Chris@0
|
1 # encoding: utf-8
|
Chris@0
|
2 module CodeRay
|
Chris@0
|
3 module Scanners
|
Chris@0
|
4
|
Chris@0
|
5 module Ruby::Patterns # :nodoc:
|
Chris@0
|
6
|
Chris@0
|
7 RESERVED_WORDS = %w[
|
Chris@0
|
8 and def end in or unless begin
|
Chris@0
|
9 defined? ensure module redo super until
|
Chris@0
|
10 BEGIN break do next rescue then
|
Chris@0
|
11 when END case else for retry
|
Chris@0
|
12 while alias class elsif if not return
|
Chris@0
|
13 undef yield
|
Chris@0
|
14 ]
|
Chris@0
|
15
|
Chris@0
|
16 DEF_KEYWORDS = %w[ def ]
|
Chris@0
|
17 UNDEF_KEYWORDS = %w[ undef ]
|
Chris@0
|
18 ALIAS_KEYWORDS = %w[ alias ]
|
Chris@0
|
19 MODULE_KEYWORDS = %w[class module]
|
Chris@0
|
20 DEF_NEW_STATE = WordList.new(:initial).
|
Chris@0
|
21 add(DEF_KEYWORDS, :def_expected).
|
Chris@0
|
22 add(UNDEF_KEYWORDS, :undef_expected).
|
Chris@0
|
23 add(ALIAS_KEYWORDS, :alias_expected).
|
Chris@0
|
24 add(MODULE_KEYWORDS, :module_expected)
|
Chris@0
|
25
|
Chris@0
|
26 PREDEFINED_CONSTANTS = %w[
|
Chris@0
|
27 nil true false self
|
Chris@0
|
28 DATA ARGV ARGF __FILE__ __LINE__
|
Chris@0
|
29 ]
|
Chris@0
|
30
|
Chris@0
|
31 IDENT_KIND = WordList.new(:ident).
|
Chris@0
|
32 add(RESERVED_WORDS, :reserved).
|
Chris@0
|
33 add(PREDEFINED_CONSTANTS, :pre_constant)
|
Chris@0
|
34
|
Chris@0
|
35 IDENT = 'ä'[/[[:alpha:]]/] == 'ä' ? /[[:alpha:]_][[:alnum:]_]*/ : /[^\W\d]\w*/
|
Chris@0
|
36
|
Chris@0
|
37 METHOD_NAME = / #{IDENT} [?!]? /ox
|
Chris@0
|
38 METHOD_NAME_OPERATOR = /
|
Chris@0
|
39 \*\*? # multiplication and power
|
Chris@0
|
40 | [-+~]@? # plus, minus, tilde with and without at sign
|
Chris@0
|
41 | [\/%&|^`] # division, modulo or format strings, and, or, xor, system
|
Chris@0
|
42 | \[\]=? # array getter and setter
|
Chris@0
|
43 | << | >> # append or shift left, shift right
|
Chris@0
|
44 | <=?>? | >=? # comparison, rocket operator
|
Chris@0
|
45 | ===? | =~ # simple equality, case equality, match
|
Chris@0
|
46 | ![~=@]? # negation with and without at sign, not-equal and not-match
|
Chris@0
|
47 /ox
|
Chris@0
|
48 METHOD_NAME_EX = / #{IDENT} (?:[?!]|=(?!>))? | #{METHOD_NAME_OPERATOR} /ox
|
Chris@0
|
49 INSTANCE_VARIABLE = / @ #{IDENT} /ox
|
Chris@0
|
50 CLASS_VARIABLE = / @@ #{IDENT} /ox
|
Chris@0
|
51 OBJECT_VARIABLE = / @@? #{IDENT} /ox
|
Chris@0
|
52 GLOBAL_VARIABLE = / \$ (?: #{IDENT} | [1-9]\d* | 0\w* | [~&+`'=\/,;_.<>!@$?*":\\] | -[a-zA-Z_0-9] ) /ox
|
Chris@0
|
53 PREFIX_VARIABLE = / #{GLOBAL_VARIABLE} | #{OBJECT_VARIABLE} /ox
|
Chris@0
|
54 VARIABLE = / @?@? #{IDENT} | #{GLOBAL_VARIABLE} /ox
|
Chris@0
|
55
|
Chris@0
|
56 QUOTE_TO_TYPE = {
|
Chris@0
|
57 '`' => :shell,
|
Chris@0
|
58 '/'=> :regexp,
|
Chris@0
|
59 }
|
Chris@0
|
60 QUOTE_TO_TYPE.default = :string
|
Chris@0
|
61
|
Chris@0
|
62 REGEXP_MODIFIERS = /[mixounse]*/
|
Chris@0
|
63 REGEXP_SYMBOLS = /[|?*+(){}\[\].^$]/
|
Chris@0
|
64
|
Chris@0
|
65 DECIMAL = /\d+(?:_\d+)*/
|
Chris@0
|
66 OCTAL = /0_?[0-7]+(?:_[0-7]+)*/
|
Chris@0
|
67 HEXADECIMAL = /0x[0-9A-Fa-f]+(?:_[0-9A-Fa-f]+)*/
|
Chris@0
|
68 BINARY = /0b[01]+(?:_[01]+)*/
|
Chris@0
|
69
|
Chris@0
|
70 EXPONENT = / [eE] [+-]? #{DECIMAL} /ox
|
Chris@0
|
71 FLOAT_SUFFIX = / #{EXPONENT} | \. #{DECIMAL} #{EXPONENT}? /ox
|
Chris@0
|
72 FLOAT_OR_INT = / #{DECIMAL} (?: #{FLOAT_SUFFIX} () )? /ox
|
Chris@0
|
73 NUMERIC = / (?: (?=0) (?: #{OCTAL} | #{HEXADECIMAL} | #{BINARY} ) | #{FLOAT_OR_INT} ) /ox
|
Chris@0
|
74
|
Chris@0
|
75 SYMBOL = /
|
Chris@0
|
76 :
|
Chris@0
|
77 (?:
|
Chris@0
|
78 #{METHOD_NAME_EX}
|
Chris@0
|
79 | #{PREFIX_VARIABLE}
|
Chris@0
|
80 | ['"]
|
Chris@0
|
81 )
|
Chris@0
|
82 /ox
|
Chris@0
|
83 METHOD_NAME_OR_SYMBOL = / #{METHOD_NAME_EX} | #{SYMBOL} /ox
|
Chris@0
|
84
|
Chris@0
|
85 SIMPLE_ESCAPE = /
|
Chris@0
|
86 [abefnrstv]
|
Chris@0
|
87 | [0-7]{1,3}
|
Chris@0
|
88 | x[0-9A-Fa-f]{1,2}
|
Chris@0
|
89 | .?
|
Chris@0
|
90 /mx
|
Chris@0
|
91
|
Chris@0
|
92 CONTROL_META_ESCAPE = /
|
Chris@0
|
93 (?: M-|C-|c )
|
Chris@0
|
94 (?: \\ (?: M-|C-|c ) )*
|
Chris@0
|
95 (?: [^\\] | \\ #{SIMPLE_ESCAPE} )?
|
Chris@0
|
96 /mox
|
Chris@0
|
97
|
Chris@0
|
98 ESCAPE = /
|
Chris@0
|
99 #{CONTROL_META_ESCAPE} | #{SIMPLE_ESCAPE}
|
Chris@0
|
100 /mox
|
Chris@0
|
101
|
Chris@0
|
102 CHARACTER = /
|
Chris@0
|
103 \?
|
Chris@0
|
104 (?:
|
Chris@0
|
105 [^\s\\]
|
Chris@0
|
106 | \\ #{ESCAPE}
|
Chris@0
|
107 )
|
Chris@0
|
108 /mox
|
Chris@0
|
109
|
Chris@0
|
110 # NOTE: This is not completely correct, but
|
Chris@0
|
111 # nobody needs heredoc delimiters ending with \n.
|
Chris@0
|
112 HEREDOC_OPEN = /
|
Chris@0
|
113 << (-)? # $1 = float
|
Chris@0
|
114 (?:
|
Chris@0
|
115 ( [A-Za-z_0-9]+ ) # $2 = delim
|
Chris@0
|
116 |
|
Chris@0
|
117 ( ["'`\/] ) # $3 = quote, type
|
Chris@0
|
118 ( [^\n]*? ) \3 # $4 = delim
|
Chris@0
|
119 )
|
Chris@0
|
120 /mx
|
Chris@0
|
121
|
Chris@0
|
122 RUBYDOC = /
|
Chris@0
|
123 =begin (?!\S)
|
Chris@0
|
124 .*?
|
Chris@0
|
125 (?: \Z | ^=end (?!\S) [^\n]* )
|
Chris@0
|
126 /mx
|
Chris@0
|
127
|
Chris@0
|
128 DATA = /
|
Chris@0
|
129 __END__$
|
Chris@0
|
130 .*?
|
Chris@0
|
131 (?: \Z | (?=^\#CODE) )
|
Chris@0
|
132 /mx
|
Chris@0
|
133
|
Chris@0
|
134 # Checks for a valid value to follow. This enables
|
Chris@0
|
135 # value_expected in method calls without parentheses.
|
Chris@0
|
136 VALUE_FOLLOWS = /
|
Chris@0
|
137 (?>[ \t\f\v]+)
|
Chris@0
|
138 (?:
|
Chris@0
|
139 [%\/][^\s=]
|
Chris@0
|
140 | <<-?\S
|
Chris@0
|
141 | [-+] \d
|
Chris@0
|
142 | #{CHARACTER}
|
Chris@0
|
143 )
|
Chris@0
|
144 /x
|
Chris@0
|
145 KEYWORDS_EXPECTING_VALUE = WordList.new.add(%w[
|
Chris@0
|
146 and end in or unless begin
|
Chris@0
|
147 defined? ensure redo super until
|
Chris@0
|
148 break do next rescue then
|
Chris@0
|
149 when case else for retry
|
Chris@0
|
150 while elsif if not return
|
Chris@0
|
151 yield
|
Chris@0
|
152 ])
|
Chris@0
|
153
|
Chris@0
|
154 RUBYDOC_OR_DATA = / #{RUBYDOC} | #{DATA} /xo
|
Chris@0
|
155
|
Chris@0
|
156 RDOC_DATA_START = / ^=begin (?!\S) | ^__END__$ /x
|
Chris@0
|
157
|
Chris@0
|
158 FANCY_START_CORRECT = / % ( [qQwWxsr] | (?![a-zA-Z0-9]) ) ([^a-zA-Z0-9]) /mx
|
Chris@0
|
159
|
Chris@0
|
160 FancyStringType = {
|
Chris@0
|
161 'q' => [:string, false],
|
Chris@0
|
162 'Q' => [:string, true],
|
Chris@0
|
163 'r' => [:regexp, true],
|
Chris@0
|
164 's' => [:symbol, false],
|
Chris@0
|
165 'x' => [:shell, true]
|
Chris@0
|
166 }
|
Chris@0
|
167 FancyStringType['w'] = FancyStringType['q']
|
Chris@0
|
168 FancyStringType['W'] = FancyStringType[''] = FancyStringType['Q']
|
Chris@0
|
169
|
Chris@0
|
170 class StringState < Struct.new :type, :interpreted, :delim, :heredoc,
|
Chris@0
|
171 :paren, :paren_depth, :pattern, :next_state
|
Chris@0
|
172
|
Chris@0
|
173 CLOSING_PAREN = Hash[ *%w[
|
Chris@0
|
174 ( )
|
Chris@0
|
175 [ ]
|
Chris@0
|
176 < >
|
Chris@0
|
177 { }
|
Chris@0
|
178 ] ]
|
Chris@0
|
179
|
Chris@0
|
180 CLOSING_PAREN.each { |k,v| k.freeze; v.freeze } # debug, if I try to change it with <<
|
Chris@0
|
181 OPENING_PAREN = CLOSING_PAREN.invert
|
Chris@0
|
182
|
Chris@0
|
183 STRING_PATTERN = Hash.new do |h, k|
|
Chris@0
|
184 delim, interpreted = *k
|
Chris@0
|
185 delim_pattern = Regexp.escape(delim.dup) # dup: workaround for old Ruby
|
Chris@0
|
186 if closing_paren = CLOSING_PAREN[delim]
|
Chris@0
|
187 delim_pattern = delim_pattern[0..-1] if defined? JRUBY_VERSION # JRuby fix
|
Chris@0
|
188 delim_pattern << Regexp.escape(closing_paren)
|
Chris@0
|
189 end
|
Chris@0
|
190 delim_pattern << '\\\\' unless delim == '\\'
|
Chris@0
|
191
|
Chris@0
|
192 special_escapes =
|
Chris@0
|
193 case interpreted
|
Chris@0
|
194 when :regexp_symbols
|
Chris@0
|
195 '| ' + REGEXP_SYMBOLS.source
|
Chris@0
|
196 when :words
|
Chris@0
|
197 '| \s'
|
Chris@0
|
198 end
|
Chris@0
|
199
|
Chris@0
|
200 h[k] =
|
Chris@0
|
201 if interpreted and not delim == '#'
|
Chris@0
|
202 / (?= [#{delim_pattern}] | \# [{$@] #{special_escapes} ) /mx
|
Chris@0
|
203 else
|
Chris@0
|
204 / (?= [#{delim_pattern}] #{special_escapes} ) /mx
|
Chris@0
|
205 end
|
Chris@0
|
206 end
|
Chris@0
|
207
|
Chris@0
|
208 HEREDOC_PATTERN = Hash.new do |h, k|
|
Chris@0
|
209 delim, interpreted, indented = *k
|
Chris@0
|
210 delim_pattern = Regexp.escape(delim.dup) # dup: workaround for old Ruby
|
Chris@0
|
211 delim_pattern = / \n #{ '(?>[\ \t]*)' if indented } #{ Regexp.new delim_pattern } $ /x
|
Chris@0
|
212 h[k] =
|
Chris@0
|
213 if interpreted
|
Chris@0
|
214 / (?= #{delim_pattern}() | \\ | \# [{$@] ) /mx # $1 set == end of heredoc
|
Chris@0
|
215 else
|
Chris@0
|
216 / (?= #{delim_pattern}() | \\ ) /mx
|
Chris@0
|
217 end
|
Chris@0
|
218 end
|
Chris@0
|
219
|
Chris@0
|
220 def initialize kind, interpreted, delim, heredoc = false
|
Chris@0
|
221 if heredoc
|
Chris@0
|
222 pattern = HEREDOC_PATTERN[ [delim, interpreted, heredoc == :indented] ]
|
Chris@0
|
223 delim = nil
|
Chris@0
|
224 else
|
Chris@0
|
225 pattern = STRING_PATTERN[ [delim, interpreted] ]
|
Chris@0
|
226 if paren = CLOSING_PAREN[delim]
|
Chris@0
|
227 delim, paren = paren, delim
|
Chris@0
|
228 paren_depth = 1
|
Chris@0
|
229 end
|
Chris@0
|
230 end
|
Chris@0
|
231 super kind, interpreted, delim, heredoc, paren, paren_depth, pattern, :initial
|
Chris@0
|
232 end
|
Chris@0
|
233 end unless defined? StringState
|
Chris@0
|
234
|
Chris@0
|
235 end
|
Chris@0
|
236
|
Chris@0
|
237 end
|
Chris@0
|
238 end
|