Skip to content

Commit a491854

Browse files
p-zubietamarijnh
p-zubieta
authored and
marijnh
committed
[julia mode] Refactor some code into tokens, use electricInput
1 parent be795a9 commit a491854

File tree

1 file changed

+89
-61
lines changed

1 file changed

+89
-61
lines changed

‎mode/julia/julia.js‎

Lines changed: 89 additions & 61 deletions
Original file line numberDiff line numberDiff line change
@@ -14,27 +14,34 @@
1414
CodeMirror.defineMode("julia", function(_conf, parserConf) {
1515
var ERRORCLASS = 'error';
1616

17-
function wordRegexp(words) {
18-
return new RegExp("^((" + words.join(")|(") + "))\\b");
17+
function wordRegexp(words, end) {
18+
if (typeof end === 'undefined') { end = "\\b"; }
19+
return new RegExp("^((" + words.join(")|(") + "))" + end);
1920
}
2021

22+
var octChar = "\\\\[0-7]{1,3}";
23+
var hexChar = "\\\\x[A-Fa-f0-9]{1,2}";
24+
var specialChar = "\\\\[abfnrtv0%?'\"\\\\]";
25+
var singleChar = "([^\\u0027\\u005C\\uD800-\\uDFFF]|[\\uD800-\\uDFFF][\\uDC00-\\uDFFF])";
2126
var operators = parserConf.operators || /^\.?[|&^\\%*+\-<>!=\/]=?|\?|~|:|\$|\.[<>]|<<=?|>>>?=?|\.[<>=]=|->?|\/\/|\bin\b(?!\()|[\u2208\u2209](?!\()/;
2227
var delimiters = parserConf.delimiters || /^[;,()[\]{}]/;
23-
var identifiers = parserConf.identifiers || /^[_A-Za-z\u00A1-\uFFFF][_A-Za-z0-9\u00A1-\uFFFF]*!*/;
28+
var identifiers = parserConf.identifiers || /^[_A-Za-z\u00A1-\uFFFF][\w\u00A1-\uFFFF]*!*/;
29+
var charsList = [octChar, hexChar, specialChar, singleChar];
2430
var blockOpeners = ["begin", "function", "type", "immutable", "let", "macro", "for", "while", "quote", "if", "else", "elseif", "try", "finally", "catch", "do"];
2531
var blockClosers = ["end", "else", "elseif", "catch", "finally"];
2632
var keywordList = ['if', 'else', 'elseif', 'while', 'for', 'begin', 'let', 'end', 'do', 'try', 'catch', 'finally', 'return', 'break', 'continue', 'global', 'local', 'const', 'export', 'import', 'importall', 'using', 'function', 'macro', 'module', 'baremodule', 'type', 'immutable', 'quote', 'typealias', 'abstract', 'bitstype'];
2733
var builtinList = ['true', 'false', 'nothing', 'NaN', 'Inf'];
2834

2935
//var stringPrefixes = new RegExp("^[br]?('|\")")
30-
var stringPrefixes = /^(`|'|"{3}|([brv]?"))/;
36+
var stringPrefixes = /^(`|"{3}|([brv]?"))/;
37+
var chars = wordRegexp(charsList, "'");
3138
var keywords = wordRegexp(keywordList);
3239
var builtins = wordRegexp(builtinList);
3340
var openers = wordRegexp(blockOpeners);
3441
var closers = wordRegexp(blockClosers);
35-
var macro = /^@[_A-Za-z][_A-Za-z0-9]*/;
36-
var symbol = /^:[_A-Za-z\u00A1-\uFFFF][_A-Za-z0-9\u00A1-\uFFFF]*!*/;
37-
var typeAnnotation = /^::[^.,;"{()=$\s]+({[^}]*}+)*/;
42+
var macro = /^@[_A-Za-z][\w]*/;
43+
var symbol = /^:[_A-Za-z\u00A1-\uFFFF][\w\u00A1-\uFFFF]*!*/;
44+
var typeAnnotation = /^::[^,;"{()=$\s]+({[^}]*}+)*/;
3845

3946
function inArray(state) {
4047
var ch = currentScope(state);
@@ -53,19 +60,10 @@ CodeMirror.defineMode("julia", function(_conf, parserConf) {
5360

5461
// tokenizers
5562
function tokenBase(stream, state) {
56-
//Handle multiline comments
57-
if (stream.match(/^#=\s*/)) {
58-
state.scopes.push('#=');
59-
}
60-
if (currentScope(state) == '#=' && stream.match(/^=#/)) {
61-
state.scopes.pop();
62-
return 'comment';
63-
}
64-
if (state.scopes.indexOf('#=') >= 0) {
65-
if (!stream.match(/.*?(?=(#=|=#))/)) {
66-
stream.skipToEnd();
67-
}
68-
return 'comment';
63+
// Handle multiline comments
64+
if (stream.match(/^#=/, false)) {
65+
state.tokenize = tokenComment;
66+
return state.tokenize(stream, state);
6967
}
7068

7169
// Handle scope changes
@@ -100,6 +98,10 @@ CodeMirror.defineMode("julia", function(_conf, parserConf) {
10098
state.scopes.push('[');
10199
}
102100

101+
if (ch === '(') {
102+
state.scopes.push('(');
103+
}
104+
103105
var scope = currentScope(state);
104106

105107
if (scope == '[' && ch === ']') {
@@ -137,33 +139,20 @@ CodeMirror.defineMode("julia", function(_conf, parserConf) {
137139
// Handle Number Literals
138140
if (stream.match(/^[0-9\.]/, false)) {
139141
var imMatcher = RegExp(/^im\b/);
140-
var floatLiteral = false;
142+
var numberLiteral = false;
141143
// Floats
142-
if (stream.match(/^\d*\.(?!\.)\d+([ef][\+\-]?\d+)?/i)) { floatLiteral = true; }
143-
if (stream.match(/^\d+\.(?!\.)\d*/)) { floatLiteral = true; }
144-
if (stream.match(/^\.\d+/)) { floatLiteral = true; }
145-
if (stream.match(/^0x\.[0-9a-f]+p[\+\-]?\d+/i)) { floatLiteral = true; }
146-
if (floatLiteral) {
147-
// Float literals may be "imaginary"
148-
stream.match(imMatcher);
149-
state.leavingExpr = true;
150-
return 'number';
151-
}
144+
if (stream.match(/^\d*\.(?!\.)\d*([Eef][\+\-]?\d+)?/i)) { numberLiteral = true; }
145+
if (stream.match(/^\d+\.(?!\.)\d*/)) { numberLiteral = true; }
146+
if (stream.match(/^\.\d+/)) { numberLiteral = true; }
147+
if (stream.match(/^0x\.[0-9a-f]+p[\+\-]?\d+/i)) { numberLiteral = true; }
152148
// Integers
153-
var intLiteral = false;
154-
// Hex
155-
if (stream.match(/^0x[0-9a-f]+/i)) { intLiteral = true; }
156-
// Binary
157-
if (stream.match(/^0b[01]+/i)) { intLiteral = true; }
158-
// Octal
159-
if (stream.match(/^0o[0-7]+/i)) { intLiteral = true; }
160-
// Decimal
161-
if (stream.match(/^[1-9]\d*(e[\+\-]?\d+)?/)) {
162-
intLiteral = true;
163-
}
149+
if (stream.match(/^0x[0-9a-f]+/i)) { numberLiteral = true; } // Hex
150+
if (stream.match(/^0b[01]+/i)) { numberLiteral = true; } // Binary
151+
if (stream.match(/^0o[0-7]+/i)) { numberLiteral = true; } // Octal
152+
if (stream.match(/^[1-9]\d*(e[\+\-]?\d+)?/)) { numberLiteral = true; } // Decimal
164153
// Zero by itself with no other piece of number.
165-
if (stream.match(/^0(?![\dx])/i)) { intLiteral = true; }
166-
if (intLiteral) {
154+
if (stream.match(/^0(?![\dx])/i)) { numberLiteral = true; }
155+
if (numberLiteral) {
167156
// Integer literals may be "long"
168157
stream.match(imMatcher);
169158
state.leavingExpr = true;
@@ -194,6 +183,12 @@ CodeMirror.defineMode("julia", function(_conf, parserConf) {
194183
return 'operator';
195184
}
196185

186+
// Handle Chars
187+
if (stream.match(/^'/)) {
188+
state.tokenize = tokenChar;
189+
return state.tokenize(stream, state);
190+
}
191+
197192
// Handle Strings
198193
if (stream.match(stringPrefixes)) {
199194
state.tokenize = tokenStringFactory(stream.current());
@@ -269,7 +264,7 @@ CodeMirror.defineMode("julia", function(_conf, parserConf) {
269264
// over two or more lines.
270265
if (stream.match(/^$/g, false)) {
271266
stream.backUp(state.charsAdvanced);
272-
while (state.scopes.length > state.firstParenPos + 1)
267+
while (state.scopes.length > state.firstParenPos)
273268
state.scopes.pop();
274269
state.firstParenPos = -1;
275270
state.charsAdvanced = 0;
@@ -279,33 +274,65 @@ CodeMirror.defineMode("julia", function(_conf, parserConf) {
279274
return callOrDef(stream, state);
280275
}
281276

277+
function tokenComment(stream, state) {
278+
if (stream.match(/^#=/)) {
279+
state.weakScopes++;
280+
}
281+
if (!stream.match(/.*?(?=(#=|=#))/)) {
282+
stream.skipToEnd();
283+
}
284+
if (stream.match(/^=#/)) {
285+
state.weakScopes--;
286+
if (state.weakScopes == 0)
287+
state.tokenize = tokenBase;
288+
}
289+
return 'comment';
290+
}
291+
292+
function tokenChar(stream, state) {
293+
var isChar = false, match;
294+
if (stream.match(chars)) {
295+
isChar = true;
296+
} else if (match = stream.match(/\\u([a-f0-9]{1,4})(?=')/i)) {
297+
var value = parseInt(match[1], 16);
298+
if (value <= 55295 || value >= 57344) { // (U+0,U+D7FF), (U+E000,U+FFFF)
299+
isChar = true;
300+
stream.next();
301+
}
302+
} else if (match = stream.match(/\\U([A-Fa-f0-9]{5,8})(?=')/)) {
303+
var value = parseInt(match[1], 16);
304+
if (value <= 1114111) { // U+10FFFF
305+
isChar = true;
306+
stream.next();
307+
}
308+
}
309+
if (isChar) {
310+
state.leavingExpr = true;
311+
state.tokenize = tokenBase;
312+
return 'string';
313+
}
314+
if (!stream.match(/^[^']+(?=')/)) { stream.skipToEnd(); }
315+
if (stream.match(/^'/)) { state.tokenize = tokenBase; }
316+
return ERRORCLASS;
317+
}
318+
282319
function tokenStringFactory(delimiter) {
283320
while ('bruv'.indexOf(delimiter.charAt(0).toLowerCase()) >= 0) {
284321
delimiter = delimiter.substr(1);
285322
}
286-
var singleline = delimiter == "'";
287323
var OUTCLASS = 'string';
288324

289325
function tokenString(stream, state) {
290326
while (!stream.eol()) {
291-
stream.eatWhile(/[^'"\\]/);
327+
stream.eatWhile(/[^"\\]/);
292328
if (stream.eat('\\')) {
293329
stream.next();
294-
if (singleline && stream.eol()) {
295-
return OUTCLASS;
296-
}
297330
} else if (stream.match(delimiter)) {
298331
state.tokenize = tokenBase;
332+
state.leavingExpr = true;
299333
return OUTCLASS;
300334
} else {
301-
stream.eat(/['"]/);
302-
}
303-
}
304-
if (singleline) {
305-
if (parserConf.singleLineStringErrors) {
306-
return ERRORCLASS;
307-
} else {
308-
state.tokenize = tokenBase;
335+
stream.eat(/["]/);
309336
}
310337
}
311338
return OUTCLASS;
@@ -319,6 +346,7 @@ CodeMirror.defineMode("julia", function(_conf, parserConf) {
319346
return {
320347
tokenize: tokenBase,
321348
scopes: [],
349+
weakScopes: 0,
322350
lastToken: null,
323351
leavingExpr: false,
324352
isDefinition: false,
@@ -345,15 +373,15 @@ CodeMirror.defineMode("julia", function(_conf, parserConf) {
345373

346374
indent: function(state, textAfter) {
347375
var delta = 0;
348-
if (textAfter == "end" || textAfter == "]" || textAfter == "}" || textAfter == "else" || textAfter == "elseif" || textAfter == "catch" || textAfter == "finally") {
376+
if (textAfter == "]" || textAfter == ")" || textAfter == "end" || textAfter == "else" || textAfter == "elseif" || textAfter == "catch" || textAfter == "finally") {
349377
delta = -1;
350378
}
351379
return (state.scopes.length + delta) * _conf.indentUnit;
352380
},
353381

382+
electricInput: /(end|else(if)?|catch|finally)$/,
354383
lineComment: "#",
355-
fold: "indent",
356-
electricChars: "edlsifyh]}"
384+
fold: "indent"
357385
};
358386
return external;
359387
});

0 commit comments

Comments
 (0)