Chris@909
|
1 # encoding: utf-8
|
Chris@909
|
2 module CodeRay
|
Chris@909
|
3 module Scanners
|
Chris@909
|
4
|
Chris@909
|
5 # Clojure scanner by Licenser.
|
Chris@909
|
6 class Clojure < Scanner
|
Chris@909
|
7
|
Chris@909
|
8 register_for :clojure
|
Chris@909
|
9 file_extension 'clj'
|
Chris@909
|
10
|
Chris@909
|
11 SPECIAL_FORMS = %w[
|
Chris@909
|
12 def if do let quote var fn loop recur throw try catch monitor-enter monitor-exit .
|
Chris@909
|
13 new
|
Chris@909
|
14 ] # :nodoc:
|
Chris@909
|
15
|
Chris@909
|
16 CORE_FORMS = %w[
|
Chris@909
|
17 + - -> ->> .. / * <= < = == >= > accessor aclone add-classpath add-watch
|
Chris@909
|
18 agent agent-error agent-errors aget alength alias all-ns alter alter-meta!
|
Chris@909
|
19 alter-var-root amap ancestors and apply areduce array-map aset aset-boolean
|
Chris@909
|
20 aset-byte aset-char aset-double aset-float aset-int aset-long aset-short
|
Chris@909
|
21 assert assoc assoc! assoc-in associative? atom await await-for bases bean
|
Chris@909
|
22 bigdec bigint binding bit-and bit-and-not bit-clear bit-flip bit-not bit-or
|
Chris@909
|
23 bit-set bit-shift-left bit-shift-right bit-test bit-xor boolean boolean-array
|
Chris@909
|
24 booleans bound-fn bound-fn* bound? butlast byte byte-array bytes case cast char
|
Chris@909
|
25 char-array char-escape-string char-name-string char? chars class class?
|
Chris@909
|
26 clear-agent-errors clojure-version coll? comment commute comp comparator
|
Chris@909
|
27 compare compare-and-set! compile complement concat cond condp conj conj!
|
Chris@909
|
28 cons constantly construct-proxy contains? count counted? create-ns
|
Chris@909
|
29 create-struct cycle dec decimal? declare definline defmacro defmethod defmulti
|
Chris@909
|
30 defn defn- defonce defprotocol defrecord defstruct deftype delay delay?
|
Chris@909
|
31 deliver denominator deref derive descendants disj disj! dissoc dissoc!
|
Chris@909
|
32 distinct distinct? doall doc dorun doseq dosync dotimes doto double
|
Chris@909
|
33 double-array doubles drop drop-last drop-while empty empty? ensure
|
Chris@909
|
34 enumeration-seq error-handler error-mode eval even? every? extend
|
Chris@909
|
35 extend-protocol extend-type extenders extends? false? ffirst file-seq
|
Chris@909
|
36 filter find find-doc find-ns find-var first float float-array float?
|
Chris@909
|
37 floats flush fn fn? fnext for force format future future-call future-cancel
|
Chris@909
|
38 future-cancelled? future-done? future? gen-class gen-interface gensym get
|
Chris@909
|
39 get-in get-method get-proxy-class get-thread-bindings get-validator hash
|
Chris@909
|
40 hash-map hash-set identical? identity if-let if-not ifn? import in-ns
|
Chris@909
|
41 inc init-proxy instance? int int-array integer? interleave intern
|
Chris@909
|
42 interpose into into-array ints io! isa? iterate iterator-seq juxt key
|
Chris@909
|
43 keys keyword keyword? last lazy-cat lazy-seq let letfn line-seq list list*
|
Chris@909
|
44 list? load load-file load-reader load-string loaded-libs locking long
|
Chris@909
|
45 long-array longs loop macroexpand macroexpand-1 make-array make-hierarchy
|
Chris@909
|
46 map map? mapcat max max-key memfn memoize merge merge-with meta methods
|
Chris@909
|
47 min min-key mod name namespace neg? newline next nfirst nil? nnext not
|
Chris@909
|
48 not-any? not-empty not-every? not= ns ns-aliases ns-imports ns-interns
|
Chris@909
|
49 ns-map ns-name ns-publics ns-refers ns-resolve ns-unalias ns-unmap nth
|
Chris@909
|
50 nthnext num number? numerator object-array odd? or parents partial
|
Chris@909
|
51 partition pcalls peek persistent! pmap pop pop! pop-thread-bindings
|
Chris@909
|
52 pos? pr pr-str prefer-method prefers print print-namespace-doc
|
Chris@909
|
53 print-str printf println println-str prn prn-str promise proxy
|
Chris@909
|
54 proxy-mappings proxy-super push-thread-bindings pvalues quot rand
|
Chris@909
|
55 rand-int range ratio? rationalize re-find re-groups re-matcher
|
Chris@909
|
56 re-matches re-pattern re-seq read read-line read-string reduce ref
|
Chris@909
|
57 ref-history-count ref-max-history ref-min-history ref-set refer
|
Chris@909
|
58 refer-clojure reify release-pending-sends rem remove remove-all-methods
|
Chris@909
|
59 remove-method remove-ns remove-watch repeat repeatedly replace replicate
|
Chris@909
|
60 require reset! reset-meta! resolve rest restart-agent resultset-seq
|
Chris@909
|
61 reverse reversible? rseq rsubseq satisfies? second select-keys send
|
Chris@909
|
62 send-off seq seq? seque sequence sequential? set set-error-handler!
|
Chris@909
|
63 set-error-mode! set-validator! set? short short-array shorts
|
Chris@909
|
64 shutdown-agents slurp some sort sort-by sorted-map sorted-map-by
|
Chris@909
|
65 sorted-set sorted-set-by sorted? special-form-anchor special-symbol?
|
Chris@909
|
66 split-at split-with str string? struct struct-map subs subseq subvec
|
Chris@909
|
67 supers swap! symbol symbol? sync syntax-symbol-anchor take take-last
|
Chris@909
|
68 take-nth take-while test the-ns thread-bound? time to-array to-array-2d
|
Chris@909
|
69 trampoline transient tree-seq true? type unchecked-add unchecked-dec
|
Chris@909
|
70 unchecked-divide unchecked-inc unchecked-multiply unchecked-negate
|
Chris@909
|
71 unchecked-remainder unchecked-subtract underive update-in update-proxy
|
Chris@909
|
72 use val vals var-get var-set var? vary-meta vec vector vector-of vector?
|
Chris@909
|
73 when when-first when-let when-not while with-bindings with-bindings*
|
Chris@909
|
74 with-in-str with-local-vars with-meta with-open with-out-str
|
Chris@909
|
75 with-precision xml-seq zero? zipmap
|
Chris@909
|
76 ] # :nodoc:
|
Chris@909
|
77
|
Chris@909
|
78 PREDEFINED_CONSTANTS = %w[
|
Chris@909
|
79 true false nil *1 *2 *3 *agent* *clojure-version* *command-line-args*
|
Chris@909
|
80 *compile-files* *compile-path* *e *err* *file* *flush-on-newline*
|
Chris@909
|
81 *in* *ns* *out* *print-dup* *print-length* *print-level* *print-meta*
|
Chris@909
|
82 *print-readably* *read-eval* *warn-on-reflection*
|
Chris@909
|
83 ] # :nodoc:
|
Chris@909
|
84
|
Chris@909
|
85 IDENT_KIND = WordList.new(:ident).
|
Chris@909
|
86 add(SPECIAL_FORMS, :keyword).
|
Chris@909
|
87 add(CORE_FORMS, :keyword).
|
Chris@909
|
88 add(PREDEFINED_CONSTANTS, :predefined_constant)
|
Chris@909
|
89
|
Chris@909
|
90 KEYWORD_NEXT_TOKEN_KIND = WordList.new(nil).
|
Chris@909
|
91 add(%w[ def defn defn- definline defmacro defmulti defmethod defstruct defonce declare ], :function).
|
Chris@909
|
92 add(%w[ ns ], :namespace).
|
Chris@909
|
93 add(%w[ defprotocol defrecord ], :class)
|
Chris@909
|
94
|
Chris@909
|
95 BASIC_IDENTIFIER = /[a-zA-Z$%*\/_+!?&<>\-=]=?[a-zA-Z0-9$&*+!\/_?<>\-\#]*/
|
Chris@909
|
96 IDENTIFIER = /(?!-\d)(?:(?:#{BASIC_IDENTIFIER}\.)*#{BASIC_IDENTIFIER}(?:\/#{BASIC_IDENTIFIER})?\.?)|\.\.?/
|
Chris@909
|
97 SYMBOL = /::?#{IDENTIFIER}/o
|
Chris@909
|
98 DIGIT = /\d/
|
Chris@909
|
99 DIGIT10 = DIGIT
|
Chris@909
|
100 DIGIT16 = /[0-9a-f]/i
|
Chris@909
|
101 DIGIT8 = /[0-7]/
|
Chris@909
|
102 DIGIT2 = /[01]/
|
Chris@909
|
103 RADIX16 = /\#x/i
|
Chris@909
|
104 RADIX8 = /\#o/i
|
Chris@909
|
105 RADIX2 = /\#b/i
|
Chris@909
|
106 RADIX10 = /\#d/i
|
Chris@909
|
107 EXACTNESS = /#i|#e/i
|
Chris@909
|
108 SIGN = /[\+-]?/
|
Chris@909
|
109 EXP_MARK = /[esfdl]/i
|
Chris@909
|
110 EXP = /#{EXP_MARK}#{SIGN}#{DIGIT}+/
|
Chris@909
|
111 SUFFIX = /#{EXP}?/
|
Chris@909
|
112 PREFIX10 = /#{RADIX10}?#{EXACTNESS}?|#{EXACTNESS}?#{RADIX10}?/
|
Chris@909
|
113 PREFIX16 = /#{RADIX16}#{EXACTNESS}?|#{EXACTNESS}?#{RADIX16}/
|
Chris@909
|
114 PREFIX8 = /#{RADIX8}#{EXACTNESS}?|#{EXACTNESS}?#{RADIX8}/
|
Chris@909
|
115 PREFIX2 = /#{RADIX2}#{EXACTNESS}?|#{EXACTNESS}?#{RADIX2}/
|
Chris@909
|
116 UINT10 = /#{DIGIT10}+#*/
|
Chris@909
|
117 UINT16 = /#{DIGIT16}+#*/
|
Chris@909
|
118 UINT8 = /#{DIGIT8}+#*/
|
Chris@909
|
119 UINT2 = /#{DIGIT2}+#*/
|
Chris@909
|
120 DECIMAL = /#{DIGIT10}+#+\.#*#{SUFFIX}|#{DIGIT10}+\.#{DIGIT10}*#*#{SUFFIX}|\.#{DIGIT10}+#*#{SUFFIX}|#{UINT10}#{EXP}/
|
Chris@909
|
121 UREAL10 = /#{UINT10}\/#{UINT10}|#{DECIMAL}|#{UINT10}/
|
Chris@909
|
122 UREAL16 = /#{UINT16}\/#{UINT16}|#{UINT16}/
|
Chris@909
|
123 UREAL8 = /#{UINT8}\/#{UINT8}|#{UINT8}/
|
Chris@909
|
124 UREAL2 = /#{UINT2}\/#{UINT2}|#{UINT2}/
|
Chris@909
|
125 REAL10 = /#{SIGN}#{UREAL10}/
|
Chris@909
|
126 REAL16 = /#{SIGN}#{UREAL16}/
|
Chris@909
|
127 REAL8 = /#{SIGN}#{UREAL8}/
|
Chris@909
|
128 REAL2 = /#{SIGN}#{UREAL2}/
|
Chris@909
|
129 IMAG10 = /i|#{UREAL10}i/
|
Chris@909
|
130 IMAG16 = /i|#{UREAL16}i/
|
Chris@909
|
131 IMAG8 = /i|#{UREAL8}i/
|
Chris@909
|
132 IMAG2 = /i|#{UREAL2}i/
|
Chris@909
|
133 COMPLEX10 = /#{REAL10}@#{REAL10}|#{REAL10}\+#{IMAG10}|#{REAL10}-#{IMAG10}|\+#{IMAG10}|-#{IMAG10}|#{REAL10}/
|
Chris@909
|
134 COMPLEX16 = /#{REAL16}@#{REAL16}|#{REAL16}\+#{IMAG16}|#{REAL16}-#{IMAG16}|\+#{IMAG16}|-#{IMAG16}|#{REAL16}/
|
Chris@909
|
135 COMPLEX8 = /#{REAL8}@#{REAL8}|#{REAL8}\+#{IMAG8}|#{REAL8}-#{IMAG8}|\+#{IMAG8}|-#{IMAG8}|#{REAL8}/
|
Chris@909
|
136 COMPLEX2 = /#{REAL2}@#{REAL2}|#{REAL2}\+#{IMAG2}|#{REAL2}-#{IMAG2}|\+#{IMAG2}|-#{IMAG2}|#{REAL2}/
|
Chris@909
|
137 NUM10 = /#{PREFIX10}?#{COMPLEX10}/
|
Chris@909
|
138 NUM16 = /#{PREFIX16}#{COMPLEX16}/
|
Chris@909
|
139 NUM8 = /#{PREFIX8}#{COMPLEX8}/
|
Chris@909
|
140 NUM2 = /#{PREFIX2}#{COMPLEX2}/
|
Chris@909
|
141 NUM = /#{NUM10}|#{NUM16}|#{NUM8}|#{NUM2}/
|
Chris@909
|
142
|
Chris@909
|
143 protected
|
Chris@909
|
144
|
Chris@909
|
145 def scan_tokens encoder, options
|
Chris@909
|
146
|
Chris@909
|
147 state = :initial
|
Chris@909
|
148 kind = nil
|
Chris@909
|
149
|
Chris@909
|
150 until eos?
|
Chris@909
|
151
|
Chris@909
|
152 case state
|
Chris@909
|
153 when :initial
|
Chris@909
|
154 if match = scan(/ \s+ | \\\n | , /x)
|
Chris@909
|
155 encoder.text_token match, :space
|
Chris@909
|
156 elsif match = scan(/['`\(\[\)\]\{\}]|\#[({]|~@?|[@\^]/)
|
Chris@909
|
157 encoder.text_token match, :operator
|
Chris@909
|
158 elsif match = scan(/;.*/)
|
Chris@909
|
159 encoder.text_token match, :comment # TODO: recognize (comment ...) too
|
Chris@909
|
160 elsif match = scan(/\#?\\(?:newline|space|.?)/)
|
Chris@909
|
161 encoder.text_token match, :char
|
Chris@909
|
162 elsif match = scan(/\#[ft]/)
|
Chris@909
|
163 encoder.text_token match, :predefined_constant
|
Chris@909
|
164 elsif match = scan(/#{IDENTIFIER}/o)
|
Chris@909
|
165 kind = IDENT_KIND[match]
|
Chris@909
|
166 encoder.text_token match, kind
|
Chris@909
|
167 if rest? && kind == :keyword
|
Chris@909
|
168 if kind = KEYWORD_NEXT_TOKEN_KIND[match]
|
Chris@909
|
169 encoder.text_token match, :space if match = scan(/\s+/o)
|
Chris@909
|
170 encoder.text_token match, kind if match = scan(/#{IDENTIFIER}/o)
|
Chris@909
|
171 end
|
Chris@909
|
172 end
|
Chris@909
|
173 elsif match = scan(/#{SYMBOL}/o)
|
Chris@909
|
174 encoder.text_token match, :symbol
|
Chris@909
|
175 elsif match = scan(/\./)
|
Chris@909
|
176 encoder.text_token match, :operator
|
Chris@909
|
177 elsif match = scan(/ \# \^ #{IDENTIFIER} /ox)
|
Chris@909
|
178 encoder.text_token match, :type
|
Chris@909
|
179 elsif match = scan(/ (\#)? " /x)
|
Chris@909
|
180 state = self[1] ? :regexp : :string
|
Chris@909
|
181 encoder.begin_group state
|
Chris@909
|
182 encoder.text_token match, :delimiter
|
Chris@909
|
183 elsif match = scan(/#{NUM}/o) and not matched.empty?
|
Chris@909
|
184 encoder.text_token match, match[/[.e\/]/i] ? :float : :integer
|
Chris@909
|
185 else
|
Chris@909
|
186 encoder.text_token getch, :error
|
Chris@909
|
187 end
|
Chris@909
|
188
|
Chris@909
|
189 when :string, :regexp
|
Chris@909
|
190 if match = scan(/[^"\\]+|\\.?/)
|
Chris@909
|
191 encoder.text_token match, :content
|
Chris@909
|
192 elsif match = scan(/"/)
|
Chris@909
|
193 encoder.text_token match, :delimiter
|
Chris@909
|
194 encoder.end_group state
|
Chris@909
|
195 state = :initial
|
Chris@909
|
196 else
|
Chris@909
|
197 raise_inspect "else case \" reached; %p not handled." % peek(1),
|
Chris@909
|
198 encoder, state
|
Chris@909
|
199 end
|
Chris@909
|
200
|
Chris@909
|
201 else
|
Chris@909
|
202 raise 'else case reached'
|
Chris@909
|
203
|
Chris@909
|
204 end
|
Chris@909
|
205
|
Chris@909
|
206 end
|
Chris@909
|
207
|
Chris@909
|
208 if [:string, :regexp].include? state
|
Chris@909
|
209 encoder.end_group state
|
Chris@909
|
210 end
|
Chris@909
|
211
|
Chris@909
|
212 encoder
|
Chris@909
|
213
|
Chris@909
|
214 end
|
Chris@909
|
215 end
|
Chris@909
|
216 end
|
Chris@909
|
217 end |