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.
# File lib/multi_json/vendor/okjson.rb, line 43 def decode(s) ts = lex(s) v, ts = textparse(ts) if ts.length > 0 raise Error, 'trailing garbage' end v 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.
# File lib/multi_json/vendor/okjson.rb, line 62 def encode(x) case x when Hash then objenc(x) when Array then arrenc(x) else valenc(x) end end
# File lib/multi_json/vendor/okjson.rb, line 71 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 if x.respond_to?(:to_json) x.to_json else raise Error, "cannot encode #{x.class}: #{x.inspect}" end end end
# File lib/multi_json/vendor/okjson.rb, line 291 def abbrev(s) t = s[0,10] p = t['`'] t = t[0,p] if p t = t + '...' if t.length < s.length '`' + t + '`' end
# File lib/multi_json/vendor/okjson.rb, line 434 def arrenc(a) '[' + a.map{|x| valenc(x)}.join(',') + ']' end
Parses an “array” in the sense of RFC 4627. Returns the parsed value and any trailing tokens.
# File lib/multi_json/vendor/okjson.rb, line 175 def arrparse(ts) ts = eat('[', ts) arr = [] if ts[0][0] == ']' return arr, ts[1..-1] end v, ts = valparse(ts) arr << v if ts[0][0] == ']' return arr, ts[1..-1] end loop do ts = eat(',', ts) v, ts = valparse(ts) arr << v if ts[0][0] == ']' return arr, ts[1..-1] end end end
# File lib/multi_json/vendor/okjson.rb, line 203 def eat(typ, ts) if ts[0][0] != typ raise Error, "expected #{typ} (got #{ts[0].inspect})" end ts[1..-1] end
# File lib/multi_json/vendor/okjson.rb, line 263 def falsetok(s); s[0,5] == 'false' ? [:val, 'false', false] : [] end
# File lib/multi_json/vendor/okjson.rb, line 398 def hexdec4(s) if s.length != 4 raise Error, 'short' end (nibble(s[0])<<12) | (nibble(s[1])<<8) | (nibble(s[2])<<4) | nibble(s[3]) end
# File lib/multi_json/vendor/okjson.rb, line 439 def keyenc(k) case k when String then strenc(k) else raise Error, "Hash key is not a string: #{k.inspect}" end end
Scans s and returns a list of json tokens, excluding white space (as defined in RFC 4627).
# File lib/multi_json/vendor/okjson.rb, line 213 def lex(s) ts = [] while s.length > 0 typ, lexeme, val = tok(s) if typ == nil raise Error, "invalid character at #{s[0,10].inspect}" end if typ != :space ts << [typ, lexeme, val] end s = s[lexeme.length..-1] end ts end
# File lib/multi_json/vendor/okjson.rb, line 419 def nibble(c) if 0 <= c && c <= 9 then c.ord - 0.ord elsif a <= c && c <= z then c.ord - a.ord + 10 elsif A <= c && c <= Z then c.ord - A.ord + 10 else raise Error, "invalid hex code #{c}" end end
# File lib/multi_json/vendor/okjson.rb, line 261 def nulltok(s); s[0,4] == 'null' ? [:val, 'null', nil] : [] end
# File lib/multi_json/vendor/okjson.rb, line 486 def numenc(x) if ((x.nan? || x.infinite?) rescue false) raise Error, "Numeric cannot be represented: #{x}" end "#{x}" end
# File lib/multi_json/vendor/okjson.rb, line 266 def numtok(s) m = /-?([1-9][0-9]+|[0-9])([.][0-9]+)?([eE][+-]?[0-9]+)?/.match(s) if m && m.begin(0) == 0 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**Integer(m[3][1..-1]))] end else [] end end
# File lib/multi_json/vendor/okjson.rb, line 429 def objenc(x) '{' + x.map{|k,v| keyenc(k) + ':' + valenc(v)}.join(',') + '}' end
Parses an “object” in the sense of RFC 4627. Returns the parsed value and any trailing tokens.
# File lib/multi_json/vendor/okjson.rb, line 132 def objparse(ts) ts = eat('{', ts) obj = {} if ts[0][0] == '}' return obj, ts[1..-1] end k, v, ts = pairparse(ts) obj[k] = v if ts[0][0] == '}' return obj, ts[1..-1] end loop do ts = eat(',', ts) k, v, ts = pairparse(ts) obj[k] = v if ts[0][0] == '}' return obj, ts[1..-1] end end end
Parses a “member” in the sense of RFC 4627. Returns the parsed values and any trailing tokens.
# File lib/multi_json/vendor/okjson.rb, line 162 def pairparse(ts) (typ, _, k), ts = ts[0], ts[1..-1] if typ != :str raise Error, "unexpected #{k.inspect}" end ts = eat(':', ts) v, ts = valparse(ts) [k, v, ts] end
# File lib/multi_json/vendor/okjson.rb, line 564 def rubydoesenc? ::String.method_defined?(:force_encoding) end
# File lib/multi_json/vendor/okjson.rb, line 448 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 t.write(c) rescue t.write(Ustrerr) end elsif Spc <= c && 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
# File lib/multi_json/vendor/okjson.rb, line 282 def strtok(s) m = /"([^"\]|\["\/\bfnrt]|\u[0-9a-fA-F]{4})*"/.match(s) if ! m raise Error, "invalid string literal at #{abbrev(s)}" end [:str, m[0], unquote(m[0])] end
# File lib/multi_json/vendor/okjson.rb, line 406 def subst(u1, u2) if Usurr1 <= u1 && u1 < Usurr2 && Usurr2 <= u2 && u2 < Usurr3 return ((u1-Usurr1)<<10) | (u2-Usurr2) + Usurrself end return Ucharerr end
# File lib/multi_json/vendor/okjson.rb, line 414 def surrogate?(u) Usurr1 <= u && u < Usurr3 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.
# File lib/multi_json/vendor/okjson.rb, line 97 def textparse(ts) if ts.length <= 0 raise Error, 'empty' end typ, _, val = ts[0] case typ when '{' then objparse(ts) when '[' then arrparse(ts) else raise Error, "unexpected #{val.inspect}" 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.
# File lib/multi_json/vendor/okjson.rb, line 242 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
# File lib/multi_json/vendor/okjson.rb, line 262 def truetok(s); s[0,4] == 'true' ? [:val, 'true', true] : [] 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.
# File lib/multi_json/vendor/okjson.rb, line 499 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 || Utag2 <= c1 # 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 || Utag2 <= c2 # 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 || Utag2 <= c3 # 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) return 1 end
Encodes unicode character u as UTF-8 bytes in string a at position i. Returns the number of bytes written.
# File lib/multi_json/vendor/okjson.rb, line 375 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
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.
# File lib/multi_json/vendor/okjson.rb, line 303 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. if rubydoesenc? a.force_encoding('UTF-8') end r, w = 0, 0 while r < q.length c = q[r] if c == \ r += 1 if r >= q.length raise Error, "string literal ends with a \"\\\": \"#{q}\"" end 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 if 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 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
Parses a “value” in the sense of RFC 4627. Returns the parsed value and any trailing tokens.
# File lib/multi_json/vendor/okjson.rb, line 114 def valparse(ts) if ts.length <= 0 raise Error, 'empty' end typ, _, val = ts[0] case typ when '{' then objparse(ts) when '[' then arrparse(ts) when :val,:str then [val, ts[1..-1]] else raise Error, "unexpected #{val.inspect}" end end