1 # -*- coding: utf-8 -*- |
|
2 """ |
|
3 pygments.lexers.rust |
|
4 ~~~~~~~~~~~~~~~~~~~~ |
|
5 |
|
6 Lexers for the Rust language. |
|
7 |
|
8 :copyright: Copyright 2006-2021 by the Pygments team, see AUTHORS. |
|
9 :license: BSD, see LICENSE for details. |
|
10 """ |
|
11 |
|
12 from pygments.lexer import RegexLexer, include, bygroups, words, default |
|
13 from pygments.token import Text, Comment, Operator, Keyword, Name, String, \ |
|
14 Number, Punctuation, Whitespace |
|
15 |
|
16 __all__ = ['RustLexer'] |
|
17 |
|
18 |
|
19 class RustLexer(RegexLexer): |
|
20 """ |
|
21 Lexer for the Rust programming language (version 1.47). |
|
22 |
|
23 .. versionadded:: 1.6 |
|
24 """ |
|
25 name = 'Rust' |
|
26 filenames = ['*.rs', '*.rs.in'] |
|
27 aliases = ['rust', 'rs'] |
|
28 mimetypes = ['text/rust', 'text/x-rust'] |
|
29 |
|
30 keyword_types = (words(( |
|
31 'u8', 'u16', 'u32', 'u64', 'u128', 'i8', 'i16', 'i32', 'i64', 'i128', |
|
32 'usize', 'isize', 'f32', 'f64', 'char', 'str', 'bool', |
|
33 ), suffix=r'\b'), Keyword.Type) |
|
34 |
|
35 builtin_funcs_types = (words(( |
|
36 'Copy', 'Send', 'Sized', 'Sync', 'Unpin', |
|
37 'Drop', 'Fn', 'FnMut', 'FnOnce', 'drop', |
|
38 'Box', 'ToOwned', 'Clone', |
|
39 'PartialEq', 'PartialOrd', 'Eq', 'Ord', |
|
40 'AsRef', 'AsMut', 'Into', 'From', 'Default', |
|
41 'Iterator', 'Extend', 'IntoIterator', 'DoubleEndedIterator', |
|
42 'ExactSizeIterator', |
|
43 'Option', 'Some', 'None', |
|
44 'Result', 'Ok', 'Err', |
|
45 'String', 'ToString', 'Vec', |
|
46 ), suffix=r'\b'), Name.Builtin) |
|
47 |
|
48 builtin_macros = (words(( |
|
49 'asm', 'assert', 'assert_eq', 'assert_ne', 'cfg', 'column', |
|
50 'compile_error', 'concat', 'concat_idents', 'dbg', 'debug_assert', |
|
51 'debug_assert_eq', 'debug_assert_ne', 'env', 'eprint', 'eprintln', |
|
52 'file', 'format', 'format_args', 'format_args_nl', 'global_asm', |
|
53 'include', 'include_bytes', 'include_str', |
|
54 'is_aarch64_feature_detected', |
|
55 'is_arm_feature_detected', |
|
56 'is_mips64_feature_detected', |
|
57 'is_mips_feature_detected', |
|
58 'is_powerpc64_feature_detected', |
|
59 'is_powerpc_feature_detected', |
|
60 'is_x86_feature_detected', |
|
61 'line', 'llvm_asm', 'log_syntax', 'macro_rules', 'matches', |
|
62 'module_path', 'option_env', 'panic', 'print', 'println', 'stringify', |
|
63 'thread_local', 'todo', 'trace_macros', 'unimplemented', 'unreachable', |
|
64 'vec', 'write', 'writeln', |
|
65 ), suffix=r'!'), Name.Function.Magic) |
|
66 |
|
67 tokens = { |
|
68 'root': [ |
|
69 # rust allows a file to start with a shebang, but if the first line |
|
70 # starts with #![ then it's not a shebang but a crate attribute. |
|
71 (r'#![^[\r\n].*$', Comment.Preproc), |
|
72 default('base'), |
|
73 ], |
|
74 'base': [ |
|
75 # Whitespace and Comments |
|
76 (r'\n', Whitespace), |
|
77 (r'\s+', Whitespace), |
|
78 (r'//!.*?\n', String.Doc), |
|
79 (r'///(\n|[^/].*?\n)', String.Doc), |
|
80 (r'//(.*?)\n', Comment.Single), |
|
81 (r'/\*\*(\n|[^/*])', String.Doc, 'doccomment'), |
|
82 (r'/\*!', String.Doc, 'doccomment'), |
|
83 (r'/\*', Comment.Multiline, 'comment'), |
|
84 |
|
85 # Macro parameters |
|
86 (r"""\$([a-zA-Z_]\w*|\(,?|\),?|,?)""", Comment.Preproc), |
|
87 # Keywords |
|
88 (words(('as', 'async', 'await', 'box', 'const', 'crate', 'dyn', |
|
89 'else', 'extern', 'for', 'if', 'impl', 'in', 'loop', |
|
90 'match', 'move', 'mut', 'pub', 'ref', 'return', 'static', |
|
91 'super', 'trait', 'unsafe', 'use', 'where', 'while'), |
|
92 suffix=r'\b'), Keyword), |
|
93 (words(('abstract', 'become', 'do', 'final', 'macro', 'override', |
|
94 'priv', 'typeof', 'try', 'unsized', 'virtual', 'yield'), |
|
95 suffix=r'\b'), Keyword.Reserved), |
|
96 (r'(true|false)\b', Keyword.Constant), |
|
97 (r'self\b', Name.Builtin.Pseudo), |
|
98 (r'mod\b', Keyword, 'modname'), |
|
99 (r'let\b', Keyword.Declaration), |
|
100 (r'fn\b', Keyword, 'funcname'), |
|
101 (r'(struct|enum|type|union)\b', Keyword, 'typename'), |
|
102 (r'(default)(\s+)(type|fn)\b', bygroups(Keyword, Text, Keyword)), |
|
103 keyword_types, |
|
104 (r'[sS]elf\b', Name.Builtin.Pseudo), |
|
105 # Prelude (taken from Rust's src/libstd/prelude.rs) |
|
106 builtin_funcs_types, |
|
107 builtin_macros, |
|
108 # Path seperators, so types don't catch them. |
|
109 (r'::\b', Text), |
|
110 # Types in positions. |
|
111 (r'(?::|->)', Text, 'typename'), |
|
112 # Labels |
|
113 (r'(break|continue)(\s*)(\'[A-Za-z_]\w*)?', |
|
114 bygroups(Keyword, Text.Whitespace, Name.Label)), |
|
115 |
|
116 # Character literals |
|
117 (r"""'(\\['"\\nrt]|\\x[0-7][0-9a-fA-F]|\\0""" |
|
118 r"""|\\u\{[0-9a-fA-F]{1,6}\}|.)'""", |
|
119 String.Char), |
|
120 (r"""b'(\\['"\\nrt]|\\x[0-9a-fA-F]{2}|\\0""" |
|
121 r"""|\\u\{[0-9a-fA-F]{1,6}\}|.)'""", |
|
122 String.Char), |
|
123 |
|
124 # Binary literals |
|
125 (r'0b[01_]+', Number.Bin, 'number_lit'), |
|
126 # Octal literals |
|
127 (r'0o[0-7_]+', Number.Oct, 'number_lit'), |
|
128 # Hexadecimal literals |
|
129 (r'0[xX][0-9a-fA-F_]+', Number.Hex, 'number_lit'), |
|
130 # Decimal literals |
|
131 (r'[0-9][0-9_]*(\.[0-9_]+[eE][+\-]?[0-9_]+|' |
|
132 r'\.[0-9_]*(?!\.)|[eE][+\-]?[0-9_]+)', Number.Float, |
|
133 'number_lit'), |
|
134 (r'[0-9][0-9_]*', Number.Integer, 'number_lit'), |
|
135 |
|
136 # String literals |
|
137 (r'b"', String, 'bytestring'), |
|
138 (r'"', String, 'string'), |
|
139 (r'b?r(#*)".*?"\1', String), |
|
140 |
|
141 # Lifetime names |
|
142 (r"'", Operator, 'lifetime'), |
|
143 |
|
144 # Operators and Punctuation |
|
145 (r'\.\.=?', Operator), |
|
146 (r'[{}()\[\],.;]', Punctuation), |
|
147 (r'[+\-*/%&|<>^!~@=:?]', Operator), |
|
148 |
|
149 # Identifiers |
|
150 (r'[a-zA-Z_]\w*', Name), |
|
151 # Raw identifiers |
|
152 (r'r#[a-zA-Z_]\w*', Name), |
|
153 |
|
154 # Attributes |
|
155 (r'#!?\[', Comment.Preproc, 'attribute['), |
|
156 ], |
|
157 'comment': [ |
|
158 (r'[^*/]+', Comment.Multiline), |
|
159 (r'/\*', Comment.Multiline, '#push'), |
|
160 (r'\*/', Comment.Multiline, '#pop'), |
|
161 (r'[*/]', Comment.Multiline), |
|
162 ], |
|
163 'doccomment': [ |
|
164 (r'[^*/]+', String.Doc), |
|
165 (r'/\*', String.Doc, '#push'), |
|
166 (r'\*/', String.Doc, '#pop'), |
|
167 (r'[*/]', String.Doc), |
|
168 ], |
|
169 'modname': [ |
|
170 (r'\s+', Text), |
|
171 (r'[a-zA-Z_]\w*', Name.Namespace, '#pop'), |
|
172 default('#pop'), |
|
173 ], |
|
174 'funcname': [ |
|
175 (r'\s+', Text), |
|
176 (r'[a-zA-Z_]\w*', Name.Function, '#pop'), |
|
177 default('#pop'), |
|
178 ], |
|
179 'typename': [ |
|
180 (r'\s+', Text), |
|
181 (r'&', Keyword.Pseudo), |
|
182 (r"'", Operator, 'lifetime'), |
|
183 builtin_funcs_types, |
|
184 keyword_types, |
|
185 (r'[a-zA-Z_]\w*', Name.Class, '#pop'), |
|
186 default('#pop'), |
|
187 ], |
|
188 'lifetime': [ |
|
189 (r"(static|_)", Name.Builtin), |
|
190 (r"[a-zA-Z_]+\w*", Name.Attribute), |
|
191 default('#pop'), |
|
192 ], |
|
193 'number_lit': [ |
|
194 (r'[ui](8|16|32|64|size)', Keyword, '#pop'), |
|
195 (r'f(32|64)', Keyword, '#pop'), |
|
196 default('#pop'), |
|
197 ], |
|
198 'string': [ |
|
199 (r'"', String, '#pop'), |
|
200 (r"""\\['"\\nrt]|\\x[0-7][0-9a-fA-F]|\\0""" |
|
201 r"""|\\u\{[0-9a-fA-F]{1,6}\}""", String.Escape), |
|
202 (r'[^\\"]+', String), |
|
203 (r'\\', String), |
|
204 ], |
|
205 'bytestring': [ |
|
206 (r"""\\x[89a-fA-F][0-9a-fA-F]""", String.Escape), |
|
207 include('string'), |
|
208 ], |
|
209 'attribute_common': [ |
|
210 (r'"', String, 'string'), |
|
211 (r'\[', Comment.Preproc, 'attribute['), |
|
212 (r'\(', Comment.Preproc, 'attribute('), |
|
213 ], |
|
214 'attribute[': [ |
|
215 include('attribute_common'), |
|
216 (r'\];?', Comment.Preproc, '#pop'), |
|
217 (r'[^"\]]+', Comment.Preproc), |
|
218 ], |
|
219 'attribute(': [ |
|
220 include('attribute_common'), |
|
221 (r'\);?', Comment.Preproc, '#pop'), |
|
222 (r'[^")]+', Comment.Preproc), |
|
223 ], |
|
224 } |
|