1 | /**
|
2 | * @fileoverview Abstraction of JavaScript source code.
|
3 | * @author Nicholas C. Zakas
|
4 | */
|
5 | ;
|
6 |
|
7 | //------------------------------------------------------------------------------
|
8 | // Requirements
|
9 | //------------------------------------------------------------------------------
|
10 |
|
11 | const TokenStore = require("../token-store"),
|
12 | Traverser = require("./traverser"),
|
13 | astUtils = require("../ast-utils"),
|
14 | lodash = require("lodash");
|
15 |
|
16 | //------------------------------------------------------------------------------
|
17 | // Private
|
18 | //------------------------------------------------------------------------------
|
19 |
|
20 | /**
|
21 | * Validates that the given AST has the required information.
|
22 | * @param {ASTNode} ast The Program node of the AST to check.
|
23 | * @throws {Error} If the AST doesn't contain the correct information.
|
24 | * @returns {void}
|
25 | * @private
|
26 | */
|
27 | function validate(ast) {
|
28 | if (!ast.tokens) {
|
29 | throw new Error("AST is missing the tokens array.");
|
30 | }
|
31 |
|
32 | if (!ast.comments) {
|
33 | throw new Error("AST is missing the comments array.");
|
34 | }
|
35 |
|
36 | if (!ast.loc) {
|
37 | throw new Error("AST is missing location information.");
|
38 | }
|
39 |
|
40 | if (!ast.range) {
|
41 | throw new Error("AST is missing range information");
|
42 | }
|
43 | }
|
44 |
|
45 | /**
|
46 | * Check to see if its a ES6 export declaration.
|
47 | * @param {ASTNode} astNode An AST node.
|
48 | * @returns {boolean} whether the given node represents an export declaration.
|
49 | * @private
|
50 | */
|
51 | function looksLikeExport(astNode) {
|
52 | return astNode.type === "ExportDefaultDeclaration" || astNode.type === "ExportNamedDeclaration" ||
|
53 | astNode.type === "ExportAllDeclaration" || astNode.type === "ExportSpecifier";
|
54 | }
|
55 |
|
56 | /**
|
57 | * Merges two sorted lists into a larger sorted list in O(n) time.
|
58 | * @param {Token[]} tokens The list of tokens.
|
59 | * @param {Token[]} comments The list of comments.
|
60 | * @returns {Token[]} A sorted list of tokens and comments.
|
61 | * @private
|
62 | */
|
63 | function sortedMerge(tokens, comments) {
|
64 | const result = [];
|
65 | let tokenIndex = 0;
|
66 | let commentIndex = 0;
|
67 |
|
68 | while (tokenIndex < tokens.length || commentIndex < comments.length) {
|
69 | if (commentIndex >= comments.length || tokenIndex < tokens.length && tokens[tokenIndex].range[0] < comments[commentIndex].range[0]) {
|
70 | result.push(tokens[tokenIndex++]);
|
71 | } else {
|
72 | result.push(comments[commentIndex++]);
|
73 | }
|
74 | }
|
75 |
|
76 | return result;
|
77 | }
|
78 |
|
79 | //------------------------------------------------------------------------------
|
80 | // Public Interface
|
81 | //------------------------------------------------------------------------------
|
82 |
|
83 | class SourceCode extends TokenStore {
|
84 |
|
85 | /**
|
86 | * Represents parsed source code.
|
87 | * @param {string|Object} textOrConfig - The source code text or config object.
|
88 | * @param {string} textOrConfig.text - The source code text.
|
89 | * @param {ASTNode} textOrConfig.ast - The Program node of the AST representing the code. This AST should be created from the text that BOM was stripped.
|
90 | * @param {Object|null} textOrConfig.parserServices - The parser srevices.
|
91 | * @param {ScopeManager|null} textOrConfig.scopeManager - The scope of this source code.
|
92 | * @param {Object|null} textOrConfig.visitorKeys - The visitor keys to traverse AST.
|
93 | * @param {ASTNode} [ast] - The Program node of the AST representing the code. This AST should be created from the text that BOM was stripped.
|
94 | * @constructor
|
95 | */
|
96 | constructor(textOrConfig, ast) {
|
97 | let text, parserServices, scopeManager, visitorKeys;
|
98 |
|
99 | // Process overloading.
|
100 | if (typeof textOrConfig === "string") {
|
101 | text = textOrConfig;
|
102 | } else if (typeof textOrConfig === "object" && textOrConfig !== null) {
|
103 | text = textOrConfig.text;
|
104 | ast = textOrConfig.ast;
|
105 | parserServices = textOrConfig.parserServices;
|
106 | scopeManager = textOrConfig.scopeManager;
|
107 | visitorKeys = textOrConfig.visitorKeys;
|
108 | }
|
109 |
|
110 | validate(ast);
|
111 | super(ast.tokens, ast.comments);
|
112 |
|
113 | /**
|
114 | * The flag to indicate that the source code has Unicode BOM.
|
115 | * @type boolean
|
116 | */
|
117 | this.hasBOM = (text.charCodeAt(0) === 0xFEFF);
|
118 |
|
119 | /**
|
120 | * The original text source code.
|
121 | * BOM was stripped from this text.
|
122 | * @type string
|
123 | */
|
124 | this.text = (this.hasBOM ? text.slice(1) : text);
|
125 |
|
126 | /**
|
127 | * The parsed AST for the source code.
|
128 | * @type ASTNode
|
129 | */
|
130 | this.ast = ast;
|
131 |
|
132 | /**
|
133 | * The parser services of this source code.
|
134 | * @type {Object}
|
135 | */
|
136 | this.parserServices = parserServices || {};
|
137 |
|
138 | /**
|
139 | * The scope of this source code.
|
140 | * @type {ScopeManager|null}
|
141 | */
|
142 | this.scopeManager = scopeManager || null;
|
143 |
|
144 | /**
|
145 | * The visitor keys to traverse AST.
|
146 | * @type {Object}
|
147 | */
|
148 | this.visitorKeys = visitorKeys || Traverser.DEFAULT_VISITOR_KEYS;
|
149 |
|
150 | // Check the source text for the presence of a shebang since it is parsed as a standard line comment.
|
151 | const shebangMatched = this.text.match(astUtils.SHEBANG_MATCHER);
|
152 | const hasShebang = shebangMatched && ast.comments.length && ast.comments[0].value === shebangMatched[1];
|
153 |
|
154 | if (hasShebang) {
|
155 | ast.comments[0].type = "Shebang";
|
156 | }
|
157 |
|
158 | this.tokensAndComments = sortedMerge(ast.tokens, ast.comments);
|
159 |
|
160 | /**
|
161 | * The source code split into lines according to ECMA-262 specification.
|
162 | * This is done to avoid each rule needing to do so separately.
|
163 | * @type string[]
|
164 | */
|
165 | this.lines = [];
|
166 | this.lineStartIndices = [0];
|
167 |
|
168 | const lineEndingPattern = astUtils.createGlobalLinebreakMatcher();
|
169 | let match;
|
170 |
|
171 | /*
|
172 | * Previously, this was implemented using a regex that
|
173 | * matched a sequence of non-linebreak characters followed by a
|
174 | * linebreak, then adding the lengths of the matches. However,
|
175 | * this caused a catastrophic backtracking issue when the end
|
176 | * of a file contained a large number of non-newline characters.
|
177 | * To avoid this, the current implementation just matches newlines
|
178 | * and uses match.index to get the correct line start indices.
|
179 | */
|
180 | while ((match = lineEndingPattern.exec(this.text))) {
|
181 | this.lines.push(this.text.slice(this.lineStartIndices[this.lineStartIndices.length - 1], match.index));
|
182 | this.lineStartIndices.push(match.index + match[0].length);
|
183 | }
|
184 | this.lines.push(this.text.slice(this.lineStartIndices[this.lineStartIndices.length - 1]));
|
185 |
|
186 | // Cache for comments found using getComments().
|
187 | this._commentCache = new WeakMap();
|
188 |
|
189 | // don't allow modification of this object
|
190 | Object.freeze(this);
|
191 | Object.freeze(this.lines);
|
192 | }
|
193 |
|
194 | /**
|
195 | * Split the source code into multiple lines based on the line delimiters.
|
196 | * @param {string} text Source code as a string.
|
197 | * @returns {string[]} Array of source code lines.
|
198 | * @public
|
199 | */
|
200 | static splitLines(text) {
|
201 | return text.split(astUtils.createGlobalLinebreakMatcher());
|
202 | }
|
203 |
|
204 | /**
|
205 | * Gets the source code for the given node.
|
206 | * @param {ASTNode=} node The AST node to get the text for.
|
207 | * @param {int=} beforeCount The number of characters before the node to retrieve.
|
208 | * @param {int=} afterCount The number of characters after the node to retrieve.
|
209 | * @returns {string} The text representing the AST node.
|
210 | * @public
|
211 | */
|
212 | getText(node, beforeCount, afterCount) {
|
213 | if (node) {
|
214 | return this.text.slice(Math.max(node.range[0] - (beforeCount || 0), 0),
|
215 | node.range[1] + (afterCount || 0));
|
216 | }
|
217 | return this.text;
|
218 | }
|
219 |
|
220 | /**
|
221 | * Gets the entire source text split into an array of lines.
|
222 | * @returns {Array} The source text as an array of lines.
|
223 | * @public
|
224 | */
|
225 | getLines() {
|
226 | return this.lines;
|
227 | }
|
228 |
|
229 | /**
|
230 | * Retrieves an array containing all comments in the source code.
|
231 | * @returns {ASTNode[]} An array of comment nodes.
|
232 | * @public
|
233 | */
|
234 | getAllComments() {
|
235 | return this.ast.comments;
|
236 | }
|
237 |
|
238 | /**
|
239 | * Gets all comments for the given node.
|
240 | * @param {ASTNode} node The AST node to get the comments for.
|
241 | * @returns {Object} An object containing a leading and trailing array
|
242 | * of comments indexed by their position.
|
243 | * @public
|
244 | */
|
245 | getComments(node) {
|
246 | if (this._commentCache.has(node)) {
|
247 | return this._commentCache.get(node);
|
248 | }
|
249 |
|
250 | const comments = {
|
251 | leading: [],
|
252 | trailing: []
|
253 | };
|
254 |
|
255 | /*
|
256 | * Return all comments as leading comments of the Program node when
|
257 | * there is no executable code.
|
258 | */
|
259 | if (node.type === "Program") {
|
260 | if (node.body.length === 0) {
|
261 | comments.leading = node.comments;
|
262 | }
|
263 | } else {
|
264 |
|
265 | /*
|
266 | * Return comments as trailing comments of nodes that only contain
|
267 | * comments (to mimic the comment attachment behavior present in Espree).
|
268 | */
|
269 | if ((node.type === "BlockStatement" || node.type === "ClassBody") && node.body.length === 0 ||
|
270 | node.type === "ObjectExpression" && node.properties.length === 0 ||
|
271 | node.type === "ArrayExpression" && node.elements.length === 0 ||
|
272 | node.type === "SwitchStatement" && node.cases.length === 0
|
273 | ) {
|
274 | comments.trailing = this.getTokens(node, {
|
275 | includeComments: true,
|
276 | filter: astUtils.isCommentToken
|
277 | });
|
278 | }
|
279 |
|
280 | /*
|
281 | * Iterate over tokens before and after node and collect comment tokens.
|
282 | * Do not include comments that exist outside of the parent node
|
283 | * to avoid duplication.
|
284 | */
|
285 | let currentToken = this.getTokenBefore(node, { includeComments: true });
|
286 |
|
287 | while (currentToken && astUtils.isCommentToken(currentToken)) {
|
288 | if (node.parent && (currentToken.start < node.parent.start)) {
|
289 | break;
|
290 | }
|
291 | comments.leading.push(currentToken);
|
292 | currentToken = this.getTokenBefore(currentToken, { includeComments: true });
|
293 | }
|
294 |
|
295 | comments.leading.reverse();
|
296 |
|
297 | currentToken = this.getTokenAfter(node, { includeComments: true });
|
298 |
|
299 | while (currentToken && astUtils.isCommentToken(currentToken)) {
|
300 | if (node.parent && (currentToken.end > node.parent.end)) {
|
301 | break;
|
302 | }
|
303 | comments.trailing.push(currentToken);
|
304 | currentToken = this.getTokenAfter(currentToken, { includeComments: true });
|
305 | }
|
306 | }
|
307 |
|
308 | this._commentCache.set(node, comments);
|
309 | return comments;
|
310 | }
|
311 |
|
312 | /**
|
313 | * Retrieves the JSDoc comment for a given node.
|
314 | * @param {ASTNode} node The AST node to get the comment for.
|
315 | * @returns {Token|null} The Block comment token containing the JSDoc comment
|
316 | * for the given node or null if not found.
|
317 | * @public
|
318 | */
|
319 | getJSDocComment(node) {
|
320 |
|
321 | /**
|
322 | * Checks for the presence of a JSDoc comment for the given node and returns it.
|
323 | * @param {ASTNode} astNode The AST node to get the comment for.
|
324 | * @returns {Token|null} The Block comment token containing the JSDoc comment
|
325 | * for the given node or null if not found.
|
326 | * @private
|
327 | */
|
328 | const findJSDocComment = astNode => {
|
329 | const tokenBefore = this.getTokenBefore(astNode, { includeComments: true });
|
330 |
|
331 | if (
|
332 | tokenBefore &&
|
333 | astUtils.isCommentToken(tokenBefore) &&
|
334 | tokenBefore.type === "Block" &&
|
335 | tokenBefore.value.charAt(0) === "*" &&
|
336 | astNode.loc.start.line - tokenBefore.loc.end.line <= 1
|
337 | ) {
|
338 | return tokenBefore;
|
339 | }
|
340 |
|
341 | return null;
|
342 | };
|
343 | let parent = node.parent;
|
344 |
|
345 | switch (node.type) {
|
346 | case "ClassDeclaration":
|
347 | case "FunctionDeclaration":
|
348 | return findJSDocComment(looksLikeExport(parent) ? parent : node);
|
349 |
|
350 | case "ClassExpression":
|
351 | return findJSDocComment(parent.parent);
|
352 |
|
353 | case "ArrowFunctionExpression":
|
354 | case "FunctionExpression":
|
355 | if (parent.type !== "CallExpression" && parent.type !== "NewExpression") {
|
356 | while (
|
357 | !this.getCommentsBefore(parent).length &&
|
358 | !/Function/.test(parent.type) &&
|
359 | parent.type !== "MethodDefinition" &&
|
360 | parent.type !== "Property"
|
361 | ) {
|
362 | parent = parent.parent;
|
363 |
|
364 | if (!parent) {
|
365 | break;
|
366 | }
|
367 | }
|
368 |
|
369 | if (parent && parent.type !== "FunctionDeclaration" && parent.type !== "Program") {
|
370 | return findJSDocComment(parent);
|
371 | }
|
372 | }
|
373 |
|
374 | return findJSDocComment(node);
|
375 |
|
376 | // falls through
|
377 | default:
|
378 | return null;
|
379 | }
|
380 | }
|
381 |
|
382 | /**
|
383 | * Gets the deepest node containing a range index.
|
384 | * @param {int} index Range index of the desired node.
|
385 | * @returns {ASTNode} The node if found or null if not found.
|
386 | * @public
|
387 | */
|
388 | getNodeByRangeIndex(index) {
|
389 | let result = null,
|
390 | resultParent = null;
|
391 |
|
392 | Traverser.traverse(this.ast, {
|
393 | visitorKeys: this.visitorKeys,
|
394 | enter(node, parent) {
|
395 | if (node.range[0] <= index && index < node.range[1]) {
|
396 | result = node;
|
397 | resultParent = parent;
|
398 | } else {
|
399 | this.skip();
|
400 | }
|
401 | },
|
402 | leave(node) {
|
403 | if (node === result) {
|
404 | this.break();
|
405 | }
|
406 | }
|
407 | });
|
408 |
|
409 | return result ? Object.assign({ parent: resultParent }, result) : null;
|
410 | }
|
411 |
|
412 | /**
|
413 | * Determines if two tokens have at least one whitespace character
|
414 | * between them. This completely disregards comments in making the
|
415 | * determination, so comments count as zero-length substrings.
|
416 | * @param {Token} first The token to check after.
|
417 | * @param {Token} second The token to check before.
|
418 | * @returns {boolean} True if there is only space between tokens, false
|
419 | * if there is anything other than whitespace between tokens.
|
420 | * @public
|
421 | */
|
422 | isSpaceBetweenTokens(first, second) {
|
423 | const text = this.text.slice(first.range[1], second.range[0]);
|
424 |
|
425 | return /\s/.test(text.replace(/\/\*.*?\*\//g, ""));
|
426 | }
|
427 |
|
428 | /**
|
429 | * Converts a source text index into a (line, column) pair.
|
430 | * @param {number} index The index of a character in a file
|
431 | * @returns {Object} A {line, column} location object with a 0-indexed column
|
432 | * @public
|
433 | */
|
434 | getLocFromIndex(index) {
|
435 | if (typeof index !== "number") {
|
436 | throw new TypeError("Expected `index` to be a number.");
|
437 | }
|
438 |
|
439 | if (index < 0 || index > this.text.length) {
|
440 | throw new RangeError(`Index out of range (requested index ${index}, but source text has length ${this.text.length}).`);
|
441 | }
|
442 |
|
443 | /*
|
444 | * For an argument of this.text.length, return the location one "spot" past the last character
|
445 | * of the file. If the last character is a linebreak, the location will be column 0 of the next
|
446 | * line; otherwise, the location will be in the next column on the same line.
|
447 | *
|
448 | * See getIndexFromLoc for the motivation for this special case.
|
449 | */
|
450 | if (index === this.text.length) {
|
451 | return { line: this.lines.length, column: this.lines[this.lines.length - 1].length };
|
452 | }
|
453 |
|
454 | /*
|
455 | * To figure out which line rangeIndex is on, determine the last index at which rangeIndex could
|
456 | * be inserted into lineIndices to keep the list sorted.
|
457 | */
|
458 | const lineNumber = lodash.sortedLastIndex(this.lineStartIndices, index);
|
459 |
|
460 | return { line: lineNumber, column: index - this.lineStartIndices[lineNumber - 1] };
|
461 | }
|
462 |
|
463 | /**
|
464 | * Converts a (line, column) pair into a range index.
|
465 | * @param {Object} loc A line/column location
|
466 | * @param {number} loc.line The line number of the location (1-indexed)
|
467 | * @param {number} loc.column The column number of the location (0-indexed)
|
468 | * @returns {number} The range index of the location in the file.
|
469 | * @public
|
470 | */
|
471 | getIndexFromLoc(loc) {
|
472 | if (typeof loc !== "object" || typeof loc.line !== "number" || typeof loc.column !== "number") {
|
473 | throw new TypeError("Expected `loc` to be an object with numeric `line` and `column` properties.");
|
474 | }
|
475 |
|
476 | if (loc.line <= 0) {
|
477 | throw new RangeError(`Line number out of range (line ${loc.line} requested). Line numbers should be 1-based.`);
|
478 | }
|
479 |
|
480 | if (loc.line > this.lineStartIndices.length) {
|
481 | throw new RangeError(`Line number out of range (line ${loc.line} requested, but only ${this.lineStartIndices.length} lines present).`);
|
482 | }
|
483 |
|
484 | const lineStartIndex = this.lineStartIndices[loc.line - 1];
|
485 | const lineEndIndex = loc.line === this.lineStartIndices.length ? this.text.length : this.lineStartIndices[loc.line];
|
486 | const positionIndex = lineStartIndex + loc.column;
|
487 |
|
488 | /*
|
489 | * By design, getIndexFromLoc({ line: lineNum, column: 0 }) should return the start index of
|
490 | * the given line, provided that the line number is valid element of this.lines. Since the
|
491 | * last element of this.lines is an empty string for files with trailing newlines, add a
|
492 | * special case where getting the index for the first location after the end of the file
|
493 | * will return the length of the file, rather than throwing an error. This allows rules to
|
494 | * use getIndexFromLoc consistently without worrying about edge cases at the end of a file.
|
495 | */
|
496 | if (
|
497 | loc.line === this.lineStartIndices.length && positionIndex > lineEndIndex ||
|
498 | loc.line < this.lineStartIndices.length && positionIndex >= lineEndIndex
|
499 | ) {
|
500 | throw new RangeError(`Column number out of range (column ${loc.column} requested, but the length of line ${loc.line} is ${lineEndIndex - lineStartIndex}).`);
|
501 | }
|
502 |
|
503 | return positionIndex;
|
504 | }
|
505 | }
|
506 |
|
507 | module.exports = SourceCode;
|