scripts/CodeMirror/mode/python/python.js (view raw)
1// CodeMirror, copyright (c) by Marijn Haverbeke and others
2// Distributed under an MIT license: https://codemirror.net/LICENSE
3
4(function(mod) {
5 if (typeof exports == "object" && typeof module == "object") // CommonJS
6 mod(require("../../lib/codemirror"));
7 else if (typeof define == "function" && define.amd) // AMD
8 define(["../../lib/codemirror"], mod);
9 else // Plain browser env
10 mod(CodeMirror);
11})(function(CodeMirror) {
12 "use strict";
13
14 function wordRegexp(words) {
15 return new RegExp("^((" + words.join(")|(") + "))\\b");
16 }
17
18 var wordOperators = wordRegexp(["and", "or", "not", "is"]);
19 var commonKeywords = ["as", "assert", "break", "class", "continue",
20 "def", "del", "elif", "else", "except", "finally",
21 "for", "from", "global", "if", "import",
22 "lambda", "pass", "raise", "return",
23 "try", "while", "with", "yield", "in"];
24 var commonBuiltins = ["abs", "all", "any", "bin", "bool", "bytearray", "callable", "chr",
25 "classmethod", "compile", "complex", "delattr", "dict", "dir", "divmod",
26 "enumerate", "eval", "filter", "float", "format", "frozenset",
27 "getattr", "globals", "hasattr", "hash", "help", "hex", "id",
28 "input", "int", "isinstance", "issubclass", "iter", "len",
29 "list", "locals", "map", "max", "memoryview", "min", "next",
30 "object", "oct", "open", "ord", "pow", "property", "range",
31 "repr", "reversed", "round", "set", "setattr", "slice",
32 "sorted", "staticmethod", "str", "sum", "super", "tuple",
33 "type", "vars", "zip", "__import__", "NotImplemented",
34 "Ellipsis", "__debug__"];
35 CodeMirror.registerHelper("hintWords", "python", commonKeywords.concat(commonBuiltins));
36
37 function top(state) {
38 return state.scopes[state.scopes.length - 1];
39 }
40
41 CodeMirror.defineMode("python", function(conf, parserConf) {
42 var ERRORCLASS = "error";
43
44 var delimiters = parserConf.delimiters || parserConf.singleDelimiters || /^[\(\)\[\]\{\}@,:`=;\.\\]/;
45 // (Backwards-compatibility with old, cumbersome config system)
46 var operators = [parserConf.singleOperators, parserConf.doubleOperators, parserConf.doubleDelimiters, parserConf.tripleDelimiters,
47 parserConf.operators || /^([-+*/%\/&|^]=?|[<>=]+|\/\/=?|\*\*=?|!=|[~!@]|\.\.\.)/]
48 for (var i = 0; i < operators.length; i++) if (!operators[i]) operators.splice(i--, 1)
49
50 var hangingIndent = parserConf.hangingIndent || conf.indentUnit;
51
52 var myKeywords = commonKeywords, myBuiltins = commonBuiltins;
53 if (parserConf.extra_keywords != undefined)
54 myKeywords = myKeywords.concat(parserConf.extra_keywords);
55
56 if (parserConf.extra_builtins != undefined)
57 myBuiltins = myBuiltins.concat(parserConf.extra_builtins);
58
59 var py3 = !(parserConf.version && Number(parserConf.version) < 3)
60 if (py3) {
61 // since http://legacy.python.org/dev/peps/pep-0465/ @ is also an operator
62 var identifiers = parserConf.identifiers|| /^[_A-Za-z\u00A1-\uFFFF][_A-Za-z0-9\u00A1-\uFFFF]*/;
63 myKeywords = myKeywords.concat(["nonlocal", "False", "True", "None", "async", "await"]);
64 myBuiltins = myBuiltins.concat(["ascii", "bytes", "exec", "print"]);
65 var stringPrefixes = new RegExp("^(([rbuf]|(br)|(fr))?('{3}|\"{3}|['\"]))", "i");
66 } else {
67 var identifiers = parserConf.identifiers|| /^[_A-Za-z][_A-Za-z0-9]*/;
68 myKeywords = myKeywords.concat(["exec", "print"]);
69 myBuiltins = myBuiltins.concat(["apply", "basestring", "buffer", "cmp", "coerce", "execfile",
70 "file", "intern", "long", "raw_input", "reduce", "reload",
71 "unichr", "unicode", "xrange", "False", "True", "None"]);
72 var stringPrefixes = new RegExp("^(([rubf]|(ur)|(br))?('{3}|\"{3}|['\"]))", "i");
73 }
74 var keywords = wordRegexp(myKeywords);
75 var builtins = wordRegexp(myBuiltins);
76
77 // tokenizers
78 function tokenBase(stream, state) {
79 var sol = stream.sol() && state.lastToken != "\\"
80 if (sol) state.indent = stream.indentation()
81 // Handle scope changes
82 if (sol && top(state).type == "py") {
83 var scopeOffset = top(state).offset;
84 if (stream.eatSpace()) {
85 var lineOffset = stream.indentation();
86 if (lineOffset > scopeOffset)
87 pushPyScope(state);
88 else if (lineOffset < scopeOffset && dedent(stream, state) && stream.peek() != "#")
89 state.errorToken = true;
90 return null;
91 } else {
92 var style = tokenBaseInner(stream, state);
93 if (scopeOffset > 0 && dedent(stream, state))
94 style += " " + ERRORCLASS;
95 return style;
96 }
97 }
98 return tokenBaseInner(stream, state);
99 }
100
101 function tokenBaseInner(stream, state, inFormat) {
102 if (stream.eatSpace()) return null;
103
104 // Handle Comments
105 if (!inFormat && stream.match(/^#.*/)) return "comment";
106
107 // Handle Number Literals
108 if (stream.match(/^[0-9\.]/, false)) {
109 var floatLiteral = false;
110 // Floats
111 if (stream.match(/^[\d_]*\.\d+(e[\+\-]?\d+)?/i)) { floatLiteral = true; }
112 if (stream.match(/^[\d_]+\.\d*/)) { floatLiteral = true; }
113 if (stream.match(/^\.\d+/)) { floatLiteral = true; }
114 if (floatLiteral) {
115 // Float literals may be "imaginary"
116 stream.eat(/J/i);
117 return "number";
118 }
119 // Integers
120 var intLiteral = false;
121 // Hex
122 if (stream.match(/^0x[0-9a-f_]+/i)) intLiteral = true;
123 // Binary
124 if (stream.match(/^0b[01_]+/i)) intLiteral = true;
125 // Octal
126 if (stream.match(/^0o[0-7_]+/i)) intLiteral = true;
127 // Decimal
128 if (stream.match(/^[1-9][\d_]*(e[\+\-]?[\d_]+)?/)) {
129 // Decimal literals may be "imaginary"
130 stream.eat(/J/i);
131 // TODO - Can you have imaginary longs?
132 intLiteral = true;
133 }
134 // Zero by itself with no other piece of number.
135 if (stream.match(/^0(?![\dx])/i)) intLiteral = true;
136 if (intLiteral) {
137 // Integer literals may be "long"
138 stream.eat(/L/i);
139 return "number";
140 }
141 }
142
143 // Handle Strings
144 if (stream.match(stringPrefixes)) {
145 var isFmtString = stream.current().toLowerCase().indexOf('f') !== -1;
146 if (!isFmtString) {
147 state.tokenize = tokenStringFactory(stream.current(), state.tokenize);
148 return state.tokenize(stream, state);
149 } else {
150 state.tokenize = formatStringFactory(stream.current(), state.tokenize);
151 return state.tokenize(stream, state);
152 }
153 }
154
155 for (var i = 0; i < operators.length; i++)
156 if (stream.match(operators[i])) return "operator"
157
158 if (stream.match(delimiters)) return "punctuation";
159
160 if (state.lastToken == "." && stream.match(identifiers))
161 return "property";
162
163 if (stream.match(keywords) || stream.match(wordOperators))
164 return "keyword";
165
166 if (stream.match(builtins))
167 return "builtin";
168
169 if (stream.match(/^(self|cls)\b/))
170 return "variable-2";
171
172 if (stream.match(identifiers)) {
173 if (state.lastToken == "def" || state.lastToken == "class")
174 return "def";
175 return "variable";
176 }
177
178 // Handle non-detected items
179 stream.next();
180 return inFormat ? null :ERRORCLASS;
181 }
182
183 function formatStringFactory(delimiter, tokenOuter) {
184 while ("rubf".indexOf(delimiter.charAt(0).toLowerCase()) >= 0)
185 delimiter = delimiter.substr(1);
186
187 var singleline = delimiter.length == 1;
188 var OUTCLASS = "string";
189
190 function tokenNestedExpr(depth) {
191 return function(stream, state) {
192 var inner = tokenBaseInner(stream, state, true)
193 if (inner == "punctuation") {
194 if (stream.current() == "{") {
195 state.tokenize = tokenNestedExpr(depth + 1)
196 } else if (stream.current() == "}") {
197 if (depth > 1) state.tokenize = tokenNestedExpr(depth - 1)
198 else state.tokenize = tokenString
199 }
200 }
201 return inner
202 }
203 }
204
205 function tokenString(stream, state) {
206 while (!stream.eol()) {
207 stream.eatWhile(/[^'"\{\}\\]/);
208 if (stream.eat("\\")) {
209 stream.next();
210 if (singleline && stream.eol())
211 return OUTCLASS;
212 } else if (stream.match(delimiter)) {
213 state.tokenize = tokenOuter;
214 return OUTCLASS;
215 } else if (stream.match('{{')) {
216 // ignore {{ in f-str
217 return OUTCLASS;
218 } else if (stream.match('{', false)) {
219 // switch to nested mode
220 state.tokenize = tokenNestedExpr(0)
221 if (stream.current()) return OUTCLASS;
222 else return state.tokenize(stream, state)
223 } else if (stream.match('}}')) {
224 return OUTCLASS;
225 } else if (stream.match('}')) {
226 // single } in f-string is an error
227 return ERRORCLASS;
228 } else {
229 stream.eat(/['"]/);
230 }
231 }
232 if (singleline) {
233 if (parserConf.singleLineStringErrors)
234 return ERRORCLASS;
235 else
236 state.tokenize = tokenOuter;
237 }
238 return OUTCLASS;
239 }
240 tokenString.isString = true;
241 return tokenString;
242 }
243
244 function tokenStringFactory(delimiter, tokenOuter) {
245 while ("rubf".indexOf(delimiter.charAt(0).toLowerCase()) >= 0)
246 delimiter = delimiter.substr(1);
247
248 var singleline = delimiter.length == 1;
249 var OUTCLASS = "string";
250
251 function tokenString(stream, state) {
252 while (!stream.eol()) {
253 stream.eatWhile(/[^'"\\]/);
254 if (stream.eat("\\")) {
255 stream.next();
256 if (singleline && stream.eol())
257 return OUTCLASS;
258 } else if (stream.match(delimiter)) {
259 state.tokenize = tokenOuter;
260 return OUTCLASS;
261 } else {
262 stream.eat(/['"]/);
263 }
264 }
265 if (singleline) {
266 if (parserConf.singleLineStringErrors)
267 return ERRORCLASS;
268 else
269 state.tokenize = tokenOuter;
270 }
271 return OUTCLASS;
272 }
273 tokenString.isString = true;
274 return tokenString;
275 }
276
277 function pushPyScope(state) {
278 while (top(state).type != "py") state.scopes.pop()
279 state.scopes.push({offset: top(state).offset + conf.indentUnit,
280 type: "py",
281 align: null})
282 }
283
284 function pushBracketScope(stream, state, type) {
285 var align = stream.match(/^([\s\[\{\(]|#.*)*$/, false) ? null : stream.column() + 1
286 state.scopes.push({offset: state.indent + hangingIndent,
287 type: type,
288 align: align})
289 }
290
291 function dedent(stream, state) {
292 var indented = stream.indentation();
293 while (state.scopes.length > 1 && top(state).offset > indented) {
294 if (top(state).type != "py") return true;
295 state.scopes.pop();
296 }
297 return top(state).offset != indented;
298 }
299
300 function tokenLexer(stream, state) {
301 if (stream.sol()) state.beginningOfLine = true;
302
303 var style = state.tokenize(stream, state);
304 var current = stream.current();
305
306 // Handle decorators
307 if (state.beginningOfLine && current == "@")
308 return stream.match(identifiers, false) ? "meta" : py3 ? "operator" : ERRORCLASS;
309
310 if (/\S/.test(current)) state.beginningOfLine = false;
311
312 if ((style == "variable" || style == "builtin")
313 && state.lastToken == "meta")
314 style = "meta";
315
316 // Handle scope changes.
317 if (current == "pass" || current == "return")
318 state.dedent += 1;
319
320 if (current == "lambda") state.lambda = true;
321 if (current == ":" && !state.lambda && top(state).type == "py")
322 pushPyScope(state);
323
324 if (current.length == 1 && !/string|comment/.test(style)) {
325 var delimiter_index = "[({".indexOf(current);
326 if (delimiter_index != -1)
327 pushBracketScope(stream, state, "])}".slice(delimiter_index, delimiter_index+1));
328
329 delimiter_index = "])}".indexOf(current);
330 if (delimiter_index != -1) {
331 if (top(state).type == current) state.indent = state.scopes.pop().offset - hangingIndent
332 else return ERRORCLASS;
333 }
334 }
335 if (state.dedent > 0 && stream.eol() && top(state).type == "py") {
336 if (state.scopes.length > 1) state.scopes.pop();
337 state.dedent -= 1;
338 }
339
340 return style;
341 }
342
343 var external = {
344 startState: function(basecolumn) {
345 return {
346 tokenize: tokenBase,
347 scopes: [{offset: basecolumn || 0, type: "py", align: null}],
348 indent: basecolumn || 0,
349 lastToken: null,
350 lambda: false,
351 dedent: 0
352 };
353 },
354
355 token: function(stream, state) {
356 var addErr = state.errorToken;
357 if (addErr) state.errorToken = false;
358 var style = tokenLexer(stream, state);
359
360 if (style && style != "comment")
361 state.lastToken = (style == "keyword" || style == "punctuation") ? stream.current() : style;
362 if (style == "punctuation") style = null;
363
364 if (stream.eol() && state.lambda)
365 state.lambda = false;
366 return addErr ? style + " " + ERRORCLASS : style;
367 },
368
369 indent: function(state, textAfter) {
370 if (state.tokenize != tokenBase)
371 return state.tokenize.isString ? CodeMirror.Pass : 0;
372
373 var scope = top(state), closing = scope.type == textAfter.charAt(0)
374 if (scope.align != null)
375 return scope.align - (closing ? 1 : 0)
376 else
377 return scope.offset - (closing ? hangingIndent : 0)
378 },
379
380 electricInput: /^\s*[\}\]\)]$/,
381 closeBrackets: {triples: "'\""},
382 lineComment: "#",
383 fold: "indent"
384 };
385 return external;
386 });
387
388 CodeMirror.defineMIME("text/x-python", "python");
389
390 var words = function(str) { return str.split(" "); };
391
392 CodeMirror.defineMIME("text/x-cython", {
393 name: "python",
394 extra_keywords: words("by cdef cimport cpdef ctypedef enum except "+
395 "extern gil include nogil property public "+
396 "readonly struct union DEF IF ELIF ELSE")
397 });
398
399});