1"""
2 pygments.lexers.csound
3 ~~~~~~~~~~~~~~~~~~~~~~
4
5 Lexers for Csound languages.
6
7 :copyright: Copyright 2006-2025 by the Pygments team, see AUTHORS.
8 :license: BSD, see LICENSE for details.
9"""
10
11import re
12
13from pygments.lexer import RegexLexer, bygroups, default, include, using, words
14from pygments.token import Comment, Error, Keyword, Name, Number, Operator, Punctuation, \
15 String, Text, Whitespace
16from pygments.lexers._csound_builtins import OPCODES, DEPRECATED_OPCODES, REMOVED_OPCODES
17from pygments.lexers.html import HtmlLexer
18from pygments.lexers.python import PythonLexer
19from pygments.lexers.scripting import LuaLexer
20
21__all__ = ['CsoundScoreLexer', 'CsoundOrchestraLexer', 'CsoundDocumentLexer']
22
23newline = (r'((?:(?:;|//).*)*)(\n)', bygroups(Comment.Single, Text))
24
25
26class CsoundLexer(RegexLexer):
27 url = 'https://csound.com/'
28
29 tokens = {
30 'whitespace': [
31 (r'[ \t]+', Whitespace),
32 (r'/[*](?:.|\n)*?[*]/', Comment.Multiline),
33 (r'(?:;|//).*$', Comment.Single),
34 (r'(\\)(\n)', bygroups(Text, Whitespace))
35 ],
36
37 'preprocessor directives': [
38 (r'#(?:e(?:nd(?:if)?|lse)\b|##)|@@?[ \t]*\d+', Comment.Preproc),
39 (r'#includestr', Comment.Preproc, 'includestr directive'),
40 (r'#include', Comment.Preproc, 'include directive'),
41 (r'#[ \t]*define', Comment.Preproc, 'define directive'),
42 (r'#(?:ifn?def|undef)\b', Comment.Preproc, 'macro directive')
43 ],
44
45 'include directive': [
46 include('whitespace'),
47 (r'([^ \t]).*?\1', String, '#pop')
48 ],
49 'includestr directive': [
50 include('whitespace'),
51 (r'"', String, ('#pop', 'quoted string'))
52 ],
53
54 'define directive': [
55 (r'\n', Whitespace),
56 include('whitespace'),
57 (r'([A-Z_a-z]\w*)(\()', bygroups(Comment.Preproc, Punctuation),
58 ('#pop', 'macro parameter name list')),
59 (r'[A-Z_a-z]\w*', Comment.Preproc, ('#pop', 'before macro body'))
60 ],
61 'macro parameter name list': [
62 include('whitespace'),
63 (r'[A-Z_a-z]\w*', Comment.Preproc),
64 (r"['#]", Punctuation),
65 (r'\)', Punctuation, ('#pop', 'before macro body'))
66 ],
67 'before macro body': [
68 (r'\n', Whitespace),
69 include('whitespace'),
70 (r'#', Punctuation, ('#pop', 'macro body'))
71 ],
72 'macro body': [
73 (r'(?:\\(?!#)|[^#\\]|\n)+', Comment.Preproc),
74 (r'\\#', Comment.Preproc),
75 (r'(?<!\\)#', Punctuation, '#pop')
76 ],
77
78 'macro directive': [
79 include('whitespace'),
80 (r'[A-Z_a-z]\w*', Comment.Preproc, '#pop')
81 ],
82
83 'macro uses': [
84 (r'(\$[A-Z_a-z]\w*\.?)(\()', bygroups(Comment.Preproc, Punctuation),
85 'macro parameter value list'),
86 (r'\$[A-Z_a-z]\w*(?:\.|\b)', Comment.Preproc)
87 ],
88 'macro parameter value list': [
89 (r'(?:[^\'#"{()]|\{(?!\{))+', Comment.Preproc),
90 (r"['#]", Punctuation),
91 (r'"', String, 'macro parameter value quoted string'),
92 (r'\{\{', String, 'macro parameter value braced string'),
93 (r'\(', Comment.Preproc, 'macro parameter value parenthetical'),
94 (r'\)', Punctuation, '#pop')
95 ],
96 'macro parameter value quoted string': [
97 (r"\\[#'()]", Comment.Preproc),
98 (r"[#'()]", Error),
99 include('quoted string')
100 ],
101 'macro parameter value braced string': [
102 (r"\\[#'()]", Comment.Preproc),
103 (r"[#'()]", Error),
104 include('braced string')
105 ],
106 'macro parameter value parenthetical': [
107 (r'(?:[^\\()]|\\\))+', Comment.Preproc),
108 (r'\(', Comment.Preproc, '#push'),
109 (r'\)', Comment.Preproc, '#pop')
110 ],
111
112 'whitespace and macro uses': [
113 include('whitespace'),
114 include('macro uses')
115 ],
116
117 'numbers': [
118 (r'\d+[Ee][+-]?\d+|(\d+\.\d*|\d*\.\d+)([Ee][+-]?\d+)?', Number.Float),
119 (r'(0[Xx])([0-9A-Fa-f]+)', bygroups(Keyword.Type, Number.Hex)),
120 (r'\d+', Number.Integer)
121 ],
122
123 'quoted string': [
124 (r'"', String, '#pop'),
125 (r'[^"$]+', String),
126 include('macro uses'),
127 (r'[$]', String)
128 ],
129
130 'braced string': [
131 # Do nothing. This must be defined in subclasses.
132 ]
133 }
134
135
136class CsoundScoreLexer(CsoundLexer):
137 """
138 For `Csound <https://csound.com>`_ scores.
139 """
140
141 name = 'Csound Score'
142 aliases = ['csound-score', 'csound-sco']
143 filenames = ['*.sco']
144 version_added = '2.1'
145
146 tokens = {
147 'root': [
148 (r'\n', Whitespace),
149 include('whitespace and macro uses'),
150 include('preprocessor directives'),
151
152 (r'[aBbCdefiqstvxy]', Keyword),
153 # There is also a w statement that is generated internally and should not be
154 # used; see https://github.com/csound/csound/issues/750.
155
156 (r'z', Keyword.Constant),
157 # z is a constant equal to 800,000,000,000. 800 billion seconds is about
158 # 25,367.8 years. See also
159 # https://csound.com/docs/manual/ScoreTop.html and
160 # https://github.com/csound/csound/search?q=stof+path%3AEngine+filename%3Asread.c.
161
162 (r'([nNpP][pP])(\d+)', bygroups(Keyword, Number.Integer)),
163
164 (r'[mn]', Keyword, 'mark statement'),
165
166 include('numbers'),
167 (r'[!+\-*/^%&|<>#~.]', Operator),
168 (r'[()\[\]]', Punctuation),
169 (r'"', String, 'quoted string'),
170 (r'\{', Comment.Preproc, 'loop after left brace'),
171 ],
172
173 'mark statement': [
174 include('whitespace and macro uses'),
175 (r'[A-Z_a-z]\w*', Name.Label),
176 (r'\n', Whitespace, '#pop')
177 ],
178
179 'loop after left brace': [
180 include('whitespace and macro uses'),
181 (r'\d+', Number.Integer, ('#pop', 'loop after repeat count')),
182 ],
183 'loop after repeat count': [
184 include('whitespace and macro uses'),
185 (r'[A-Z_a-z]\w*', Comment.Preproc, ('#pop', 'loop'))
186 ],
187 'loop': [
188 (r'\}', Comment.Preproc, '#pop'),
189 include('root')
190 ],
191
192 # Braced strings are not allowed in Csound scores, but this is needed because the
193 # superclass includes it.
194 'braced string': [
195 (r'\}\}', String, '#pop'),
196 (r'[^}]|\}(?!\})', String)
197 ]
198 }
199
200
201class CsoundOrchestraLexer(CsoundLexer):
202 """
203 For `Csound <https://csound.com>`_ orchestras.
204 """
205
206 name = 'Csound Orchestra'
207 aliases = ['csound', 'csound-orc']
208 filenames = ['*.orc', '*.udo']
209 version_added = '2.1'
210
211 user_defined_opcodes = set()
212
213 def opcode_name_callback(lexer, match):
214 opcode = match.group(0)
215 lexer.user_defined_opcodes.add(opcode)
216 yield match.start(), Name.Function, opcode
217
218 def name_callback(lexer, match):
219 type_annotation_token = Keyword.Type
220
221 name = match.group(1)
222 if name in OPCODES or name in DEPRECATED_OPCODES or name in REMOVED_OPCODES:
223 yield match.start(), Name.Builtin, name
224 elif name in lexer.user_defined_opcodes:
225 yield match.start(), Name.Function, name
226 else:
227 type_annotation_token = Name
228 name_match = re.search(r'^(g?[afikSw])(\w+)', name)
229 if name_match:
230 yield name_match.start(1), Keyword.Type, name_match.group(1)
231 yield name_match.start(2), Name, name_match.group(2)
232 else:
233 yield match.start(), Name, name
234
235 if match.group(2):
236 yield match.start(2), Punctuation, match.group(2)
237 yield match.start(3), type_annotation_token, match.group(3)
238
239 tokens = {
240 'root': [
241 (r'\n', Whitespace),
242
243 (r'^([ \t]*)(\w+)(:)([ \t]+|$)', bygroups(Whitespace, Name.Label, Punctuation, Whitespace)),
244
245 include('whitespace and macro uses'),
246 include('preprocessor directives'),
247
248 (r'\binstr\b', Keyword.Declaration, 'instrument numbers and identifiers'),
249 (r'\bopcode\b', Keyword.Declaration, 'after opcode keyword'),
250 (r'\b(?:end(?:in|op))\b', Keyword.Declaration),
251
252 include('partial statements')
253 ],
254
255 'partial statements': [
256 (r'\b(?:0dbfs|A4|k(?:r|smps)|nchnls(?:_i)?|sr)\b', Name.Variable.Global),
257
258 include('numbers'),
259
260 (r'\+=|-=|\*=|/=|<<|>>|<=|>=|==|!=|&&|\|\||[~¬]|[=!+\-*/^%&|<>#?:]', Operator),
261 (r'[(),\[\]]', Punctuation),
262
263 (r'"', String, 'quoted string'),
264 (r'\{\{', String, 'braced string'),
265
266 (words((
267 'do', 'else', 'elseif', 'endif', 'enduntil', 'fi', 'if', 'ithen', 'kthen',
268 'od', 'then', 'until', 'while',
269 ), prefix=r'\b', suffix=r'\b'), Keyword),
270 (words(('return', 'rireturn'), prefix=r'\b', suffix=r'\b'), Keyword.Pseudo),
271
272 (r'\b[ik]?goto\b', Keyword, 'goto label'),
273 (r'\b(r(?:einit|igoto)|tigoto)(\(|\b)', bygroups(Keyword.Pseudo, Punctuation),
274 'goto label'),
275 (r'\b(c(?:g|in?|k|nk?)goto)(\(|\b)', bygroups(Keyword.Pseudo, Punctuation),
276 ('goto label', 'goto argument')),
277 (r'\b(timout)(\(|\b)', bygroups(Keyword.Pseudo, Punctuation),
278 ('goto label', 'goto argument', 'goto argument')),
279 (r'\b(loop_[gl][et])(\(|\b)', bygroups(Keyword.Pseudo, Punctuation),
280 ('goto label', 'goto argument', 'goto argument', 'goto argument')),
281
282 (r'\bprintk?s\b', Name.Builtin, 'prints opcode'),
283 (r'\b(?:readscore|scoreline(?:_i)?)\b', Name.Builtin, 'Csound score opcode'),
284 (r'\bpyl?run[it]?\b', Name.Builtin, 'Python opcode'),
285 (r'\blua_(?:exec|opdef)\b', Name.Builtin, 'Lua opcode'),
286 (r'\bp\d+\b', Name.Variable.Instance),
287 (r'\b([A-Z_a-z]\w*)(?:(:)([A-Za-z]))?\b', name_callback)
288 ],
289
290 'instrument numbers and identifiers': [
291 include('whitespace and macro uses'),
292 (r'\d+|[A-Z_a-z]\w*', Name.Function),
293 (r'[+,]', Punctuation),
294 (r'\n', Whitespace, '#pop')
295 ],
296
297 'after opcode keyword': [
298 include('whitespace and macro uses'),
299 (r'[A-Z_a-z]\w*', opcode_name_callback, ('#pop', 'opcode type signatures')),
300 (r'\n', Whitespace, '#pop')
301 ],
302 'opcode type signatures': [
303 include('whitespace and macro uses'),
304
305 # https://github.com/csound/csound/search?q=XIDENT+path%3AEngine+filename%3Acsound_orc.lex
306 (r'0|[afijkKoOpPStV\[\]]+', Keyword.Type),
307
308 (r',', Punctuation),
309 (r'\n', Whitespace, '#pop')
310 ],
311
312 'quoted string': [
313 (r'"', String, '#pop'),
314 (r'[^\\"$%)]+', String),
315 include('macro uses'),
316 include('escape sequences'),
317 include('format specifiers'),
318 (r'[\\$%)]', String)
319 ],
320 'braced string': [
321 (r'\}\}', String, '#pop'),
322 (r'(?:[^\\%)}]|\}(?!\}))+', String),
323 include('escape sequences'),
324 include('format specifiers'),
325 (r'[\\%)]', String)
326 ],
327 'escape sequences': [
328 # https://github.com/csound/csound/search?q=unquote_string+path%3AEngine+filename%3Acsound_orc_compile.c
329 (r'\\(?:[\\abnrt"]|[0-7]{1,3})', String.Escape)
330 ],
331 # Format specifiers are highlighted in all strings, even though only
332 # fprintks https://csound.com/docs/manual/fprintks.html
333 # fprints https://csound.com/docs/manual/fprints.html
334 # printf/printf_i https://csound.com/docs/manual/printf.html
335 # printks https://csound.com/docs/manual/printks.html
336 # prints https://csound.com/docs/manual/prints.html
337 # sprintf https://csound.com/docs/manual/sprintf.html
338 # sprintfk https://csound.com/docs/manual/sprintfk.html
339 # work with strings that contain format specifiers. In addition, these opcodes’
340 # handling of format specifiers is inconsistent:
341 # - fprintks and fprints accept %a and %A specifiers, and accept %s specifiers
342 # starting in Csound 6.15.0.
343 # - printks and prints accept %a and %A specifiers, but don’t accept %s
344 # specifiers.
345 # - printf, printf_i, sprintf, and sprintfk don’t accept %a and %A specifiers,
346 # but accept %s specifiers.
347 # See https://github.com/csound/csound/issues/747 for more information.
348 'format specifiers': [
349 (r'%[#0\- +]*\d*(?:\.\d+)?[AE-GXac-giosux]', String.Interpol),
350 (r'%%', String.Escape)
351 ],
352
353 'goto argument': [
354 include('whitespace and macro uses'),
355 (r',', Punctuation, '#pop'),
356 include('partial statements')
357 ],
358 'goto label': [
359 include('whitespace and macro uses'),
360 (r'\w+', Name.Label, '#pop'),
361 default('#pop')
362 ],
363
364 'prints opcode': [
365 include('whitespace and macro uses'),
366 (r'"', String, 'prints quoted string'),
367 default('#pop')
368 ],
369 'prints quoted string': [
370 (r'\\\\[aAbBnNrRtT]', String.Escape),
371 (r'%[!nNrRtT]|[~^]{1,2}', String.Escape),
372 include('quoted string')
373 ],
374
375 'Csound score opcode': [
376 include('whitespace and macro uses'),
377 (r'"', String, 'quoted string'),
378 (r'\{\{', String, 'Csound score'),
379 (r'\n', Whitespace, '#pop')
380 ],
381 'Csound score': [
382 (r'\}\}', String, '#pop'),
383 (r'([^}]+)|\}(?!\})', using(CsoundScoreLexer))
384 ],
385
386 'Python opcode': [
387 include('whitespace and macro uses'),
388 (r'"', String, 'quoted string'),
389 (r'\{\{', String, 'Python'),
390 (r'\n', Whitespace, '#pop')
391 ],
392 'Python': [
393 (r'\}\}', String, '#pop'),
394 (r'([^}]+)|\}(?!\})', using(PythonLexer))
395 ],
396
397 'Lua opcode': [
398 include('whitespace and macro uses'),
399 (r'"', String, 'quoted string'),
400 (r'\{\{', String, 'Lua'),
401 (r'\n', Whitespace, '#pop')
402 ],
403 'Lua': [
404 (r'\}\}', String, '#pop'),
405 (r'([^}]+)|\}(?!\})', using(LuaLexer))
406 ]
407 }
408
409
410class CsoundDocumentLexer(RegexLexer):
411 """
412 For Csound documents.
413 """
414
415 name = 'Csound Document'
416 aliases = ['csound-document', 'csound-csd']
417 filenames = ['*.csd']
418 url = 'https://csound.com'
419 version_added = '2.1'
420
421 # These tokens are based on those in XmlLexer in pygments/lexers/html.py. Making
422 # CsoundDocumentLexer a subclass of XmlLexer rather than RegexLexer may seem like a
423 # better idea, since Csound Document files look like XML files. However, Csound
424 # Documents can contain Csound comments (preceded by //, for example) before and
425 # after the root element, unescaped bitwise AND & and less than < operators, etc. In
426 # other words, while Csound Document files look like XML files, they may not actually
427 # be XML files.
428 tokens = {
429 'root': [
430 (r'/[*](.|\n)*?[*]/', Comment.Multiline),
431 (r'(?:;|//).*$', Comment.Single),
432 (r'[^/;<]+|/(?!/)', Text),
433
434 (r'<\s*CsInstruments', Name.Tag, ('orchestra', 'tag')),
435 (r'<\s*CsScore', Name.Tag, ('score', 'tag')),
436 (r'<\s*[Hh][Tt][Mm][Ll]', Name.Tag, ('HTML', 'tag')),
437
438 (r'<\s*[\w:.-]+', Name.Tag, 'tag'),
439 (r'<\s*/\s*[\w:.-]+\s*>', Name.Tag)
440 ],
441
442 'orchestra': [
443 (r'<\s*/\s*CsInstruments\s*>', Name.Tag, '#pop'),
444 (r'(.|\n)+?(?=<\s*/\s*CsInstruments\s*>)', using(CsoundOrchestraLexer))
445 ],
446 'score': [
447 (r'<\s*/\s*CsScore\s*>', Name.Tag, '#pop'),
448 (r'(.|\n)+?(?=<\s*/\s*CsScore\s*>)', using(CsoundScoreLexer))
449 ],
450 'HTML': [
451 (r'<\s*/\s*[Hh][Tt][Mm][Ll]\s*>', Name.Tag, '#pop'),
452 (r'(.|\n)+?(?=<\s*/\s*[Hh][Tt][Mm][Ll]\s*>)', using(HtmlLexer))
453 ],
454
455 'tag': [
456 (r'\s+', Whitespace),
457 (r'[\w.:-]+\s*=', Name.Attribute, 'attr'),
458 (r'/?\s*>', Name.Tag, '#pop')
459 ],
460 'attr': [
461 (r'\s+', Whitespace),
462 (r'".*?"', String, '#pop'),
463 (r"'.*?'", String, '#pop'),
464 (r'[^\s>]+', String, '#pop')
465 ]
466 }