1 | 'use strict';
|
---|
2 |
|
---|
3 | const stringify = require('./stringify');
|
---|
4 |
|
---|
5 | /**
|
---|
6 | * Constants
|
---|
7 | */
|
---|
8 |
|
---|
9 | const {
|
---|
10 | MAX_LENGTH,
|
---|
11 | CHAR_BACKSLASH, /* \ */
|
---|
12 | CHAR_BACKTICK, /* ` */
|
---|
13 | CHAR_COMMA, /* , */
|
---|
14 | CHAR_DOT, /* . */
|
---|
15 | CHAR_LEFT_PARENTHESES, /* ( */
|
---|
16 | CHAR_RIGHT_PARENTHESES, /* ) */
|
---|
17 | CHAR_LEFT_CURLY_BRACE, /* { */
|
---|
18 | CHAR_RIGHT_CURLY_BRACE, /* } */
|
---|
19 | CHAR_LEFT_SQUARE_BRACKET, /* [ */
|
---|
20 | CHAR_RIGHT_SQUARE_BRACKET, /* ] */
|
---|
21 | CHAR_DOUBLE_QUOTE, /* " */
|
---|
22 | CHAR_SINGLE_QUOTE, /* ' */
|
---|
23 | CHAR_NO_BREAK_SPACE,
|
---|
24 | CHAR_ZERO_WIDTH_NOBREAK_SPACE
|
---|
25 | } = require('./constants');
|
---|
26 |
|
---|
27 | /**
|
---|
28 | * parse
|
---|
29 | */
|
---|
30 |
|
---|
31 | const parse = (input, options = {}) => {
|
---|
32 | if (typeof input !== 'string') {
|
---|
33 | throw new TypeError('Expected a string');
|
---|
34 | }
|
---|
35 |
|
---|
36 | let opts = options || {};
|
---|
37 | let max = typeof opts.maxLength === 'number' ? Math.min(MAX_LENGTH, opts.maxLength) : MAX_LENGTH;
|
---|
38 | if (input.length > max) {
|
---|
39 | throw new SyntaxError(`Input length (${input.length}), exceeds max characters (${max})`);
|
---|
40 | }
|
---|
41 |
|
---|
42 | let ast = { type: 'root', input, nodes: [] };
|
---|
43 | let stack = [ast];
|
---|
44 | let block = ast;
|
---|
45 | let prev = ast;
|
---|
46 | let brackets = 0;
|
---|
47 | let length = input.length;
|
---|
48 | let index = 0;
|
---|
49 | let depth = 0;
|
---|
50 | let value;
|
---|
51 | let memo = {};
|
---|
52 |
|
---|
53 | /**
|
---|
54 | * Helpers
|
---|
55 | */
|
---|
56 |
|
---|
57 | const advance = () => input[index++];
|
---|
58 | const push = node => {
|
---|
59 | if (node.type === 'text' && prev.type === 'dot') {
|
---|
60 | prev.type = 'text';
|
---|
61 | }
|
---|
62 |
|
---|
63 | if (prev && prev.type === 'text' && node.type === 'text') {
|
---|
64 | prev.value += node.value;
|
---|
65 | return;
|
---|
66 | }
|
---|
67 |
|
---|
68 | block.nodes.push(node);
|
---|
69 | node.parent = block;
|
---|
70 | node.prev = prev;
|
---|
71 | prev = node;
|
---|
72 | return node;
|
---|
73 | };
|
---|
74 |
|
---|
75 | push({ type: 'bos' });
|
---|
76 |
|
---|
77 | while (index < length) {
|
---|
78 | block = stack[stack.length - 1];
|
---|
79 | value = advance();
|
---|
80 |
|
---|
81 | /**
|
---|
82 | * Invalid chars
|
---|
83 | */
|
---|
84 |
|
---|
85 | if (value === CHAR_ZERO_WIDTH_NOBREAK_SPACE || value === CHAR_NO_BREAK_SPACE) {
|
---|
86 | continue;
|
---|
87 | }
|
---|
88 |
|
---|
89 | /**
|
---|
90 | * Escaped chars
|
---|
91 | */
|
---|
92 |
|
---|
93 | if (value === CHAR_BACKSLASH) {
|
---|
94 | push({ type: 'text', value: (options.keepEscaping ? value : '') + advance() });
|
---|
95 | continue;
|
---|
96 | }
|
---|
97 |
|
---|
98 | /**
|
---|
99 | * Right square bracket (literal): ']'
|
---|
100 | */
|
---|
101 |
|
---|
102 | if (value === CHAR_RIGHT_SQUARE_BRACKET) {
|
---|
103 | push({ type: 'text', value: '\\' + value });
|
---|
104 | continue;
|
---|
105 | }
|
---|
106 |
|
---|
107 | /**
|
---|
108 | * Left square bracket: '['
|
---|
109 | */
|
---|
110 |
|
---|
111 | if (value === CHAR_LEFT_SQUARE_BRACKET) {
|
---|
112 | brackets++;
|
---|
113 |
|
---|
114 | let closed = true;
|
---|
115 | let next;
|
---|
116 |
|
---|
117 | while (index < length && (next = advance())) {
|
---|
118 | value += next;
|
---|
119 |
|
---|
120 | if (next === CHAR_LEFT_SQUARE_BRACKET) {
|
---|
121 | brackets++;
|
---|
122 | continue;
|
---|
123 | }
|
---|
124 |
|
---|
125 | if (next === CHAR_BACKSLASH) {
|
---|
126 | value += advance();
|
---|
127 | continue;
|
---|
128 | }
|
---|
129 |
|
---|
130 | if (next === CHAR_RIGHT_SQUARE_BRACKET) {
|
---|
131 | brackets--;
|
---|
132 |
|
---|
133 | if (brackets === 0) {
|
---|
134 | break;
|
---|
135 | }
|
---|
136 | }
|
---|
137 | }
|
---|
138 |
|
---|
139 | push({ type: 'text', value });
|
---|
140 | continue;
|
---|
141 | }
|
---|
142 |
|
---|
143 | /**
|
---|
144 | * Parentheses
|
---|
145 | */
|
---|
146 |
|
---|
147 | if (value === CHAR_LEFT_PARENTHESES) {
|
---|
148 | block = push({ type: 'paren', nodes: [] });
|
---|
149 | stack.push(block);
|
---|
150 | push({ type: 'text', value });
|
---|
151 | continue;
|
---|
152 | }
|
---|
153 |
|
---|
154 | if (value === CHAR_RIGHT_PARENTHESES) {
|
---|
155 | if (block.type !== 'paren') {
|
---|
156 | push({ type: 'text', value });
|
---|
157 | continue;
|
---|
158 | }
|
---|
159 | block = stack.pop();
|
---|
160 | push({ type: 'text', value });
|
---|
161 | block = stack[stack.length - 1];
|
---|
162 | continue;
|
---|
163 | }
|
---|
164 |
|
---|
165 | /**
|
---|
166 | * Quotes: '|"|`
|
---|
167 | */
|
---|
168 |
|
---|
169 | if (value === CHAR_DOUBLE_QUOTE || value === CHAR_SINGLE_QUOTE || value === CHAR_BACKTICK) {
|
---|
170 | let open = value;
|
---|
171 | let next;
|
---|
172 |
|
---|
173 | if (options.keepQuotes !== true) {
|
---|
174 | value = '';
|
---|
175 | }
|
---|
176 |
|
---|
177 | while (index < length && (next = advance())) {
|
---|
178 | if (next === CHAR_BACKSLASH) {
|
---|
179 | value += next + advance();
|
---|
180 | continue;
|
---|
181 | }
|
---|
182 |
|
---|
183 | if (next === open) {
|
---|
184 | if (options.keepQuotes === true) value += next;
|
---|
185 | break;
|
---|
186 | }
|
---|
187 |
|
---|
188 | value += next;
|
---|
189 | }
|
---|
190 |
|
---|
191 | push({ type: 'text', value });
|
---|
192 | continue;
|
---|
193 | }
|
---|
194 |
|
---|
195 | /**
|
---|
196 | * Left curly brace: '{'
|
---|
197 | */
|
---|
198 |
|
---|
199 | if (value === CHAR_LEFT_CURLY_BRACE) {
|
---|
200 | depth++;
|
---|
201 |
|
---|
202 | let dollar = prev.value && prev.value.slice(-1) === '$' || block.dollar === true;
|
---|
203 | let brace = {
|
---|
204 | type: 'brace',
|
---|
205 | open: true,
|
---|
206 | close: false,
|
---|
207 | dollar,
|
---|
208 | depth,
|
---|
209 | commas: 0,
|
---|
210 | ranges: 0,
|
---|
211 | nodes: []
|
---|
212 | };
|
---|
213 |
|
---|
214 | block = push(brace);
|
---|
215 | stack.push(block);
|
---|
216 | push({ type: 'open', value });
|
---|
217 | continue;
|
---|
218 | }
|
---|
219 |
|
---|
220 | /**
|
---|
221 | * Right curly brace: '}'
|
---|
222 | */
|
---|
223 |
|
---|
224 | if (value === CHAR_RIGHT_CURLY_BRACE) {
|
---|
225 | if (block.type !== 'brace') {
|
---|
226 | push({ type: 'text', value });
|
---|
227 | continue;
|
---|
228 | }
|
---|
229 |
|
---|
230 | let type = 'close';
|
---|
231 | block = stack.pop();
|
---|
232 | block.close = true;
|
---|
233 |
|
---|
234 | push({ type, value });
|
---|
235 | depth--;
|
---|
236 |
|
---|
237 | block = stack[stack.length - 1];
|
---|
238 | continue;
|
---|
239 | }
|
---|
240 |
|
---|
241 | /**
|
---|
242 | * Comma: ','
|
---|
243 | */
|
---|
244 |
|
---|
245 | if (value === CHAR_COMMA && depth > 0) {
|
---|
246 | if (block.ranges > 0) {
|
---|
247 | block.ranges = 0;
|
---|
248 | let open = block.nodes.shift();
|
---|
249 | block.nodes = [open, { type: 'text', value: stringify(block) }];
|
---|
250 | }
|
---|
251 |
|
---|
252 | push({ type: 'comma', value });
|
---|
253 | block.commas++;
|
---|
254 | continue;
|
---|
255 | }
|
---|
256 |
|
---|
257 | /**
|
---|
258 | * Dot: '.'
|
---|
259 | */
|
---|
260 |
|
---|
261 | if (value === CHAR_DOT && depth > 0 && block.commas === 0) {
|
---|
262 | let siblings = block.nodes;
|
---|
263 |
|
---|
264 | if (depth === 0 || siblings.length === 0) {
|
---|
265 | push({ type: 'text', value });
|
---|
266 | continue;
|
---|
267 | }
|
---|
268 |
|
---|
269 | if (prev.type === 'dot') {
|
---|
270 | block.range = [];
|
---|
271 | prev.value += value;
|
---|
272 | prev.type = 'range';
|
---|
273 |
|
---|
274 | if (block.nodes.length !== 3 && block.nodes.length !== 5) {
|
---|
275 | block.invalid = true;
|
---|
276 | block.ranges = 0;
|
---|
277 | prev.type = 'text';
|
---|
278 | continue;
|
---|
279 | }
|
---|
280 |
|
---|
281 | block.ranges++;
|
---|
282 | block.args = [];
|
---|
283 | continue;
|
---|
284 | }
|
---|
285 |
|
---|
286 | if (prev.type === 'range') {
|
---|
287 | siblings.pop();
|
---|
288 |
|
---|
289 | let before = siblings[siblings.length - 1];
|
---|
290 | before.value += prev.value + value;
|
---|
291 | prev = before;
|
---|
292 | block.ranges--;
|
---|
293 | continue;
|
---|
294 | }
|
---|
295 |
|
---|
296 | push({ type: 'dot', value });
|
---|
297 | continue;
|
---|
298 | }
|
---|
299 |
|
---|
300 | /**
|
---|
301 | * Text
|
---|
302 | */
|
---|
303 |
|
---|
304 | push({ type: 'text', value });
|
---|
305 | }
|
---|
306 |
|
---|
307 | // Mark imbalanced braces and brackets as invalid
|
---|
308 | do {
|
---|
309 | block = stack.pop();
|
---|
310 |
|
---|
311 | if (block.type !== 'root') {
|
---|
312 | block.nodes.forEach(node => {
|
---|
313 | if (!node.nodes) {
|
---|
314 | if (node.type === 'open') node.isOpen = true;
|
---|
315 | if (node.type === 'close') node.isClose = true;
|
---|
316 | if (!node.nodes) node.type = 'text';
|
---|
317 | node.invalid = true;
|
---|
318 | }
|
---|
319 | });
|
---|
320 |
|
---|
321 | // get the location of the block on parent.nodes (block's siblings)
|
---|
322 | let parent = stack[stack.length - 1];
|
---|
323 | let index = parent.nodes.indexOf(block);
|
---|
324 | // replace the (invalid) block with it's nodes
|
---|
325 | parent.nodes.splice(index, 1, ...block.nodes);
|
---|
326 | }
|
---|
327 | } while (stack.length > 0);
|
---|
328 |
|
---|
329 | push({ type: 'eos' });
|
---|
330 | return ast;
|
---|
331 | };
|
---|
332 |
|
---|
333 | module.exports = parse;
|
---|