1<lexer>
2 <config>
3 <name>Python</name>
4 <alias>python</alias>
5 <alias>py</alias>
6 <alias>sage</alias>
7 <alias>python3</alias>
8 <alias>py3</alias>
9 <filename>*.py</filename>
10 <filename>*.pyi</filename>
11 <filename>*.pyw</filename>
12 <filename>*.jy</filename>
13 <filename>*.sage</filename>
14 <filename>*.sc</filename>
15 <filename>SConstruct</filename>
16 <filename>SConscript</filename>
17 <filename>*.bzl</filename>
18 <filename>BUCK</filename>
19 <filename>BUILD</filename>
20 <filename>BUILD.bazel</filename>
21 <filename>WORKSPACE</filename>
22 <filename>*.tac</filename>
23 <mime_type>text/x-python</mime_type>
24 <mime_type>application/x-python</mime_type>
25 <mime_type>text/x-python3</mime_type>
26 <mime_type>application/x-python3</mime_type>
27 </config>
28 <rules>
29 <state name="numbers">
30 <rule pattern="(\d(?:_?\d)*\.(?:\d(?:_?\d)*)?|(?:\d(?:_?\d)*)?\.\d(?:_?\d)*)([eE][+-]?\d(?:_?\d)*)?">
31 <token type="LiteralNumberFloat"/>
32 </rule>
33 <rule pattern="\d(?:_?\d)*[eE][+-]?\d(?:_?\d)*j?">
34 <token type="LiteralNumberFloat"/>
35 </rule>
36 <rule pattern="0[oO](?:_?[0-7])+">
37 <token type="LiteralNumberOct"/>
38 </rule>
39 <rule pattern="0[bB](?:_?[01])+">
40 <token type="LiteralNumberBin"/>
41 </rule>
42 <rule pattern="0[xX](?:_?[a-fA-F0-9])+">
43 <token type="LiteralNumberHex"/>
44 </rule>
45 <rule pattern="\d(?:_?\d)*">
46 <token type="LiteralNumberInteger"/>
47 </rule>
48 </state>
49 <state name="expr">
50 <rule pattern="(?i)(rf|fr)(""")">
51 <bygroups>
52 <token type="LiteralStringAffix"/>
53 <token type="LiteralStringDouble"/>
54 </bygroups>
55 <combined state="rfstringescape" state="tdqf"/>
56 </rule>
57 <rule pattern="(?i)(rf|fr)(''')">
58 <bygroups>
59 <token type="LiteralStringAffix"/>
60 <token type="LiteralStringSingle"/>
61 </bygroups>
62 <combined state="rfstringescape" state="tsqf"/>
63 </rule>
64 <rule pattern="(?i)(rf|fr)(")">
65 <bygroups>
66 <token type="LiteralStringAffix"/>
67 <token type="LiteralStringDouble"/>
68 </bygroups>
69 <combined state="rfstringescape" state="dqf"/>
70 </rule>
71 <rule pattern="(?i)(rf|fr)(')">
72 <bygroups>
73 <token type="LiteralStringAffix"/>
74 <token type="LiteralStringSingle"/>
75 </bygroups>
76 <combined state="rfstringescape" state="sqf"/>
77 </rule>
78 <rule pattern="([fF])(""")">
79 <bygroups>
80 <token type="LiteralStringAffix"/>
81 <token type="LiteralStringDouble"/>
82 </bygroups>
83 <combined state="fstringescape" state="tdqf"/>
84 </rule>
85 <rule pattern="([fF])(''')">
86 <bygroups>
87 <token type="LiteralStringAffix"/>
88 <token type="LiteralStringSingle"/>
89 </bygroups>
90 <combined state="fstringescape" state="tsqf"/>
91 </rule>
92 <rule pattern="([fF])(")">
93 <bygroups>
94 <token type="LiteralStringAffix"/>
95 <token type="LiteralStringDouble"/>
96 </bygroups>
97 <combined state="fstringescape" state="dqf"/>
98 </rule>
99 <rule pattern="([fF])(')">
100 <bygroups>
101 <token type="LiteralStringAffix"/>
102 <token type="LiteralStringSingle"/>
103 </bygroups>
104 <combined state="fstringescape" state="sqf"/>
105 </rule>
106 <rule pattern="(?i)(rb|br|r)(""")">
107 <bygroups>
108 <token type="LiteralStringAffix"/>
109 <token type="LiteralStringDouble"/>
110 </bygroups>
111 <push state="tdqs"/>
112 </rule>
113 <rule pattern="(?i)(rb|br|r)(''')">
114 <bygroups>
115 <token type="LiteralStringAffix"/>
116 <token type="LiteralStringSingle"/>
117 </bygroups>
118 <push state="tsqs"/>
119 </rule>
120 <rule pattern="(?i)(rb|br|r)(")">
121 <bygroups>
122 <token type="LiteralStringAffix"/>
123 <token type="LiteralStringDouble"/>
124 </bygroups>
125 <push state="dqs"/>
126 </rule>
127 <rule pattern="(?i)(rb|br|r)(')">
128 <bygroups>
129 <token type="LiteralStringAffix"/>
130 <token type="LiteralStringSingle"/>
131 </bygroups>
132 <push state="sqs"/>
133 </rule>
134 <rule pattern="([uUbB]?)(""")">
135 <bygroups>
136 <token type="LiteralStringAffix"/>
137 <token type="LiteralStringDouble"/>
138 </bygroups>
139 <combined state="stringescape" state="tdqs"/>
140 </rule>
141 <rule pattern="([uUbB]?)(''')">
142 <bygroups>
143 <token type="LiteralStringAffix"/>
144 <token type="LiteralStringSingle"/>
145 </bygroups>
146 <combined state="stringescape" state="tsqs"/>
147 </rule>
148 <rule pattern="([uUbB]?)(")">
149 <bygroups>
150 <token type="LiteralStringAffix"/>
151 <token type="LiteralStringDouble"/>
152 </bygroups>
153 <combined state="stringescape" state="dqs"/>
154 </rule>
155 <rule pattern="([uUbB]?)(')">
156 <bygroups>
157 <token type="LiteralStringAffix"/>
158 <token type="LiteralStringSingle"/>
159 </bygroups>
160 <combined state="stringescape" state="sqs"/>
161 </rule>
162 <rule pattern="[^\S\n]+">
163 <token type="Text"/>
164 </rule>
165 <rule>
166 <include state="numbers"/>
167 </rule>
168 <rule pattern="!=|==|<<|>>|:=|[-~+/*%=<>&^|.]">
169 <token type="Operator"/>
170 </rule>
171 <rule pattern="[]{}:(),;[]">
172 <token type="Punctuation"/>
173 </rule>
174 <rule pattern="(in|is|and|or|not)\b">
175 <token type="OperatorWord"/>
176 </rule>
177 <rule>
178 <include state="expr-keywords"/>
179 </rule>
180 <rule>
181 <include state="builtins"/>
182 </rule>
183 <rule>
184 <include state="magicfuncs"/>
185 </rule>
186 <rule>
187 <include state="magicvars"/>
188 </rule>
189 <rule>
190 <include state="name"/>
191 </rule>
192 </state>
193 <state name="fstrings-double">
194 <rule pattern="\}">
195 <token type="LiteralStringInterpol"/>
196 </rule>
197 <rule pattern="\{">
198 <token type="LiteralStringInterpol"/>
199 <push state="expr-inside-fstring"/>
200 </rule>
201 <rule pattern="[^\\\'"{}\n]+">
202 <token type="LiteralStringDouble"/>
203 </rule>
204 <rule pattern="[\'"\\]">
205 <token type="LiteralStringDouble"/>
206 </rule>
207 </state>
208 <state name="keywords">
209 <rule pattern="(yield from|nonlocal|continue|finally|except|lambda|assert|global|return|raise|yield|while|break|await|async|pass|else|elif|with|try|for|del|as|if)\b">
210 <token type="Keyword"/>
211 </rule>
212 <rule pattern="(False|True|None)\b">
213 <token type="KeywordConstant"/>
214 </rule>
215 </state>
216 <state name="dqs">
217 <rule pattern=""">
218 <token type="LiteralStringDouble"/>
219 <pop depth="1"/>
220 </rule>
221 <rule pattern="\\\\|\\"|\\\n">
222 <token type="LiteralStringEscape"/>
223 </rule>
224 <rule>
225 <include state="strings-double"/>
226 </rule>
227 </state>
228 <state name="fromimport">
229 <rule pattern="(\s+)(import)\b">
230 <bygroups>
231 <token type="Text"/>
232 <token type="KeywordNamespace"/>
233 </bygroups>
234 <pop depth="1"/>
235 </rule>
236 <rule pattern="\.">
237 <token type="NameNamespace"/>
238 </rule>
239 <rule pattern="None\b">
240 <token type="NameBuiltinPseudo"/>
241 <pop depth="1"/>
242 </rule>
243 <rule pattern="[_\p{L}][_\p{L}\p{N}]*">
244 <token type="NameNamespace"/>
245 </rule>
246 <rule>
247 <pop depth="1"/>
248 </rule>
249 </state>
250 <state name="builtins">
251 <rule pattern="(?<!\.)(staticmethod|classmethod|memoryview|__import__|issubclass|isinstance|frozenset|bytearray|enumerate|reversed|property|compile|complex|delattr|hasattr|setattr|globals|getattr|divmod|filter|locals|format|object|sorted|slice|print|bytes|range|input|tuple|round|super|float|eval|list|dict|repr|type|vars|hash|next|bool|open|iter|oct|pow|min|zip|max|map|bin|len|set|any|dir|all|abs|str|sum|chr|int|hex|ord|id)\b">
252 <token type="NameBuiltin"/>
253 </rule>
254 <rule pattern="(?<!\.)(self|Ellipsis|NotImplemented|cls)\b">
255 <token type="NameBuiltinPseudo"/>
256 </rule>
257 <rule pattern="(?<!\.)(PendingDeprecationWarning|ConnectionAbortedError|ConnectionRefusedError|UnicodeTranslateError|ConnectionResetError|ModuleNotFoundError|NotImplementedError|FloatingPointError|StopAsyncIteration|UnicodeDecodeError|DeprecationWarning|UnicodeEncodeError|NotADirectoryError|ProcessLookupError|ZeroDivisionError|IsADirectoryError|FileNotFoundError|UnboundLocalError|KeyboardInterrupt|ChildProcessError|EnvironmentError|IndentationError|InterruptedError|BlockingIOError|ArithmeticError|ConnectionError|BrokenPipeError|FileExistsError|ResourceWarning|PermissionError|RuntimeWarning|ReferenceError|AttributeError|AssertionError|UnicodeWarning|RecursionError|StopIteration|BaseException|OverflowError|SyntaxWarning|FutureWarning|GeneratorExit|ImportWarning|UnicodeError|TimeoutError|WindowsError|RuntimeError|BytesWarning|SystemError|UserWarning|MemoryError|ImportError|LookupError|BufferError|SyntaxError|SystemExit|ValueError|IndexError|NameError|Exception|TypeError|TabError|EOFError|KeyError|VMSError|Warning|OSError|IOError)\b">
258 <token type="NameException"/>
259 </rule>
260 </state>
261 <state name="classname">
262 <rule pattern="[_\p{L}][_\p{L}\p{N}]*">
263 <token type="NameClass"/>
264 <pop depth="1"/>
265 </rule>
266 </state>
267 <state name="import">
268 <rule pattern="(\s+)(as)(\s+)">
269 <bygroups>
270 <token type="Text"/>
271 <token type="Keyword"/>
272 <token type="Text"/>
273 </bygroups>
274 </rule>
275 <rule pattern="\.">
276 <token type="NameNamespace"/>
277 </rule>
278 <rule pattern="[_\p{L}][_\p{L}\p{N}]*">
279 <token type="NameNamespace"/>
280 </rule>
281 <rule pattern="(\s*)(,)(\s*)">
282 <bygroups>
283 <token type="Text"/>
284 <token type="Operator"/>
285 <token type="Text"/>
286 </bygroups>
287 </rule>
288 <rule>
289 <pop depth="1"/>
290 </rule>
291 </state>
292 <state name="tsqs">
293 <rule pattern="'''">
294 <token type="LiteralStringSingle"/>
295 <pop depth="1"/>
296 </rule>
297 <rule>
298 <include state="strings-single"/>
299 </rule>
300 <rule pattern="\n">
301 <token type="LiteralStringSingle"/>
302 </rule>
303 </state>
304 <state name="strings-double">
305 <rule pattern="%(\(\w+\))?[-#0 +]*([0-9]+|[*])?(\.([0-9]+|[*]))?[hlL]?[E-GXc-giorsaux%]">
306 <token type="LiteralStringInterpol"/>
307 </rule>
308 <rule pattern="\{((\w+)((\.\w+)|(\[[^\]]+\]))*)?(\![sra])?(\:(.?[<>=\^])?[-+ ]?#?0?(\d+)?,?(\.\d+)?[E-GXb-gnosx%]?)?\}">
309 <token type="LiteralStringInterpol"/>
310 </rule>
311 <rule pattern="[^\\\'"%{\n]+">
312 <token type="LiteralStringDouble"/>
313 </rule>
314 <rule pattern="[\'"\\]">
315 <token type="LiteralStringDouble"/>
316 </rule>
317 <rule pattern="%|(\{{1,2})">
318 <token type="LiteralStringDouble"/>
319 </rule>
320 </state>
321 <state name="tdqf">
322 <rule pattern=""""">
323 <token type="LiteralStringDouble"/>
324 <pop depth="1"/>
325 </rule>
326 <rule>
327 <include state="fstrings-double"/>
328 </rule>
329 <rule pattern="\n">
330 <token type="LiteralStringDouble"/>
331 </rule>
332 </state>
333 <state name="expr-inside-fstring-inner">
334 <rule pattern="[{([]">
335 <token type="Punctuation"/>
336 <push state="expr-inside-fstring-inner"/>
337 </rule>
338 <rule pattern="[])}]">
339 <token type="Punctuation"/>
340 <pop depth="1"/>
341 </rule>
342 <rule pattern="\s+">
343 <token type="Text"/>
344 </rule>
345 <rule>
346 <include state="expr"/>
347 </rule>
348 </state>
349 <state name="sqs">
350 <rule pattern="'">
351 <token type="LiteralStringSingle"/>
352 <pop depth="1"/>
353 </rule>
354 <rule pattern="\\\\|\\'|\\\n">
355 <token type="LiteralStringEscape"/>
356 </rule>
357 <rule>
358 <include state="strings-single"/>
359 </rule>
360 </state>
361 <state name="funcname">
362 <rule>
363 <include state="magicfuncs"/>
364 </rule>
365 <rule pattern="[_\p{L}][_\p{L}\p{N}]*">
366 <token type="NameFunction"/>
367 <pop depth="1"/>
368 </rule>
369 <rule>
370 <pop depth="1"/>
371 </rule>
372 </state>
373 <state name="expr-keywords">
374 <rule pattern="(yield from|async for|lambda|yield|await|else|for|if)\b">
375 <token type="Keyword"/>
376 </rule>
377 <rule pattern="(False|True|None)\b">
378 <token type="KeywordConstant"/>
379 </rule>
380 </state>
381 <state name="name">
382 <rule pattern="@[_\p{L}][_\p{L}\p{N}]*">
383 <token type="NameDecorator"/>
384 </rule>
385 <rule pattern="@">
386 <token type="Operator"/>
387 </rule>
388 <rule pattern="[_\p{L}][_\p{L}\p{N}]*">
389 <token type="Name"/>
390 </rule>
391 </state>
392 <state name="magicfuncs">
393 <rule pattern="(__instancecheck__|__subclasscheck__|__getattribute__|__length_hint__|__rfloordiv__|__ifloordiv__|__itruediv__|__contains__|__floordiv__|__rtruediv__|__reversed__|__setitem__|__complex__|__rdivmod__|__delattr__|__rmatmul__|__ilshift__|__prepare__|__delitem__|__rrshift__|__imatmul__|__rlshift__|__setattr__|__truediv__|__getitem__|__missing__|__getattr__|__irshift__|__rshift__|__format__|__invert__|__matmul__|__divmod__|__delete__|__aenter__|__lshift__|__await__|__bytes__|__anext__|__aiter__|__aexit__|__round__|__float__|__enter__|__index__|__iadd__|__ipow__|__rpow__|__iter__|__init__|__ixor__|__rmul__|__rmod__|__imul__|__imod__|__iand__|__hash__|__rsub__|__exit__|__rxor__|__bool__|__call__|__rand__|__next__|__radd__|__isub__|__repr__|__set__|__add__|__new__|__neg__|__xor__|__and__|__mul__|__mod__|__sub__|__len__|__str__|__ror__|__ior__|__pos__|__del__|__get__|__dir__|__abs__|__int__|__pow__|__eq__|__gt__|__le__|__lt__|__ne__|__or__|__ge__)\b">
394 <token type="NameFunctionMagic"/>
395 </rule>
396 </state>
397 <state name="root">
398 <rule pattern="\n">
399 <token type="Text"/>
400 </rule>
401 <rule pattern="^(\s*)([rRuUbB]{,2})("""(?:.|\n)*?""")">
402 <bygroups>
403 <token type="Text"/>
404 <token type="LiteralStringAffix"/>
405 <token type="LiteralStringDoc"/>
406 </bygroups>
407 </rule>
408 <rule pattern="^(\s*)([rRuUbB]{,2})('''(?:.|\n)*?''')">
409 <bygroups>
410 <token type="Text"/>
411 <token type="LiteralStringAffix"/>
412 <token type="LiteralStringDoc"/>
413 </bygroups>
414 </rule>
415 <rule pattern="\A#!.+$">
416 <token type="CommentHashbang"/>
417 </rule>
418 <rule pattern="#.*$">
419 <token type="CommentSingle"/>
420 </rule>
421 <rule pattern="\\\n">
422 <token type="Text"/>
423 </rule>
424 <rule pattern="\\">
425 <token type="Text"/>
426 </rule>
427 <rule>
428 <include state="keywords"/>
429 </rule>
430 <rule pattern="(def)((?:\s|\\\s)+)">
431 <bygroups>
432 <token type="Keyword"/>
433 <token type="Text"/>
434 </bygroups>
435 <push state="funcname"/>
436 </rule>
437 <rule pattern="(class)((?:\s|\\\s)+)">
438 <bygroups>
439 <token type="Keyword"/>
440 <token type="Text"/>
441 </bygroups>
442 <push state="classname"/>
443 </rule>
444 <rule pattern="(from)((?:\s|\\\s)+)">
445 <bygroups>
446 <token type="KeywordNamespace"/>
447 <token type="Text"/>
448 </bygroups>
449 <push state="fromimport"/>
450 </rule>
451 <rule pattern="(import)((?:\s|\\\s)+)">
452 <bygroups>
453 <token type="KeywordNamespace"/>
454 <token type="Text"/>
455 </bygroups>
456 <push state="import"/>
457 </rule>
458 <rule>
459 <include state="expr"/>
460 </rule>
461 </state>
462 <state name="fstrings-single">
463 <rule pattern="\}">
464 <token type="LiteralStringInterpol"/>
465 </rule>
466 <rule pattern="\{">
467 <token type="LiteralStringInterpol"/>
468 <push state="expr-inside-fstring"/>
469 </rule>
470 <rule pattern="[^\\\'"{}\n]+">
471 <token type="LiteralStringSingle"/>
472 </rule>
473 <rule pattern="[\'"\\]">
474 <token type="LiteralStringSingle"/>
475 </rule>
476 </state>
477 <state name="magicvars">
478 <rule pattern="(__annotations__|__kwdefaults__|__qualname__|__objclass__|__defaults__|__closure__|__globals__|__weakref__|__module__|__class__|__bases__|__slots__|__file__|__code__|__name__|__func__|__dict__|__self__|__mro__|__doc__)\b">
479 <token type="NameVariableMagic"/>
480 </rule>
481 </state>
482 <state name="tsqf">
483 <rule pattern="'''">
484 <token type="LiteralStringSingle"/>
485 <pop depth="1"/>
486 </rule>
487 <rule>
488 <include state="fstrings-single"/>
489 </rule>
490 <rule pattern="\n">
491 <token type="LiteralStringSingle"/>
492 </rule>
493 </state>
494 <state name="strings-single">
495 <rule pattern="%(\(\w+\))?[-#0 +]*([0-9]+|[*])?(\.([0-9]+|[*]))?[hlL]?[E-GXc-giorsaux%]">
496 <token type="LiteralStringInterpol"/>
497 </rule>
498 <rule pattern="\{((\w+)((\.\w+)|(\[[^\]]+\]))*)?(\![sra])?(\:(.?[<>=\^])?[-+ ]?#?0?(\d+)?,?(\.\d+)?[E-GXb-gnosx%]?)?\}">
499 <token type="LiteralStringInterpol"/>
500 </rule>
501 <rule pattern="[^\\\'"%{\n]+">
502 <token type="LiteralStringSingle"/>
503 </rule>
504 <rule pattern="[\'"\\]">
505 <token type="LiteralStringSingle"/>
506 </rule>
507 <rule pattern="%|(\{{1,2})">
508 <token type="LiteralStringSingle"/>
509 </rule>
510 </state>
511 <state name="rfstringescape">
512 <rule pattern="\{\{">
513 <token type="LiteralStringEscape"/>
514 </rule>
515 <rule pattern="\}\}">
516 <token type="LiteralStringEscape"/>
517 </rule>
518 </state>
519 <state name="sqf">
520 <rule pattern="'">
521 <token type="LiteralStringSingle"/>
522 <pop depth="1"/>
523 </rule>
524 <rule pattern="\\\\|\\'|\\\n">
525 <token type="LiteralStringEscape"/>
526 </rule>
527 <rule>
528 <include state="fstrings-single"/>
529 </rule>
530 </state>
531 <state name="dqf">
532 <rule pattern=""">
533 <token type="LiteralStringDouble"/>
534 <pop depth="1"/>
535 </rule>
536 <rule pattern="\\\\|\\"|\\\n">
537 <token type="LiteralStringEscape"/>
538 </rule>
539 <rule>
540 <include state="fstrings-double"/>
541 </rule>
542 </state>
543 <state name="expr-inside-fstring">
544 <rule pattern="[{([]">
545 <token type="Punctuation"/>
546 <push state="expr-inside-fstring-inner"/>
547 </rule>
548 <rule pattern="(=\s*)?(\![sraf])?\}">
549 <token type="LiteralStringInterpol"/>
550 <pop depth="1"/>
551 </rule>
552 <rule pattern="(=\s*)?(\![sraf])?:">
553 <token type="LiteralStringInterpol"/>
554 <pop depth="1"/>
555 </rule>
556 <rule pattern="\s+">
557 <token type="Text"/>
558 </rule>
559 <rule>
560 <include state="expr"/>
561 </rule>
562 </state>
563 <state name="tdqs">
564 <rule pattern=""""">
565 <token type="LiteralStringDouble"/>
566 <pop depth="1"/>
567 </rule>
568 <rule>
569 <include state="strings-double"/>
570 </rule>
571 <rule pattern="\n">
572 <token type="LiteralStringDouble"/>
573 </rule>
574 </state>
575 <state name="fstringescape">
576 <rule>
577 <include state="rfstringescape"/>
578 </rule>
579 <rule>
580 <include state="stringescape"/>
581 </rule>
582 </state>
583 <state name="stringescape">
584 <rule pattern="\\([\\abfnrtv"\']|\n|N\{.*?\}|u[a-fA-F0-9]{4}|U[a-fA-F0-9]{8}|x[a-fA-F0-9]{2}|[0-7]{1,3})">
585 <token type="LiteralStringEscape"/>
586 </rule>
587 </state>
588 </rules>
589</lexer>