # => true

Related: see Querying.

static VALUE
rb_str_oct(VALUE str)
{
    return rb_str_to_inum(str, -8, FALSE);
}

Interprets the leading substring of self as octal, binary, decimal, or hexadecimal, possibly signed; returns their value as an integer.

In brief:

# Interpreted as octal.
'777'.oct   # => 511
'777x'.oct  # => 511
'0777'.oct  # => 511
'0o777'.oct # => 511
'-777'.oct  # => -511
# Not interpreted as octal.
'0b111'.oct # => 7     # Interpreted as binary.
'0d999'.oct # => 999   # Interpreted as decimal.
'0xfff'.oct # => 4095  # Interpreted as hexadecimal.

The leading substring is interpreted as octal when it begins with:

• One or more character representing octal digits (each in the range '0'..'7'); the string to be interpreted ends at the first character that does not represent an octal digit:

  '7'.oct      @ => 7
  '11'.oct     # => 9
  '777'.oct    # => 511
  '0777'.oct   # => 511
  '7778'.oct   # => 511
  '777x'.oct   # => 511

• '0o', followed by one or more octal digits:

  '0o777'.oct  # => 511
  '0o7778'.oct # => 511
  

The leading substring is not interpreted as octal when it begins with:

• '0b', followed by one or more characters representing binary digits (each in the range '0'..'1'); the string to be interpreted ends at the first character that does not represent a binary digit. the string is interpreted as binary digits (base 2):

  '0b111'.oct  # => 7
  '0b1112'.oct # => 7
  

• '0d', followed by one or more characters representing decimal digits (each in the range '0'..'9'); the string to be interpreted ends at the first character that does not represent a decimal digit. the string is interpreted as decimal digits (base 10):

  '0d999'.oct  # => 999
  '0d999x'.oct # => 999
  

• '0x', followed by one or more characters representing hexadecimal digits (each in one of the ranges '0'..'9', 'a'..'f', or 'A'..'F'); the string to be interpreted ends at the first character that does not represent a hexadecimal digit. the string is interpreted as hexadecimal digits (base 16):

  '0xfff'.oct  # => 4095
  '0xfffg'.oct # => 4095
  

Any of the above may prefixed with '-', which negates the interpreted value:

'-777'.oct   # => -511
'-0777'.oct  # => -511
'-0b111'.oct # => -7
'-0xfff'.oct # => -4095

For any substring not described above, returns zero:

'foo'.oct      # => 0
''.oct         # => 0

Related: see Converting to Non-String.

static VALUE
rb_str_ord(VALUE s)
{
    unsigned int c;

    c = rb_enc_codepoint(RSTRING_PTR(s), RSTRING_END(s), STR_ENC_GET(s));
    return UINT2NUM(c);
}

Returns the integer ordinal of the first character of self:

'h'.ord         # => 104
'hello'.ord     # => 104
'тест'.ord      # => 1090
'こんにちは'.ord  # => 12371

Related: see Converting to Non-String.

static VALUE
rb_str_partition(VALUE str, VALUE sep)
{
    long pos;

    sep = get_pat_quoted(sep, 0);
    if (RB_TYPE_P(sep, T_REGEXP)) {
        if (rb_reg_search(sep, str, 0, 0) < 0) {
            goto failed;
        }
        VALUE match = rb_backref_get();
        struct re_registers *regs = RMATCH_REGS(match);

        pos = BEG(0);
        sep = rb_str_subseq(str, pos, END(0) - pos);
    }
    else {
        pos = rb_str_index(str, sep, 0);
        if (pos < 0) goto failed;
    }
    return rb_ary_new3(3, rb_str_subseq(str, 0, pos),
                          sep,
                          rb_str_subseq(str, pos+RSTRING_LEN(sep),
                                             RSTRING_LEN(str)-pos-RSTRING_LEN(sep)));

  failed:
    return rb_ary_new3(3, str_duplicate(rb_cString, str), str_new_empty_String(str), str_new_empty_String(str));
}

Returns a 3-element array of substrings of self.

If pattern is matched, returns the array:

[pre_match, first_match, post_match]

where:

• first_match is the first-found matching substring.

• pre_match and post_match are the preceding and following substrings.

If pattern is not matched, returns the array:

[self.dup, "", ""]

Note that in the examples below, a returned string 'hello' is a copy of self, not self.

If pattern is a Regexp, performs the equivalent of self.match(pattern) (also setting matched-data variables):

'hello'.partition(/h/)  # => ["", "h", "ello"]
'hello'.partition(/l/)  # => ["he", "l", "lo"]
'hello'.partition(/l+/) # => ["he", "ll", "o"]
'hello'.partition(/o/)  # => ["hell", "o", ""]
'hello'.partition(/^/)  # => ["", "", "hello"]
'hello'.partition(//)   # => ["", "", "hello"]
'hello'.partition(/$/)  # => ["hello", "", ""]
'hello'.partition(/x/)  # => ["hello", "", ""]

If pattern is not a Regexp, converts it to a string (if it is not already one), then performs the equivalent of self.index(pattern) (and does not set matched-data global variables):

'hello'.partition('h')     # => ["", "h", "ello"]
'hello'.partition('l')     # => ["he", "l", "lo"]
'hello'.partition('ll')    # => ["he", "ll", "o"]
'hello'.partition('o')     # => ["hell", "o", ""]
'hello'.partition('')      # => ["", "", "hello"]
'hello'.partition('x')     # => ["hello", "", ""]
'тест'.partition('т')      # => ["", "т", "ест"]
'こんにちは'.partition('に') # => ["こん", "に", "ちは"]

Related: see Converting to Non-String.

static VALUE
rb_str_prepend_multi(int argc, VALUE *argv, VALUE str)
{
    str_modifiable(str);

    if (argc == 1) {
        rb_str_update(str, 0L, 0L, argv[0]);
    }
    else if (argc > 1) {
        int i;
        VALUE arg_str = rb_str_tmp_new(0);
        rb_enc_copy(arg_str, str);
        for (i = 0; i < argc; i++) {
            rb_str_append(arg_str, argv[i]);
        }
        rb_str_update(str, 0L, 0L, arg_str);
    }

    return str;
}

Prefixes to self the concatenation of the given other_strings; returns self:

'baz'.prepend('foo', 'bar') # => "foobarbaz"

Related: see Modifying.

VALUE
rb_str_replace(VALUE str, VALUE str2)
{
    str_modifiable(str);
    if (str == str2) return str;

    StringValue(str2);
    str_discard(str);
    return str_replace(str, str2);
}

Replaces the contents of self with the contents of other_string; returns self:

s = 'foo'        # => "foo"
s.replace('bar') # => "bar"

Related: see Modifying.

static VALUE
rb_str_reverse(VALUE str)
{
    rb_encoding *enc;
    VALUE rev;
    char *s, *e, *p;
    int cr;

    if (RSTRING_LEN(str) <= 1) return str_duplicate(rb_cString, str);
    enc = STR_ENC_GET(str);
    rev = rb_str_new(0, RSTRING_LEN(str));
    s = RSTRING_PTR(str); e = RSTRING_END(str);
    p = RSTRING_END(rev);
    cr = ENC_CODERANGE(str);

    if (RSTRING_LEN(str) > 1) {
        if (single_byte_optimizable(str)) {
            while (s < e) {
                *--p = *s++;
            }
        }
        else if (cr == ENC_CODERANGE_VALID) {
            while (s < e) {
                int clen = rb_enc_fast_mbclen(s, e, enc);

                p -= clen;
                memcpy(p, s, clen);
                s += clen;
            }
        }
        else {
            cr = rb_enc_asciicompat(enc) ?
                ENC_CODERANGE_7BIT : ENC_CODERANGE_VALID;
            while (s < e) {
                int clen = rb_enc_mbclen(s, e, enc);

                if (clen > 1 || (*s & 0x80)) cr = ENC_CODERANGE_UNKNOWN;
                p -= clen;
                memcpy(p, s, clen);
                s += clen;
            }
        }
    }
    STR_SET_LEN(rev, RSTRING_LEN(str));
    str_enc_copy_direct(rev, str);
    ENC_CODERANGE_SET(rev, cr);

    return rev;
}

Returns a new string with the characters from self in reverse order.

'drawer'.reverse       # => "reward"
'reviled'.reverse      # => "deliver"
'stressed'.reverse     # => "desserts"
'semordnilaps'.reverse # => "spalindromes"

Related: see Converting to New String.

static VALUE
rb_str_reverse_bang(VALUE str)
{
    if (RSTRING_LEN(str) > 1) {
        if (single_byte_optimizable(str)) {
            char *s, *e, c;

            str_modify_keep_cr(str);
            s = RSTRING_PTR(str);
            e = RSTRING_END(str) - 1;
            while (s < e) {
                c = *s;
                *s++ = *e;
                *e-- = c;
            }
        }
        else {
            str_shared_replace(str, rb_str_reverse(str));
        }
    }
    else {
        str_modify_keep_cr(str);
    }
    return str;
}

Returns self with its characters reversed:

'drawer'.reverse!       # => "reward"
'reviled'.reverse!      # => "deliver"
'stressed'.reverse!     # => "desserts"
'semordnilaps'.reverse! # => "spalindromes"

Related: see Modifying.

static VALUE
rb_str_rindex_m(int argc, VALUE *argv, VALUE str)
{
    VALUE sub;
    VALUE initpos;
    rb_encoding *enc = STR_ENC_GET(str);
    long pos, len = str_strlen(str, enc); /* str's enc */

    if (rb_scan_args(argc, argv, "11", &sub, &initpos) == 2) {
        pos = NUM2LONG(initpos);
        if (pos < 0 && (pos += len) < 0) {
            if (RB_TYPE_P(sub, T_REGEXP)) {
                rb_backref_set(Qnil);
            }
            return Qnil;
        }
        if (pos > len) pos = len;
    }
    else {
        pos = len;
    }

    if (RB_TYPE_P(sub, T_REGEXP)) {
        /* enc = rb_enc_check(str, sub); */
        pos = str_offset(RSTRING_PTR(str), RSTRING_END(str), pos,
                         enc, single_byte_optimizable(str));

        if (rb_reg_search(sub, str, pos, 1) >= 0) {
            VALUE match = rb_backref_get();
            struct re_registers *regs = RMATCH_REGS(match);
            pos = rb_str_sublen(str, BEG(0));
            return LONG2NUM(pos);
        }
    }
    else {
        StringValue(sub);
        pos = rb_str_rindex(str, sub, pos);
        if (pos >= 0) {
            pos = rb_str_sublen(str, pos);
            return LONG2NUM(pos);
        }
    }
    return Qnil;
}

Returns the integer position of the last substring that matches the given argument pattern, or nil if none found.

When pattern is a string, returns the index of the last matching substring in self:

'foo'.rindex('f')       # => 0
'foo'.rindex('o')       # => 2
'foo'.rindex('oo'       # => 1
'foo'.rindex('ooo')     # => nil
'тест'.rindex('т')      # => 3
'こんにちは'.rindex('ち') # => 3

When pattern is a Regexp, returns the index of the last match in self:

'foo'.rindex(/f/)   # => 0
'foo'.rindex(/o/)   # => 2
'foo'.rindex(/oo/)  # => 1
'foo'.rindex(/ooo/) # => nil

When offset is non-negative, it specifies the maximum starting position in the string to end the search:

'foo'.rindex('o', 0) # => nil
'foo'.rindex('o', 1) # => 1
'foo'.rindex('o', 2) # => 2
'foo'.rindex('o', 3) # => 2

With negative integer argument offset, selects the search position by counting backward from the end of self:

'foo'.rindex('o', -1) # => 2
'foo'.rindex('o', -2) # => 1
'foo'.rindex('o', -3) # => nil
'foo'.rindex('o', -4) # => nil

The last match means starting at the possible last position, not the last of longest matches:

'foo'.rindex(/o+/) # => 2
$~                 # => #<MatchData "o">

To get the last longest match, combine with negative lookbehind:

'foo'.rindex(/(?<!o)o+/) # => 1
$~                       # => #<MatchData "oo">

Or String#index with negative lookforward.

'foo'.index(/o+(?!.*o)/) # => 1
$~                       # => #<MatchData "oo">

Related: see Querying.

static VALUE
rb_str_rjust(int argc, VALUE *argv, VALUE str)
{
    return rb_str_justify(argc, argv, str, 'r');
}

Returns a right-justified copy of self.

If integer argument width is greater than the size (in characters) of self, returns a new string of length width that is a copy of self, right justified and padded on the left with pad_string:

'hello'.rjust(10)       # => "     hello"
'hello  '.rjust(10)     # => "   hello  "
'hello'.rjust(10, 'ab') # => "ababahello"
'тест'.rjust(10)        # => "      тест"
'こんにちは'.rjust(10)    # => "     こんにちは"

If width <= self.size, returns a copy of self:

'hello'.rjust(5, 'ab')  # => "hello"
'hello'.rjust(1, 'ab')  # => "hello"

Related: see Converting to New String.

static VALUE
rb_str_rpartition(VALUE str, VALUE sep)
{
    long pos = RSTRING_LEN(str);

    sep = get_pat_quoted(sep, 0);
    if (RB_TYPE_P(sep, T_REGEXP)) {
        if (rb_reg_search(sep, str, pos, 1) < 0) {
            goto failed;
        }
        VALUE match = rb_backref_get();
        struct re_registers *regs = RMATCH_REGS(match);

        pos = BEG(0);
        sep = rb_str_subseq(str, pos, END(0) - pos);
    }
    else {
        pos = rb_str_sublen(str, pos);
        pos = rb_str_rindex(str, sep, pos);
        if (pos < 0) {
            goto failed;
        }
    }

    return rb_ary_new3(3, rb_str_subseq(str, 0, pos),
                          sep,
                          rb_str_subseq(str, pos+RSTRING_LEN(sep),
                                        RSTRING_LEN(str)-pos-RSTRING_LEN(sep)));
  failed:
    return rb_ary_new3(3, str_new_empty_String(str), str_new_empty_String(str), str_duplicate(rb_cString, str));
}

Returns a 3-element array of substrings of self.

Searches self for a match of pattern, seeking the last match.

If pattern is not matched, returns the array:

["", "", self.dup]

If pattern is matched, returns the array:

[pre_match, last_match, post_match]

where:

• last_match is the last-found matching substring.

• pre_match and post_match are the preceding and following substrings.

The pattern used is:

• pattern itself, if it is a Regexp.

• Regexp.quote(pattern), if pattern is a string.

Note that in the examples below, a returned string 'hello' is a copy of self, not self.

If pattern is a Regexp, searches for the last matching substring (also setting matched-data global variables):

'hello'.rpartition(/l/)     # => ["hel", "l", "o"]
'hello'.rpartition(/ll/)    # => ["he", "ll", "o"]
'hello'.rpartition(/h/)     # => ["", "h", "ello"]
'hello'.rpartition(/o/)     # => ["hell", "o", ""]
'hello'.rpartition(//)      # => ["hello", "", ""]
'hello'.rpartition(/x/)     # => ["", "", "hello"]
'тест'.rpartition(/т/)      # => ["тес", "т", ""]
'こんにちは'.rpartition(/に/) # => ["こん", "に", "ちは"]

If pattern is not a Regexp, converts it to a string (if it is not already one), then searches for the last matching substring (and does not set matched-data global variables):

'hello'.rpartition('l')     # => ["hel", "l", "o"]
'hello'.rpartition('ll')    # => ["he", "ll", "o"]
'hello'.rpartition('h')     # => ["", "h", "ello"]
'hello'.rpartition('o')     # => ["hell", "o", ""]
'hello'.rpartition('')      # => ["hello", "", ""]
'тест'.rpartition('т')      # => ["тес", "т", ""]
'こんにちは'.rpartition('に') # => ["こん", "に", "ちは"]

Related: see Converting to Non-String.

static VALUE
rb_str_rstrip(int argc, VALUE *argv, VALUE str)
{
    rb_encoding *enc;
    char *start;
    long olen, roffset;

    enc = STR_ENC_GET(str);
    RSTRING_GETMEM(str, start, olen);
    if (argc > 0) {
        char table[TR_TABLE_SIZE];
        VALUE del = 0, nodel = 0;

        tr_setup_table_multi(table, &del, &nodel, str, argc, argv);
        roffset = rstrip_offset_table(str, start, start+olen, enc, table, del, nodel);
    }
    else {
        roffset = rstrip_offset(str, start, start+olen, enc);
    }
    if (roffset <= 0) return str_duplicate(rb_cString, str);
    return rb_str_subseq(str, 0, olen-roffset);
}

Returns a copy of self with trailing whitespace removed; see Whitespace in Strings:

whitespace = "\x00\t\n\v\f\r "
s = whitespace + 'abc' + whitespace
s        # => "\u0000\t\n\v\f\r abc\u0000\t\n\v\f\r "
s.rstrip # => "\u0000\t\n\v\f\r abc"

If selectors are given, removes characters of selectors from the end of self:

s = "---abc+++"
s.rstrip("+") # => "---abc"

selectors must be valid character selectors (see Character Selectors), and may use any of its valid forms, including negation, ranges, and escapes:

"01234abc56789".rstrip("0-9") # "01234abc"
"01234abc56789".rstrip("0-9", "^4-6") # "01234abc56"

Related: see Converting to New String.

static VALUE
rb_str_rstrip_bang(int argc, VALUE *argv, VALUE str)
{
    rb_encoding *enc;
    char *start;
    long olen, roffset;

    str_modify_keep_cr(str);
    enc = STR_ENC_GET(str);
    RSTRING_GETMEM(str, start, olen);
    if (argc > 0) {
        char table[TR_TABLE_SIZE];
        VALUE del = 0, nodel = 0;

        tr_setup_table_multi(table, &del, &nodel, str, argc, argv);
        roffset = rstrip_offset_table(str, start, start+olen, enc, table, del, nodel);
    }
    else {
        roffset = rstrip_offset(str, start, start+olen, enc);
    }
    if (roffset > 0) {
        long len = olen - roffset;

        STR_SET_LEN(str, len);
        TERM_FILL(start+len, rb_enc_mbminlen(enc));
        return str;
    }
    return Qnil;
}

Like String#rstrip, except that:

• Performs stripping in self (not in a copy of self).

• Returns self if any characters are stripped, nil otherwise.

Related: see Modifying.

static VALUE
rb_str_scan(VALUE str, VALUE pat)
{
    VALUE result;
    long start = 0;
    long last = -1, prev = 0;
    char *p = RSTRING_PTR(str); long len = RSTRING_LEN(str);

    pat = get_pat_quoted(pat, 1);
    mustnot_broken(str);
    if (!rb_block_given_p()) {
        VALUE ary = rb_ary_new();

        while (!NIL_P(result = scan_once(str, pat, &start, 0))) {
            last = prev;
            prev = start;
            rb_ary_push(ary, result);
        }
        if (last >= 0) rb_pat_search(pat, str, last, 1);
        else rb_backref_set(Qnil);
        return ary;
    }

    while (!NIL_P(result = scan_once(str, pat, &start, 1))) {
        last = prev;
        prev = start;
        rb_yield(result);
        str_mod_check(str, p, len);
    }
    if (last >= 0) rb_pat_search(pat, str, last, 1);
    return str;
}

Matches a pattern against self:

• If pattern is a Regexp, the pattern used is pattern itself.

• If pattern is a string, the pattern used is Regexp.quote(pattern).

Generates a collection of matching results and updates regexp-related global variables:

• If the pattern contains no groups, each result is a matched substring.

• If the pattern contains groups, each result is an array containing a matched substring for each group.

With no block given, returns an array of the results:

'cruel world'.scan(/\w+/)      # => ["cruel", "world"]
'cruel world'.scan(/.../)      # => ["cru", "el ", "wor"]
'cruel world'.scan(/(...)/)    # => [["cru"], ["el "], ["wor"]]
'cruel world'.scan(/(..)(..)/) # => [["cr", "ue"], ["l ", "wo"]]
'тест'.scan(/../)              # => ["те", "ст"]
'こんにちは'.scan(/../)         # => ["こん", "にち"]
'abracadabra'.scan('ab')       # => ["ab", "ab"]
'abracadabra'.scan('nosuch')   # => []

With a block given, calls the block with each result; returns self:

'cruel world'.scan(/\w+/) {|w| p w }
# => "cruel"
# => "world"
'cruel world'.scan(/(.)(.)/) {|x, y| p [x, y] }
# => ["c", "r"]
# => ["u", "e"]
# => ["l", " "]
# => ["w", "o"]
# => ["r", "l"]

Related: see Converting to Non-String.

static VALUE
str_scrub(int argc, VALUE *argv, VALUE str)
{
    VALUE repl = argc ? (rb_check_arity(argc, 0, 1), argv[0]) : Qnil;
    VALUE new = rb_str_scrub(str, repl);
    return NIL_P(new) ? str_duplicate(rb_cString, str): new;
}

Returns a copy of self with each invalid byte sequence replaced by the given replacement_string.

With no block given, replaces each invalid sequence with the given default_replacement_string (by default, "�" for a Unicode encoding, '?' otherwise):

"foo\x81\x81bar"scrub                             # => "foo��bar"
"foo\x81\x81bar".force_encoding('US-ASCII').scrub # => "foo??bar"
"foo\x81\x81bar".scrub('xyzzy')                   # => "fooxyzzyxyzzybar"

With a block given, calls the block with each invalid sequence, and replaces that sequence with the return value of the block:

"foo\x81\x81bar".scrub {|sequence| p sequence; 'XYZZY' } # => "fooXYZZYXYZZYbar"

Output :

"\x81"
"\x81"

Related: see Converting to New String.

static VALUE
str_scrub_bang(int argc, VALUE *argv, VALUE str)
{
    VALUE repl = argc ? (rb_check_arity(argc, 0, 1), argv[0]) : Qnil;
    VALUE new = rb_str_scrub(str, repl);
    if (!NIL_P(new)) rb_str_replace(str, new);
    return str;
}

Like String#scrub, except that:

• Any replacements are made in self.

• Returns self.

Related: see Modifying.

VALUE
rb_str_setbyte(VALUE str, VALUE index, VALUE value)
{
    long pos = NUM2LONG(index);
    long len = RSTRING_LEN(str);
    char *ptr, *head, *left = 0;
    rb_encoding *enc;
    int cr = ENC_CODERANGE_UNKNOWN, width, nlen;

    if (pos < -len || len <= pos)
        rb_raise(rb_eIndexError, "index %ld out of string", pos);
    if (pos < 0)
        pos += len;

    VALUE v = rb_to_int(value);
    VALUE w = rb_int_and(v, INT2FIX(0xff));
    char byte = (char)(NUM2INT(w) & 0xFF);

    if (!str_independent(str))
        str_make_independent(str);
    enc = STR_ENC_GET(str);
    head = RSTRING_PTR(str);
    ptr = &head[pos];
    if (!STR_EMBED_P(str)) {
        cr = ENC_CODERANGE(str);
        switch (cr) {
          case ENC_CODERANGE_7BIT:
            left = ptr;
            *ptr = byte;
            if (ISASCII(byte)) goto end;
            nlen = rb_enc_precise_mbclen(left, head+len, enc);
            if (!MBCLEN_CHARFOUND_P(nlen))
                ENC_CODERANGE_SET(str, ENC_CODERANGE_BROKEN);
            else
                ENC_CODERANGE_SET(str, ENC_CODERANGE_VALID);
            goto end;
          case ENC_CODERANGE_VALID:
            left = rb_enc_left_char_head(head, ptr, head+len, enc);
            width = rb_enc_precise_mbclen(left, head+len, enc);
            *ptr = byte;
            nlen = rb_enc_precise_mbclen(left, head+len, enc);
            if (!MBCLEN_CHARFOUND_P(nlen))
                ENC_CODERANGE_SET(str, ENC_CODERANGE_BROKEN);
            else if (MBCLEN_CHARFOUND_LEN(nlen) != width || ISASCII(byte))
                ENC_CODERANGE_CLEAR(str);
            goto end;
        }
    }
    ENC_CODERANGE_CLEAR(str);
    *ptr = byte;

  end:
    return value;
}

Sets the byte at zero-based offset index to the value of the given integer; returns integer:

s = 'xyzzy'
s.setbyte(2, 129) # => 129
s                 # => "xy\x81zy"

Related: see Modifying.

# File lib/shellwords.rb, line 238
def shellescape
  Shellwords.escape(self)
end

Escapes str so that it can be safely used in a Bourne shell command line.

See Shellwords.shellescape for details.

# File lib/shellwords.rb, line 227
def shellsplit
  Shellwords.split(self)
end

Splits str into an array of tokens in the same way the UNIX Bourne shell does.

See Shellwords.shellsplit for details.

static VALUE
rb_str_slice_bang(int argc, VALUE *argv, VALUE str)
{
    VALUE result = Qnil;
    VALUE indx;
    long beg, len = 1;
    char *p;

    rb_check_arity(argc, 1, 2);
    str_modify_keep_cr(str);
    indx = argv[0];
    if (RB_TYPE_P(indx, T_REGEXP)) {
        if (rb_reg_search(indx, str, 0, 0) < 0) return Qnil;
        VALUE match = rb_backref_get();
        struct re_registers *regs = RMATCH_REGS(match);
        int nth = 0;
        if (argc > 1 && (nth = rb_reg_backref_number(match, argv[1])) < 0) {
            if ((nth += regs->num_regs) <= 0) return Qnil;
        }
        else if (nth >= regs->num_regs) return Qnil;
        beg = BEG(nth);
        len = END(nth) - beg;
        goto subseq;
    }
    else if (argc == 2) {
        beg = NUM2LONG(indx);
        len = NUM2LONG(argv[1]);
        goto num_index;
    }
    else if (FIXNUM_P(indx)) {
        beg = FIX2LONG(indx);
        if (!(p = rb_str_subpos(str, beg, &len))) return Qnil;
        if (!len) return Qnil;
        beg = p - RSTRING_PTR(str);
        goto subseq;
    }
    else if (RB_TYPE_P(indx, T_STRING)) {
        beg = rb_str_index(str, indx, 0);
        if (beg == -1) return Qnil;
        len = RSTRING_LEN(indx);
        result = str_duplicate(rb_cString, indx);
        goto squash;
    }
    else {
        switch (rb_range_beg_len(indx, &beg, &len, str_strlen(str, NULL), 0)) {
          case Qnil:
            return Qnil;
          case Qfalse:
            beg = NUM2LONG(indx);
            if (!(p = rb_str_subpos(str, beg, &len))) return Qnil;
            if (!len) return Qnil;
            beg = p - RSTRING_PTR(str);
            goto subseq;
          default:
            goto num_index;
        }
    }

  num_index:
    if (!(p = rb_str_subpos(str, beg, &len))) return Qnil;
    beg = p - RSTRING_PTR(str);

  subseq:
    result = rb_str_new(RSTRING_PTR(str)+beg, len);
    rb_enc_cr_str_copy_for_substr(result, str);

  squash:
    if (len > 0) {
        if (beg == 0) {
            rb_str_drop_bytes(str, len);
        }
        else {
            char *sptr = RSTRING_PTR(str);
            long slen = RSTRING_LEN(str);
            if (beg + len > slen) /* pathological check */
                len = slen - beg;
            memmove(sptr + beg,
                    sptr + beg + len,
                    slen - (beg + len));
            slen -= len;
            STR_SET_LEN(str, slen);
            TERM_FILL(&sptr[slen], TERM_LEN(str));
        }
    }
    return result;
}

Like String#[] (and its alias String#slice), except that:

• Performs substitutions in self (not in a copy of self).

• Returns the removed substring if any modifications were made, nil otherwise.

A few examples:

s = 'hello'
s.slice!('e') # => "e"
s             # => "hllo"
s.slice!('e') # => nil
s             # => "hllo"

Related: see Modifying.

static VALUE
rb_str_split_m(int argc, VALUE *argv, VALUE str)
{
    rb_encoding *enc;
    VALUE spat;
    VALUE limit;
    split_type_t split_type;
    long beg, end, i = 0, empty_count = -1;
    int lim = 0;
    VALUE result, tmp;

    result = rb_block_given_p() ? Qfalse : Qnil;
    if (rb_scan_args(argc, argv, "02", &spat, &limit) == 2) {
        lim = NUM2INT(limit);
        if (lim <= 0) limit = Qnil;
        else if (lim == 1) {
            if (RSTRING_LEN(str) == 0)
                return result ? rb_ary_new2(0) : str;
            tmp = str_duplicate(rb_cString, str);
            if (!result) {
                rb_yield(tmp);
                return str;
            }
            return rb_ary_new3(1, tmp);
        }
        i = 1;
    }
    if (NIL_P(limit) && !lim) empty_count = 0;

    enc = STR_ENC_GET(str);
    split_type = SPLIT_TYPE_REGEXP;
    if (!NIL_P(spat)) {
        spat = get_pat_quoted(spat, 0);
    }
    else if (NIL_P(spat = rb_fs)) {
        split_type = SPLIT_TYPE_AWK;
    }
    else if (!(spat = rb_fs_check(spat))) {
        rb_raise(rb_eTypeError, "value of $; must be String or Regexp");
    }
    else {
        rb_category_warn(RB_WARN_CATEGORY_DEPRECATED, "$; is set to non-nil value");
    }
    if (split_type != SPLIT_TYPE_AWK) {
        switch (BUILTIN_TYPE(spat)) {
          case T_REGEXP:
            rb_reg_options(spat); /* check if uninitialized */
            tmp = RREGEXP_SRC(spat);
            split_type = literal_split_pattern(tmp, SPLIT_TYPE_REGEXP);
            if (split_type == SPLIT_TYPE_AWK) {
                spat = tmp;
                split_type = SPLIT_TYPE_STRING;
            }
            break;

          case T_STRING:
            mustnot_broken(spat);
            split_type = literal_split_pattern(spat, SPLIT_TYPE_STRING);
            break;

          default:
            UNREACHABLE_RETURN(Qnil);
        }
    }

#define SPLIT_STR(beg, len) ( \
        empty_count = split_string(result, str, beg, len, empty_count), \
        str_mod_check(str, str_start, str_len))

    beg = 0;
    char *ptr = RSTRING_PTR(str);
    char *const str_start = ptr;
    const long str_len = RSTRING_LEN(str);
    char *const eptr = str_start + str_len;
    if (split_type == SPLIT_TYPE_AWK) {
        char *bptr = ptr;
        int skip = 1;
        unsigned int c;

        if (result) result = rb_ary_new();
        end = beg;
        if (is_ascii_string(str)) {
            while (ptr < eptr) {
                c = (unsigned char)*ptr++;
                if (skip) {
                    if (ascii_isspace(c)) {
                        beg = ptr - bptr;
                    }
                    else {
                        end = ptr - bptr;
                        skip = 0;
                        if (!NIL_P(limit) && lim <= i) break;
                    }
                }
                else if (ascii_isspace(c)) {
                    SPLIT_STR(beg, end-beg);
                    skip = 1;
                    beg = ptr - bptr;
                    if (!NIL_P(limit)) ++i;
                }
                else {
                    end = ptr - bptr;
                }
            }
        }
        else {
            while (ptr < eptr) {
                int n;

                c = rb_enc_codepoint_len(ptr, eptr, &n, enc);
                ptr += n;
                if (skip) {
                    if (rb_isspace(c)) {
                        beg = ptr - bptr;
                    }
                    else {
                        end = ptr - bptr;
                        skip = 0;
                        if (!NIL_P(limit) && lim <= i) break;
                    }
                }
                else if (rb_isspace(c)) {
                    SPLIT_STR(beg, end-beg);
                    skip = 1;
                    beg = ptr - bptr;
                    if (!NIL_P(limit)) ++i;
                }
                else {
                    end = ptr - bptr;
                }
            }
        }
    }
    else if (split_type == SPLIT_TYPE_STRING) {
        char *substr_start = ptr;
        char *sptr = RSTRING_PTR(spat);
        long slen = RSTRING_LEN(spat);

        if (result) result = rb_ary_new();
        mustnot_broken(str);
        enc = rb_enc_check(str, spat);
        while (ptr < eptr &&
               (end = rb_memsearch(sptr, slen, ptr, eptr - ptr, enc)) >= 0) {
            /* Check we are at the start of a char */
            char *t = rb_enc_right_char_head(ptr, ptr + end, eptr, enc);
            if (t != ptr + end) {
                ptr = t;
                continue;
            }
            SPLIT_STR(substr_start - str_start, (ptr+end) - substr_start);
            str_mod_check(spat, sptr, slen);
            ptr += end + slen;
            substr_start = ptr;
            if (!NIL_P(limit) && lim <= ++i) break;
        }
        beg = ptr - str_start;
    }
    else if (split_type == SPLIT_TYPE_CHARS) {
        int n;

        if (result) result = rb_ary_new_capa(RSTRING_LEN(str));
        mustnot_broken(str);
        enc = rb_enc_get(str);
        while (ptr < eptr &&
               (n = rb_enc_precise_mbclen(ptr, eptr, enc)) > 0) {
            SPLIT_STR(ptr - str_start, n);
            ptr += n;
            if (!NIL_P(limit) && lim <= ++i) break;
        }
        beg = ptr - str_start;
    }
    else {
        if (result) result = rb_ary_new();
        long len = RSTRING_LEN(str);
        long start = beg;
        long idx;
        int last_null = 0;
        struct re_registers *regs;
        VALUE match = 0;

        for (; rb_reg_search(spat, str, start, 0) >= 0;
             (match ? (rb_match_unbusy(match), rb_backref_set(match)) : (void)0)) {
            match = rb_backref_get();
            if (!result) rb_match_busy(match);
            regs = RMATCH_REGS(match);
            end = BEG(0);
            if (start == end && BEG(0) == END(0)) {
                if (!ptr) {
                    SPLIT_STR(0, 0);
                    break;
                }
                else if (last_null == 1) {
                    SPLIT_STR(beg, rb_enc_fast_mbclen(ptr+beg, eptr, enc));
                    beg = start;
                }
                else {
                    if (start == len)
                        start++;
                    else
                        start += rb_enc_fast_mbclen(ptr+start,eptr,enc);
                    last_null = 1;
                    continue;
                }
            }
            else {
                SPLIT_STR(beg, end-beg);
                beg = start = END(0);
            }
            last_null = 0;

            for (idx=1; idx < regs->num_regs; idx++) {
                if (BEG(idx) == -1) continue;
                SPLIT_STR(BEG(idx), END(idx)-BEG(idx));
            }
            if (!NIL_P(limit) && lim <= ++i) break;
        }
        if (match) rb_match_unbusy(match);
    }
    if (RSTRING_LEN(str) > 0 && (!NIL_P(limit) || RSTRING_LEN(str) > beg || lim < 0)) {
        SPLIT_STR(beg, RSTRING_LEN(str)-beg);
    }

    return result ? result : str;
}

Creates an array of substrings by splitting self at each occurrence of the given field separator field_sep.

With no arguments given, splits using the field separator $;, whose default value is nil.

With no block given, returns the array of substrings:

'abracadabra'.split('a') # => ["", "br", "c", "d", "br"]

When field_sep is nil or ' ' (a single space), splits at each sequence of whitespace:

'foo bar baz'.split(nil)          # => ["foo", "bar", "baz"]
'foo bar baz'.split(' ')          # => ["foo", "bar", "baz"]
"foo \n\tbar\t\n  baz".split(' ') # => ["foo", "bar", "baz"]
'foo  bar   baz'.split(' ')       # => ["foo", "bar", "baz"]
''.split(' ')                     # => []

When field_sep is an empty string, splits at every character:

'abracadabra'.split('') # => ["a", "b", "r", "a", "c", "a", "d", "a", "b", "r", "a"]
''.split('')            # => []
'тест'.split('')        # => ["т", "е", "с", "т"]
'こんにちは'.split('')   # => ["こ", "ん", "に", "ち", "は"]

When field_sep is a non-empty string and different from ' ' (a single space), uses that string as the separator:

'abracadabra'.split('a')  # => ["", "br", "c", "d", "br"]
'abracadabra'.split('ab') # => ["", "racad", "ra"]
''.split('a')             # => []
'тест'.split('т')         # => ["", "ес"]
'こんにちは'.split('に')    # => ["こん", "ちは"]

When field_sep is a Regexp, splits at each occurrence of a matching substring:

'abracadabra'.split(/ab/) # => ["", "racad", "ra"]
'1 + 1 == 2'.split(/\W+/) # => ["1", "1", "2"]
'abracadabra'.split(//)   # => ["a", "b", "r", "a", "c", "a", "d", "a", "b", "r", "a"]

If the Regexp contains groups, their matches are included in the returned array:

'1:2:3'.split(/(:)()()/, 2) # => ["1", ":", "", "", "2:3"]

Argument limit sets a limit on the size of the returned array; it also determines whether trailing empty strings are included in the returned array.

When limit is zero, there is no limit on the size of the array, but trailing empty strings are omitted:

'abracadabra'.split('', 0)  # => ["a", "b", "r", "a", "c", "a", "d", "a", "b", "r", "a"]
'abracadabra'.split('a', 0) # => ["", "br", "c", "d", "br"]  # Empty string after last 'a' omitted.

When limit is a positive integer, there is a limit on the size of the array (no more than n - 1 splits occur), and trailing empty strings are included:

'abracadabra'.split('', 3)   # => ["a", "b", "racadabra"]
'abracadabra'.split('a', 3)  # => ["", "br", "cadabra"]
'abracadabra'.split('', 30)  # => ["a", "b", "r", "a", "c", "a", "d", "a", "b", "r", "a", ""]
'abracadabra'.split('a', 30) # => ["", "br", "c", "d", "br", ""]
'abracadabra'.split('', 1)   # => ["abracadabra"]
'abracadabra'.split('a', 1)  # => ["abracadabra"]

When limit is negative, there is no limit on the size of the array, and trailing empty strings are omitted:

'abracadabra'.split('', -1)  # => ["a", "b", "r", "a", "c", "a", "d", "a", "b", "r", "a", ""]
'abracadabra'.split('a', -1) # => ["", "br", "c", "d", "br", ""]

If a block is given, it is called with each substring and returns self:

'foo bar baz'.split(' ') {|substring| p substring }

Output :

"foo"
"bar"
"baz"

Note that the above example is functionally equivalent to:

'foo bar baz'.split(' ').each {|substring| p substring }

Output :

"foo"
"bar"
"baz"

But the latter:

• Has poorer performance because it creates an intermediate array.

• Returns an array (instead of self).

Related: see Converting to Non-String.

static VALUE
rb_str_squeeze(int argc, VALUE *argv, VALUE str)
{
    str = str_duplicate(rb_cString, str);
    rb_str_squeeze_bang(argc, argv, str);
    return str;
}

Returns a copy of self with each tuple (doubling, tripling, etc.) of specified characters “squeezed” down to a single character.

The tuples to be squeezed are specified by arguments selectors, each of which is a string; see Character Selectors.

A single argument may be a single character:

'Noooooo!'.squeeze('o')      # => "No!"
'foo  bar  baz'.squeeze(' ') # => "foo bar baz"
'Mississippi'.squeeze('s')   # => "Misisippi"
'Mississippi'.squeeze('p')   # => "Mississipi"
'Mississippi'.squeeze('x')   # => "Mississippi"  # Unused selector character is ignored.
'бессонница'.squeeze('с')    # => "бесонница"
'бессонница'.squeeze('н')    # => "бессоница"

A single argument may be a string of characters:

'Mississippi'.squeeze('sp')       # => "Misisipi"
'Mississippi'.squeeze('ps')       # => "Misisipi"   # Order doesn't matter.
'Mississippi'.squeeze('nonsense') # => "Misisippi"  # Unused selector characters are ignored.

A single argument may be a range of characters:

'Mississippi'.squeeze('a-p') # => "Mississipi"
'Mississippi'.squeeze('q-z') # => "Misisippi"
'Mississippi'.squeeze('a-z') # => "Misisipi"

Multiple arguments are allowed; see Multiple Character Selectors.

Related: see Converting to New String.

static VALUE
rb_str_squeeze_bang(int argc, VALUE *argv, VALUE str)
{
    char squeez[TR_TABLE_SIZE];
    rb_encoding *enc = 0;
    VALUE del = 0, nodel = 0;
    unsigned char *s, *send, *t;
    int i, modify = 0;
    int ascompat, singlebyte = single_byte_optimizable(str);
    unsigned int save;

    if (argc == 0) {
        enc = STR_ENC_GET(str);
    }
    else {
        for (i=0; i<argc; i++) {
            VALUE s = argv[i];

            StringValue(s);
            enc = rb_enc_check(str, s);
            if (singlebyte && !single_byte_optimizable(s))
                singlebyte = 0;
            tr_setup_table(s, squeez, i==0, &del, &nodel, enc);
        }
    }

    str_modify_keep_cr(str);
    s = t = (unsigned char *)RSTRING_PTR(str);
    if (!s || RSTRING_LEN(str) == 0) return Qnil;
    send = (unsigned char *)RSTRING_END(str);
    save = -1;
    ascompat = rb_enc_asciicompat(enc);

    if (singlebyte) {
        while (s < send) {
            unsigned int c = *s++;
            if (c != save || (argc > 0 && !squeez[c])) {
                *t++ = save = c;
            }
        }
    }
    else {
        while (s < send) {
            unsigned int c;
            int clen;

            if (ascompat && (c = *s) < 0x80) {
                if (c != save || (argc > 0 && !squeez[c])) {
                    *t++ = save = c;
                }
                s++;
            }
            else {
                c = rb_enc_codepoint_len((char *)s, (char *)send, &clen, enc);

                if (c != save || (argc > 0 && !tr_find(c, squeez, del, nodel))) {
                    if (t != s) rb_enc_mbcput(c, t, enc);
                    save = c;
                    t += clen;
                }
                s += clen;
            }
        }
    }

    TERM_FILL((char *)t, TERM_LEN(str));
    if ((char *)t - RSTRING_PTR(str) != RSTRING_LEN(str)) {
        STR_SET_LEN(str, (char *)t - RSTRING_PTR(str));
        modify = 1;
    }

    if (modify) return str;
    return Qnil;
}

Like String#squeeze, except that:

• Characters are squeezed in self (not in a copy of self).

• Returns self if any changes are made, nil otherwise.

Related: See Modifying.

static VALUE
rb_str_start_with(int argc, VALUE *argv, VALUE str)
{
    int i;

    for (i=0; i<argc; i++) {
        VALUE tmp = argv[i];
        if (RB_TYPE_P(tmp, T_REGEXP)) {
            if (rb_reg_start_with_p(tmp, str))
                return Qtrue;
        }
        else {
            const char *p, *s, *e;
            long slen, tlen;
            rb_encoding *enc;

            StringValue(tmp);
            enc = rb_enc_check(str, tmp);
            if ((tlen = RSTRING_LEN(tmp)) == 0) return Qtrue;
            if ((slen = RSTRING_LEN(str)) < tlen) continue;
            p = RSTRING_PTR(str);
            e = p + slen;
            s = p + tlen;
            if (!at_char_right_boundary(p, s, e, enc))
                continue;
            if (memcmp(p, RSTRING_PTR(tmp), tlen) == 0)
                return Qtrue;
        }
    }
    return Qfalse;
}

Returns whether self starts with any of the given patterns.

For each argument, the pattern used is:

• The pattern itself, if it is a Regexp.

• Regexp.quote(pattern), if it is a string.

Returns true if any pattern matches the beginning, false otherwise:

'hello'.start_with?('hell')               # => true
'hello'.start_with?(/H/i)                 # => true
'hello'.start_with?('heaven', 'hell')     # => true
'hello'.start_with?('heaven', 'paradise') # => false
'тест'.start_with?('т')                   # => true
'こんにちは'.start_with?('こ')              # => true

Related: see Querying.

static VALUE
rb_str_strip(int argc, VALUE *argv, VALUE str)
{
    char *start;
    long olen, loffset, roffset;
    rb_encoding *enc = STR_ENC_GET(str);

    RSTRING_GETMEM(str, start, olen);

    if (argc > 0) {
        char table[TR_TABLE_SIZE];
        VALUE del = 0, nodel = 0;

        tr_setup_table_multi(table, &del, &nodel, str, argc, argv);
        loffset = lstrip_offset_table(str, start, start+olen, enc, table, del, nodel);
        roffset = rstrip_offset_table(str, start+loffset, start+olen, enc, table, del, nodel);
    }
    else {
        loffset = lstrip_offset(str, start, start+olen, enc);
        roffset = rstrip_offset(str, start+loffset, start+olen, enc);
    }

    if (loffset <= 0 && roffset <= 0) return str_duplicate(rb_cString, str);
    return rb_str_subseq(str, loffset, olen-loffset-roffset);
}

Returns a copy of self with leading and trailing whitespace removed; see Whitespace in Strings:

whitespace = "\x00\t\n\v\f\r "
s = whitespace + 'abc' + whitespace
# => "\u0000\t\n\v\f\r abc\u0000\t\n\v\f\r "
s.strip # => "abc"

If selectors are given, removes characters of selectors from both ends of self:

s = "---abc+++"
s.strip("-+") # => "abc"
s.strip("+-") # => "abc"

selectors must be valid character selectors (see Character Selectors), and may use any of its valid forms, including negation, ranges, and escapes:

"01234abc56789".strip("0-9") # "abc"
"01234abc56789".strip("0-9", "^4-6") # "4abc56"

Related: see Converting to New String.

static VALUE
rb_str_strip_bang(int argc, VALUE *argv, VALUE str)
{
    char *start;
    long olen, loffset, roffset;
    rb_encoding *enc;

    str_modify_keep_cr(str);
    enc = STR_ENC_GET(str);
    RSTRING_GETMEM(str, start, olen);

    if (argc > 0) {
        char table[TR_TABLE_SIZE];
        VALUE del = 0, nodel = 0;

        tr_setup_table_multi(table, &del, &nodel, str, argc, argv);
        loffset = lstrip_offset_table(str, start, start+olen, enc, table, del, nodel);
        roffset = rstrip_offset_table(str, start+loffset, start+olen, enc, table, del, nodel);
    }
    else {
        loffset = lstrip_offset(str, start, start+olen, enc);
        roffset = rstrip_offset(str, start+loffset, start+olen, enc);
    }

    if (loffset > 0 || roffset > 0) {
        long len = olen-roffset;
        if (loffset > 0) {
            len -= loffset;
            memmove(start, start + loffset, len);
        }
        STR_SET_LEN(str, len);
        TERM_FILL(start+len, rb_enc_mbminlen(enc));
        return str;
    }
    return Qnil;
}

Like String#strip, except that:

• Any modifications are made to self.

• Returns self if any modification are made, nil otherwise.

Related: see Modifying.

static VALUE
rb_str_sub(int argc, VALUE *argv, VALUE str)
{
    str = str_duplicate(rb_cString, str);
    rb_str_sub_bang(argc, argv, str);
    return str;
}

Returns a copy of self, possibly with a substring replaced.

Argument pattern may be a string or a Regexp; argument replacement may be a string or a Hash.

Varying types for the argument values makes this method very versatile.

Below are some simple examples; for many more examples, see Substitution Methods.

With arguments pattern and string replacement given, replaces the first matching substring with the given replacement string:

s = 'abracadabra'       # => "abracadabra"
s.sub('bra', 'xyzzy')   # => "axyzzycadabra"
s.sub(/bra/, 'xyzzy')   # => "axyzzycadabra"
s.sub('nope', 'xyzzy')  # => "abracadabra"

With arguments pattern and hash replacement given, replaces the first matching substring with a value from the given replacement hash, or removes it:

h = {'a' => 'A', 'b' => 'B', 'c' => 'C'}
s.sub('b', h)  # => "aBracadabra"
s.sub(/b/, h)  # => "aBracadabra"
s.sub(/d/, h)  # => "abracaabra"  # 'd' removed.

With argument pattern and a block given, calls the block with each matching substring; replaces that substring with the block’s return value:

s.sub('b') {|match| match.upcase } # => "aBracadabra"

Related: see Converting to New String.

static VALUE
rb_str_sub_bang(int argc, VALUE *argv, VALUE str)
{
    VALUE pat, repl, hash = Qnil;
    int iter = 0;
    long plen;
    int min_arity = rb_block_given_p() ? 1 : 2;
    long beg;

    rb_check_arity(argc, min_arity, 2);
    if (argc == 1) {
        iter = 1;
    }
    else {
        repl = argv[1];
        hash = rb_check_hash_type(argv[1]);
        if (NIL_P(hash)) {
            StringValue(repl);
        }
    }

    pat = get_pat_quoted(argv[0], 1);

    str_modifiable(str);
    beg = rb_pat_search(pat, str, 0, 1);
    if (beg >= 0) {
        rb_encoding *enc;
        int cr = ENC_CODERANGE(str);
        long beg0, end0;
        VALUE match, match0 = Qnil;
        struct re_registers *regs;
        char *p, *rp;
        long len, rlen;

        match = rb_backref_get();
        regs = RMATCH_REGS(match);
        if (RB_TYPE_P(pat, T_STRING)) {
            beg0 = beg;
            end0 = beg0 + RSTRING_LEN(pat);
            match0 = pat;
        }
        else {
            beg0 = BEG(0);
            end0 = END(0);
            if (iter) match0 = rb_reg_nth_match(0, match);
        }

        if (iter || !NIL_P(hash)) {
            p = RSTRING_PTR(str); len = RSTRING_LEN(str);

            if (iter) {
                repl = rb_obj_as_string(rb_yield(match0));
            }
            else {
                repl = rb_hash_aref(hash, rb_str_subseq(str, beg0, end0 - beg0));
                repl = rb_obj_as_string(repl);
            }
            str_mod_check(str, p, len);
            rb_check_frozen(str);
        }
        else {
            repl = rb_reg_regsub(repl, str, regs, RB_TYPE_P(pat, T_STRING) ? Qnil : pat);
        }

        enc = rb_enc_compatible(str, repl);
        if (!enc) {
            rb_encoding *str_enc = STR_ENC_GET(str);
            p = RSTRING_PTR(str); len = RSTRING_LEN(str);
            if (coderange_scan(p, beg0, str_enc) != ENC_CODERANGE_7BIT ||
                coderange_scan(p+end0, len-end0, str_enc) != ENC_CODERANGE_7BIT) {
                rb_raise(rb_eEncCompatError, "incompatible character encodings: %s and %s",
                         rb_enc_inspect_name(str_enc),
                         rb_enc_inspect_name(STR_ENC_GET(repl)));
            }
            enc = STR_ENC_GET(repl);
        }
        rb_str_modify(str);
        rb_enc_associate(str, enc);
        if (ENC_CODERANGE_UNKNOWN < cr && cr < ENC_CODERANGE_BROKEN) {
            int cr2 = ENC_CODERANGE(repl);
            if (cr2 == ENC_CODERANGE_BROKEN ||
                (cr == ENC_CODERANGE_VALID && cr2 == ENC_CODERANGE_7BIT))
                cr = ENC_CODERANGE_UNKNOWN;
            else
                cr = cr2;
        }
        plen = end0 - beg0;
        rlen = RSTRING_LEN(repl);
        len = RSTRING_LEN(str);
        if (rlen > plen) {
            RESIZE_CAPA(str, len + rlen - plen);
        }
        p = RSTRING_PTR(str);
        if (rlen != plen) {
            memmove(p + beg0 + rlen, p + beg0 + plen, len - beg0 - plen);
        }
        rp = RSTRING_PTR(repl);
        memmove(p + beg0, rp, rlen);
        len += rlen - plen;
        STR_SET_LEN(str, len);
        TERM_FILL(&RSTRING_PTR(str)[len], TERM_LEN(str));
        ENC_CODERANGE_SET(str, cr);

        RB_GC_GUARD(match);

        return str;
    }
    return Qnil;
}

Like String#sub, except that:

• Changes are made to self, not to copy of self.

• Returns self if any changes are made, nil otherwise.

Related: see Modifying.

VALUE
rb_str_succ(VALUE orig)
{
    VALUE str;
    str = rb_str_new(RSTRING_PTR(orig), RSTRING_LEN(orig));
    rb_enc_cr_str_copy_for_substr(str, orig);
    return str_succ(str);
}

Returns the successor to self. The successor is calculated by incrementing characters.

The first character to be incremented is the rightmost alphanumeric: or, if no alphanumerics, the rightmost character:

'THX1138'.succ   # => "THX1139"
'<<koala>>'.succ # => "<<koalb>>"
'***'.succ       # => '**+'
'тест'.succ      # => "тесу"
'こんにちは'.succ  # => "こんにちば"

The successor to a digit is another digit, “carrying” to the next-left character for a “rollover” from 9 to 0, and prepending another digit if necessary:

'00'.succ # => "01"
'09'.succ # => "10"
'99'.succ # => "100"

The successor to a letter is another letter of the same case, carrying to the next-left character for a rollover, and prepending another same-case letter if necessary:

'aa'.succ # => "ab"
'az'.succ # => "ba"
'zz'.succ # => "aaa"
'AA'.succ # => "AB"
'AZ'.succ # => "BA"
'ZZ'.succ # => "AAA"

The successor to a non-alphanumeric character is the next character in the underlying character set’s collating sequence, carrying to the next-left character for a rollover, and prepending another character if necessary:

s = 0.chr * 3   # => "\x00\x00\x00"
s.succ        # => "\x00\x00\x01"
s = 255.chr * 3 # => "\xFF\xFF\xFF"
s.succ        # => "\x01\x00\x00\x00"

Carrying can occur between and among mixtures of alphanumeric characters:

s = 'zz99zz99' # => "zz99zz99"
s.succ         # => "aaa00aa00"
s = '99zz99zz' # => "99zz99zz"
s.succ         # => "100aa00aa"

The successor to an empty String is a new empty String:

''.succ # => ""

Related: see Converting to New String.

static VALUE
rb_str_succ_bang(VALUE str)
{
    rb_str_modify(str);
    str_succ(str);
    return str;
}

Like String#succ, but modifies self in place; returns self.

Related: see Modifying.

static VALUE
rb_str_sum(int argc, VALUE *argv, VALUE str)
{
    int bits = 16;
    char *ptr, *p, *pend;
    long len;
    VALUE sum = INT2FIX(0);
    unsigned long sum0 = 0;

    if (rb_check_arity(argc, 0, 1) && (bits = NUM2INT(argv[0])) < 0) {
        bits = 0;
    }
    ptr = p = RSTRING_PTR(str);
    len = RSTRING_LEN(str);
    pend = p + len;

    while (p < pend) {
        if (FIXNUM_MAX - UCHAR_MAX < sum0) {
            sum = rb_funcall(sum, '+', 1, LONG2FIX(sum0));
            str_mod_check(str, ptr, len);
            sum0 = 0;
        }
        sum0 += (unsigned char)*p;
        p++;
    }

    if (bits == 0) {
        if (sum0) {
            sum = rb_funcall(sum, '+', 1, LONG2FIX(sum0));
        }
    }
    else {
        if (sum == INT2FIX(0)) {
            if (bits < (int)sizeof(long)*CHAR_BIT) {
                sum0 &= (((unsigned long)1)<<bits)-1;
            }
            sum = LONG2FIX(sum0);
        }
        else {
            VALUE mod;

            if (sum0) {
                sum = rb_funcall(sum, '+', 1, LONG2FIX(sum0));
            }

            mod = rb_funcall(INT2FIX(1), idLTLT, 1, INT2FIX(bits));
            mod = rb_funcall(mod, '-', 1, INT2FIX(1));
            sum = rb_funcall(sum, '&', 1, mod);
        }
    }
    return sum;
}

Returns a basic n-bit checksum of the characters in self; the checksum is the sum of the binary value of each byte in self, modulo 2**n - 1:

'hello'.sum     # => 532
'hello'.sum(4)  # => 4
'hello'.sum(64) # => 532
'тест'.sum      # => 1405
'こんにちは'.sum  # => 2582

This is not a particularly strong checksum.

Related: see Querying.

static VALUE
rb_str_swapcase(int argc, VALUE *argv, VALUE str)
{
    rb_encoding *enc;
    OnigCaseFoldType flags = ONIGENC_CASE_UPCASE | ONIGENC_CASE_DOWNCASE;
    VALUE ret;

    flags = check_case_options(argc, argv, flags);
    enc = str_true_enc(str);
    if (RSTRING_LEN(str) == 0 || !RSTRING_PTR(str)) return str_duplicate(rb_cString, str);
    if (flags&ONIGENC_CASE_ASCII_ONLY) {
        ret = rb_str_new(0, RSTRING_LEN(str));
        rb_str_ascii_casemap(str, ret, &flags, enc);
    }
    else {
        ret = rb_str_casemap(str, &flags, enc);
    }
    return ret;
}

Returns a string containing the characters in self, with cases reversed:

• Each uppercase character is downcased.

• Each lowercase character is upcased.

Examples:

'Hello'.swapcase        # => "hELLO"
'Straße'.swapcase       # => "sTRASSE"
'Привет'.swapcase       # => "пРИВЕТ"
'RubyGems.org'.swapcase # => "rUBYgEMS.ORG"

The sizes of self and the upcased result may differ:

s = 'Straße'
s.size          # => 6
s.swapcase      # => "sTRASSE"
s.swapcase.size # => 7

Some characters (and some character sets) do not have upcase and downcase versions; see Case Mapping:

s = '1, 2, 3, ...'
s.swapcase == s # => true
s = 'こんにちは'
s.swapcase == s # => true

The casing is affected by the given mapping, which may be :ascii, :fold, or :turkic; see Case Mappings.

Related: see Converting to New String.

static VALUE
rb_str_swapcase_bang(int argc, VALUE *argv, VALUE str)
{
    rb_encoding *enc;
    OnigCaseFoldType flags = ONIGENC_CASE_UPCASE | ONIGENC_CASE_DOWNCASE;

    flags = check_case_options(argc, argv, flags);
    str_modify_keep_cr(str);
    enc = str_true_enc(str);
    if (flags&ONIGENC_CASE_ASCII_ONLY)
        rb_str_ascii_casemap(str, str, &flags, enc);
    else
        str_shared_replace(str, rb_str_casemap(str, &flags, enc));

    if (ONIGENC_CASE_MODIFIED&flags) return str;
    return Qnil;
}

Like String#swapcase, except that:

• Changes are made to self, not to copy of self.

• Returns self if any changes are made, nil otherwise.

Related: see Modifying.

static VALUE
string_to_c(VALUE self)
{
    VALUE num;

    rb_must_asciicompat(self);

    (void)parse_comp(rb_str_fill_terminator(self, 1), FALSE, &num);

    return num;
}

Returns a Complex object: parses the leading substring of self to extract two numeric values that become the coordinates of the complex object.

The substring is interpreted as containing either rectangular coordinates (real and imaginary parts) or polar coordinates (magnitude and angle parts), depending on an included or implied “separator” character:

• '+', '-', or no separator: rectangular coordinates.

• '@': polar coordinates.

In Brief

In these examples, we use method Complex#rect to display rectangular coordinates, and method Complex#polar to display polar coordinates.

# Rectangular coordinates.

# Real-only: no separator; imaginary part is zero.
'9'.to_c.rect         # => [9, 0]         # Integer.
'-9'.to_c.rect        # => [-9, 0]        # Integer (negative).
'2.5'.to_c.rect       # => [2.5, 0]       # Float.
'1.23e-14'.to_c.rect  # => [1.23e-14, 0]  # Float with exponent.
'2.5/1'.to_c.rect     # => [(5/2), 0]     # Rational.

# Some things are ignored.
'foo1'.to_c.rect      # => [0, 0]         # Unparsed entire substring.
'1foo'.to_c.rect      # => [1, 0]         # Unparsed trailing substring.
' 1 '.to_c.rect       # => [1, 0]         # Leading and trailing whitespace.
*
# Imaginary only: trailing 'i' required; real part is zero.
'9i'.to_c.rect        # => [0, 9]
'-9i'.to_c.rect       # => [0, -9]
'2.5i'.to_c.rect      # => [0, 2.5]
'1.23e-14i'.to_c.rect # => [0, 1.23e-14]
'2.5/1i'.to_c.rect    # => [0, (5/2)]

# Real and imaginary; '+' or '-' separator; trailing 'i' required.
'2+3i'.to_c.rect      # => [2, 3]
'-2-3i'.to_c.rect     # => [-2, -3]
'2.5+3i'.to_c.rect    # => [2.5, 3]
'2.5+3/2i'.to_c.rect  # => [2.5, (3/2)]

# Polar coordinates; '@' separator; magnitude required.
'1.0@0'.to_c.polar             # => [1.0, 0.0]
'1.0@'.to_c.polar              # => [1.0, 0.0]
"1.0@#{Math::PI}".to_c.polar   # => [1.0, 3.141592653589793]
"1.0@#{Math::PI/2}".to_c.polar # => [1.0, 1.5707963267948966]

Parsed Values

The parsing may be thought of as searching for numeric literals embedded in the substring.

This section shows how the method parses numeric values from leading substrings. The examples show real-only or imaginary-only parsing; the parsing is the same for each part.

'1foo'.to_c # => (1+0i)      # Ignores trailing unparsed characters.
' 1 '.to_c  # => (1+0i)      # Ignores leading and trailing whitespace.
'x1'.to_c   # => (0+0i)      # Finds no leading numeric.

# Integer literal embedded in the substring.
'1'.to_c       # => (1+0i)
'-1'.to_c      # => (-1+0i)
'1i'.to_c      # => (0+1i)

# Integer literals that don't work.
'0b100'.to_c   # => (0+0i)   # Not parsed as binary.
'0o100'.to_c   # => (0+0i)   # Not parsed as octal.
'0d100'.to_c   # => (0+0i)   # Not parsed as decimal.
'0x100'.to_c   # => (0+0i)   # Not parsed as hexadecimal.
'010'.to_c     # => (10+0i)  # Not parsed as octal.

# Float literals:
'3.14'.to_c    # => (3.14+0i)
'3.14i'.to_c   # => (0+3.14i)
'1.23e4'.to_c  # => (12300.0+0i)
'1.23e+4'.to_c # => (12300.0+0i)
'1.23e-4'.to_c # => (0.000123+0i)

# Rational literals:
'1/2'.to_c     # => ((1/2)+0i)
'-1/2'.to_c    # => ((-1/2)+0i)
'1/2r'.to_c    # => ((1/2)+0i)
'-1/2r'.to_c   # => ((-1/2)+0i)

Rectangular Coordinates

With separator '+' or '-', or with no separator, interprets the values as rectangular coordinates: real and imaginary.

With no separator, assigns a single value to either the real or the imaginary part:

 ''.to_c  # => (0+0i)  # Defaults to zero.
'1'.to_c  # => (1+0i)  # Real (no trailing 'i').
'1i'.to_c # => (0+1i)  # Imaginary (trailing 'i').
'i'.to_c  # => (0+1i)  # Special case (imaginary 1).

With separator '+', both parts positive (or zero):

# Without trailing 'i'.
'+'.to_c    # => (0+0i)  # No values: defaults to zero.
'+1'.to_c   # => (1+0i)  # Value after '+': real only.
'1+'.to_c   # => (1+0i)  # Value before '+': real only.
'2+1'.to_c  # => (2+0i)  # Values before and after '+': real and imaginary.
# With trailing 'i'.
'+1i'.to_c  # => (0+1i)  # Value after '+': imaginary only.
'2+i'.to_c  # => (2+1i)  # Value before '+': real and imaginary 1.
'2+1i'.to_c # => (2+1i)  # Values before and after '+': real and imaginary.

With separator '-', negative imaginary part:

# Without trailing 'i'.
'-'.to_c    # => (0+0i)   # No values: defaults to zero.
'-1'.to_c   # => (-1+0i)  # Value after '-': negative real, zero imaginary.
'1-'.to_c   # => (1+0i)   # Value before '-': positive real, zero imaginary.
'2-1'.to_c  # => (2+0i)   # Values before and after '-': positive real, zero imaginary.
# With trailing 'i'.
'-1i'.to_c  # => (0-1i)   # Value after '-': negative real, zero imaginary.
'2-i'.to_c  # => (2-1i)   # Value before '-': positive real, negative imaginary.
'2-1i'.to_c # => (2-1i)   # Values before and after '-': positive real, negative imaginary.

Note that the suffixed character 'i' may instead be one of 'I', 'j', or 'J', with the same effect.

Polar Coordinates

With separator '@') interprets the values as polar coordinates: magnitude and angle.

'2@'.to_c.polar  # => [2, 0.0]    # Value before '@': magnitude only.
 # Values before and after '@': magnitude and angle.
'2@1'.to_c.polar # => [2.0, 1.0]
"1.0@#{Math::PI/2}".to_c # => (0.0+1i)
"1.0@#{Math::PI}".to_c   # => (-1+0.0i)
# Magnitude not given: defaults to zero.
'@'.to_c.polar   # => [0, 0.0]
'@1'.to_c.polar  # => [0, 0.0]

'1.0@0'.to_c             # => (1+0.0i)

Note that in all cases, the suffixed character 'i' may instead be one of 'I', 'j', 'J', with the same effect.

See Converting to Non-String.

static VALUE
rb_str_to_f(VALUE str)
{
    return DBL2NUM(rb_str_to_dbl(str, FALSE));
}

Returns the result of interpreting leading characters in +self+ as a Float:

  '3.14159'.to_f  # => 3.14159
  '1.234e-2'.to_f # => 0.01234

Characters past a leading valid number are ignored:

  '3.14 (pi to two places)'.to_f # => 3.14

Returns zero if there is no leading valid number:

  'abcdef'.to_f # => 0.0

See Converting to Non-String.

static VALUE
rb_str_to_i(int argc, VALUE *argv, VALUE str)
{
    int base = 10;

    if (rb_check_arity(argc, 0, 1) && (base = NUM2INT(argv[0])) < 0) {
        rb_raise(rb_eArgError, "invalid radix %d", base);
    }
    return rb_str_to_inum(str, base, FALSE);
}

Returns the result of interpreting leading characters in self as an integer in the given base; base must be either 0 or in range (2..36):

'123456'.to_i     # => 123456
'123def'.to_i(16) # => 1195503

With base zero given, string object may contain leading characters to specify the actual base:

'123def'.to_i(0)   # => 123
'0123def'.to_i(0)  # => 83
'0b123def'.to_i(0) # => 1
'0o123def'.to_i(0) # => 83
'0d123def'.to_i(0) # => 123
'0x123def'.to_i(0) # => 1195503

Characters past a leading valid number (in the given base) are ignored:

'12.345'.to_i   # => 12
'12345'.to_i(2) # => 1

Returns zero if there is no leading valid number:

'abcdef'.to_i # => 0
'2'.to_i(2)   # => 0

Related: see Converting to Non-String.

# File ext/json/lib/json/add/string.rb, line 32
def to_json_raw(...)
  to_json_raw_object.to_json(...)
end

This method creates a JSON text from the result of a call to to_json_raw_object of this String.

# File ext/json/lib/json/add/string.rb, line 21
def to_json_raw_object
  {
    JSON.create_id => self.class.name,
    "raw" => unpack("C*"),
  }
end

This method creates a raw object hash, that can be nested into other data structures and will be generated as a raw string. This method should be used, if you want to convert raw strings to JSON instead of UTF-8 strings, e. g. binary data.

static VALUE
string_to_r(VALUE self)
{
    VALUE num;

    rb_must_asciicompat(self);

    num = parse_rat(RSTRING_PTR(self), RSTRING_END(self), 0, TRUE);

    if (RB_FLOAT_TYPE_P(num) && !FLOAT_ZERO_P(num))
        rb_raise(rb_eFloatDomainError, "Infinity");
    return num;
}

Returns the result of interpreting leading characters in self as a rational value:

'123'.to_r       # => (123/1)   # Integer literal.
'300/2'.to_r     # => (150/1)   # Rational literal.
'-9.2'.to_r      # => (-46/5)   # Float literal.
'-9.2e2'.to_r    # => (-920/1)  # Float literal.

Ignores leading and trailing whitespace, and trailing non-numeric characters:

' 2 '.to_r       # => (2/1)
'21-Jun-09'.to_r # => (21/1)

Returns Rational zero if there are no leading numeric characters.

'BWV 1079'.to_r  # => (0/1)

NOTE: '0.3'.to_r is equivalent to 3/10r, but is different from 0.3.to_r:

'0.3'.to_r # => (3/10)
3/10r      # => (3/10)
0.3.to_r   # => (5404319552844595/18014398509481984)

Related: see Converting to Non-String.

static VALUE
rb_str_to_s(VALUE str)
{
    if (rb_obj_class(str) != rb_cString) {
        return str_duplicate(rb_cString, str);
    }
    return str;
}

Returns self if self is a String, or self converted to a String if self is a subclass of String.

Related: see Converting to New String.

static VALUE
rb_str_tr(VALUE str, VALUE src, VALUE repl)
{
    str = str_duplicate(rb_cString, str);
    tr_trans(str, src, repl, 0);
    return str;
}

Returns a copy of self with each character specified by string selector translated to the corresponding character in string replacements. The correspondence is positional:

• Each occurrence of the first character specified by selector is translated to the first character in replacements.

• Each occurrence of the second character specified by selector is translated to the second character in replacements.

• And so on.

Example:

'hello'.tr('el', 'ip') #=> "hippo"

If replacements is shorter than selector, it is implicitly padded with its own last character:

'hello'.tr('aeiou', '-')   # => "h-ll-"
'hello'.tr('aeiou', 'AA-') # => "hAll-"

Arguments selector and replacements must be valid character selectors (see Character Selectors), and may use any of its valid forms, including negation, ranges, and escapes:

'hello'.tr('^aeiou', '-')       # => "-e--o"     # Negation.
'ibm'.tr('b-z', 'a-z')          # => "hal"       # Range.
'hel^lo'.tr('\^aeiou', '-')     # => "h-l-l-"    # Escaped leading caret.
'i-b-m'.tr('b\-z', 'a-z')       # => "ibabm"     # Escaped embedded hyphen.
'foo\\bar'.tr('ab\\', 'XYZ')    # => "fooZYXr"   # Escaped backslash.

Related: see Converting to New String.

static VALUE
rb_str_tr_bang(VALUE str, VALUE src, VALUE repl)
{
    return tr_trans(str, src, repl, 0);
}

Like String#tr, except:

• Performs substitutions in self (not in a copy of self).

• Returns self if any modifications were made, nil otherwise.

Related: Modifying.

static VALUE
rb_str_tr_s(VALUE str, VALUE src, VALUE repl)
{
    str = str_duplicate(rb_cString, str);
    tr_trans(str, src, repl, 1);
    return str;
}

Like String#tr, except:

• Also squeezes the modified portions of the translated string; see String#squeeze.

• Returns the translated and squeezed string.

Examples:

'hello'.tr_s('l', 'r')   #=> "hero"
'hello'.tr_s('el', '-')  #=> "h-o"
'hello'.tr_s('el', 'hx') #=> "hhxo"

Related: see Converting to New String.

static VALUE
rb_str_tr_s_bang(VALUE str, VALUE src, VALUE repl)
{
    return tr_trans(str, src, repl, 1);
}

Like String#tr_s, except:

• Modifies self in place (not a copy of self).

• Returns self if any changes were made, nil otherwise.

Related: Modifying.

static VALUE
str_undump(VALUE str)
{
    const char *s = RSTRING_PTR(str);
    const char *s_end = RSTRING_END(str);
    rb_encoding *enc = rb_enc_get(str);
    VALUE undumped = rb_enc_str_new(s, 0L, enc);
    bool utf8 = false;
    bool binary = false;
    int w;

    rb_must_asciicompat(str);
    if (rb_str_is_ascii_only_p(str) == Qfalse) {
        rb_raise(rb_eRuntimeError, "non-ASCII character detected");
    }
    if (!str_null_check(str, &w)) {
        rb_raise(rb_eRuntimeError, "string contains null byte");
    }
    if (RSTRING_LEN(str) < 2) goto invalid_format;
    if (*s != '"') goto invalid_format;

    /* strip '"' at the start */
    s++;

    for (;;) {
        if (s >= s_end) {
            rb_raise(rb_eRuntimeError, "unterminated dumped string");
        }

        if (*s == '"') {
            /* epilogue */
            s++;
            if (s == s_end) {
                /* ascii compatible dumped string */
                break;
            }
            else {
                static const char force_encoding_suffix[] = ".force_encoding(\""; /* "\")" */
                static const char dup_suffix[] = ".dup";
                const char *encname;
                int encidx;
                ptrdiff_t size;

                /* check separately for strings dumped by older versions */
                size = sizeof(dup_suffix) - 1;
                if (s_end - s > size && memcmp(s, dup_suffix, size) == 0) s += size;

                size = sizeof(force_encoding_suffix) - 1;
                if (s_end - s <= size) goto invalid_format;
                if (memcmp(s, force_encoding_suffix, size) != 0) goto invalid_format;
                s += size;

                if (utf8) {
                    rb_raise(rb_eRuntimeError, "dumped string contained Unicode escape but used force_encoding");
                }

                encname = s;
                s = memchr(s, '"', s_end-s);
                size = s - encname;
                if (!s) goto invalid_format;
                if (s_end - s != 2) goto invalid_format;
                if (s[0] != '"' || s[1] != ')') goto invalid_format;

                encidx = rb_enc_find_index2(encname, (long)size);
                if (encidx < 0) {
                    rb_raise(rb_eRuntimeError, "dumped string has unknown encoding name");
                }
                rb_enc_associate_index(undumped, encidx);
            }
            break;
        }

        if (*s == '\\') {
            s++;
            if (s >= s_end) {
                rb_raise(rb_eRuntimeError, "invalid escape");
            }
            undump_after_backslash(undumped, &s, s_end, &enc, &utf8, &binary);
        }
        else {
            rb_str_cat(undumped, s++, 1);
        }
    }

    RB_GC_GUARD(str);

    return undumped;
invalid_format:
    rb_raise(rb_eRuntimeError, "invalid dumped string; not wrapped with '\"' nor '\"...\".force_encoding(\"...\")' form");
}

Inverse of String#dump; returns a copy of self with changes of the kinds made by String#dump “undone.”

Related: see Converting to New String.

static VALUE
rb_str_unicode_normalize(int argc, VALUE *argv, VALUE str)
{
    return unicode_normalize_common(argc, argv, str, id_normalize);
}

Returns a copy of self with Unicode normalization applied.

Argument form must be one of the following symbols (see Unicode normalization forms):

• :nfc: Canonical decomposition, followed by canonical composition.

• :nfd: Canonical decomposition.

• :nfkc: Compatibility decomposition, followed by canonical composition.

• :nfkd: Compatibility decomposition.

The encoding of self must be one of:

• Encoding::UTF_8.

• Encoding::UTF_16BE.

• Encoding::UTF_16LE.

• Encoding::UTF_32BE.

• Encoding::UTF_32LE.

• Encoding::GB18030.

• Encoding::UCS_2BE.

• Encoding::UCS_4BE.

Examples:

"a\u0300".unicode_normalize       # => "à"  # Lowercase 'a' with grave accens.
"a\u0300".unicode_normalize(:nfd) # => "à"  # Same.

Related: see Converting to New String.

static VALUE
rb_str_unicode_normalize_bang(int argc, VALUE *argv, VALUE str)
{
    return rb_str_replace(str, unicode_normalize_common(argc, argv, str, id_normalize));
}

Like String#unicode_normalize, except that the normalization is performed on self (not on a copy of self).

Related: see Modifying.

static VALUE
rb_str_unicode_normalized_p(int argc, VALUE *argv, VALUE str)
{
    return unicode_normalize_common(argc, argv, str, id_normalized_p);
}

Returns whether self is in the given form of Unicode normalization; see String#unicode_normalize.

The form must be one of :nfc, :nfd, :nfkc, or :nfkd.

Examples:

"a\u0300".unicode_normalized?       # => false
"a\u0300".unicode_normalized?(:nfd) # => true
"\u00E0".unicode_normalized?        # => true
"\u00E0".unicode_normalized?(:nfd)  # => false

Raises an exception if self is not in a Unicode encoding:

s = "\xE0".force_encoding(Encoding::ISO_8859_1)
s.unicode_normalized? # Raises Encoding::CompatibilityError

Related: see Querying.

# File pack.rb, line 25
def unpack(fmt, offset: 0)
  Primitive.attr! :use_block
  Primitive.pack_unpack(fmt, offset)
end

Extracts data from self to form new objects; see Packed Data.

With a block given, calls the block with each unpacked object.

With no block given, returns an array containing the unpacked objects.

Related: see Converting to Non-String.

# File pack.rb, line 37
def unpack1(fmt, offset: 0)
  Primitive.pack_unpack1(fmt, offset)
end

Like String#unpack with no block, but unpacks and returns only the first extracted object. See Packed Data.

Related: see Converting to Non-String.

static VALUE
rb_str_upcase(int argc, VALUE *argv, VALUE str)
{
    rb_encoding *enc;
    OnigCaseFoldType flags = ONIGENC_CASE_UPCASE;
    VALUE ret;

    flags = check_case_options(argc, argv, flags);
    enc = str_true_enc(str);
    if (case_option_single_p(flags, enc, str)) {
        ret = rb_str_new(RSTRING_PTR(str), RSTRING_LEN(str));
        str_enc_copy_direct(ret, str);
        upcase_single(ret);
    }
    else if (flags&ONIGENC_CASE_ASCII_ONLY) {
        ret = rb_str_new(0, RSTRING_LEN(str));
        rb_str_ascii_casemap(str, ret, &flags, enc);
    }
    else {
        ret = rb_str_casemap(str, &flags, enc);
    }

    return ret;
}

Returns a new string containing the upcased characters in self:

'hello'.upcase        # => "HELLO"
'straße'.upcase       # => "STRASSE"
'привет'.upcase       # => "ПРИВЕТ"
'RubyGems.org'.upcase # => "RUBYGEMS.ORG"

The sizes of self and the upcased result may differ:

s = 'Straße'
s.size        # => 6
s.upcase      # => "STRASSE"
s.upcase.size # => 7

Some characters (and some character sets) do not have upcase and downcase versions; see Case Mapping:

s = '1, 2, 3, ...'
s.upcase == s # => true
s = 'こんにちは'
s.upcase == s # => true

The casing is affected by the given mapping, which may be :ascii, :fold, or :turkic; see Case Mappings.

Related: see Converting to New String.

static VALUE
rb_str_upcase_bang(int argc, VALUE *argv, VALUE str)
{
    rb_encoding *enc;
    OnigCaseFoldType flags = ONIGENC_CASE_UPCASE;

    flags = check_case_options(argc, argv, flags);
    str_modify_keep_cr(str);
    enc = str_true_enc(str);
    if (case_option_single_p(flags, enc, str)) {
        if (upcase_single(str))
            flags |= ONIGENC_CASE_MODIFIED;
    }
    else if (flags&ONIGENC_CASE_ASCII_ONLY)
        rb_str_ascii_casemap(str, str, &flags, enc);
    else
        str_shared_replace(str, rb_str_casemap(str, &flags, enc));

    if (ONIGENC_CASE_MODIFIED&flags) return str;
    return Qnil;
}

Like String#upcase, except that:

• Changes character casings in self (not in a copy of self).

• Returns self if any changes are made, nil otherwise.

Related: See Modifying.

static VALUE
rb_str_upto(int argc, VALUE *argv, VALUE beg)
{
    VALUE end, exclusive;

    rb_scan_args(argc, argv, "11", &end, &exclusive);
    RETURN_ENUMERATOR(beg, argc, argv);
    return rb_str_upto_each(beg, end, RTEST(exclusive), str_upto_i, Qnil);
}

With a block given, calls the block with each String value returned by successive calls to String#succ; the first value is self, the next is self.succ, and so on; the sequence terminates when value other_string is reached; returns self:

a = []
'a'.upto('f') {|c| a.push(c) }
a # => ["a", "b", "c", "d", "e", "f"]

a = []
'Ж'.upto('П') {|c| a.push(c) }
a # => ["Ж", "З", "И", "Й", "К", "Л", "М", "Н", "О", "П"]

a = []
'よ'.upto('ろ') {|c| a.push(c) }
a # => ["よ", "ら", "り", "る", "れ", "ろ"]

a = []
'a8'.upto('b6') {|c| a.push(c) }
a # => ["a8", "a9", "b0", "b1", "b2", "b3", "b4", "b5", "b6"]

If argument exclusive is given as a truthy object, the last value is omitted:

a = []
'a'.upto('f', true) {|c| a.push(c) }
a # => ["a", "b", "c", "d", "e"]

If other_string would not be reached, does not call the block:

'25'.upto('5') {|s| fail s }
'aa'.upto('a') {|s| fail s }

With no block given, returns a new Enumerator:

'a8'.upto('b6') # => #<Enumerator: "a8":upto("b6")>

Related: see Iterating.

static VALUE
rb_str_valid_encoding_p(VALUE str)
{
    int cr = rb_enc_str_coderange(str);

    return RBOOL(cr != ENC_CODERANGE_BROKEN);
}

Returns whether self is encoded correctly:

s = 'Straße'
s.valid_encoding?                                 # => true
s.encoding                                        # => #<Encoding:UTF-8>
s.force_encoding(Encoding::ASCII).valid_encoding? # => false

Related: see Querying.