module MultiJson::OkJson
Some parts adapted from golang.org/src/encoding/json/decode.go and golang.org/src/unicode/utf8/utf8.go
Constants
- ParseError
- Spc
- Uchar1max
- Uchar2max
- Uchar3max
- Ucharerr
- Umask2
- Umask3
- Umask4
- Umaskx
- Unesc
- Upstream
- Ustrerr
- Usurr1
- Usurr2
- Usurr3
- Usurrself
- Utag2
- Utag3
- Utag4
- Utag5
- Utagx
Public Instance Methods
Source
# File lib/multi_json/vendor/okjson.rb, line 41 def decode(s) ts = lex(s) v, ts = textparse(ts) raise Error, "trailing garbage" unless ts.empty? v end
Decodes a json document in string s and returns the corresponding ruby value. String s must be valid UTF-8. If you have a string in some other encoding, convert it first.
String values in the resulting structure will be UTF-8.
Source
# File lib/multi_json/vendor/okjson.rb, line 58 def encode(x) case x when Hash then objenc(x) when Array then arrenc(x) else raise Error, "root value must be an Array or a Hash" end end
Encodes x into a json text. It may contain only Array, Hash, String, Numeric, true, false, nil. (Note, this list excludes Symbol.) X itself must be an Array or a Hash. No other value can be encoded, and an error will be raised if x contains any other value, such as Nan, Infinity, Symbol, and Proc, or if a Hash key is not a String. Strings contained in x must be valid UTF-8.
Source
# File lib/multi_json/vendor/okjson.rb, line 67 def valenc(x) case x when Hash then objenc(x) when Array then arrenc(x) when String then strenc(x) when Numeric then numenc(x) when true then "true" when false then "false" when nil then "null" else raise Error, "cannot encode #{x.class}: #{x.inspect}" unless x.respond_to?(:to_json) x.to_json end end
Private Instance Methods
Source
# File lib/multi_json/vendor/okjson.rb, line 253 def abbrev(s) t = s[0, 10] p = t["`"] t = t[0, p] if p t += "..." if t.length < s.length "`" + t + "`" end
Source
# File lib/multi_json/vendor/okjson.rb, line 381 def arrenc(a) "[" + a.map { |x| valenc(x) }.join(",") + "]" end
Source
# File lib/multi_json/vendor/okjson.rb, line 153 def arrparse(ts) ts = eat("[", ts) arr = [] return arr, ts[1..] if ts[0][0] == "]" v, ts = valparse(ts) arr << v return arr, ts[1..] if ts[0][0] == "]" loop do ts = eat(",", ts) v, ts = valparse(ts) arr << v return arr, ts[1..] if ts[0][0] == "]" end end
Parses an “array” in the sense of RFC 4627. Returns the parsed value and any trailing tokens.
Source
# File lib/multi_json/vendor/okjson.rb, line 174 def eat(typ, ts) raise Error, "expected #{typ} (got #{ts[0].inspect})" if ts[0][0] != typ ts[1..] end
Source
# File lib/multi_json/vendor/okjson.rb, line 229 def falsetok(s) = (s[0, 5] == "false") ? [:val, "false", false] : [] def numtok(s) m = /(-?(?:[1-9][0-9]+|[0-9]))([.][0-9]+)?([eE][+-]?[0-9]+)?/.match(s) if m&.begin(0)&.zero? if !m[2] && !m[3] [:val, m[0], Integer(m[0])] elsif m[2] [:val, m[0], Float(m[0])] else [:val, m[0], Integer(m[1]) * (10**m[3][1..].to_i(10))] end else [] end end def strtok(s) m = %r{"([^"\\]|\\["/\\bfnrt]|\\u[0-9a-fA-F]{4})*"}.match(s) raise Error, "invalid string literal at #{abbrev(s)}" unless m [:str, m[0], unquote(m[0])] end def abbrev(s) t = s[0, 10] p = t["`"] t = t[0, p] if p t += "..." if t.length < s.length "`" + t + "`" end # Converts a quoted json string literal q into a UTF-8-encoded string. # The rules are different than for Ruby, so we cannot use eval. # Unquote will raise an error if q contains control characters. def unquote(q) q = q[1...-1] a = q.dup # allocate a big enough string # In ruby >= 1.9, a[w] is a codepoint, not a byte. a.force_encoding("UTF-8") if rubydoesenc? r = 0 w = 0 while r < q.length c = q[r] if c == "\\" r += 1 raise Error, "string literal ends with a \"\\\": \"#{q}\"" if r >= q.length case q[r] when '"', "\\", "/", "'" a[w] = q[r] r += 1 w += 1 when "b", "f", "n", "r", "t" a[w] = Unesc[q[r]] r += 1 w += 1 when "u" r += 1 uchar = begin hexdec4(q[r, 4]) rescue RuntimeError => e raise Error, "invalid escape sequence \\u#{q[r, 4]}: #{e}" end r += 4 if surrogate?(uchar) && (q.length >= r + 6) uchar1 = hexdec4(q[r + 2, 4]) uchar = subst(uchar, uchar1) if uchar != Ucharerr # A valid pair; consume. r += 6 end end if rubydoesenc? a[w] = "" << uchar w += 1 else w += ucharenc(a, w, uchar) end else raise Error, "invalid escape char #{q[r]} in \"#{q}\"" end elsif c == '"' || c < Spc raise Error, "invalid character in string literal \"#{q}\"" else # Copy anything else byte-for-byte. # Valid UTF-8 will remain valid UTF-8. # Invalid UTF-8 will remain invalid UTF-8. # In ruby >= 1.9, c is a codepoint, not a byte, # in which case this is still what we want. a[w] = c r += 1 w += 1 end end a[0, w] end # Encodes unicode character u as UTF-8 # bytes in string a at position i. # Returns the number of bytes written. def ucharenc(a, i, u) if u <= Uchar1max a[i] = (u & 0xff).chr 1 elsif u <= Uchar2max a[i + 0] = (Utag2 | ((u >> 6) & 0xff)).chr a[i + 1] = (Utagx | (u & Umaskx)).chr 2 elsif u <= Uchar3max a[i + 0] = (Utag3 | ((u >> 12) & 0xff)).chr a[i + 1] = (Utagx | ((u >> 6) & Umaskx)).chr a[i + 2] = (Utagx | (u & Umaskx)).chr 3 else a[i + 0] = (Utag4 | ((u >> 18) & 0xff)).chr a[i + 1] = (Utagx | ((u >> 12) & Umaskx)).chr a[i + 2] = (Utagx | ((u >> 6) & Umaskx)).chr a[i + 3] = (Utagx | (u & Umaskx)).chr 4 end end def hexdec4(s) raise Error, "short" if s.length != 4 (nibble(s[0]) << 12) | (nibble(s[1]) << 8) | (nibble(s[2]) << 4) | nibble(s[3]) end def subst(u1, u2) return ((u1 - Usurr1) << 10) | ((u2 - Usurr2) + Usurrself) if u1 >= Usurr1 && u1 < Usurr2 && u2 >= Usurr2 && u2 < Usurr3 Ucharerr end def surrogate?(u) u >= Usurr1 && u < Usurr3 end def nibble(c) if c >= "0" && c <= "9" then c.ord - "0".ord elsif c >= "a" && c <= "z" then c.ord - "a".ord + 10 elsif c >= "A" && c <= "Z" then c.ord - "A".ord + 10 else raise Error, "invalid hex code #{c}" end end def objenc(x) "{" + x.map { |k, v| keyenc(k) + ":" + valenc(v) }.join(",") + "}" end def arrenc(a) "[" + a.map { |x| valenc(x) }.join(",") + "]" end def keyenc(k) case k when String then strenc(k) else raise Error, "Hash key is not a string: #{k.inspect}" end end def strenc(s) t = StringIO.new t.putc('"') r = 0 while r < s.length case s[r] when '"' then t.print('\\"') when "\\" then t.print("\\\\") when "\b" then t.print('\\b') when "\f" then t.print('\\f') when "\n" then t.print('\\n') when "\r" then t.print('\\r') when "\t" then t.print('\\t') else c = s[r] # In ruby >= 1.9, s[r] is a codepoint, not a byte. if rubydoesenc? begin # c.ord will raise an error if c is invalid UTF-8 c = "\\u%04x" % [c.ord] if c.ord < Spc.ord t.write(c) rescue t.write(Ustrerr) end elsif c < Spc t.write("\\u%04x" % c) elsif c >= Spc && c <= "~" t.putc(c) else n = ucharcopy(t, s, r) # ensure valid UTF-8 output r += n - 1 # r is incremented below end end r += 1 end t.putc('"') t.string end def numenc(x) raise Error, "Numeric cannot be represented: #{x}" if (x.nan? || x.infinite? rescue false) x.to_s end # Copies the valid UTF-8 bytes of a single character # from string s at position i to I/O object t, and # returns the number of bytes copied. # If no valid UTF-8 char exists at position i, # ucharcopy writes Ustrerr and returns 1. def ucharcopy(t, s, i) n = s.length - i raise Utf8Error if n < 1 c0 = s[i].ord # 1-byte, 7-bit sequence? if c0 < Utagx t.putc(c0) return 1 end raise Utf8Error if c0 < Utag2 # unexpected continuation byte? raise Utf8Error if n < 2 # need continuation byte c1 = s[i + 1].ord raise Utf8Error if c1 < Utagx || c1 >= Utag2 # 2-byte, 11-bit sequence? if c0 < Utag3 raise Utf8Error if (((c0 & Umask2) << 6) | (c1 & Umaskx)) <= Uchar1max t.putc(c0) t.putc(c1) return 2 end # need second continuation byte raise Utf8Error if n < 3 c2 = s[i + 2].ord raise Utf8Error if c2 < Utagx || c2 >= Utag2 # 3-byte, 16-bit sequence? if c0 < Utag4 u = ((c0 & Umask3) << 12) | ((c1 & Umaskx) << 6) | (c2 & Umaskx) raise Utf8Error if u <= Uchar2max t.putc(c0) t.putc(c1) t.putc(c2) return 3 end # need third continuation byte raise Utf8Error if n < 4 c3 = s[i + 3].ord raise Utf8Error if c3 < Utagx || c3 >= Utag2 # 4-byte, 21-bit sequence? if c0 < Utag5 u = ((c0 & Umask4) << 18) | ((c1 & Umaskx) << 12) | ((c2 & Umaskx) << 6) | (c3 & Umaskx) raise Utf8Error if u <= Uchar3max t.putc(c0) t.putc(c1) t.putc(c2) t.putc(c3) return 4 end raise Utf8Error rescue Utf8Error t.write(Ustrerr) 1 end def rubydoesenc? ::String.method_defined?(:force_encoding) end class Utf8Error < ::StandardError end class Error < ::StandardError end Utagx = 0b1000_0000 Utag2 = 0b1100_0000 Utag3 = 0b1110_0000 Utag4 = 0b1111_0000 Utag5 = 0b1111_1000 Umaskx = 0b0011_1111 Umask2 = 0b0001_1111 Umask3 = 0b0000_1111 Umask4 = 0b0000_0111 Uchar1max = (1 << 7) - 1 Uchar2max = (1 << 11) - 1 Uchar3max = (1 << 16) - 1 Ucharerr = 0xFFFD # unicode "replacement char" Ustrerr = "\xef\xbf\xbd" # unicode "replacement char" Usurrself = 0x10000 Usurr1 = 0xd800 Usurr2 = 0xdc00 Usurr3 = 0xe000 Spc = " "[0] Unesc = {"b" => "\b", "f" => "\f", "n" => "\n", "r" => "\r", "t" => "\t"} end
Source
# File lib/multi_json/vendor/okjson.rb, line 352 def hexdec4(s) raise Error, "short" if s.length != 4 (nibble(s[0]) << 12) | (nibble(s[1]) << 8) | (nibble(s[2]) << 4) | nibble(s[3]) end
Source
# File lib/multi_json/vendor/okjson.rb, line 385 def keyenc(k) case k when String then strenc(k) else raise Error, "Hash key is not a string: #{k.inspect}" end end
Source
# File lib/multi_json/vendor/okjson.rb, line 182 def lex(s) ts = [] until s.empty? typ, lexeme, val = tok(s) raise Error, "invalid character at #{s[0, 10].inspect}" if typ.nil? ts << [typ, lexeme, val] if typ != :space s = s[lexeme.length..] end ts end
Scans s and returns a list of json tokens, excluding white space (as defined in RFC 4627).
Source
# File lib/multi_json/vendor/okjson.rb, line 368 def nibble(c) if c >= "0" && c <= "9" then c.ord - "0".ord elsif c >= "a" && c <= "z" then c.ord - "a".ord + 10 elsif c >= "A" && c <= "Z" then c.ord - "A".ord + 10 else raise Error, "invalid hex code #{c}" end end
Source
# File lib/multi_json/vendor/okjson.rb, line 225 def nulltok(s) = (s[0, 4] == "null") ? [:val, "null", nil] : [] def truetok(s) = (s[0, 4] == "true") ? [:val, "true", true] : [] def falsetok(s) = (s[0, 5] == "false") ? [:val, "false", false] : [] def numtok(s) m = /(-?(?:[1-9][0-9]+|[0-9]))([.][0-9]+)?([eE][+-]?[0-9]+)?/.match(s) if m&.begin(0)&.zero? if !m[2] && !m[3] [:val, m[0], Integer(m[0])] elsif m[2] [:val, m[0], Float(m[0])] else [:val, m[0], Integer(m[1]) * (10**m[3][1..].to_i(10))] end else [] end end def strtok(s) m = %r{"([^"\\]|\\["/\\bfnrt]|\\u[0-9a-fA-F]{4})*"}.match(s) raise Error, "invalid string literal at #{abbrev(s)}" unless m [:str, m[0], unquote(m[0])] end def abbrev(s) t = s[0, 10] p = t["`"] t = t[0, p] if p t += "..." if t.length < s.length "`" + t + "`" end # Converts a quoted json string literal q into a UTF-8-encoded string. # The rules are different than for Ruby, so we cannot use eval. # Unquote will raise an error if q contains control characters. def unquote(q) q = q[1...-1] a = q.dup # allocate a big enough string # In ruby >= 1.9, a[w] is a codepoint, not a byte. a.force_encoding("UTF-8") if rubydoesenc? r = 0 w = 0 while r < q.length c = q[r] if c == "\\" r += 1 raise Error, "string literal ends with a \"\\\": \"#{q}\"" if r >= q.length case q[r] when '"', "\\", "/", "'" a[w] = q[r] r += 1 w += 1 when "b", "f", "n", "r", "t" a[w] = Unesc[q[r]] r += 1 w += 1 when "u" r += 1 uchar = begin hexdec4(q[r, 4]) rescue RuntimeError => e raise Error, "invalid escape sequence \\u#{q[r, 4]}: #{e}" end r += 4 if surrogate?(uchar) && (q.length >= r + 6) uchar1 = hexdec4(q[r + 2, 4]) uchar = subst(uchar, uchar1) if uchar != Ucharerr # A valid pair; consume. r += 6 end end if rubydoesenc? a[w] = "" << uchar w += 1 else w += ucharenc(a, w, uchar) end else raise Error, "invalid escape char #{q[r]} in \"#{q}\"" end elsif c == '"' || c < Spc raise Error, "invalid character in string literal \"#{q}\"" else # Copy anything else byte-for-byte. # Valid UTF-8 will remain valid UTF-8. # Invalid UTF-8 will remain invalid UTF-8. # In ruby >= 1.9, c is a codepoint, not a byte, # in which case this is still what we want. a[w] = c r += 1 w += 1 end end a[0, w] end # Encodes unicode character u as UTF-8 # bytes in string a at position i. # Returns the number of bytes written. def ucharenc(a, i, u) if u <= Uchar1max a[i] = (u & 0xff).chr 1 elsif u <= Uchar2max a[i + 0] = (Utag2 | ((u >> 6) & 0xff)).chr a[i + 1] = (Utagx | (u & Umaskx)).chr 2 elsif u <= Uchar3max a[i + 0] = (Utag3 | ((u >> 12) & 0xff)).chr a[i + 1] = (Utagx | ((u >> 6) & Umaskx)).chr a[i + 2] = (Utagx | (u & Umaskx)).chr 3 else a[i + 0] = (Utag4 | ((u >> 18) & 0xff)).chr a[i + 1] = (Utagx | ((u >> 12) & Umaskx)).chr a[i + 2] = (Utagx | ((u >> 6) & Umaskx)).chr a[i + 3] = (Utagx | (u & Umaskx)).chr 4 end end def hexdec4(s) raise Error, "short" if s.length != 4 (nibble(s[0]) << 12) | (nibble(s[1]) << 8) | (nibble(s[2]) << 4) | nibble(s[3]) end def subst(u1, u2) return ((u1 - Usurr1) << 10) | ((u2 - Usurr2) + Usurrself) if u1 >= Usurr1 && u1 < Usurr2 && u2 >= Usurr2 && u2 < Usurr3 Ucharerr end def surrogate?(u) u >= Usurr1 && u < Usurr3 end def nibble(c) if c >= "0" && c <= "9" then c.ord - "0".ord elsif c >= "a" && c <= "z" then c.ord - "a".ord + 10 elsif c >= "A" && c <= "Z" then c.ord - "A".ord + 10 else raise Error, "invalid hex code #{c}" end end def objenc(x) "{" + x.map { |k, v| keyenc(k) + ":" + valenc(v) }.join(",") + "}" end def arrenc(a) "[" + a.map { |x| valenc(x) }.join(",") + "]" end def keyenc(k) case k when String then strenc(k) else raise Error, "Hash key is not a string: #{k.inspect}" end end def strenc(s) t = StringIO.new t.putc('"') r = 0 while r < s.length case s[r] when '"' then t.print('\\"') when "\\" then t.print("\\\\") when "\b" then t.print('\\b') when "\f" then t.print('\\f') when "\n" then t.print('\\n') when "\r" then t.print('\\r') when "\t" then t.print('\\t') else c = s[r] # In ruby >= 1.9, s[r] is a codepoint, not a byte. if rubydoesenc? begin # c.ord will raise an error if c is invalid UTF-8 c = "\\u%04x" % [c.ord] if c.ord < Spc.ord t.write(c) rescue t.write(Ustrerr) end elsif c < Spc t.write("\\u%04x" % c) elsif c >= Spc && c <= "~" t.putc(c) else n = ucharcopy(t, s, r) # ensure valid UTF-8 output r += n - 1 # r is incremented below end end r += 1 end t.putc('"') t.string end def numenc(x) raise Error, "Numeric cannot be represented: #{x}" if (x.nan? || x.infinite? rescue false) x.to_s end # Copies the valid UTF-8 bytes of a single character # from string s at position i to I/O object t, and # returns the number of bytes copied. # If no valid UTF-8 char exists at position i, # ucharcopy writes Ustrerr and returns 1. def ucharcopy(t, s, i) n = s.length - i raise Utf8Error if n < 1 c0 = s[i].ord # 1-byte, 7-bit sequence? if c0 < Utagx t.putc(c0) return 1 end raise Utf8Error if c0 < Utag2 # unexpected continuation byte? raise Utf8Error if n < 2 # need continuation byte c1 = s[i + 1].ord raise Utf8Error if c1 < Utagx || c1 >= Utag2 # 2-byte, 11-bit sequence? if c0 < Utag3 raise Utf8Error if (((c0 & Umask2) << 6) | (c1 & Umaskx)) <= Uchar1max t.putc(c0) t.putc(c1) return 2 end # need second continuation byte raise Utf8Error if n < 3 c2 = s[i + 2].ord raise Utf8Error if c2 < Utagx || c2 >= Utag2 # 3-byte, 16-bit sequence? if c0 < Utag4 u = ((c0 & Umask3) << 12) | ((c1 & Umaskx) << 6) | (c2 & Umaskx) raise Utf8Error if u <= Uchar2max t.putc(c0) t.putc(c1) t.putc(c2) return 3 end # need third continuation byte raise Utf8Error if n < 4 c3 = s[i + 3].ord raise Utf8Error if c3 < Utagx || c3 >= Utag2 # 4-byte, 21-bit sequence? if c0 < Utag5 u = ((c0 & Umask4) << 18) | ((c1 & Umaskx) << 12) | ((c2 & Umaskx) << 6) | (c3 & Umaskx) raise Utf8Error if u <= Uchar3max t.putc(c0) t.putc(c1) t.putc(c2) t.putc(c3) return 4 end raise Utf8Error rescue Utf8Error t.write(Ustrerr) 1 end def rubydoesenc? ::String.method_defined?(:force_encoding) end class Utf8Error < ::StandardError end class Error < ::StandardError end Utagx = 0b1000_0000 Utag2 = 0b1100_0000 Utag3 = 0b1110_0000 Utag4 = 0b1111_0000 Utag5 = 0b1111_1000 Umaskx = 0b0011_1111 Umask2 = 0b0001_1111 Umask3 = 0b0000_1111 Umask4 = 0b0000_0111 Uchar1max = (1 << 7) - 1 Uchar2max = (1 << 11) - 1 Uchar3max = (1 << 16) - 1 Ucharerr = 0xFFFD # unicode "replacement char" Ustrerr = "\xef\xbf\xbd" # unicode "replacement char" Usurrself = 0x10000 Usurr1 = 0xd800 Usurr2 = 0xdc00 Usurr3 = 0xe000 Spc = " "[0] Unesc = {"b" => "\b", "f" => "\f", "n" => "\n", "r" => "\r", "t" => "\t"} end
Source
# File lib/multi_json/vendor/okjson.rb, line 433 def numenc(x) raise Error, "Numeric cannot be represented: #{x}" if (x.nan? || x.infinite? rescue false) x.to_s end
Source
# File lib/multi_json/vendor/okjson.rb, line 231 def numtok(s) m = /(-?(?:[1-9][0-9]+|[0-9]))([.][0-9]+)?([eE][+-]?[0-9]+)?/.match(s) if m&.begin(0)&.zero? if !m[2] && !m[3] [:val, m[0], Integer(m[0])] elsif m[2] [:val, m[0], Float(m[0])] else [:val, m[0], Integer(m[1]) * (10**m[3][1..].to_i(10))] end else [] end end
Source
# File lib/multi_json/vendor/okjson.rb, line 377 def objenc(x) "{" + x.map { |k, v| keyenc(k) + ":" + valenc(v) }.join(",") + "}" end
Source
# File lib/multi_json/vendor/okjson.rb, line 118 def objparse(ts) ts = eat("{", ts) obj = {} return obj, ts[1..] if ts[0][0] == "}" k, v, ts = pairparse(ts) obj[k] = v return obj, ts[1..] if ts[0][0] == "}" loop do ts = eat(",", ts) k, v, ts = pairparse(ts) obj[k] = v return obj, ts[1..] if ts[0][0] == "}" end end
Parses an “object” in the sense of RFC 4627. Returns the parsed value and any trailing tokens.
Source
# File lib/multi_json/vendor/okjson.rb, line 141 def pairparse(ts) (typ, _, k) = ts[0] ts = ts[1..] raise Error, "unexpected #{k.inspect}" if typ != :str ts = eat(":", ts) v, ts = valparse(ts) [k, v, ts] end
Parses a “member” in the sense of RFC 4627. Returns the parsed values and any trailing tokens.
Source
# File lib/multi_json/vendor/okjson.rb, line 513 def rubydoesenc? ::String.method_defined?(:force_encoding) end
Source
# File lib/multi_json/vendor/okjson.rb, line 393 def strenc(s) t = StringIO.new t.putc('"') r = 0 while r < s.length case s[r] when '"' then t.print('\\"') when "\\" then t.print("\\\\") when "\b" then t.print('\\b') when "\f" then t.print('\\f') when "\n" then t.print('\\n') when "\r" then t.print('\\r') when "\t" then t.print('\\t') else c = s[r] # In ruby >= 1.9, s[r] is a codepoint, not a byte. if rubydoesenc? begin # c.ord will raise an error if c is invalid UTF-8 c = "\\u%04x" % [c.ord] if c.ord < Spc.ord t.write(c) rescue t.write(Ustrerr) end elsif c < Spc t.write("\\u%04x" % c) elsif c >= Spc && c <= "~" t.putc(c) else n = ucharcopy(t, s, r) # ensure valid UTF-8 output r += n - 1 # r is incremented below end end r += 1 end t.putc('"') t.string end
Source
# File lib/multi_json/vendor/okjson.rb, line 246 def strtok(s) m = %r{"([^"\\]|\\["/\\bfnrt]|\\u[0-9a-fA-F]{4})*"}.match(s) raise Error, "invalid string literal at #{abbrev(s)}" unless m [:str, m[0], unquote(m[0])] end
Source
# File lib/multi_json/vendor/okjson.rb, line 358 def subst(u1, u2) return ((u1 - Usurr1) << 10) | ((u2 - Usurr2) + Usurrself) if u1 >= Usurr1 && u1 < Usurr2 && u2 >= Usurr2 && u2 < Usurr3 Ucharerr end
Source
# File lib/multi_json/vendor/okjson.rb, line 364 def surrogate?(u) u >= Usurr1 && u < Usurr3 end
Source
# File lib/multi_json/vendor/okjson.rb, line 89 def textparse(ts) raise Error, "empty" if ts.length <= 0 typ, _, val = ts[0] case typ when "{" then objparse(ts) when "[" then arrparse(ts) else raise Error, "unexpected #{val.inspect}" end end
Parses a “json text” in the sense of RFC 4627. Returns the parsed value and any trailing tokens. Note: this is almost the same as valparse, except that it does not accept atomic values.
Source
# File lib/multi_json/vendor/okjson.rb, line 207 def tok(s) case s[0] when "{" then ["{", s[0, 1], s[0, 1]] when "}" then ["}", s[0, 1], s[0, 1]] when ":" then [":", s[0, 1], s[0, 1]] when "," then [",", s[0, 1], s[0, 1]] when "[" then ["[", s[0, 1], s[0, 1]] when "]" then ["]", s[0, 1], s[0, 1]] when "n" then nulltok(s) when "t" then truetok(s) when "f" then falsetok(s) when '"' then strtok(s) when Spc, "\t", "\n", "\r" then [:space, s[0, 1], s[0, 1]] else numtok(s) end end
Scans the first token in s and returns a 3-element list, or nil if s does not begin with a valid token.
The first list element is one of ‘{’, ‘}’, ‘:’, ‘,’, ‘[’, ‘]’, :val, :str, and :space.
The second element is the lexeme.
The third element is the value of the token for :val and :str, otherwise it is the lexeme.
Source
# File lib/multi_json/vendor/okjson.rb, line 227 def truetok(s) = (s[0, 4] == "true") ? [:val, "true", true] : [] def falsetok(s) = (s[0, 5] == "false") ? [:val, "false", false] : [] def numtok(s) m = /(-?(?:[1-9][0-9]+|[0-9]))([.][0-9]+)?([eE][+-]?[0-9]+)?/.match(s) if m&.begin(0)&.zero? if !m[2] && !m[3] [:val, m[0], Integer(m[0])] elsif m[2] [:val, m[0], Float(m[0])] else [:val, m[0], Integer(m[1]) * (10**m[3][1..].to_i(10))] end else [] end end def strtok(s) m = %r{"([^"\\]|\\["/\\bfnrt]|\\u[0-9a-fA-F]{4})*"}.match(s) raise Error, "invalid string literal at #{abbrev(s)}" unless m [:str, m[0], unquote(m[0])] end def abbrev(s) t = s[0, 10] p = t["`"] t = t[0, p] if p t += "..." if t.length < s.length "`" + t + "`" end # Converts a quoted json string literal q into a UTF-8-encoded string. # The rules are different than for Ruby, so we cannot use eval. # Unquote will raise an error if q contains control characters. def unquote(q) q = q[1...-1] a = q.dup # allocate a big enough string # In ruby >= 1.9, a[w] is a codepoint, not a byte. a.force_encoding("UTF-8") if rubydoesenc? r = 0 w = 0 while r < q.length c = q[r] if c == "\\" r += 1 raise Error, "string literal ends with a \"\\\": \"#{q}\"" if r >= q.length case q[r] when '"', "\\", "/", "'" a[w] = q[r] r += 1 w += 1 when "b", "f", "n", "r", "t" a[w] = Unesc[q[r]] r += 1 w += 1 when "u" r += 1 uchar = begin hexdec4(q[r, 4]) rescue RuntimeError => e raise Error, "invalid escape sequence \\u#{q[r, 4]}: #{e}" end r += 4 if surrogate?(uchar) && (q.length >= r + 6) uchar1 = hexdec4(q[r + 2, 4]) uchar = subst(uchar, uchar1) if uchar != Ucharerr # A valid pair; consume. r += 6 end end if rubydoesenc? a[w] = "" << uchar w += 1 else w += ucharenc(a, w, uchar) end else raise Error, "invalid escape char #{q[r]} in \"#{q}\"" end elsif c == '"' || c < Spc raise Error, "invalid character in string literal \"#{q}\"" else # Copy anything else byte-for-byte. # Valid UTF-8 will remain valid UTF-8. # Invalid UTF-8 will remain invalid UTF-8. # In ruby >= 1.9, c is a codepoint, not a byte, # in which case this is still what we want. a[w] = c r += 1 w += 1 end end a[0, w] end # Encodes unicode character u as UTF-8 # bytes in string a at position i. # Returns the number of bytes written. def ucharenc(a, i, u) if u <= Uchar1max a[i] = (u & 0xff).chr 1 elsif u <= Uchar2max a[i + 0] = (Utag2 | ((u >> 6) & 0xff)).chr a[i + 1] = (Utagx | (u & Umaskx)).chr 2 elsif u <= Uchar3max a[i + 0] = (Utag3 | ((u >> 12) & 0xff)).chr a[i + 1] = (Utagx | ((u >> 6) & Umaskx)).chr a[i + 2] = (Utagx | (u & Umaskx)).chr 3 else a[i + 0] = (Utag4 | ((u >> 18) & 0xff)).chr a[i + 1] = (Utagx | ((u >> 12) & Umaskx)).chr a[i + 2] = (Utagx | ((u >> 6) & Umaskx)).chr a[i + 3] = (Utagx | (u & Umaskx)).chr 4 end end def hexdec4(s) raise Error, "short" if s.length != 4 (nibble(s[0]) << 12) | (nibble(s[1]) << 8) | (nibble(s[2]) << 4) | nibble(s[3]) end def subst(u1, u2) return ((u1 - Usurr1) << 10) | ((u2 - Usurr2) + Usurrself) if u1 >= Usurr1 && u1 < Usurr2 && u2 >= Usurr2 && u2 < Usurr3 Ucharerr end def surrogate?(u) u >= Usurr1 && u < Usurr3 end def nibble(c) if c >= "0" && c <= "9" then c.ord - "0".ord elsif c >= "a" && c <= "z" then c.ord - "a".ord + 10 elsif c >= "A" && c <= "Z" then c.ord - "A".ord + 10 else raise Error, "invalid hex code #{c}" end end def objenc(x) "{" + x.map { |k, v| keyenc(k) + ":" + valenc(v) }.join(",") + "}" end def arrenc(a) "[" + a.map { |x| valenc(x) }.join(",") + "]" end def keyenc(k) case k when String then strenc(k) else raise Error, "Hash key is not a string: #{k.inspect}" end end def strenc(s) t = StringIO.new t.putc('"') r = 0 while r < s.length case s[r] when '"' then t.print('\\"') when "\\" then t.print("\\\\") when "\b" then t.print('\\b') when "\f" then t.print('\\f') when "\n" then t.print('\\n') when "\r" then t.print('\\r') when "\t" then t.print('\\t') else c = s[r] # In ruby >= 1.9, s[r] is a codepoint, not a byte. if rubydoesenc? begin # c.ord will raise an error if c is invalid UTF-8 c = "\\u%04x" % [c.ord] if c.ord < Spc.ord t.write(c) rescue t.write(Ustrerr) end elsif c < Spc t.write("\\u%04x" % c) elsif c >= Spc && c <= "~" t.putc(c) else n = ucharcopy(t, s, r) # ensure valid UTF-8 output r += n - 1 # r is incremented below end end r += 1 end t.putc('"') t.string end def numenc(x) raise Error, "Numeric cannot be represented: #{x}" if (x.nan? || x.infinite? rescue false) x.to_s end # Copies the valid UTF-8 bytes of a single character # from string s at position i to I/O object t, and # returns the number of bytes copied. # If no valid UTF-8 char exists at position i, # ucharcopy writes Ustrerr and returns 1. def ucharcopy(t, s, i) n = s.length - i raise Utf8Error if n < 1 c0 = s[i].ord # 1-byte, 7-bit sequence? if c0 < Utagx t.putc(c0) return 1 end raise Utf8Error if c0 < Utag2 # unexpected continuation byte? raise Utf8Error if n < 2 # need continuation byte c1 = s[i + 1].ord raise Utf8Error if c1 < Utagx || c1 >= Utag2 # 2-byte, 11-bit sequence? if c0 < Utag3 raise Utf8Error if (((c0 & Umask2) << 6) | (c1 & Umaskx)) <= Uchar1max t.putc(c0) t.putc(c1) return 2 end # need second continuation byte raise Utf8Error if n < 3 c2 = s[i + 2].ord raise Utf8Error if c2 < Utagx || c2 >= Utag2 # 3-byte, 16-bit sequence? if c0 < Utag4 u = ((c0 & Umask3) << 12) | ((c1 & Umaskx) << 6) | (c2 & Umaskx) raise Utf8Error if u <= Uchar2max t.putc(c0) t.putc(c1) t.putc(c2) return 3 end # need third continuation byte raise Utf8Error if n < 4 c3 = s[i + 3].ord raise Utf8Error if c3 < Utagx || c3 >= Utag2 # 4-byte, 21-bit sequence? if c0 < Utag5 u = ((c0 & Umask4) << 18) | ((c1 & Umaskx) << 12) | ((c2 & Umaskx) << 6) | (c3 & Umaskx) raise Utf8Error if u <= Uchar3max t.putc(c0) t.putc(c1) t.putc(c2) t.putc(c3) return 4 end raise Utf8Error rescue Utf8Error t.write(Ustrerr) 1 end def rubydoesenc? ::String.method_defined?(:force_encoding) end class Utf8Error < ::StandardError end class Error < ::StandardError end Utagx = 0b1000_0000 Utag2 = 0b1100_0000 Utag3 = 0b1110_0000 Utag4 = 0b1111_0000 Utag5 = 0b1111_1000 Umaskx = 0b0011_1111 Umask2 = 0b0001_1111 Umask3 = 0b0000_1111 Umask4 = 0b0000_0111 Uchar1max = (1 << 7) - 1 Uchar2max = (1 << 11) - 1 Uchar3max = (1 << 16) - 1 Ucharerr = 0xFFFD # unicode "replacement char" Ustrerr = "\xef\xbf\xbd" # unicode "replacement char" Usurrself = 0x10000 Usurr1 = 0xd800 Usurr2 = 0xdc00 Usurr3 = 0xe000 Spc = " "[0] Unesc = {"b" => "\b", "f" => "\f", "n" => "\n", "r" => "\r", "t" => "\t"} end end
Source
# File lib/multi_json/vendor/okjson.rb, line 444 def ucharcopy(t, s, i) n = s.length - i raise Utf8Error if n < 1 c0 = s[i].ord # 1-byte, 7-bit sequence? if c0 < Utagx t.putc(c0) return 1 end raise Utf8Error if c0 < Utag2 # unexpected continuation byte? raise Utf8Error if n < 2 # need continuation byte c1 = s[i + 1].ord raise Utf8Error if c1 < Utagx || c1 >= Utag2 # 2-byte, 11-bit sequence? if c0 < Utag3 raise Utf8Error if (((c0 & Umask2) << 6) | (c1 & Umaskx)) <= Uchar1max t.putc(c0) t.putc(c1) return 2 end # need second continuation byte raise Utf8Error if n < 3 c2 = s[i + 2].ord raise Utf8Error if c2 < Utagx || c2 >= Utag2 # 3-byte, 16-bit sequence? if c0 < Utag4 u = ((c0 & Umask3) << 12) | ((c1 & Umaskx) << 6) | (c2 & Umaskx) raise Utf8Error if u <= Uchar2max t.putc(c0) t.putc(c1) t.putc(c2) return 3 end # need third continuation byte raise Utf8Error if n < 4 c3 = s[i + 3].ord raise Utf8Error if c3 < Utagx || c3 >= Utag2 # 4-byte, 21-bit sequence? if c0 < Utag5 u = ((c0 & Umask4) << 18) | ((c1 & Umaskx) << 12) | ((c2 & Umaskx) << 6) | (c3 & Umaskx) raise Utf8Error if u <= Uchar3max t.putc(c0) t.putc(c1) t.putc(c2) t.putc(c3) return 4 end raise Utf8Error rescue Utf8Error t.write(Ustrerr) 1 end
Copies the valid UTF-8 bytes of a single character from string s at position i to I/O object t, and returns the number of bytes copied. If no valid UTF-8 char exists at position i, ucharcopy writes Ustrerr
and returns 1.
Source
# File lib/multi_json/vendor/okjson.rb, line 330 def ucharenc(a, i, u) if u <= Uchar1max a[i] = (u & 0xff).chr 1 elsif u <= Uchar2max a[i + 0] = (Utag2 | ((u >> 6) & 0xff)).chr a[i + 1] = (Utagx | (u & Umaskx)).chr 2 elsif u <= Uchar3max a[i + 0] = (Utag3 | ((u >> 12) & 0xff)).chr a[i + 1] = (Utagx | ((u >> 6) & Umaskx)).chr a[i + 2] = (Utagx | (u & Umaskx)).chr 3 else a[i + 0] = (Utag4 | ((u >> 18) & 0xff)).chr a[i + 1] = (Utagx | ((u >> 12) & Umaskx)).chr a[i + 2] = (Utagx | ((u >> 6) & Umaskx)).chr a[i + 3] = (Utagx | (u & Umaskx)).chr 4 end end
Encodes unicode character u as UTF-8 bytes in string a at position i. Returns the number of bytes written.
Source
# File lib/multi_json/vendor/okjson.rb, line 264 def unquote(q) q = q[1...-1] a = q.dup # allocate a big enough string # In ruby >= 1.9, a[w] is a codepoint, not a byte. a.force_encoding("UTF-8") if rubydoesenc? r = 0 w = 0 while r < q.length c = q[r] if c == "\\" r += 1 raise Error, "string literal ends with a \"\\\": \"#{q}\"" if r >= q.length case q[r] when '"', "\\", "/", "'" a[w] = q[r] r += 1 w += 1 when "b", "f", "n", "r", "t" a[w] = Unesc[q[r]] r += 1 w += 1 when "u" r += 1 uchar = begin hexdec4(q[r, 4]) rescue RuntimeError => e raise Error, "invalid escape sequence \\u#{q[r, 4]}: #{e}" end r += 4 if surrogate?(uchar) && (q.length >= r + 6) uchar1 = hexdec4(q[r + 2, 4]) uchar = subst(uchar, uchar1) if uchar != Ucharerr # A valid pair; consume. r += 6 end end if rubydoesenc? a[w] = "" << uchar w += 1 else w += ucharenc(a, w, uchar) end else raise Error, "invalid escape char #{q[r]} in \"#{q}\"" end elsif c == '"' || c < Spc raise Error, "invalid character in string literal \"#{q}\"" else # Copy anything else byte-for-byte. # Valid UTF-8 will remain valid UTF-8. # Invalid UTF-8 will remain invalid UTF-8. # In ruby >= 1.9, c is a codepoint, not a byte, # in which case this is still what we want. a[w] = c r += 1 w += 1 end end a[0, w] end
Converts a quoted json string literal q into a UTF-8-encoded string. The rules are different than for Ruby, so we cannot use eval. Unquote will raise an error if q contains control characters.
Source
# File lib/multi_json/vendor/okjson.rb, line 103 def valparse(ts) raise Error, "empty" if ts.length <= 0 typ, _, val = ts[0] case typ when "{" then objparse(ts) when "[" then arrparse(ts) when :val, :str then [val, ts[1..]] else raise Error, "unexpected #{val.inspect}" end end
Parses a “value” in the sense of RFC 4627. Returns the parsed value and any trailing tokens.