Mercurial

comparison third_party/highlight/languages/python.js @ 173:827c6ac504cd hg-web

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Merged in default here.
author MrJuneJune <me@mrjunejune.com>
date Mon, 19 Jan 2026 18:59:10 -0800
parents 2db6253f355d
children
comparison
equal deleted inserted replaced
151:c033667da5f9 173:827c6ac504cd
1 /*! `python` grammar compiled for Highlight.js 11.11.1 */
2 (function(){
3 var hljsGrammar = (function () {
4 'use strict';
5
6 /*
7 Language: Python
8 Description: Python is an interpreted, object-oriented, high-level programming language with dynamic semantics.
9 Website: https://www.python.org
10 Category: common
11 */
12
13 function python(hljs) {
14 const regex = hljs.regex;
15 const IDENT_RE = /[\p{XID_Start}_]\p{XID_Continue}*/u;
16 const RESERVED_WORDS = [
17 'and',
18 'as',
19 'assert',
20 'async',
21 'await',
22 'break',
23 'case',
24 'class',
25 'continue',
26 'def',
27 'del',
28 'elif',
29 'else',
30 'except',
31 'finally',
32 'for',
33 'from',
34 'global',
35 'if',
36 'import',
37 'in',
38 'is',
39 'lambda',
40 'match',
41 'nonlocal|10',
42 'not',
43 'or',
44 'pass',
45 'raise',
46 'return',
47 'try',
48 'while',
49 'with',
50 'yield'
51 ];
52
53 const BUILT_INS = [
54 '__import__',
55 'abs',
56 'all',
57 'any',
58 'ascii',
59 'bin',
60 'bool',
61 'breakpoint',
62 'bytearray',
63 'bytes',
64 'callable',
65 'chr',
66 'classmethod',
67 'compile',
68 'complex',
69 'delattr',
70 'dict',
71 'dir',
72 'divmod',
73 'enumerate',
74 'eval',
75 'exec',
76 'filter',
77 'float',
78 'format',
79 'frozenset',
80 'getattr',
81 'globals',
82 'hasattr',
83 'hash',
84 'help',
85 'hex',
86 'id',
87 'input',
88 'int',
89 'isinstance',
90 'issubclass',
91 'iter',
92 'len',
93 'list',
94 'locals',
95 'map',
96 'max',
97 'memoryview',
98 'min',
99 'next',
100 'object',
101 'oct',
102 'open',
103 'ord',
104 'pow',
105 'print',
106 'property',
107 'range',
108 'repr',
109 'reversed',
110 'round',
111 'set',
112 'setattr',
113 'slice',
114 'sorted',
115 'staticmethod',
116 'str',
117 'sum',
118 'super',
119 'tuple',
120 'type',
121 'vars',
122 'zip'
123 ];
124
125 const LITERALS = [
126 '__debug__',
127 'Ellipsis',
128 'False',
129 'None',
130 'NotImplemented',
131 'True'
132 ];
133
134 // https://docs.python.org/3/library/typing.html
135 // TODO: Could these be supplemented by a CamelCase matcher in certain
136 // contexts, leaving these remaining only for relevance hinting?
137 const TYPES = [
138 "Any",
139 "Callable",
140 "Coroutine",
141 "Dict",
142 "List",
143 "Literal",
144 "Generic",
145 "Optional",
146 "Sequence",
147 "Set",
148 "Tuple",
149 "Type",
150 "Union"
151 ];
152
153 const KEYWORDS = {
154 $pattern: /[A-Za-z]\w+|__\w+__/,
155 keyword: RESERVED_WORDS,
156 built_in: BUILT_INS,
157 literal: LITERALS,
158 type: TYPES
159 };
160
161 const PROMPT = {
162 className: 'meta',
163 begin: /^(>>>|\.\.\.) /
164 };
165
166 const SUBST = {
167 className: 'subst',
168 begin: /\{/,
169 end: /\}/,
170 keywords: KEYWORDS,
171 illegal: /#/
172 };
173
174 const LITERAL_BRACKET = {
175 begin: /\{\{/,
176 relevance: 0
177 };
178
179 const STRING = {
180 className: 'string',
181 contains: [ hljs.BACKSLASH_ESCAPE ],
182 variants: [
183 {
184 begin: /([uU]|[bB]|[rR]|[bB][rR]|[rR][bB])?'''/,
185 end: /'''/,
186 contains: [
187 hljs.BACKSLASH_ESCAPE,
188 PROMPT
189 ],
190 relevance: 10
191 },
192 {
193 begin: /([uU]|[bB]|[rR]|[bB][rR]|[rR][bB])?"""/,
194 end: /"""/,
195 contains: [
196 hljs.BACKSLASH_ESCAPE,
197 PROMPT
198 ],
199 relevance: 10
200 },
201 {
202 begin: /([fF][rR]|[rR][fF]|[fF])'''/,
203 end: /'''/,
204 contains: [
205 hljs.BACKSLASH_ESCAPE,
206 PROMPT,
207 LITERAL_BRACKET,
208 SUBST
209 ]
210 },
211 {
212 begin: /([fF][rR]|[rR][fF]|[fF])"""/,
213 end: /"""/,
214 contains: [
215 hljs.BACKSLASH_ESCAPE,
216 PROMPT,
217 LITERAL_BRACKET,
218 SUBST
219 ]
220 },
221 {
222 begin: /([uU]|[rR])'/,
223 end: /'/,
224 relevance: 10
225 },
226 {
227 begin: /([uU]|[rR])"/,
228 end: /"/,
229 relevance: 10
230 },
231 {
232 begin: /([bB]|[bB][rR]|[rR][bB])'/,
233 end: /'/
234 },
235 {
236 begin: /([bB]|[bB][rR]|[rR][bB])"/,
237 end: /"/
238 },
239 {
240 begin: /([fF][rR]|[rR][fF]|[fF])'/,
241 end: /'/,
242 contains: [
243 hljs.BACKSLASH_ESCAPE,
244 LITERAL_BRACKET,
245 SUBST
246 ]
247 },
248 {
249 begin: /([fF][rR]|[rR][fF]|[fF])"/,
250 end: /"/,
251 contains: [
252 hljs.BACKSLASH_ESCAPE,
253 LITERAL_BRACKET,
254 SUBST
255 ]
256 },
257 hljs.APOS_STRING_MODE,
258 hljs.QUOTE_STRING_MODE
259 ]
260 };
261
262 // https://docs.python.org/3.9/reference/lexical_analysis.html#numeric-literals
263 const digitpart = '[0-9](_?[0-9])*';
264 const pointfloat = `(\\b(${digitpart}))?\\.(${digitpart})|\\b(${digitpart})\\.`;
265 // Whitespace after a number (or any lexical token) is needed only if its absence
266 // would change the tokenization
267 // https://docs.python.org/3.9/reference/lexical_analysis.html#whitespace-between-tokens
268 // We deviate slightly, requiring a word boundary or a keyword
269 // to avoid accidentally recognizing *prefixes* (e.g., `0` in `0x41` or `08` or `0__1`)
270 const lookahead = `\\b|${RESERVED_WORDS.join('|')}`;
271 const NUMBER = {
272 className: 'number',
273 relevance: 0,
274 variants: [
275 // exponentfloat, pointfloat
276 // https://docs.python.org/3.9/reference/lexical_analysis.html#floating-point-literals
277 // optionally imaginary
278 // https://docs.python.org/3.9/reference/lexical_analysis.html#imaginary-literals
279 // Note: no leading \b because floats can start with a decimal point
280 // and we don't want to mishandle e.g. `fn(.5)`,
281 // no trailing \b for pointfloat because it can end with a decimal point
282 // and we don't want to mishandle e.g. `0..hex()`; this should be safe
283 // because both MUST contain a decimal point and so cannot be confused with
284 // the interior part of an identifier
285 {
286 begin: `(\\b(${digitpart})|(${pointfloat}))[eE][+-]?(${digitpart})[jJ]?(?=${lookahead})`
287 },
288 {
289 begin: `(${pointfloat})[jJ]?`
290 },
291
292 // decinteger, bininteger, octinteger, hexinteger
293 // https://docs.python.org/3.9/reference/lexical_analysis.html#integer-literals
294 // optionally "long" in Python 2
295 // https://docs.python.org/2.7/reference/lexical_analysis.html#integer-and-long-integer-literals
296 // decinteger is optionally imaginary
297 // https://docs.python.org/3.9/reference/lexical_analysis.html#imaginary-literals
298 {
299 begin: `\\b([1-9](_?[0-9])*|0+(_?0)*)[lLjJ]?(?=${lookahead})`
300 },
301 {
302 begin: `\\b0[bB](_?[01])+[lL]?(?=${lookahead})`
303 },
304 {
305 begin: `\\b0[oO](_?[0-7])+[lL]?(?=${lookahead})`
306 },
307 {
308 begin: `\\b0[xX](_?[0-9a-fA-F])+[lL]?(?=${lookahead})`
309 },
310
311 // imagnumber (digitpart-based)
312 // https://docs.python.org/3.9/reference/lexical_analysis.html#imaginary-literals
313 {
314 begin: `\\b(${digitpart})[jJ](?=${lookahead})`
315 }
316 ]
317 };
318 const COMMENT_TYPE = {
319 className: "comment",
320 begin: regex.lookahead(/# type:/),
321 end: /$/,
322 keywords: KEYWORDS,
323 contains: [
324 { // prevent keywords from coloring `type`
325 begin: /# type:/
326 },
327 // comment within a datatype comment includes no keywords
328 {
329 begin: /#/,
330 end: /\b\B/,
331 endsWithParent: true
332 }
333 ]
334 };
335 const PARAMS = {
336 className: 'params',
337 variants: [
338 // Exclude params in functions without params
339 {
340 className: "",
341 begin: /\(\s*\)/,
342 skip: true
343 },
344 {
345 begin: /\(/,
346 end: /\)/,
347 excludeBegin: true,
348 excludeEnd: true,
349 keywords: KEYWORDS,
350 contains: [
351 'self',
352 PROMPT,
353 NUMBER,
354 STRING,
355 hljs.HASH_COMMENT_MODE
356 ]
357 }
358 ]
359 };
360 SUBST.contains = [
361 STRING,
362 NUMBER,
363 PROMPT
364 ];
365
366 return {
367 name: 'Python',
368 aliases: [
369 'py',
370 'gyp',
371 'ipython'
372 ],
373 unicodeRegex: true,
374 keywords: KEYWORDS,
375 illegal: /(<\/|\?)|=>/,
376 contains: [
377 PROMPT,
378 NUMBER,
379 {
380 // very common convention
381 scope: 'variable.language',
382 match: /\bself\b/
383 },
384 {
385 // eat "if" prior to string so that it won't accidentally be
386 // labeled as an f-string
387 beginKeywords: "if",
388 relevance: 0
389 },
390 { match: /\bor\b/, scope: "keyword" },
391 STRING,
392 COMMENT_TYPE,
393 hljs.HASH_COMMENT_MODE,
394 {
395 match: [
396 /\bdef/, /\s+/,
397 IDENT_RE,
398 ],
399 scope: {
400 1: "keyword",
401 3: "title.function"
402 },
403 contains: [ PARAMS ]
404 },
405 {
406 variants: [
407 {
408 match: [
409 /\bclass/, /\s+/,
410 IDENT_RE, /\s*/,
411 /\(\s*/, IDENT_RE,/\s*\)/
412 ],
413 },
414 {
415 match: [
416 /\bclass/, /\s+/,
417 IDENT_RE
418 ],
419 }
420 ],
421 scope: {
422 1: "keyword",
423 3: "title.class",
424 6: "title.class.inherited",
425 }
426 },
427 {
428 className: 'meta',
429 begin: /^[\t ]*@/,
430 end: /(?=#)|$/,
431 contains: [
432 NUMBER,
433 PARAMS,
434 STRING
435 ]
436 }
437 ]
438 };
439 }
440
441 return python;
442
443 })();
444
445 hljs.registerLanguage('python', hljsGrammar);
446 })();