Shortcuts on this page
r m x toggle line displays
j k next/prev highlighted chunk
0 (zero) top of page
1 (one) first highlighted chunk
1"""
2 pygments.lexers.crystal
3 ~~~~~~~~~~~~~~~~~~~~~~~
5 Lexer for Crystal.
7 :copyright: Copyright 2006-2025 by the Pygments team, see AUTHORS.
8 :license: BSD, see LICENSE for details.
9"""
11import re
13from pygments.lexer import ExtendedRegexLexer, include, bygroups, default, \
14 words, line_re
15from pygments.token import Comment, Operator, Keyword, Name, String, Number, \
16 Punctuation, Error, Whitespace
18__all__ = ['CrystalLexer']
21CRYSTAL_OPERATORS = [
22 '!=', '!~', '!', '%', '&&', '&', '**', '*', '+', '-', '/', '<=>', '<<', '<=', '<',
23 '===', '==', '=~', '=', '>=', '>>', '>', '[]=', '[]?', '[]', '^', '||', '|', '~'
24]
27class CrystalLexer(ExtendedRegexLexer):
28 """
29 For Crystal source code.
30 """
32 name = 'Crystal'
33 url = 'https://crystal-lang.org'
34 aliases = ['cr', 'crystal']
35 filenames = ['*.cr']
36 mimetypes = ['text/x-crystal']
37 version_added = '2.2'
39 flags = re.DOTALL | re.MULTILINE
41 def heredoc_callback(self, match, ctx):
42 # okay, this is the hardest part of parsing Crystal...
43 # match: 1 = <<-?, 2 = quote? 3 = name 4 = quote? 5 = rest of line
45 start = match.start(1)
46 yield start, Operator, match.group(1) # <<-?
47 yield match.start(2), String.Heredoc, match.group(2) # quote ", ', `
48 yield match.start(3), String.Delimiter, match.group(3) # heredoc name
49 yield match.start(4), String.Heredoc, match.group(4) # quote again
51 heredocstack = ctx.__dict__.setdefault('heredocstack', [])
52 outermost = not bool(heredocstack)
53 heredocstack.append((match.group(1) == '<<-', match.group(3)))
55 ctx.pos = match.start(5)
56 ctx.end = match.end(5)
57 # this may find other heredocs, so limit the recursion depth
58 if len(heredocstack) < 100:
59 yield from self.get_tokens_unprocessed(context=ctx)
60 else:
61 yield ctx.pos, String.Heredoc, match.group(5)
62 ctx.pos = match.end()
64 if outermost:
65 # this is the outer heredoc again, now we can process them all
66 for tolerant, hdname in heredocstack:
67 lines = []
68 for match in line_re.finditer(ctx.text, ctx.pos):
69 if tolerant:
70 check = match.group().strip()
71 else:
72 check = match.group().rstrip()
73 if check == hdname:
74 for amatch in lines:
75 yield amatch.start(), String.Heredoc, amatch.group()
76 yield match.start(), String.Delimiter, match.group()
77 ctx.pos = match.end()
78 break
79 else:
80 lines.append(match)
81 else:
82 # end of heredoc not found -- error!
83 for amatch in lines:
84 yield amatch.start(), Error, amatch.group()
85 ctx.end = len(ctx.text)
86 del heredocstack[:]
88 def gen_crystalstrings_rules():
89 states = {}
90 states['strings'] = [
91 (r'\:\w+[!?]?', String.Symbol),
92 (words(CRYSTAL_OPERATORS, prefix=r'\:'), String.Symbol),
93 (r":'(\\\\|\\[^\\]|[^'\\])*'", String.Symbol),
94 # This allows arbitrary text after '\ for simplicity
95 (r"'(\\\\|\\'|[^']|\\[^'\\]+)'", String.Char),
96 (r':"', String.Symbol, 'simple-sym'),
97 # Crystal doesn't have "symbol:"s but this simplifies function args
98 (r'([a-zA-Z_]\w*)(:)(?!:)', bygroups(String.Symbol, Punctuation)),
99 (r'"', String.Double, 'simple-string'),
100 (r'(?<!\.)`', String.Backtick, 'simple-backtick'),
101 ]
103 # double-quoted string and symbol
104 for name, ttype, end in ('string', String.Double, '"'), \
105 ('sym', String.Symbol, '"'), \
106 ('backtick', String.Backtick, '`'):
107 states['simple-'+name] = [
108 include('string-escaped' if name == 'sym' else 'string-intp-escaped'),
109 (rf'[^\\{end}#]+', ttype),
110 (r'[\\#]', ttype),
111 (end, ttype, '#pop'),
112 ]
114 # https://crystal-lang.org/docs/syntax_and_semantics/literals/string.html#percent-string-literals
115 for lbrace, rbrace, bracecc, name in \
116 ('\\{', '\\}', '{}', 'cb'), \
117 ('\\[', '\\]', '\\[\\]', 'sb'), \
118 ('\\(', '\\)', '()', 'pa'), \
119 ('<', '>', '<>', 'ab'), \
120 ('\\|', '\\|', '\\|', 'pi'):
121 states[name+'-intp-string'] = [
122 (r'\\' + lbrace, String.Other),
123 ] + (lbrace != rbrace) * [
124 (lbrace, String.Other, '#push'),
125 ] + [
126 (rbrace, String.Other, '#pop'),
127 include('string-intp-escaped'),
128 (r'[\\#' + bracecc + ']', String.Other),
129 (r'[^\\#' + bracecc + ']+', String.Other),
130 ]
131 states['strings'].append((r'%Q?' + lbrace, String.Other,
132 name+'-intp-string'))
133 states[name+'-string'] = [
134 (r'\\[\\' + bracecc + ']', String.Other),
135 ] + (lbrace != rbrace) * [
136 (lbrace, String.Other, '#push'),
137 ] + [
138 (rbrace, String.Other, '#pop'),
139 (r'[\\#' + bracecc + ']', String.Other),
140 (r'[^\\#' + bracecc + ']+', String.Other),
141 ]
142 # https://crystal-lang.org/docs/syntax_and_semantics/literals/array.html#percent-array-literals
143 states['strings'].append((r'%[qwi]' + lbrace, String.Other,
144 name+'-string'))
145 states[name+'-regex'] = [
146 (r'\\[\\' + bracecc + ']', String.Regex),
147 ] + (lbrace != rbrace) * [
148 (lbrace, String.Regex, '#push'),
149 ] + [
150 (rbrace + '[imsx]*', String.Regex, '#pop'),
151 include('string-intp'),
152 (r'[\\#' + bracecc + ']', String.Regex),
153 (r'[^\\#' + bracecc + ']+', String.Regex),
154 ]
155 states['strings'].append((r'%r' + lbrace, String.Regex,
156 name+'-regex'))
158 return states
160 tokens = {
161 'root': [
162 (r'#.*?$', Comment.Single),
163 # keywords
164 (words('''
165 abstract asm begin break case do else elsif end ensure extend if in
166 include next of private protected require rescue return select self super
167 then unless until when while with yield
168 '''.split(), suffix=r'\b'), Keyword),
169 (words('''
170 previous_def forall out uninitialized __DIR__ __FILE__ __LINE__
171 __END_LINE__
172 '''.split(), prefix=r'(?<!\.)', suffix=r'\b'), Keyword.Pseudo),
173 # https://crystal-lang.org/docs/syntax_and_semantics/is_a.html
174 (r'\.(is_a\?|nil\?|responds_to\?|as\?|as\b)', Keyword.Pseudo),
175 (words(['true', 'false', 'nil'], suffix=r'\b'), Keyword.Constant),
176 # start of function, class and module names
177 (r'(module|lib)(\s+)([a-zA-Z_]\w*(?:::[a-zA-Z_]\w*)*)',
178 bygroups(Keyword, Whitespace, Name.Namespace)),
179 (r'(def|fun|macro)(\s+)((?:[a-zA-Z_]\w*::)*)',
180 bygroups(Keyword, Whitespace, Name.Namespace), 'funcname'),
181 (r'def(?=[*%&^`~+-/\[<>=])', Keyword, 'funcname'),
182 (r'(annotation|class|struct|union|type|alias|enum)(\s+)((?:[a-zA-Z_]\w*::)*)',
183 bygroups(Keyword, Whitespace, Name.Namespace), 'classname'),
184 # https://crystal-lang.org/api/toplevel.html
185 (words('''
186 instance_sizeof offsetof pointerof sizeof typeof
187 '''.split(), prefix=r'(?<!\.)', suffix=r'\b'), Keyword.Pseudo),
188 # macros
189 (r'(?<!\.)(debugger\b|p!|pp!|record\b|spawn\b)', Name.Builtin.Pseudo),
190 # builtins
191 (words('''
192 abort at_exit caller exit gets loop main p pp print printf puts
193 raise rand read_line sleep spawn sprintf system
194 '''.split(), prefix=r'(?<!\.)', suffix=r'\b'), Name.Builtin),
195 # https://crystal-lang.org/api/Object.html#macro-summary
196 (r'(?<!\.)(((class_)?((getter|property)\b[!?]?|setter\b))|'
197 r'(def_(clone|equals|equals_and_hash|hash)|delegate|forward_missing_to)\b)',
198 Name.Builtin.Pseudo),
199 # normal heredocs
200 (r'(?<!\w)(<<-?)(["`\']?)([a-zA-Z_]\w*)(\2)(.*?\n)',
201 heredoc_callback),
202 # empty string heredocs
203 (r'(<<-?)("|\')()(\2)(.*?\n)', heredoc_callback),
204 (r'__END__', Comment.Preproc, 'end-part'),
205 # multiline regex (after keywords or assignments)
206 (r'(?:^|(?<=[=<>~!:])|'
207 r'(?<=(?:\s|;)when\s)|'
208 r'(?<=(?:\s|;)or\s)|'
209 r'(?<=(?:\s|;)and\s)|'
210 r'(?<=\.index\s)|'
211 r'(?<=\.scan\s)|'
212 r'(?<=\.sub\s)|'
213 r'(?<=\.sub!\s)|'
214 r'(?<=\.gsub\s)|'
215 r'(?<=\.gsub!\s)|'
216 r'(?<=\.match\s)|'
217 r'(?<=(?:\s|;)if\s)|'
218 r'(?<=(?:\s|;)elsif\s)|'
219 r'(?<=^when\s)|'
220 r'(?<=^index\s)|'
221 r'(?<=^scan\s)|'
222 r'(?<=^sub\s)|'
223 r'(?<=^gsub\s)|'
224 r'(?<=^sub!\s)|'
225 r'(?<=^gsub!\s)|'
226 r'(?<=^match\s)|'
227 r'(?<=^if\s)|'
228 r'(?<=^elsif\s)'
229 r')(\s*)(/)', bygroups(Whitespace, String.Regex), 'multiline-regex'),
230 # multiline regex (in method calls or subscripts)
231 (r'(?<=\(|,|\[)/', String.Regex, 'multiline-regex'),
232 # multiline regex (this time the funny no whitespace rule)
233 (r'(\s+)(/)(?![\s=])', bygroups(Whitespace, String.Regex),
234 'multiline-regex'),
235 # lex numbers and ignore following regular expressions which
236 # are division operators in fact (grrrr. i hate that. any
237 # better ideas?)
238 # since pygments 0.7 we also eat a "?" operator after numbers
239 # so that the char operator does not work. Chars are not allowed
240 # there so that you can use the ternary operator.
241 # stupid example:
242 # x>=0?n[x]:""
243 (r'(0o[0-7]+(?:_[0-7]+)*(?:_?[iu][0-9]+)?)\b(\s*)([/?])?',
244 bygroups(Number.Oct, Whitespace, Operator)),
245 (r'(0x[0-9A-Fa-f]+(?:_[0-9A-Fa-f]+)*(?:_?[iu][0-9]+)?)\b(\s*)([/?])?',
246 bygroups(Number.Hex, Whitespace, Operator)),
247 (r'(0b[01]+(?:_[01]+)*(?:_?[iu][0-9]+)?)\b(\s*)([/?])?',
248 bygroups(Number.Bin, Whitespace, Operator)),
249 # 3 separate expressions for floats because any of the 3 optional
250 # parts makes it a float
251 (r'((?:0(?![0-9])|[1-9][\d_]*)(?:\.\d[\d_]*)(?:e[+-]?[0-9]+)?'
252 r'(?:_?f[0-9]+)?)(\s*)([/?])?',
253 bygroups(Number.Float, Whitespace, Operator)),
254 (r'((?:0(?![0-9])|[1-9][\d_]*)(?:\.\d[\d_]*)?(?:e[+-]?[0-9]+)'
255 r'(?:_?f[0-9]+)?)(\s*)([/?])?',
256 bygroups(Number.Float, Whitespace, Operator)),
257 (r'((?:0(?![0-9])|[1-9][\d_]*)(?:\.\d[\d_]*)?(?:e[+-]?[0-9]+)?'
258 r'(?:_?f[0-9]+))(\s*)([/?])?',
259 bygroups(Number.Float, Whitespace, Operator)),
260 (r'(0\b|[1-9][\d]*(?:_\d+)*(?:_?[iu][0-9]+)?)\b(\s*)([/?])?',
261 bygroups(Number.Integer, Whitespace, Operator)),
262 # Names
263 (r'@@[a-zA-Z_]\w*', Name.Variable.Class),
264 (r'@[a-zA-Z_]\w*', Name.Variable.Instance),
265 (r'\$\w+', Name.Variable.Global),
266 (r'\$[!@&`\'+~=/\\,;.<>_*$?:"^-]', Name.Variable.Global),
267 (r'\$-[0adFiIlpvw]', Name.Variable.Global),
268 (r'::', Operator),
269 include('strings'),
270 # https://crystal-lang.org/reference/syntax_and_semantics/literals/char.html
271 (r'\?(\\[MC]-)*' # modifiers
272 r'(\\([\\abefnrtv#"\']|[0-7]{1,3}|x[a-fA-F0-9]{2}|u[a-fA-F0-9]{4}|u\{[a-fA-F0-9 ]+\})|\S)'
273 r'(?!\w)',
274 String.Char),
275 (r'[A-Z][A-Z_]+\b(?!::|\.)', Name.Constant),
276 # macro expansion
277 (r'\{%', String.Interpol, 'in-macro-control'),
278 (r'\{\{', String.Interpol, 'in-macro-expr'),
279 # annotations
280 (r'(@\[)(\s*)([A-Z]\w*(::[A-Z]\w*)*)',
281 bygroups(Operator, Whitespace, Name.Decorator), 'in-annot'),
282 # this is needed because Crystal attributes can look
283 # like keywords (class) or like this: ` ?!?
284 (words(CRYSTAL_OPERATORS, prefix=r'(\.|::)'),
285 bygroups(Operator, Name.Operator)),
286 (r'(\.|::)([a-zA-Z_]\w*[!?]?|[*%&^`~+\-/\[<>=])',
287 bygroups(Operator, Name)),
288 # Names can end with [!?] unless it's "!="
289 (r'[a-zA-Z_]\w*(?:[!?](?!=))?', Name),
290 (r'(\[|\]\??|\*\*|<=>?|>=|<<?|>>?|=~|===|'
291 r'!~|&&?|\|\||\.{1,3})', Operator),
292 (r'[-+/*%=<>&!^|~]=?', Operator),
293 (r'[(){};,/?:\\]', Punctuation),
294 (r'\s+', Whitespace)
295 ],
296 'funcname': [
297 (r'(?:([a-zA-Z_]\w*)(\.))?'
298 r'([a-zA-Z_]\w*[!?]?|\*\*?|[-+]@?|'
299 r'[/%&|^`~]|\[\]=?|<<|>>|<=?>|>=?|===?)',
300 bygroups(Name.Class, Operator, Name.Function), '#pop'),
301 default('#pop')
302 ],
303 'classname': [
304 (r'[A-Z_]\w*', Name.Class),
305 (r'(\()(\s*)([A-Z_]\w*)(\s*)(\))',
306 bygroups(Punctuation, Whitespace, Name.Class, Whitespace, Punctuation)),
307 default('#pop')
308 ],
309 'in-intp': [
310 (r'\{', String.Interpol, '#push'),
311 (r'\}', String.Interpol, '#pop'),
312 include('root'),
313 ],
314 'string-intp': [
315 (r'#\{', String.Interpol, 'in-intp'),
316 ],
317 'string-escaped': [
318 # https://crystal-lang.org/reference/syntax_and_semantics/literals/string.html
319 (r'\\([\\abefnrtv#"\']|[0-7]{1,3}|x[a-fA-F0-9]{2}|u[a-fA-F0-9]{4}|u\{[a-fA-F0-9 ]+\})',
320 String.Escape)
321 ],
322 'string-intp-escaped': [
323 include('string-intp'),
324 include('string-escaped'),
325 ],
326 'interpolated-regex': [
327 include('string-intp'),
328 (r'[\\#]', String.Regex),
329 (r'[^\\#]+', String.Regex),
330 ],
331 'interpolated-string': [
332 include('string-intp'),
333 (r'[\\#]', String.Other),
334 (r'[^\\#]+', String.Other),
335 ],
336 'multiline-regex': [
337 include('string-intp'),
338 (r'\\\\', String.Regex),
339 (r'\\/', String.Regex),
340 (r'[\\#]', String.Regex),
341 (r'[^\\/#]+', String.Regex),
342 (r'/[imsx]*', String.Regex, '#pop'),
343 ],
344 'end-part': [
345 (r'.+', Comment.Preproc, '#pop')
346 ],
347 'in-macro-control': [
348 (r'\{%', String.Interpol, '#push'),
349 (r'%\}', String.Interpol, '#pop'),
350 (r'(for|verbatim)\b', Keyword),
351 include('root'),
352 ],
353 'in-macro-expr': [
354 (r'\{\{', String.Interpol, '#push'),
355 (r'\}\}', String.Interpol, '#pop'),
356 include('root'),
357 ],
358 'in-annot': [
359 (r'\[', Operator, '#push'),
360 (r'\]', Operator, '#pop'),
361 include('root'),
362 ],
363 }
364 tokens.update(gen_crystalstrings_rules())