|
1 # -*- coding: utf-8 -*- |
|
2 """ |
|
3 pygments.lexers.prolog |
|
4 ~~~~~~~~~~~~~~~~~~~~~~ |
|
5 |
|
6 Lexers for Prolog and Prolog-like languages. |
|
7 |
|
8 :copyright: Copyright 2006-2014 by the Pygments team, see AUTHORS. |
|
9 :license: BSD, see LICENSE for details. |
|
10 """ |
|
11 |
|
12 import re |
|
13 |
|
14 from pygments.lexer import RegexLexer, bygroups |
|
15 from pygments.token import Text, Comment, Operator, Keyword, Name, String, \ |
|
16 Number, Punctuation |
|
17 |
|
18 __all__ = ['PrologLexer', 'LogtalkLexer'] |
|
19 |
|
20 |
|
21 class PrologLexer(RegexLexer): |
|
22 """ |
|
23 Lexer for Prolog files. |
|
24 """ |
|
25 name = 'Prolog' |
|
26 aliases = ['prolog'] |
|
27 filenames = ['*.ecl', '*.prolog', '*.pro', '*.pl'] |
|
28 mimetypes = ['text/x-prolog'] |
|
29 |
|
30 flags = re.UNICODE | re.MULTILINE |
|
31 |
|
32 tokens = { |
|
33 'root': [ |
|
34 (r'^#.*', Comment.Single), |
|
35 (r'/\*', Comment.Multiline, 'nested-comment'), |
|
36 (r'%.*', Comment.Single), |
|
37 # character literal |
|
38 (r'0\'.', String.Char), |
|
39 (r'0b[01]+', Number.Bin), |
|
40 (r'0o[0-7]+', Number.Oct), |
|
41 (r'0x[0-9a-fA-F]+', Number.Hex), |
|
42 # literal with prepended base |
|
43 (r'\d\d?\'[a-zA-Z0-9]+', Number.Integer), |
|
44 (r'(\d+\.\d*|\d*\.\d+)([eE][+-]?[0-9]+)?', Number.Float), |
|
45 (r'\d+', Number.Integer), |
|
46 (r'[\[\](){}|.,;!]', Punctuation), |
|
47 (r':-|-->', Punctuation), |
|
48 (r'"(?:\\x[0-9a-fA-F]+\\|\\u[0-9a-fA-F]{4}|\\U[0-9a-fA-F]{8}|' |
|
49 r'\\[0-7]+\\|\\["\nabcefnrstv]|[^\\"])*"', String.Double), |
|
50 (r"'(?:''|[^'])*'", String.Atom), # quoted atom |
|
51 # Needs to not be followed by an atom. |
|
52 # (r'=(?=\s|[a-zA-Z\[])', Operator), |
|
53 (r'is\b', Operator), |
|
54 (r'(<|>|=<|>=|==|=:=|=|/|//|\*|\+|-)(?=\s|[a-zA-Z0-9\[])', |
|
55 Operator), |
|
56 (r'(mod|div|not)\b', Operator), |
|
57 (r'_', Keyword), # The don't-care variable |
|
58 (r'([a-z]+)(:)', bygroups(Name.Namespace, Punctuation)), |
|
59 (u'([a-z\u00c0-\u1fff\u3040-\ud7ff\ue000-\uffef]' |
|
60 u'[\w$\u00c0-\u1fff\u3040-\ud7ff\ue000-\uffef]*)' |
|
61 u'(\\s*)(:-|-->)', |
|
62 bygroups(Name.Function, Text, Operator)), # function defn |
|
63 (u'([a-z\u00c0-\u1fff\u3040-\ud7ff\ue000-\uffef]' |
|
64 u'[\w$\u00c0-\u1fff\u3040-\ud7ff\ue000-\uffef]*)' |
|
65 u'(\\s*)(\\()', |
|
66 bygroups(Name.Function, Text, Punctuation)), |
|
67 (u'[a-z\u00c0-\u1fff\u3040-\ud7ff\ue000-\uffef]' |
|
68 u'[\w$\u00c0-\u1fff\u3040-\ud7ff\ue000-\uffef]*', |
|
69 String.Atom), # atom, characters |
|
70 # This one includes ! |
|
71 (u'[#&*+\\-./:<=>?@\\\\^~\u00a1-\u00bf\u2010-\u303f]+', |
|
72 String.Atom), # atom, graphics |
|
73 (r'[A-Z_]\w*', Name.Variable), |
|
74 (u'\\s+|[\u2000-\u200f\ufff0-\ufffe\uffef]', Text), |
|
75 ], |
|
76 'nested-comment': [ |
|
77 (r'\*/', Comment.Multiline, '#pop'), |
|
78 (r'/\*', Comment.Multiline, '#push'), |
|
79 (r'[^*/]+', Comment.Multiline), |
|
80 (r'[*/]', Comment.Multiline), |
|
81 ], |
|
82 } |
|
83 |
|
84 def analyse_text(text): |
|
85 return ':-' in text |
|
86 |
|
87 |
|
88 class LogtalkLexer(RegexLexer): |
|
89 """ |
|
90 For `Logtalk <http://logtalk.org/>`_ source code. |
|
91 |
|
92 .. versionadded:: 0.10 |
|
93 """ |
|
94 |
|
95 name = 'Logtalk' |
|
96 aliases = ['logtalk'] |
|
97 filenames = ['*.lgt', '*.logtalk'] |
|
98 mimetypes = ['text/x-logtalk'] |
|
99 |
|
100 tokens = { |
|
101 'root': [ |
|
102 # Directives |
|
103 (r'^\s*:-\s', Punctuation, 'directive'), |
|
104 # Comments |
|
105 (r'%.*?\n', Comment), |
|
106 (r'/\*(.|\n)*?\*/', Comment), |
|
107 # Whitespace |
|
108 (r'\n', Text), |
|
109 (r'\s+', Text), |
|
110 # Numbers |
|
111 (r"0'.", Number), |
|
112 (r'0b[01]+', Number.Bin), |
|
113 (r'0o[0-7]+', Number.Oct), |
|
114 (r'0x[0-9a-fA-F]+', Number.Hex), |
|
115 (r'\d+\.?\d*((e|E)(\+|-)?\d+)?', Number), |
|
116 # Variables |
|
117 (r'([A-Z_]\w*)', Name.Variable), |
|
118 # Event handlers |
|
119 (r'(after|before)(?=[(])', Keyword), |
|
120 # Message forwarding handler |
|
121 (r'forward(?=[(])', Keyword), |
|
122 # Execution-context methods |
|
123 (r'(parameter|this|se(lf|nder))(?=[(])', Keyword), |
|
124 # Reflection |
|
125 (r'(current_predicate|predicate_property)(?=[(])', Keyword), |
|
126 # DCGs and term expansion |
|
127 (r'(expand_(goal|term)|(goal|term)_expansion|phrase)(?=[(])', Keyword), |
|
128 # Entity |
|
129 (r'(abolish|c(reate|urrent))_(object|protocol|category)(?=[(])', Keyword), |
|
130 (r'(object|protocol|category)_property(?=[(])', Keyword), |
|
131 # Entity relations |
|
132 (r'co(mplements_object|nforms_to_protocol)(?=[(])', Keyword), |
|
133 (r'extends_(object|protocol|category)(?=[(])', Keyword), |
|
134 (r'imp(lements_protocol|orts_category)(?=[(])', Keyword), |
|
135 (r'(instantiat|specializ)es_class(?=[(])', Keyword), |
|
136 # Events |
|
137 (r'(current_event|(abolish|define)_events)(?=[(])', Keyword), |
|
138 # Flags |
|
139 (r'(current|set)_logtalk_flag(?=[(])', Keyword), |
|
140 # Compiling, loading, and library paths |
|
141 (r'logtalk_(compile|l(ibrary_path|oad|oad_context)|make)(?=[(])', Keyword), |
|
142 (r'\blogtalk_make\b', Keyword), |
|
143 # Database |
|
144 (r'(clause|retract(all)?)(?=[(])', Keyword), |
|
145 (r'a(bolish|ssert(a|z))(?=[(])', Keyword), |
|
146 # Control constructs |
|
147 (r'(ca(ll|tch)|throw)(?=[(])', Keyword), |
|
148 (r'(fa(il|lse)|true)\b', Keyword), |
|
149 # All solutions |
|
150 (r'((bag|set)of|f(ind|or)all)(?=[(])', Keyword), |
|
151 # Multi-threading meta-predicates |
|
152 (r'threaded(_(call|once|ignore|exit|peek|wait|notify))?(?=[(])', Keyword), |
|
153 # Term unification |
|
154 (r'(subsumes_term|unify_with_occurs_check)(?=[(])', Keyword), |
|
155 # Term creation and decomposition |
|
156 (r'(functor|arg|copy_term|numbervars|term_variables)(?=[(])', Keyword), |
|
157 # Evaluable functors |
|
158 (r'(rem|m(ax|in|od)|abs|sign)(?=[(])', Keyword), |
|
159 (r'float(_(integer|fractional)_part)?(?=[(])', Keyword), |
|
160 (r'(floor|truncate|round|ceiling)(?=[(])', Keyword), |
|
161 # Other arithmetic functors |
|
162 (r'(cos|a(cos|sin|tan)|exp|log|s(in|qrt))(?=[(])', Keyword), |
|
163 # Term testing |
|
164 (r'(var|atom(ic)?|integer|float|c(allable|ompound)|n(onvar|umber)|' |
|
165 r'ground|acyclic_term)(?=[(])', Keyword), |
|
166 # Term comparison |
|
167 (r'compare(?=[(])', Keyword), |
|
168 # Stream selection and control |
|
169 (r'(curren|se)t_(in|out)put(?=[(])', Keyword), |
|
170 (r'(open|close)(?=[(])', Keyword), |
|
171 (r'flush_output(?=[(])', Keyword), |
|
172 (r'(at_end_of_stream|flush_output)\b', Keyword), |
|
173 (r'(stream_property|at_end_of_stream|set_stream_position)(?=[(])', Keyword), |
|
174 # Character and byte input/output |
|
175 (r'(nl|(get|peek|put)_(byte|c(har|ode)))(?=[(])', Keyword), |
|
176 (r'\bnl\b', Keyword), |
|
177 # Term input/output |
|
178 (r'read(_term)?(?=[(])', Keyword), |
|
179 (r'write(q|_(canonical|term))?(?=[(])', Keyword), |
|
180 (r'(current_)?op(?=[(])', Keyword), |
|
181 (r'(current_)?char_conversion(?=[(])', Keyword), |
|
182 # Atomic term processing |
|
183 (r'atom_(length|c(hars|o(ncat|des)))(?=[(])', Keyword), |
|
184 (r'(char_code|sub_atom)(?=[(])', Keyword), |
|
185 (r'number_c(har|ode)s(?=[(])', Keyword), |
|
186 # Implementation defined hooks functions |
|
187 (r'(se|curren)t_prolog_flag(?=[(])', Keyword), |
|
188 (r'\bhalt\b', Keyword), |
|
189 (r'halt(?=[(])', Keyword), |
|
190 # Message sending operators |
|
191 (r'(::|:|\^\^)', Operator), |
|
192 # External call |
|
193 (r'[{}]', Keyword), |
|
194 # Logic and control |
|
195 (r'(ignore|once)(?=[(])', Keyword), |
|
196 (r'\brepeat\b', Keyword), |
|
197 # Sorting |
|
198 (r'(key)?sort(?=[(])', Keyword), |
|
199 # Bitwise functors |
|
200 (r'(>>|<<|/\\|\\\\|\\)', Operator), |
|
201 # Predicate aliases |
|
202 (r'\bas\b', Operator), |
|
203 # Arithemtic evaluation |
|
204 (r'\bis\b', Keyword), |
|
205 # Arithemtic comparison |
|
206 (r'(=:=|=\\=|<|=<|>=|>)', Operator), |
|
207 # Term creation and decomposition |
|
208 (r'=\.\.', Operator), |
|
209 # Term unification |
|
210 (r'(=|\\=)', Operator), |
|
211 # Term comparison |
|
212 (r'(==|\\==|@=<|@<|@>=|@>)', Operator), |
|
213 # Evaluable functors |
|
214 (r'(//|[-+*/])', Operator), |
|
215 (r'\b(e|pi|mod|rem)\b', Operator), |
|
216 # Other arithemtic functors |
|
217 (r'\b\*\*\b', Operator), |
|
218 # DCG rules |
|
219 (r'-->', Operator), |
|
220 # Control constructs |
|
221 (r'([!;]|->)', Operator), |
|
222 # Logic and control |
|
223 (r'\\+', Operator), |
|
224 # Mode operators |
|
225 (r'[?@]', Operator), |
|
226 # Existential quantifier |
|
227 (r'\^', Operator), |
|
228 # Strings |
|
229 (r'"(\\\\|\\"|[^"])*"', String), |
|
230 # Ponctuation |
|
231 (r'[()\[\],.|]', Text), |
|
232 # Atoms |
|
233 (r"[a-z]\w*", Text), |
|
234 (r"'", String, 'quoted_atom'), |
|
235 ], |
|
236 |
|
237 'quoted_atom': [ |
|
238 (r"''", String), |
|
239 (r"'", String, '#pop'), |
|
240 (r'\\([\\abfnrtv"\']|(x[a-fA-F0-9]+|[0-7]+)\\)', String.Escape), |
|
241 (r"[^\\'\n]+", String), |
|
242 (r'\\', String), |
|
243 ], |
|
244 |
|
245 'directive': [ |
|
246 # Conditional compilation directives |
|
247 (r'(el)?if(?=[(])', Keyword, 'root'), |
|
248 (r'(e(lse|ndif))[.]', Keyword, 'root'), |
|
249 # Entity directives |
|
250 (r'(category|object|protocol)(?=[(])', Keyword, 'entityrelations'), |
|
251 (r'(end_(category|object|protocol))[.]', Keyword, 'root'), |
|
252 # Predicate scope directives |
|
253 (r'(public|protected|private)(?=[(])', Keyword, 'root'), |
|
254 # Other directives |
|
255 (r'e(n(coding|sure_loaded)|xport)(?=[(])', Keyword, 'root'), |
|
256 (r'in(clude|itialization|fo)(?=[(])', Keyword, 'root'), |
|
257 (r'(built_in|dynamic|synchronized|threaded)[.]', Keyword, 'root'), |
|
258 (r'(alias|d(ynamic|iscontiguous)|m(eta_(non_terminal|predicate)|ode|ultifile)|' |
|
259 r's(et_(logtalk|prolog)_flag|ynchronized))(?=[(])', Keyword, 'root'), |
|
260 (r'op(?=[(])', Keyword, 'root'), |
|
261 (r'(c(alls|oinductive)|module|reexport|use(s|_module))(?=[(])', Keyword, 'root'), |
|
262 (r'[a-z]\w*(?=[(])', Text, 'root'), |
|
263 (r'[a-z]\w*[.]', Text, 'root'), |
|
264 ], |
|
265 |
|
266 'entityrelations': [ |
|
267 (r'(complements|extends|i(nstantiates|mp(lements|orts))|specializes)(?=[(])', Keyword), |
|
268 # Numbers |
|
269 (r"0'.", Number), |
|
270 (r'0b[01]+', Number.Bin), |
|
271 (r'0o[0-7]+', Number.Oct), |
|
272 (r'0x[0-9a-fA-F]+', Number.Hex), |
|
273 (r'\d+\.?\d*((e|E)(\+|-)?\d+)?', Number), |
|
274 # Variables |
|
275 (r'([A-Z_]\w*)', Name.Variable), |
|
276 # Atoms |
|
277 (r"[a-z]\w*", Text), |
|
278 (r"'", String, 'quoted_atom'), |
|
279 # Strings |
|
280 (r'"(\\\\|\\"|[^"])*"', String), |
|
281 # End of entity-opening directive |
|
282 (r'([)]\.)', Text, 'root'), |
|
283 # Scope operator |
|
284 (r'(::)', Operator), |
|
285 # Ponctuation |
|
286 (r'[()\[\],.|]', Text), |
|
287 # Comments |
|
288 (r'%.*?\n', Comment), |
|
289 (r'/\*(.|\n)*?\*/', Comment), |
|
290 # Whitespace |
|
291 (r'\n', Text), |
|
292 (r'\s+', Text), |
|
293 ] |
|
294 } |
|
295 |
|
296 def analyse_text(text): |
|
297 if ':- object(' in text: |
|
298 return 1.0 |
|
299 elif ':- protocol(' in text: |
|
300 return 1.0 |
|
301 elif ':- category(' in text: |
|
302 return 1.0 |
|
303 elif re.search('^:-\s[a-z]', text, re.M): |
|
304 return 0.9 |
|
305 else: |
|
306 return 0.0 |