]>
Commit | Line | Data |
---|---|---|
8bb4bdeb XL |
1 | // Copyright 2012-2014 The Rust Project Developers. See the COPYRIGHT |
2 | // file at the top-level directory of this distribution and at | |
3 | // http://rust-lang.org/COPYRIGHT. | |
4 | // | |
5 | // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or | |
6 | // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license | |
7 | // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your | |
8 | // option. This file may not be copied, modified, or distributed | |
9 | // except according to those terms. | |
10 | ||
11 | use core::cmp; | |
12 | use core::iter::Filter; | |
13 | ||
14 | use tables::word::WordCat; | |
15 | ||
16 | /// An iterator over the substrings of a string which, after splitting the string on | |
17 | /// [word boundaries](http://www.unicode.org/reports/tr29/#Word_Boundaries), | |
18 | /// contain any characters with the | |
19 | /// [Alphabetic](http://unicode.org/reports/tr44/#Alphabetic) | |
20 | /// property, or with | |
21 | /// [General_Category=Number](http://unicode.org/reports/tr44/#General_Category_Values). | |
22 | pub struct UnicodeWords<'a> { | |
23 | inner: Filter<UWordBounds<'a>, fn(&&str) -> bool>, | |
24 | } | |
25 | ||
26 | impl<'a> Iterator for UnicodeWords<'a> { | |
27 | type Item = &'a str; | |
28 | ||
29 | #[inline] | |
30 | fn next(&mut self) -> Option<&'a str> { self.inner.next() } | |
31 | } | |
32 | impl<'a> DoubleEndedIterator for UnicodeWords<'a> { | |
33 | #[inline] | |
34 | fn next_back(&mut self) -> Option<&'a str> { self.inner.next_back() } | |
35 | } | |
36 | ||
37 | /// External iterator for a string's | |
38 | /// [word boundaries](http://www.unicode.org/reports/tr29/#Word_Boundaries). | |
39 | #[derive(Clone)] | |
40 | pub struct UWordBounds<'a> { | |
41 | string: &'a str, | |
42 | cat: Option<WordCat>, | |
43 | catb: Option<WordCat>, | |
44 | } | |
45 | ||
46 | /// External iterator for word boundaries and byte offsets. | |
47 | #[derive(Clone)] | |
48 | pub struct UWordBoundIndices<'a> { | |
49 | start_offset: usize, | |
50 | iter: UWordBounds<'a>, | |
51 | } | |
52 | ||
8bb4bdeb XL |
53 | impl<'a> Iterator for UWordBoundIndices<'a> { |
54 | type Item = (usize, &'a str); | |
55 | ||
56 | #[inline] | |
57 | fn next(&mut self) -> Option<(usize, &'a str)> { | |
58 | self.iter.next().map(|s| (s.as_ptr() as usize - self.start_offset, s)) | |
59 | } | |
60 | ||
61 | #[inline] | |
62 | fn size_hint(&self) -> (usize, Option<usize>) { | |
63 | self.iter.size_hint() | |
64 | } | |
65 | } | |
66 | ||
67 | impl<'a> DoubleEndedIterator for UWordBoundIndices<'a> { | |
68 | #[inline] | |
69 | fn next_back(&mut self) -> Option<(usize, &'a str)> { | |
70 | self.iter.next_back().map(|s| (s.as_ptr() as usize - self.start_offset, s)) | |
71 | } | |
72 | } | |
73 | ||
74 | // state machine for word boundary rules | |
7cac9316 | 75 | #[derive(Clone,Copy,PartialEq,Eq)] |
8bb4bdeb XL |
76 | enum UWordBoundsState { |
77 | Start, | |
78 | Letter, | |
79 | HLetter, | |
80 | Numeric, | |
81 | Katakana, | |
82 | ExtendNumLet, | |
7cac9316 | 83 | Regional, |
8bb4bdeb | 84 | FormatExtend(FormatExtendType), |
8bb4bdeb XL |
85 | } |
86 | ||
87 | // subtypes for FormatExtend state in UWordBoundsState | |
7cac9316 | 88 | #[derive(Clone,Copy,PartialEq,Eq)] |
8bb4bdeb XL |
89 | enum FormatExtendType { |
90 | AcceptAny, | |
91 | AcceptNone, | |
92 | RequireLetter, | |
93 | RequireHLetter, | |
94 | AcceptQLetter, | |
95 | RequireNumeric, | |
96 | } | |
97 | ||
8bb4bdeb XL |
98 | impl<'a> Iterator for UWordBounds<'a> { |
99 | type Item = &'a str; | |
100 | ||
101 | #[inline] | |
102 | fn size_hint(&self) -> (usize, Option<usize>) { | |
103 | let slen = self.string.len(); | |
104 | (cmp::min(slen, 1), Some(slen)) | |
105 | } | |
106 | ||
107 | #[inline] | |
108 | fn next(&mut self) -> Option<&'a str> { | |
109 | use self::UWordBoundsState::*; | |
110 | use self::FormatExtendType::*; | |
111 | use tables::word as wd; | |
112 | if self.string.len() == 0 { | |
113 | return None; | |
114 | } | |
115 | ||
116 | let mut take_curr = true; | |
117 | let mut take_cat = true; | |
118 | let mut idx = 0; | |
119 | let mut saveidx = 0; | |
120 | let mut state = Start; | |
121 | let mut cat = wd::WC_Any; | |
122 | let mut savecat = wd::WC_Any; | |
8bb4bdeb XL |
123 | for (curr, ch) in self.string.char_indices() { |
124 | idx = curr; | |
7cac9316 | 125 | |
8bb4bdeb XL |
126 | // if there's a category cached, grab it |
127 | cat = match self.cat { | |
128 | None => wd::word_category(ch), | |
129 | _ => self.cat.take().unwrap() | |
130 | }; | |
131 | take_cat = true; | |
132 | ||
133 | // handle rule WB4 | |
7cac9316 | 134 | // just skip all format and extend chars |
8bb4bdeb XL |
135 | // note that Start is a special case: if there's a bunch of Format | Extend |
136 | // characters at the beginning of a block of text, dump them out as one unit. | |
137 | // | |
138 | // (This is not obvious from the wording of UAX#29, but if you look at the | |
139 | // test cases http://www.unicode.org/Public/UNIDATA/auxiliary/WordBreakTest.txt | |
140 | // then the "correct" interpretation of WB4 becomes apparent.) | |
7cac9316 XL |
141 | if state != Start && (cat == wd::WC_Extend || cat == wd::WC_Format) { |
142 | continue; | |
8bb4bdeb XL |
143 | } |
144 | ||
8bb4bdeb XL |
145 | state = match state { |
146 | Start if cat == wd::WC_CR => { | |
147 | idx += match self.get_next_cat(idx) { | |
148 | Some(ncat) if ncat == wd::WC_LF => 1, // rule WB3 | |
149 | _ => 0 | |
150 | }; | |
151 | break; // rule WB3a | |
152 | }, | |
153 | Start => match cat { | |
154 | wd::WC_ALetter => Letter, // rule WB5, WB6, WB9, WB13a | |
155 | wd::WC_Hebrew_Letter => HLetter, // rule WB5, WB6, WB7a, WB7b, WB9, WB13a | |
156 | wd::WC_Numeric => Numeric, // rule WB8, WB10, WB12, WB13a | |
157 | wd::WC_Katakana => Katakana, // rule WB13, WB13a | |
158 | wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13a, WB13b | |
7cac9316 | 159 | wd::WC_Regional_Indicator => Regional, // rule WB13c |
8bb4bdeb | 160 | wd::WC_LF | wd::WC_Newline => break, // rule WB3a |
8bb4bdeb XL |
161 | _ => { |
162 | if let Some(ncat) = self.get_next_cat(idx) { // rule WB4 | |
7cac9316 | 163 | if ncat == wd::WC_Format || ncat == wd::WC_Extend { |
8bb4bdeb XL |
164 | state = FormatExtend(AcceptNone); |
165 | self.cat = Some(ncat); | |
166 | continue; | |
167 | } | |
168 | } | |
7cac9316 | 169 | break; // rule WB14 |
8bb4bdeb XL |
170 | } |
171 | }, | |
8bb4bdeb XL |
172 | Letter | HLetter => match cat { |
173 | wd::WC_ALetter => Letter, // rule WB5 | |
174 | wd::WC_Hebrew_Letter => HLetter, // rule WB5 | |
175 | wd::WC_Numeric => Numeric, // rule WB9 | |
176 | wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13a | |
177 | wd::WC_Double_Quote if state == HLetter => { | |
178 | savecat = cat; | |
179 | saveidx = idx; | |
180 | FormatExtend(RequireHLetter) // rule WB7b | |
181 | }, | |
182 | wd::WC_Single_Quote if state == HLetter => { | |
183 | FormatExtend(AcceptQLetter) // rule WB7a | |
184 | }, | |
185 | wd::WC_MidLetter | wd::WC_MidNumLet | wd::WC_Single_Quote => { | |
186 | savecat = cat; | |
187 | saveidx = idx; | |
188 | FormatExtend(RequireLetter) // rule WB6 | |
189 | }, | |
190 | _ => { | |
191 | take_curr = false; | |
192 | break; | |
193 | } | |
194 | }, | |
195 | Numeric => match cat { | |
196 | wd::WC_Numeric => Numeric, // rule WB8 | |
197 | wd::WC_ALetter => Letter, // rule WB10 | |
198 | wd::WC_Hebrew_Letter => HLetter, // rule WB10 | |
199 | wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13a | |
200 | wd::WC_MidNum | wd::WC_MidNumLet | wd::WC_Single_Quote => { | |
201 | savecat = cat; | |
202 | saveidx = idx; | |
203 | FormatExtend(RequireNumeric) // rule WB12 | |
204 | }, | |
205 | _ => { | |
206 | take_curr = false; | |
207 | break; | |
208 | } | |
209 | }, | |
210 | Katakana => match cat { | |
211 | wd::WC_Katakana => Katakana, // rule WB13 | |
212 | wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13a | |
213 | _ => { | |
214 | take_curr = false; | |
215 | break; | |
216 | } | |
217 | }, | |
218 | ExtendNumLet => match cat { | |
219 | wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13a | |
220 | wd::WC_ALetter => Letter, // rule WB13b | |
221 | wd::WC_Hebrew_Letter => HLetter, // rule WB13b | |
222 | wd::WC_Numeric => Numeric, // rule WB13b | |
223 | wd::WC_Katakana => Katakana, // rule WB13b | |
224 | _ => { | |
225 | take_curr = false; | |
226 | break; | |
227 | } | |
228 | }, | |
7cac9316 XL |
229 | Regional => match cat { |
230 | wd::WC_Regional_Indicator => Regional, // rule WB13c | |
8bb4bdeb XL |
231 | _ => { |
232 | take_curr = false; | |
233 | break; | |
234 | } | |
235 | }, | |
236 | FormatExtend(t) => match t { // handle FormatExtends depending on what type | |
237 | RequireNumeric if cat == wd::WC_Numeric => Numeric, // rule WB11 | |
238 | RequireLetter | AcceptQLetter if cat == wd::WC_ALetter => Letter, // rule WB7 | |
239 | RequireLetter | AcceptQLetter if cat == wd::WC_Hebrew_Letter => HLetter, // WB7a | |
240 | RequireHLetter if cat == wd::WC_Hebrew_Letter => HLetter, // rule WB7b | |
241 | AcceptNone | AcceptQLetter => { | |
242 | take_curr = false; // emit all the Format|Extend characters | |
243 | take_cat = false; | |
244 | break; | |
245 | }, | |
246 | _ => break // rewind (in if statement below) | |
247 | } | |
248 | } | |
249 | } | |
250 | ||
251 | if let FormatExtend(t) = state { | |
252 | // we were looking for something and didn't find it; we have to back up | |
253 | if t == RequireLetter || t == RequireHLetter || t == RequireNumeric { | |
254 | idx = saveidx; | |
255 | cat = savecat; | |
256 | take_curr = false; | |
257 | } | |
258 | } | |
259 | ||
260 | self.cat = if take_curr { | |
261 | idx = idx + self.string[idx..].chars().next().unwrap().len_utf8(); | |
262 | None | |
263 | } else if take_cat { | |
264 | Some(cat) | |
265 | } else { | |
266 | None | |
267 | }; | |
268 | ||
269 | let retstr = &self.string[..idx]; | |
270 | self.string = &self.string[idx..]; | |
271 | Some(retstr) | |
272 | } | |
273 | } | |
274 | ||
275 | impl<'a> DoubleEndedIterator for UWordBounds<'a> { | |
276 | #[inline] | |
277 | fn next_back(&mut self) -> Option<&'a str> { | |
278 | use self::UWordBoundsState::*; | |
279 | use self::FormatExtendType::*; | |
280 | use tables::word as wd; | |
281 | if self.string.len() == 0 { | |
282 | return None; | |
283 | } | |
284 | ||
285 | let mut take_curr = true; | |
286 | let mut take_cat = true; | |
287 | let mut idx = self.string.len(); | |
288 | idx -= self.string.chars().next_back().unwrap().len_utf8(); | |
289 | let mut previdx = idx; | |
290 | let mut saveidx = idx; | |
291 | let mut state = Start; | |
292 | let mut savestate = Start; | |
293 | let mut cat = wd::WC_Any; | |
8bb4bdeb XL |
294 | for (curr, ch) in self.string.char_indices().rev() { |
295 | previdx = idx; | |
296 | idx = curr; | |
297 | ||
298 | // if there's a category cached, grab it | |
299 | cat = match self.catb { | |
300 | None => wd::word_category(ch), | |
301 | _ => self.catb.take().unwrap() | |
302 | }; | |
303 | take_cat = true; | |
304 | ||
305 | // backward iterator over word boundaries. Mostly the same as the forward | |
306 | // iterator, with two weirdnesses: | |
307 | // (1) If we encounter a single quote in the Start state, we have to check for a | |
308 | // Hebrew Letter immediately before it. | |
309 | // (2) Format and Extend char handling takes some gymnastics. | |
310 | ||
7cac9316 | 311 | if cat == wd::WC_Extend || cat == wd::WC_Format { |
8bb4bdeb XL |
312 | if match state { |
313 | FormatExtend(_) | Start => false, | |
314 | _ => true | |
315 | } { | |
316 | saveidx = previdx; | |
317 | savestate = state; | |
318 | state = FormatExtend(AcceptNone); | |
319 | } | |
320 | ||
321 | if state != Start { | |
322 | continue; | |
323 | } | |
324 | } else if state == FormatExtend(AcceptNone) { | |
325 | // finished a scan of some Format|Extend chars, restore previous state | |
326 | state = savestate; | |
327 | previdx = saveidx; | |
328 | take_cat = false; | |
329 | } | |
330 | ||
8bb4bdeb XL |
331 | state = match state { |
332 | Start | FormatExtend(AcceptAny) => match cat { | |
333 | wd::WC_ALetter => Letter, // rule WB5, WB7, WB10, WB13b | |
334 | wd::WC_Hebrew_Letter => HLetter, // rule WB5, WB7, WB7c, WB10, WB13b | |
335 | wd::WC_Numeric => Numeric, // rule WB8, WB9, WB11, WB13b | |
336 | wd::WC_Katakana => Katakana, // rule WB13, WB13b | |
337 | wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13a | |
7cac9316 XL |
338 | wd::WC_Regional_Indicator => Regional, // rule WB13c |
339 | wd::WC_Extend | wd::WC_Format => FormatExtend(AcceptAny), // rule WB4 | |
8bb4bdeb XL |
340 | wd::WC_Single_Quote => { |
341 | saveidx = idx; | |
342 | FormatExtend(AcceptQLetter) // rule WB7a | |
343 | }, | |
8bb4bdeb XL |
344 | wd::WC_CR | wd::WC_LF | wd::WC_Newline => { |
345 | if state == Start { | |
346 | if cat == wd::WC_LF { | |
347 | idx -= match self.get_prev_cat(idx) { | |
348 | Some(pcat) if pcat == wd::WC_CR => 1, // rule WB3 | |
349 | _ => 0 | |
350 | }; | |
351 | } | |
352 | } else { | |
353 | take_curr = false; | |
354 | } | |
355 | break; // rule WB3a | |
356 | }, | |
7cac9316 | 357 | _ => break // rule WB14 |
8bb4bdeb XL |
358 | }, |
359 | Letter | HLetter => match cat { | |
360 | wd::WC_ALetter => Letter, // rule WB5 | |
361 | wd::WC_Hebrew_Letter => HLetter, // rule WB5 | |
362 | wd::WC_Numeric => Numeric, // rule WB10 | |
363 | wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13b | |
364 | wd::WC_Double_Quote if state == HLetter => { | |
365 | saveidx = previdx; | |
366 | FormatExtend(RequireHLetter) // rule WB7c | |
367 | }, | |
368 | wd::WC_MidLetter | wd::WC_MidNumLet | wd::WC_Single_Quote => { | |
369 | saveidx = previdx; | |
370 | FormatExtend(RequireLetter) // rule WB7 | |
371 | }, | |
372 | _ => { | |
373 | take_curr = false; | |
374 | break; | |
375 | } | |
376 | }, | |
377 | Numeric => match cat { | |
378 | wd::WC_Numeric => Numeric, // rule WB8 | |
379 | wd::WC_ALetter => Letter, // rule WB9 | |
380 | wd::WC_Hebrew_Letter => HLetter, // rule WB9 | |
381 | wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13b | |
382 | wd::WC_MidNum | wd::WC_MidNumLet | wd::WC_Single_Quote => { | |
383 | saveidx = previdx; | |
384 | FormatExtend(RequireNumeric) // rule WB11 | |
385 | }, | |
386 | _ => { | |
387 | take_curr = false; | |
388 | break; | |
389 | } | |
390 | }, | |
391 | Katakana => match cat { | |
392 | wd::WC_Katakana => Katakana, // rule WB13 | |
393 | wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13b | |
394 | _ => { | |
395 | take_curr = false; | |
396 | break; | |
397 | } | |
398 | }, | |
399 | ExtendNumLet => match cat { | |
400 | wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13a | |
401 | wd::WC_ALetter => Letter, // rule WB13a | |
402 | wd::WC_Hebrew_Letter => HLetter, // rule WB13a | |
403 | wd::WC_Numeric => Numeric, // rule WB13a | |
404 | wd::WC_Katakana => Katakana, // rule WB13a | |
405 | _ => { | |
406 | take_curr = false; | |
407 | break; | |
408 | } | |
409 | }, | |
7cac9316 XL |
410 | Regional => match cat { |
411 | wd::WC_Regional_Indicator => Regional, // rule WB13c | |
8bb4bdeb XL |
412 | _ => { |
413 | take_curr = false; | |
414 | break; | |
415 | } | |
416 | }, | |
417 | FormatExtend(t) => match t { | |
418 | RequireNumeric if cat == wd::WC_Numeric => Numeric, // rule WB12 | |
419 | RequireLetter if cat == wd::WC_ALetter => Letter, // rule WB6 | |
420 | RequireLetter if cat == wd::WC_Hebrew_Letter => HLetter, // rule WB6 | |
421 | AcceptQLetter if cat == wd::WC_Hebrew_Letter => HLetter, // rule WB7a | |
422 | RequireHLetter if cat == wd::WC_Hebrew_Letter => HLetter, // rule WB7b | |
423 | _ => break // backtrack will happens | |
424 | } | |
425 | } | |
426 | } | |
427 | ||
428 | if let FormatExtend(t) = state { | |
429 | // if we required something but didn't find it, backtrack | |
430 | if t == RequireLetter || t == RequireHLetter || | |
431 | t == RequireNumeric || t == AcceptNone || t == AcceptQLetter { | |
432 | previdx = saveidx; | |
433 | take_cat = false; | |
434 | take_curr = false; | |
435 | } | |
436 | } | |
437 | ||
438 | self.catb = if take_curr { | |
439 | None | |
440 | } else { | |
441 | idx = previdx; | |
442 | if take_cat { | |
443 | Some(cat) | |
444 | } else { | |
445 | None | |
446 | } | |
447 | }; | |
448 | ||
449 | let retstr = &self.string[idx..]; | |
450 | self.string = &self.string[..idx]; | |
451 | Some(retstr) | |
452 | } | |
453 | } | |
454 | ||
455 | impl<'a> UWordBounds<'a> { | |
8bb4bdeb XL |
456 | #[inline] |
457 | fn get_next_cat(&self, idx: usize) -> Option<WordCat> { | |
458 | use tables::word as wd; | |
459 | let nidx = idx + self.string[idx..].chars().next().unwrap().len_utf8(); | |
460 | if nidx < self.string.len() { | |
461 | let nch = self.string[nidx..].chars().next().unwrap(); | |
462 | Some(wd::word_category(nch)) | |
463 | } else { | |
464 | None | |
465 | } | |
466 | } | |
467 | ||
468 | #[inline] | |
469 | fn get_prev_cat(&self, idx: usize) -> Option<WordCat> { | |
470 | use tables::word as wd; | |
471 | if idx > 0 { | |
472 | let nch = self.string[..idx].chars().next_back().unwrap(); | |
473 | Some(wd::word_category(nch)) | |
474 | } else { | |
475 | None | |
476 | } | |
477 | } | |
478 | } | |
479 | ||
480 | #[inline] | |
481 | pub fn new_word_bounds<'b>(s: &'b str) -> UWordBounds<'b> { | |
482 | UWordBounds { string: s, cat: None, catb: None } | |
483 | } | |
484 | ||
485 | #[inline] | |
486 | pub fn new_word_bound_indices<'b>(s: &'b str) -> UWordBoundIndices<'b> { | |
487 | UWordBoundIndices { start_offset: s.as_ptr() as usize, iter: new_word_bounds(s) } | |
488 | } | |
489 | ||
490 | #[inline] | |
491 | pub fn new_unicode_words<'b>(s: &'b str) -> UnicodeWords<'b> { | |
492 | use super::UnicodeSegmentation; | |
493 | use tables::util::is_alphanumeric; | |
494 | ||
495 | fn has_alphanumeric(s: &&str) -> bool { s.chars().any(|c| is_alphanumeric(c)) } | |
496 | let has_alphanumeric: fn(&&str) -> bool = has_alphanumeric; // coerce to fn pointer | |
497 | ||
498 | UnicodeWords { inner: s.split_word_bounds().filter(has_alphanumeric) } | |
499 | } |