| 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 crate::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 |
/// |
| 23 |
/// This struct is created by the [`unicode_words`] method on the [`UnicodeSegmentation`] trait. See |
| 24 |
/// its documentation for more. |
| 25 |
/// |
| 26 |
/// [`unicode_words`]: trait.UnicodeSegmentation.html#tymethod.unicode_words |
| 27 |
/// [`UnicodeSegmentation`]: trait.UnicodeSegmentation.html |
| 28 |
pub struct UnicodeWords<'a> { |
| 29 |
inner: Filter<UWordBounds<'a>, fn(&&str) -> bool>, |
| 30 |
} |
| 31 |
|
| 32 |
impl<'a> Iterator for UnicodeWords<'a> { |
| 33 |
type Item = &'a str; |
| 34 |
|
| 35 |
#[inline] |
| 36 |
fn next(&mut self) -> Option<&'a str> { |
| 37 |
self.inner.next() |
| 38 |
} |
| 39 |
|
| 40 |
#[inline] |
| 41 |
fn size_hint(&self) -> (usize, Option<usize>) { |
| 42 |
self.inner.size_hint() |
| 43 |
} |
| 44 |
} |
| 45 |
impl<'a> DoubleEndedIterator for UnicodeWords<'a> { |
| 46 |
#[inline] |
| 47 |
fn next_back(&mut self) -> Option<&'a str> { |
| 48 |
self.inner.next_back() |
| 49 |
} |
| 50 |
} |
| 51 |
|
| 52 |
/// An iterator over the substrings of a string which, after splitting the string on |
| 53 |
/// [word boundaries](http://www.unicode.org/reports/tr29/#Word_Boundaries), |
| 54 |
/// contain any characters with the |
| 55 |
/// [Alphabetic](http://unicode.org/reports/tr44/#Alphabetic) |
| 56 |
/// property, or with |
| 57 |
/// [General_Category=Number](http://unicode.org/reports/tr44/#General_Category_Values). |
| 58 |
/// This iterator also provides the byte offsets for each substring. |
| 59 |
/// |
| 60 |
/// This struct is created by the [`unicode_word_indices`] method on the [`UnicodeSegmentation`] trait. See |
| 61 |
/// its documentation for more. |
| 62 |
/// |
| 63 |
/// [`unicode_word_indices`]: trait.UnicodeSegmentation.html#tymethod.unicode_word_indices |
| 64 |
/// [`UnicodeSegmentation`]: trait.UnicodeSegmentation.html |
| 65 |
pub struct UnicodeWordIndices<'a> { |
| 66 |
inner: Filter<UWordBoundIndices<'a>, fn(&(usize, &str)) -> bool>, |
| 67 |
} |
| 68 |
|
| 69 |
impl<'a> Iterator for UnicodeWordIndices<'a> { |
| 70 |
type Item = (usize, &'a str); |
| 71 |
|
| 72 |
#[inline] |
| 73 |
fn next(&mut self) -> Option<(usize, &'a str)> { |
| 74 |
self.inner.next() |
| 75 |
} |
| 76 |
|
| 77 |
#[inline] |
| 78 |
fn size_hint(&self) -> (usize, Option<usize>) { |
| 79 |
self.inner.size_hint() |
| 80 |
} |
| 81 |
} |
| 82 |
impl<'a> DoubleEndedIterator for UnicodeWordIndices<'a> { |
| 83 |
#[inline] |
| 84 |
fn next_back(&mut self) -> Option<(usize, &'a str)> { |
| 85 |
self.inner.next_back() |
| 86 |
} |
| 87 |
} |
| 88 |
|
| 89 |
/// External iterator for a string's |
| 90 |
/// [word boundaries](http://www.unicode.org/reports/tr29/#Word_Boundaries). |
| 91 |
/// |
| 92 |
/// This struct is created by the [`split_word_bounds`] method on the [`UnicodeSegmentation`] |
| 93 |
/// trait. See its documentation for more. |
| 94 |
/// |
| 95 |
/// [`split_word_bounds`]: trait.UnicodeSegmentation.html#tymethod.split_word_bounds |
| 96 |
/// [`UnicodeSegmentation`]: trait.UnicodeSegmentation.html |
| 97 |
#[derive(Clone)] |
| 98 |
pub struct UWordBounds<'a> { |
| 99 |
string: &'a str, |
| 100 |
cat: Option<WordCat>, |
| 101 |
catb: Option<WordCat>, |
| 102 |
} |
| 103 |
|
| 104 |
/// External iterator for word boundaries and byte offsets. |
| 105 |
/// |
| 106 |
/// This struct is created by the [`split_word_bound_indices`] method on the |
| 107 |
/// [`UnicodeSegmentation`] trait. See its documentation for more. |
| 108 |
/// |
| 109 |
/// [`split_word_bound_indices`]: trait.UnicodeSegmentation.html#tymethod.split_word_bound_indices |
| 110 |
/// [`UnicodeSegmentation`]: trait.UnicodeSegmentation.html |
| 111 |
#[derive(Clone)] |
| 112 |
pub struct UWordBoundIndices<'a> { |
| 113 |
start_offset: usize, |
| 114 |
iter: UWordBounds<'a>, |
| 115 |
} |
| 116 |
|
| 117 |
impl<'a> UWordBoundIndices<'a> { |
| 118 |
#[inline] |
| 119 |
/// View the underlying data (the part yet to be iterated) as a slice of the original string. |
| 120 |
/// |
| 121 |
/// ```rust |
| 122 |
/// # use unicode_segmentation::UnicodeSegmentation; |
| 123 |
/// let mut iter = "Hello world".split_word_bound_indices(); |
| 124 |
/// assert_eq!(iter.as_str(), "Hello world"); |
| 125 |
/// iter.next(); |
| 126 |
/// assert_eq!(iter.as_str(), " world"); |
| 127 |
/// iter.next(); |
| 128 |
/// assert_eq!(iter.as_str(), "world"); |
| 129 |
/// ``` |
| 130 |
pub fn as_str(&self) -> &'a str { |
| 131 |
self.iter.as_str() |
| 132 |
} |
| 133 |
} |
| 134 |
|
| 135 |
impl<'a> Iterator for UWordBoundIndices<'a> { |
| 136 |
type Item = (usize, &'a str); |
| 137 |
|
| 138 |
#[inline] |
| 139 |
fn next(&mut self) -> Option<(usize, &'a str)> { |
| 140 |
self.iter |
| 141 |
.next() |
| 142 |
.map(|s| (s.as_ptr() as usize - self.start_offset, s)) |
| 143 |
} |
| 144 |
|
| 145 |
#[inline] |
| 146 |
fn size_hint(&self) -> (usize, Option<usize>) { |
| 147 |
self.iter.size_hint() |
| 148 |
} |
| 149 |
} |
| 150 |
|
| 151 |
impl<'a> DoubleEndedIterator for UWordBoundIndices<'a> { |
| 152 |
#[inline] |
| 153 |
fn next_back(&mut self) -> Option<(usize, &'a str)> { |
| 154 |
self.iter |
| 155 |
.next_back() |
| 156 |
.map(|s| (s.as_ptr() as usize - self.start_offset, s)) |
| 157 |
} |
| 158 |
} |
| 159 |
|
| 160 |
// state machine for word boundary rules |
| 161 |
#[derive(Clone, Copy, PartialEq, Eq, Debug)] |
| 162 |
enum UWordBoundsState { |
| 163 |
Start, |
| 164 |
Letter, |
| 165 |
HLetter, |
| 166 |
Numeric, |
| 167 |
Katakana, |
| 168 |
ExtendNumLet, |
| 169 |
Regional(RegionalState), |
| 170 |
FormatExtend(FormatExtendType), |
| 171 |
Zwj, |
| 172 |
Emoji, |
| 173 |
WSegSpace, |
| 174 |
} |
| 175 |
|
| 176 |
// subtypes for FormatExtend state in UWordBoundsState |
| 177 |
#[derive(Clone, Copy, PartialEq, Eq, Debug)] |
| 178 |
enum FormatExtendType { |
| 179 |
AcceptAny, |
| 180 |
AcceptNone, |
| 181 |
RequireLetter, |
| 182 |
RequireHLetter, |
| 183 |
AcceptQLetter, |
| 184 |
RequireNumeric, |
| 185 |
} |
| 186 |
|
| 187 |
#[derive(Clone, Copy, PartialEq, Eq, Debug)] |
| 188 |
enum RegionalState { |
| 189 |
Half, |
| 190 |
Full, |
| 191 |
Unknown, |
| 192 |
} |
| 193 |
|
| 194 |
fn is_emoji(ch: char) -> bool { |
| 195 |
use crate::tables::emoji; |
| 196 |
emoji::emoji_category(ch).2 == emoji::EmojiCat::EC_Extended_Pictographic |
| 197 |
} |
| 198 |
|
| 199 |
impl<'a> Iterator for UWordBounds<'a> { |
| 200 |
type Item = &'a str; |
| 201 |
|
| 202 |
#[inline] |
| 203 |
fn size_hint(&self) -> (usize, Option<usize>) { |
| 204 |
let slen = self.string.len(); |
| 205 |
(cmp::min(slen, 1), Some(slen)) |
| 206 |
} |
| 207 |
|
| 208 |
#[inline] |
| 209 |
fn next(&mut self) -> Option<&'a str> { |
| 210 |
use self::FormatExtendType::*; |
| 211 |
use self::UWordBoundsState::*; |
| 212 |
use crate::tables::word as wd; |
| 213 |
if self.string.len() == 0 { |
| 214 |
return None; |
| 215 |
} |
| 216 |
|
| 217 |
let mut take_curr = true; |
| 218 |
let mut take_cat = true; |
| 219 |
let mut idx = 0; |
| 220 |
let mut saveidx = 0; |
| 221 |
let mut state = Start; |
| 222 |
let mut cat = wd::WC_Any; |
| 223 |
let mut savecat = wd::WC_Any; |
| 224 |
|
| 225 |
// If extend/format/zwj were skipped. Handles precedence of WB3d over WB4 |
| 226 |
let mut skipped_format_extend = false; |
| 227 |
for (curr, ch) in self.string.char_indices() { |
| 228 |
idx = curr; |
| 229 |
// Whether or not the previous category was ZWJ |
| 230 |
// ZWJs get collapsed, so this handles precedence of WB3c over WB4 |
| 231 |
let prev_zwj = cat == wd::WC_ZWJ; |
| 232 |
// if there's a category cached, grab it |
| 233 |
cat = match self.cat { |
| 234 |
None => wd::word_category(ch).2, |
| 235 |
_ => self.cat.take().unwrap(), |
| 236 |
}; |
| 237 |
take_cat = true; |
| 238 |
|
| 239 |
// handle rule WB4 |
| 240 |
// just skip all format, extend, and zwj chars |
| 241 |
// note that Start is a special case: if there's a bunch of Format | Extend |
| 242 |
// characters at the beginning of a block of text, dump them out as one unit. |
| 243 |
// |
| 244 |
// (This is not obvious from the wording of UAX#29, but if you look at the |
| 245 |
// test cases http://www.unicode.org/Public/UNIDATA/auxiliary/WordBreakTest.txt |
| 246 |
// then the "correct" interpretation of WB4 becomes apparent.) |
| 247 |
if state != Start { |
| 248 |
match cat { |
| 249 |
wd::WC_Extend | wd::WC_Format | wd::WC_ZWJ => { |
| 250 |
skipped_format_extend = true; |
| 251 |
continue; |
| 252 |
} |
| 253 |
_ => {} |
| 254 |
} |
| 255 |
} |
| 256 |
|
| 257 |
// rule WB3c |
| 258 |
// WB4 makes all ZWJs collapse into the previous state |
| 259 |
// but you can still be in a Zwj state if you started with Zwj |
| 260 |
// |
| 261 |
// This means that an EP + Zwj will collapse into EP, which is wrong, |
| 262 |
// since EP+EP is not a boundary but EP+ZWJ+EP is |
| 263 |
// |
| 264 |
// Thus, we separately keep track of whether or not the last character |
| 265 |
// was a ZWJ. This is an additional bit of state tracked outside of the |
| 266 |
// state enum; the state enum represents the last non-zwj state encountered. |
| 267 |
// When prev_zwj is true, for the purposes of WB3c, we are in the Zwj state, |
| 268 |
// however we are in the previous state for the purposes of all other rules. |
| 269 |
if prev_zwj { |
| 270 |
if is_emoji(ch) { |
| 271 |
state = Emoji; |
| 272 |
continue; |
| 273 |
} |
| 274 |
} |
| 275 |
// Don't use `continue` in this match without updating `cat` |
| 276 |
state = match state { |
| 277 |
Start if cat == wd::WC_CR => { |
| 278 |
idx += match self.get_next_cat(idx) { |
| 279 |
Some(ncat) if ncat == wd::WC_LF => 1, // rule WB3 |
| 280 |
_ => 0, |
| 281 |
}; |
| 282 |
break; // rule WB3a |
| 283 |
} |
| 284 |
Start => match cat { |
| 285 |
wd::WC_ALetter => Letter, // rule WB5, WB6, WB9, WB13a |
| 286 |
wd::WC_Hebrew_Letter => HLetter, // rule WB5, WB6, WB7a, WB7b, WB9, WB13a |
| 287 |
wd::WC_Numeric => Numeric, // rule WB8, WB10, WB12, WB13a |
| 288 |
wd::WC_Katakana => Katakana, // rule WB13, WB13a |
| 289 |
wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13a, WB13b |
| 290 |
wd::WC_Regional_Indicator => Regional(RegionalState::Half), // rule WB13c |
| 291 |
wd::WC_LF | wd::WC_Newline => break, // rule WB3a |
| 292 |
wd::WC_ZWJ => Zwj, // rule WB3c |
| 293 |
wd::WC_WSegSpace => WSegSpace, // rule WB3d |
| 294 |
_ => { |
| 295 |
if let Some(ncat) = self.get_next_cat(idx) { |
| 296 |
// rule WB4 |
| 297 |
if ncat == wd::WC_Format || ncat == wd::WC_Extend || ncat == wd::WC_ZWJ |
| 298 |
{ |
| 299 |
state = FormatExtend(AcceptNone); |
| 300 |
self.cat = Some(ncat); |
| 301 |
continue; |
| 302 |
} |
| 303 |
} |
| 304 |
break; // rule WB999 |
| 305 |
} |
| 306 |
}, |
| 307 |
WSegSpace => match cat { |
| 308 |
wd::WC_WSegSpace if !skipped_format_extend => WSegSpace, |
| 309 |
_ => { |
| 310 |
take_curr = false; |
| 311 |
break; |
| 312 |
} |
| 313 |
}, |
| 314 |
Zwj => { |
| 315 |
// We already handle WB3c above. |
| 316 |
take_curr = false; |
| 317 |
break; |
| 318 |
} |
| 319 |
Letter | HLetter => match cat { |
| 320 |
wd::WC_ALetter => Letter, // rule WB5 |
| 321 |
wd::WC_Hebrew_Letter => HLetter, // rule WB5 |
| 322 |
wd::WC_Numeric => Numeric, // rule WB9 |
| 323 |
wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13a |
| 324 |
wd::WC_Double_Quote if state == HLetter => { |
| 325 |
savecat = cat; |
| 326 |
saveidx = idx; |
| 327 |
FormatExtend(RequireHLetter) // rule WB7b |
| 328 |
} |
| 329 |
wd::WC_Single_Quote if state == HLetter => { |
| 330 |
FormatExtend(AcceptQLetter) // rule WB7a |
| 331 |
} |
| 332 |
wd::WC_MidLetter | wd::WC_MidNumLet | wd::WC_Single_Quote => { |
| 333 |
savecat = cat; |
| 334 |
saveidx = idx; |
| 335 |
FormatExtend(RequireLetter) // rule WB6 |
| 336 |
} |
| 337 |
_ => { |
| 338 |
take_curr = false; |
| 339 |
break; |
| 340 |
} |
| 341 |
}, |
| 342 |
Numeric => match cat { |
| 343 |
wd::WC_Numeric => Numeric, // rule WB8 |
| 344 |
wd::WC_ALetter => Letter, // rule WB10 |
| 345 |
wd::WC_Hebrew_Letter => HLetter, // rule WB10 |
| 346 |
wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13a |
| 347 |
wd::WC_MidNum | wd::WC_MidNumLet | wd::WC_Single_Quote => { |
| 348 |
savecat = cat; |
| 349 |
saveidx = idx; |
| 350 |
FormatExtend(RequireNumeric) // rule WB12 |
| 351 |
} |
| 352 |
_ => { |
| 353 |
take_curr = false; |
| 354 |
break; |
| 355 |
} |
| 356 |
}, |
| 357 |
Katakana => match cat { |
| 358 |
wd::WC_Katakana => Katakana, // rule WB13 |
| 359 |
wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13a |
| 360 |
_ => { |
| 361 |
take_curr = false; |
| 362 |
break; |
| 363 |
} |
| 364 |
}, |
| 365 |
ExtendNumLet => match cat { |
| 366 |
wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13a |
| 367 |
wd::WC_ALetter => Letter, // rule WB13b |
| 368 |
wd::WC_Hebrew_Letter => HLetter, // rule WB13b |
| 369 |
wd::WC_Numeric => Numeric, // rule WB13b |
| 370 |
wd::WC_Katakana => Katakana, // rule WB13b |
| 371 |
_ => { |
| 372 |
take_curr = false; |
| 373 |
break; |
| 374 |
} |
| 375 |
}, |
| 376 |
Regional(RegionalState::Full) => { |
| 377 |
// if it reaches here we've gone too far, |
| 378 |
// a full flag can only compose with ZWJ/Extend/Format |
| 379 |
// proceeding it. |
| 380 |
take_curr = false; |
| 381 |
break; |
| 382 |
} |
| 383 |
Regional(RegionalState::Half) => match cat { |
| 384 |
wd::WC_Regional_Indicator => Regional(RegionalState::Full), // rule WB13c |
| 385 |
_ => { |
| 386 |
take_curr = false; |
| 387 |
break; |
| 388 |
} |
| 389 |
}, |
| 390 |
Regional(_) => { |
| 391 |
unreachable!("RegionalState::Unknown should not occur on forward iteration") |
| 392 |
} |
| 393 |
Emoji => { |
| 394 |
// We already handle WB3c above. If you've reached this point, the emoji sequence is over. |
| 395 |
take_curr = false; |
| 396 |
break; |
| 397 |
} |
| 398 |
FormatExtend(t) => match t { |
| 399 |
// handle FormatExtends depending on what type |
| 400 |
RequireNumeric if cat == wd::WC_Numeric => Numeric, // rule WB11 |
| 401 |
RequireLetter | AcceptQLetter if cat == wd::WC_ALetter => Letter, // rule WB7 |
| 402 |
RequireLetter | AcceptQLetter if cat == wd::WC_Hebrew_Letter => HLetter, // WB7a |
| 403 |
RequireHLetter if cat == wd::WC_Hebrew_Letter => HLetter, // rule WB7b |
| 404 |
AcceptNone | AcceptQLetter => { |
| 405 |
take_curr = false; // emit all the Format|Extend characters |
| 406 |
take_cat = false; |
| 407 |
break; |
| 408 |
} |
| 409 |
_ => break, // rewind (in if statement below) |
| 410 |
}, |
| 411 |
} |
| 412 |
} |
| 413 |
|
| 414 |
if let FormatExtend(t) = state { |
| 415 |
// we were looking for something and didn't find it; we have to back up |
| 416 |
if t == RequireLetter || t == RequireHLetter || t == RequireNumeric { |
| 417 |
idx = saveidx; |
| 418 |
cat = savecat; |
| 419 |
take_curr = false; |
| 420 |
} |
| 421 |
} |
| 422 |
|
| 423 |
self.cat = if take_curr { |
| 424 |
idx = idx + self.string[idx..].chars().next().unwrap().len_utf8(); |
| 425 |
None |
| 426 |
} else if take_cat { |
| 427 |
Some(cat) |
| 428 |
} else { |
| 429 |
None |
| 430 |
}; |
| 431 |
|
| 432 |
let retstr = &self.string[..idx]; |
| 433 |
self.string = &self.string[idx..]; |
| 434 |
Some(retstr) |
| 435 |
} |
| 436 |
} |
| 437 |
|
| 438 |
impl<'a> DoubleEndedIterator for UWordBounds<'a> { |
| 439 |
#[inline] |
| 440 |
fn next_back(&mut self) -> Option<&'a str> { |
| 441 |
use self::FormatExtendType::*; |
| 442 |
use self::UWordBoundsState::*; |
| 443 |
use crate::tables::word as wd; |
| 444 |
if self.string.len() == 0 { |
| 445 |
return None; |
| 446 |
} |
| 447 |
|
| 448 |
let mut take_curr = true; |
| 449 |
let mut take_cat = true; |
| 450 |
let mut idx = self.string.len(); |
| 451 |
idx -= self.string.chars().next_back().unwrap().len_utf8(); |
| 452 |
let mut previdx = idx; |
| 453 |
let mut saveidx = idx; |
| 454 |
let mut state = Start; |
| 455 |
let mut savestate = Start; |
| 456 |
let mut cat = wd::WC_Any; |
| 457 |
|
| 458 |
let mut skipped_format_extend = false; |
| 459 |
|
| 460 |
for (curr, ch) in self.string.char_indices().rev() { |
| 461 |
previdx = idx; |
| 462 |
idx = curr; |
| 463 |
|
| 464 |
// if there's a category cached, grab it |
| 465 |
cat = match self.catb { |
| 466 |
None => wd::word_category(ch).2, |
| 467 |
_ => self.catb.take().unwrap(), |
| 468 |
}; |
| 469 |
take_cat = true; |
| 470 |
|
| 471 |
// backward iterator over word boundaries. Mostly the same as the forward |
| 472 |
// iterator, with two weirdnesses: |
| 473 |
// (1) If we encounter a single quote in the Start state, we have to check for a |
| 474 |
// Hebrew Letter immediately before it. |
| 475 |
// (2) Format and Extend char handling takes some gymnastics. |
| 476 |
|
| 477 |
if cat == wd::WC_Extend || cat == wd::WC_Format || (cat == wd::WC_ZWJ && state != Zwj) { |
| 478 |
// WB3c has more priority so we should not |
| 479 |
// fold in that case |
| 480 |
if match state { |
| 481 |
FormatExtend(_) | Start => false, |
| 482 |
_ => true, |
| 483 |
} { |
| 484 |
saveidx = previdx; |
| 485 |
savestate = state; |
| 486 |
state = FormatExtend(AcceptNone); |
| 487 |
} |
| 488 |
|
| 489 |
if state != Start { |
| 490 |
continue; |
| 491 |
} |
| 492 |
} else if state == FormatExtend(AcceptNone) { |
| 493 |
// finished a scan of some Format|Extend chars, restore previous state |
| 494 |
state = savestate; |
| 495 |
previdx = saveidx; |
| 496 |
take_cat = false; |
| 497 |
skipped_format_extend = true; |
| 498 |
} |
| 499 |
|
| 500 |
// Don't use `continue` in this match without updating `catb` |
| 501 |
state = match state { |
| 502 |
Start | FormatExtend(AcceptAny) => match cat { |
| 503 |
_ if is_emoji(ch) => Zwj, |
| 504 |
wd::WC_ALetter => Letter, // rule WB5, WB7, WB10, WB13b |
| 505 |
wd::WC_Hebrew_Letter => HLetter, // rule WB5, WB7, WB7c, WB10, WB13b |
| 506 |
wd::WC_Numeric => Numeric, // rule WB8, WB9, WB11, WB13b |
| 507 |
wd::WC_Katakana => Katakana, // rule WB13, WB13b |
| 508 |
wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13a |
| 509 |
wd::WC_Regional_Indicator => Regional(RegionalState::Unknown), // rule WB13c |
| 510 |
// rule WB4: |
| 511 |
wd::WC_Extend | wd::WC_Format | wd::WC_ZWJ => FormatExtend(AcceptAny), |
| 512 |
wd::WC_Single_Quote => { |
| 513 |
saveidx = idx; |
| 514 |
FormatExtend(AcceptQLetter) // rule WB7a |
| 515 |
} |
| 516 |
wd::WC_WSegSpace => WSegSpace, |
| 517 |
wd::WC_CR | wd::WC_LF | wd::WC_Newline => { |
| 518 |
if state == Start { |
| 519 |
if cat == wd::WC_LF { |
| 520 |
idx -= match self.get_prev_cat(idx) { |
| 521 |
Some(pcat) if pcat == wd::WC_CR => 1, // rule WB3 |
| 522 |
_ => 0, |
| 523 |
}; |
| 524 |
} |
| 525 |
} else { |
| 526 |
take_curr = false; |
| 527 |
} |
| 528 |
break; // rule WB3a |
| 529 |
} |
| 530 |
_ => break, // rule WB999 |
| 531 |
}, |
| 532 |
Zwj => match cat { |
| 533 |
// rule WB3c |
| 534 |
wd::WC_ZWJ => FormatExtend(AcceptAny), |
| 535 |
_ => { |
| 536 |
take_curr = false; |
| 537 |
break; |
| 538 |
} |
| 539 |
}, |
| 540 |
WSegSpace => match cat { |
| 541 |
// rule WB3d |
| 542 |
wd::WC_WSegSpace if !skipped_format_extend => WSegSpace, |
| 543 |
_ => { |
| 544 |
take_curr = false; |
| 545 |
break; |
| 546 |
} |
| 547 |
}, |
| 548 |
Letter | HLetter => match cat { |
| 549 |
wd::WC_ALetter => Letter, // rule WB5 |
| 550 |
wd::WC_Hebrew_Letter => HLetter, // rule WB5 |
| 551 |
wd::WC_Numeric => Numeric, // rule WB10 |
| 552 |
wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13b |
| 553 |
wd::WC_Double_Quote if state == HLetter => { |
| 554 |
saveidx = previdx; |
| 555 |
FormatExtend(RequireHLetter) // rule WB7c |
| 556 |
} |
| 557 |
wd::WC_MidLetter | wd::WC_MidNumLet | wd::WC_Single_Quote => { |
| 558 |
saveidx = previdx; |
| 559 |
FormatExtend(RequireLetter) // rule WB7 |
| 560 |
} |
| 561 |
_ => { |
| 562 |
take_curr = false; |
| 563 |
break; |
| 564 |
} |
| 565 |
}, |
| 566 |
Numeric => match cat { |
| 567 |
wd::WC_Numeric => Numeric, // rule WB8 |
| 568 |
wd::WC_ALetter => Letter, // rule WB9 |
| 569 |
wd::WC_Hebrew_Letter => HLetter, // rule WB9 |
| 570 |
wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13b |
| 571 |
wd::WC_MidNum | wd::WC_MidNumLet | wd::WC_Single_Quote => { |
| 572 |
saveidx = previdx; |
| 573 |
FormatExtend(RequireNumeric) // rule WB11 |
| 574 |
} |
| 575 |
_ => { |
| 576 |
take_curr = false; |
| 577 |
break; |
| 578 |
} |
| 579 |
}, |
| 580 |
Katakana => match cat { |
| 581 |
wd::WC_Katakana => Katakana, // rule WB13 |
| 582 |
wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13b |
| 583 |
_ => { |
| 584 |
take_curr = false; |
| 585 |
break; |
| 586 |
} |
| 587 |
}, |
| 588 |
ExtendNumLet => match cat { |
| 589 |
wd::WC_ExtendNumLet => ExtendNumLet, // rule WB13a |
| 590 |
wd::WC_ALetter => Letter, // rule WB13a |
| 591 |
wd::WC_Hebrew_Letter => HLetter, // rule WB13a |
| 592 |
wd::WC_Numeric => Numeric, // rule WB13a |
| 593 |
wd::WC_Katakana => Katakana, // rule WB13a |
| 594 |
_ => { |
| 595 |
take_curr = false; |
| 596 |
break; |
| 597 |
} |
| 598 |
}, |
| 599 |
Regional(mut regional_state) => match cat { |
| 600 |
// rule WB13c |
| 601 |
wd::WC_Regional_Indicator => { |
| 602 |
if regional_state == RegionalState::Unknown { |
| 603 |
let count = self.string[..previdx] |
| 604 |
.chars() |
| 605 |
.rev() |
| 606 |
.map(|c| wd::word_category(c).2) |
| 607 |
.filter(|&c| { |
| 608 |
!(c == wd::WC_ZWJ || c == wd::WC_Extend || c == wd::WC_Format) |
| 609 |
}) |
| 610 |
.take_while(|&c| c == wd::WC_Regional_Indicator) |
| 611 |
.count(); |
| 612 |
regional_state = if count % 2 == 0 { |
| 613 |
RegionalState::Full |
| 614 |
} else { |
| 615 |
RegionalState::Half |
| 616 |
}; |
| 617 |
} |
| 618 |
if regional_state == RegionalState::Full { |
| 619 |
take_curr = false; |
| 620 |
break; |
| 621 |
} else { |
| 622 |
Regional(RegionalState::Full) |
| 623 |
} |
| 624 |
} |
| 625 |
_ => { |
| 626 |
take_curr = false; |
| 627 |
break; |
| 628 |
} |
| 629 |
}, |
| 630 |
Emoji => { |
| 631 |
if is_emoji(ch) { |
| 632 |
// rule WB3c |
| 633 |
Zwj |
| 634 |
} else { |
| 635 |
take_curr = false; |
| 636 |
break; |
| 637 |
} |
| 638 |
} |
| 639 |
FormatExtend(t) => match t { |
| 640 |
RequireNumeric if cat == wd::WC_Numeric => Numeric, // rule WB12 |
| 641 |
RequireLetter if cat == wd::WC_ALetter => Letter, // rule WB6 |
| 642 |
RequireLetter if cat == wd::WC_Hebrew_Letter => HLetter, // rule WB6 |
| 643 |
AcceptQLetter if cat == wd::WC_Hebrew_Letter => HLetter, // rule WB7a |
| 644 |
RequireHLetter if cat == wd::WC_Hebrew_Letter => HLetter, // rule WB7b |
| 645 |
_ => break, // backtrack will happens |
| 646 |
}, |
| 647 |
} |
| 648 |
} |
| 649 |
|
| 650 |
if let FormatExtend(t) = state { |
| 651 |
// if we required something but didn't find it, backtrack |
| 652 |
if t == RequireLetter |
| 653 |
|| t == RequireHLetter |
| 654 |
|| t == RequireNumeric |
| 655 |
|| t == AcceptNone |
| 656 |
|| t == AcceptQLetter |
| 657 |
{ |
| 658 |
previdx = saveidx; |
| 659 |
take_cat = false; |
| 660 |
take_curr = false; |
| 661 |
} |
| 662 |
} |
| 663 |
|
| 664 |
self.catb = if take_curr { |
| 665 |
None |
| 666 |
} else { |
| 667 |
idx = previdx; |
| 668 |
if take_cat { |
| 669 |
Some(cat) |
| 670 |
} else { |
| 671 |
None |
| 672 |
} |
| 673 |
}; |
| 674 |
|
| 675 |
let retstr = &self.string[idx..]; |
| 676 |
self.string = &self.string[..idx]; |
| 677 |
Some(retstr) |
| 678 |
} |
| 679 |
} |
| 680 |
|
| 681 |
impl<'a> UWordBounds<'a> { |
| 682 |
#[inline] |
| 683 |
/// View the underlying data (the part yet to be iterated) as a slice of the original string. |
| 684 |
/// |
| 685 |
/// ```rust |
| 686 |
/// # use unicode_segmentation::UnicodeSegmentation; |
| 687 |
/// let mut iter = "Hello world".split_word_bounds(); |
| 688 |
/// assert_eq!(iter.as_str(), "Hello world"); |
| 689 |
/// iter.next(); |
| 690 |
/// assert_eq!(iter.as_str(), " world"); |
| 691 |
/// iter.next(); |
| 692 |
/// assert_eq!(iter.as_str(), "world"); |
| 693 |
/// ``` |
| 694 |
pub fn as_str(&self) -> &'a str { |
| 695 |
self.string |
| 696 |
} |
| 697 |
|
| 698 |
#[inline] |
| 699 |
fn get_next_cat(&self, idx: usize) -> Option<WordCat> { |
| 700 |
use crate::tables::word as wd; |
| 701 |
let nidx = idx + self.string[idx..].chars().next().unwrap().len_utf8(); |
| 702 |
if nidx < self.string.len() { |
| 703 |
let nch = self.string[nidx..].chars().next().unwrap(); |
| 704 |
Some(wd::word_category(nch).2) |
| 705 |
} else { |
| 706 |
None |
| 707 |
} |
| 708 |
} |
| 709 |
|
| 710 |
#[inline] |
| 711 |
fn get_prev_cat(&self, idx: usize) -> Option<WordCat> { |
| 712 |
use crate::tables::word as wd; |
| 713 |
if idx > 0 { |
| 714 |
let nch = self.string[..idx].chars().next_back().unwrap(); |
| 715 |
Some(wd::word_category(nch).2) |
| 716 |
} else { |
| 717 |
None |
| 718 |
} |
| 719 |
} |
| 720 |
} |
| 721 |
|
| 722 |
#[inline] |
| 723 |
pub fn new_word_bounds<'b>(s: &'b str) -> UWordBounds<'b> { |
| 724 |
UWordBounds { |
| 725 |
string: s, |
| 726 |
cat: None, |
| 727 |
catb: None, |
| 728 |
} |
| 729 |
} |
| 730 |
|
| 731 |
#[inline] |
| 732 |
pub fn new_word_bound_indices<'b>(s: &'b str) -> UWordBoundIndices<'b> { |
| 733 |
UWordBoundIndices { |
| 734 |
start_offset: s.as_ptr() as usize, |
| 735 |
iter: new_word_bounds(s), |
| 736 |
} |
| 737 |
} |
| 738 |
|
| 739 |
#[inline] |
| 740 |
fn has_alphanumeric(s: &&str) -> bool { |
| 741 |
use crate::tables::util::is_alphanumeric; |
| 742 |
|
| 743 |
s.chars().any(|c| is_alphanumeric(c)) |
| 744 |
} |
| 745 |
|
| 746 |
#[inline] |
| 747 |
pub fn new_unicode_words<'b>(s: &'b str) -> UnicodeWords<'b> { |
| 748 |
use super::UnicodeSegmentation; |
| 749 |
|
| 750 |
UnicodeWords { |
| 751 |
inner: s.split_word_bounds().filter(has_alphanumeric), |
| 752 |
} |
| 753 |
} |
| 754 |
|
| 755 |
#[inline] |
| 756 |
pub fn new_unicode_word_indices<'b>(s: &'b str) -> UnicodeWordIndices<'b> { |
| 757 |
use super::UnicodeSegmentation; |
| 758 |
|
| 759 |
UnicodeWordIndices { |
| 760 |
inner: s |
| 761 |
.split_word_bound_indices() |
| 762 |
.filter(|(_, c)| has_alphanumeric(c)), |
| 763 |
} |
| 764 |
} |