Some parts adapted from golang.org/src/pkg/json/decode.go and golang.org/src/pkg/utf8/utf8.go
# File lib/heroku/kensa/okjson.rb, line 249 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/heroku/kensa/okjson.rb, line 440 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/heroku/kensa/okjson.rb, line 133 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
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/heroku/kensa/okjson.rb, line 41 def decode(s) ts = lex(s) v, ts = textparse(ts) if ts.length > 0 raise Error, 'trailing garbage' end v end
# File lib/heroku/kensa/okjson.rb, line 161 def eat(typ, ts) if ts[0][0] != typ raise Error, "expected #{typ} (got #{ts[0].inspect})" end ts[1..-1] end
# File lib/heroku/kensa/okjson-support.rb, line 21 def encode(x) encode_without_stringify(x.stringify_keys) end
# File lib/heroku/kensa/okjson.rb, line 223 def falsetok(s); s[0,5] == 'false' && [:val, 'false', false] end
# File lib/heroku/kensa/okjson.rb, line 360 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/heroku/kensa/okjson.rb, line 506 def hexenc4(t, u) t.putc(Hex[(u>>12)&0xf]) t.putc(Hex[(u>>8)&0xf]) t.putc(Hex[(u>>4)&0xf]) t.putc(Hex[u&0xf]) end
# File lib/heroku/kensa/okjson.rb, line 445 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/heroku/kensa/okjson.rb, line 171 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/heroku/kensa/okjson.rb, line 390 def nibble(c) case true when 00 <= c && c <= 99 then c.ord - 00.ord when aa <= c && c <= zz then c.ord - aa.ord + 10 when AA <= c && c <= ZZ then c.ord - AA.ord + 10 else raise Error, "invalid hex code #{c}" end end
# File lib/heroku/kensa/okjson.rb, line 221 def nulltok(s); s[0,4] == 'null' && [:val, 'null', nil] end
# File lib/heroku/kensa/okjson.rb, line 514 def numenc(x) if ((x.nan? || x.infinite?) rescue false) raise Error, "Numeric cannot be represented: #{x}" end "#{x}" end
# File lib/heroku/kensa/okjson.rb, line 226 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[3] && !m[2] [:val, m[0], Integer(m[1])*(10**Integer(m[3][1..-1]))] elsif m[2] [:val, m[0], Float(m[0])] else [:val, m[0], Integer(m[0])] end end end
# File lib/heroku/kensa/okjson.rb, line 435 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/heroku/kensa/okjson.rb, line 90 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/heroku/kensa/okjson.rb, line 120 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/heroku/kensa/okjson.rb, line 454 def strenc(s) t = StringIO.new t.putc("") r = 0 # In ruby >= 1.9, s[r] is a codepoint, not a byte. rubydoesenc = s.class.method_defined?(:encoding) 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] case true when Spc <= c && c <= ~~ t.putc(c) when rubydoesenc u = c.ord surrenc(t, u) else u, size = uchardec(s, r) r += size - 1 # we add one more at the bottom of the loop surrenc(t, u) end end r += 1 end t.putc("") t.string end
# File lib/heroku/kensa/okjson.rb, line 240 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/heroku/kensa/okjson.rb, line 368 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/heroku/kensa/okjson.rb, line 492 def surrenc(t, u) if u < 0x10000 t.print('\u') hexenc4(t, u) else u1, u2 = unsubst(u) t.print('\u') hexenc4(t, u1) t.print('\u') hexenc4(t, u2) end end
# File lib/heroku/kensa/okjson.rb, line 385 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/heroku/kensa/okjson.rb, line 55 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/heroku/kensa/okjson.rb, line 200 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 nn then nulltok(s) when tt then truetok(s) when ff then falsetok(s) when "" then strtok(s) when Spc then [:space, s[0,1], s[0,1]] when \t\ then [:space, s[0,1], s[0,1]] when \n\ then [:space, s[0,1], s[0,1]] when \r\ then [:space, s[0,1], s[0,1]] else numtok(s) end end
# File lib/heroku/kensa/okjson.rb, line 222 def truetok(s); s[0,4] == 'true' && [:val, 'true', true] end
Decodes unicode character u from UTF-8 bytes in string s at position i. Returns u and the number of bytes read.
# File lib/heroku/kensa/okjson.rb, line 525 def uchardec(s, i) n = s.length - i return [Ucharerr, 1] if n < 1 c0 = s[i].ord # 1-byte, 7-bit sequence? if c0 < Utagx return [c0, 1] end # unexpected continuation byte? return [Ucharerr, 1] if c0 < Utag2 # need continuation byte return [Ucharerr, 1] if n < 2 c1 = s[i+1].ord return [Ucharerr, 1] if c1 < Utagx || Utag2 <= c1 # 2-byte, 11-bit sequence? if c0 < Utag3 u = (c0&Umask2)<<6 | (c1&Umaskx) return [Ucharerr, 1] if u <= Uchar1max return [u, 2] end # need second continuation byte return [Ucharerr, 1] if n < 3 c2 = s[i+2].ord return [Ucharerr, 1] if c2 < Utagx || Utag2 <= c2 # 3-byte, 16-bit sequence? if c0 < Utag4 u = (c0&Umask3)<<12 | (c1&Umaskx)<<6 | (c2&Umaskx) return [Ucharerr, 1] if u <= Uchar2max return [u, 3] end # need third continuation byte return [Ucharerr, 1] if n < 4 c3 = s[i+3].ord return [Ucharerr, 1] 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) return [Ucharerr, 1] if u <= Uchar3max return [u, 4] end return [Ucharerr, 1] end
Encodes unicode character u as UTF-8 bytes in string a at position i. Returns the number of bytes written.
# File lib/heroku/kensa/okjson.rb, line 336 def ucharenc(a, i, u) case true when u <= Uchar1max a[i] = (u & 0xff).chr 1 when u <= Uchar2max a[i+0] = (Utag2 | ((u>>6)&0xff)).chr a[i+1] = (Utagx | (u&Umaskx)).chr 2 when 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/heroku/kensa/okjson.rb, line 261 def unquote(q) q = q[1...-1] a = q.dup # allocate a big enough string rubydoesenc = false # In ruby >= 1.9, a[w] is a codepoint, not a byte. if a.class.method_defined?(:force_encoding) a.force_encoding('UTF-8') rubydoesenc = true end r, w = 0, 0 while r < q.length c = q[r] case true when 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 bb,ff,nn,rr,tt a[w] = Unesc[q[r]] r += 1 w += 1 when uu 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 when 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
# File lib/heroku/kensa/okjson.rb, line 376 def unsubst(u) if u < Usurrself || u > Umax || surrogate?(u) return Ucharerr, Ucharerr end u -= Usurrself [Usurr1 + ((u>>10)&0x3ff), Usurr2 + (u&0x3ff)] end
# File lib/heroku/kensa/okjson.rb, line 420 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}" end end
Parses a “value” in the sense of RFC 4627. Returns the parsed value and any trailing tokens.
# File lib/heroku/kensa/okjson.rb, line 72 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
Generated with the Darkfish Rdoc Generator 2.