| 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 |
|
| 13 |
use crate::tables::grapheme::GraphemeCat; |
| 14 |
|
| 15 |
/// External iterator for grapheme clusters and byte offsets. |
| 16 |
/// |
| 17 |
/// This struct is created by the [`grapheme_indices`] method on the [`UnicodeSegmentation`] |
| 18 |
/// trait. See its documentation for more. |
| 19 |
/// |
| 20 |
/// [`grapheme_indices`]: trait.UnicodeSegmentation.html#tymethod.grapheme_indices |
| 21 |
/// [`UnicodeSegmentation`]: trait.UnicodeSegmentation.html |
| 22 |
#[derive(Clone)] |
| 23 |
pub struct GraphemeIndices<'a> { |
| 24 |
start_offset: usize, |
| 25 |
iter: Graphemes<'a>, |
| 26 |
} |
| 27 |
|
| 28 |
impl<'a> GraphemeIndices<'a> { |
| 29 |
#[inline] |
| 30 |
/// View the underlying data (the part yet to be iterated) as a slice of the original string. |
| 31 |
/// |
| 32 |
/// ```rust |
| 33 |
/// # use unicode_segmentation::UnicodeSegmentation; |
| 34 |
/// let mut iter = "abc".grapheme_indices(true); |
| 35 |
/// assert_eq!(iter.as_str(), "abc"); |
| 36 |
/// iter.next(); |
| 37 |
/// assert_eq!(iter.as_str(), "bc"); |
| 38 |
/// iter.next(); |
| 39 |
/// iter.next(); |
| 40 |
/// assert_eq!(iter.as_str(), ""); |
| 41 |
/// ``` |
| 42 |
pub fn as_str(&self) -> &'a str { |
| 43 |
self.iter.as_str() |
| 44 |
} |
| 45 |
} |
| 46 |
|
| 47 |
impl<'a> Iterator for GraphemeIndices<'a> { |
| 48 |
type Item = (usize, &'a str); |
| 49 |
|
| 50 |
#[inline] |
| 51 |
fn next(&mut self) -> Option<(usize, &'a str)> { |
| 52 |
self.iter |
| 53 |
.next() |
| 54 |
.map(|s| (s.as_ptr() as usize - self.start_offset, s)) |
| 55 |
} |
| 56 |
|
| 57 |
#[inline] |
| 58 |
fn size_hint(&self) -> (usize, Option<usize>) { |
| 59 |
self.iter.size_hint() |
| 60 |
} |
| 61 |
} |
| 62 |
|
| 63 |
impl<'a> DoubleEndedIterator for GraphemeIndices<'a> { |
| 64 |
#[inline] |
| 65 |
fn next_back(&mut self) -> Option<(usize, &'a str)> { |
| 66 |
self.iter |
| 67 |
.next_back() |
| 68 |
.map(|s| (s.as_ptr() as usize - self.start_offset, s)) |
| 69 |
} |
| 70 |
} |
| 71 |
|
| 72 |
/// External iterator for a string's |
| 73 |
/// [grapheme clusters](http://www.unicode.org/reports/tr29/#Grapheme_Cluster_Boundaries). |
| 74 |
/// |
| 75 |
/// This struct is created by the [`graphemes`] method on the [`UnicodeSegmentation`] trait. See its |
| 76 |
/// documentation for more. |
| 77 |
/// |
| 78 |
/// [`graphemes`]: trait.UnicodeSegmentation.html#tymethod.graphemes |
| 79 |
/// [`UnicodeSegmentation`]: trait.UnicodeSegmentation.html |
| 80 |
#[derive(Clone, Debug)] |
| 81 |
pub struct Graphemes<'a> { |
| 82 |
string: &'a str, |
| 83 |
cursor: GraphemeCursor, |
| 84 |
cursor_back: GraphemeCursor, |
| 85 |
} |
| 86 |
|
| 87 |
impl<'a> Graphemes<'a> { |
| 88 |
#[inline] |
| 89 |
/// View the underlying data (the part yet to be iterated) as a slice of the original string. |
| 90 |
/// |
| 91 |
/// ```rust |
| 92 |
/// # use unicode_segmentation::UnicodeSegmentation; |
| 93 |
/// let mut iter = "abc".graphemes(true); |
| 94 |
/// assert_eq!(iter.as_str(), "abc"); |
| 95 |
/// iter.next(); |
| 96 |
/// assert_eq!(iter.as_str(), "bc"); |
| 97 |
/// iter.next(); |
| 98 |
/// iter.next(); |
| 99 |
/// assert_eq!(iter.as_str(), ""); |
| 100 |
/// ``` |
| 101 |
pub fn as_str(&self) -> &'a str { |
| 102 |
&self.string[self.cursor.cur_cursor()..self.cursor_back.cur_cursor()] |
| 103 |
} |
| 104 |
} |
| 105 |
|
| 106 |
impl<'a> Iterator for Graphemes<'a> { |
| 107 |
type Item = &'a str; |
| 108 |
|
| 109 |
#[inline] |
| 110 |
fn size_hint(&self) -> (usize, Option<usize>) { |
| 111 |
let slen = self.cursor_back.cur_cursor() - self.cursor.cur_cursor(); |
| 112 |
(cmp::min(slen, 1), Some(slen)) |
| 113 |
} |
| 114 |
|
| 115 |
#[inline] |
| 116 |
fn next(&mut self) -> Option<&'a str> { |
| 117 |
let start = self.cursor.cur_cursor(); |
| 118 |
if start == self.cursor_back.cur_cursor() { |
| 119 |
return None; |
| 120 |
} |
| 121 |
let next = self.cursor.next_boundary(self.string, 0).unwrap().unwrap(); |
| 122 |
Some(&self.string[start..next]) |
| 123 |
} |
| 124 |
} |
| 125 |
|
| 126 |
impl<'a> DoubleEndedIterator for Graphemes<'a> { |
| 127 |
#[inline] |
| 128 |
fn next_back(&mut self) -> Option<&'a str> { |
| 129 |
let end = self.cursor_back.cur_cursor(); |
| 130 |
if end == self.cursor.cur_cursor() { |
| 131 |
return None; |
| 132 |
} |
| 133 |
let prev = self |
| 134 |
.cursor_back |
| 135 |
.prev_boundary(self.string, 0) |
| 136 |
.unwrap() |
| 137 |
.unwrap(); |
| 138 |
Some(&self.string[prev..end]) |
| 139 |
} |
| 140 |
} |
| 141 |
|
| 142 |
#[inline] |
| 143 |
pub fn new_graphemes<'b>(s: &'b str, is_extended: bool) -> Graphemes<'b> { |
| 144 |
let len = s.len(); |
| 145 |
Graphemes { |
| 146 |
string: s, |
| 147 |
cursor: GraphemeCursor::new(0, len, is_extended), |
| 148 |
cursor_back: GraphemeCursor::new(len, len, is_extended), |
| 149 |
} |
| 150 |
} |
| 151 |
|
| 152 |
#[inline] |
| 153 |
pub fn new_grapheme_indices<'b>(s: &'b str, is_extended: bool) -> GraphemeIndices<'b> { |
| 154 |
GraphemeIndices { |
| 155 |
start_offset: s.as_ptr() as usize, |
| 156 |
iter: new_graphemes(s, is_extended), |
| 157 |
} |
| 158 |
} |
| 159 |
|
| 160 |
// maybe unify with PairResult? |
| 161 |
// An enum describing information about a potential boundary. |
| 162 |
#[derive(PartialEq, Eq, Clone, Debug)] |
| 163 |
enum GraphemeState { |
| 164 |
// No information is known. |
| 165 |
Unknown, |
| 166 |
// It is known to not be a boundary. |
| 167 |
NotBreak, |
| 168 |
// It is known to be a boundary. |
| 169 |
Break, |
| 170 |
// The codepoint after is a Regional Indicator Symbol, so a boundary iff |
| 171 |
// it is preceded by an even number of RIS codepoints. (GB12, GB13) |
| 172 |
Regional, |
| 173 |
// The codepoint after is Extended_Pictographic, |
| 174 |
// so whether it's a boundary depends on pre-context according to GB11. |
| 175 |
Emoji, |
| 176 |
} |
| 177 |
|
| 178 |
/// Cursor-based segmenter for grapheme clusters. |
| 179 |
/// |
| 180 |
/// This allows working with ropes and other datastructures where the string is not contiguous or |
| 181 |
/// fully known at initialization time. |
| 182 |
#[derive(Clone, Debug)] |
| 183 |
pub struct GraphemeCursor { |
| 184 |
// Current cursor position. |
| 185 |
offset: usize, |
| 186 |
// Total length of the string. |
| 187 |
len: usize, |
| 188 |
// A config flag indicating whether this cursor computes legacy or extended |
| 189 |
// grapheme cluster boundaries (enables GB9a and GB9b if set). |
| 190 |
is_extended: bool, |
| 191 |
// Information about the potential boundary at `offset` |
| 192 |
state: GraphemeState, |
| 193 |
// Category of codepoint immediately preceding cursor, if known. |
| 194 |
cat_before: Option<GraphemeCat>, |
| 195 |
// Category of codepoint immediately after cursor, if known. |
| 196 |
cat_after: Option<GraphemeCat>, |
| 197 |
// If set, at least one more codepoint immediately preceding this offset |
| 198 |
// is needed to resolve whether there's a boundary at `offset`. |
| 199 |
pre_context_offset: Option<usize>, |
| 200 |
// The number of RIS codepoints preceding `offset`. If `pre_context_offset` |
| 201 |
// is set, then counts the number of RIS between that and `offset`, otherwise |
| 202 |
// is an accurate count relative to the string. |
| 203 |
ris_count: Option<usize>, |
| 204 |
// Set if a call to `prev_boundary` or `next_boundary` was suspended due |
| 205 |
// to needing more input. |
| 206 |
resuming: bool, |
| 207 |
// Cached grapheme category and associated scalar value range. |
| 208 |
grapheme_cat_cache: (u32, u32, GraphemeCat), |
| 209 |
} |
| 210 |
|
| 211 |
/// An error return indicating that not enough content was available in the |
| 212 |
/// provided chunk to satisfy the query, and that more content must be provided. |
| 213 |
#[derive(PartialEq, Eq, Debug)] |
| 214 |
pub enum GraphemeIncomplete { |
| 215 |
/// More pre-context is needed. The caller should call `provide_context` |
| 216 |
/// with a chunk ending at the offset given, then retry the query. This |
| 217 |
/// will only be returned if the `chunk_start` parameter is nonzero. |
| 218 |
PreContext(usize), |
| 219 |
|
| 220 |
/// When requesting `prev_boundary`, the cursor is moving past the beginning |
| 221 |
/// of the current chunk, so the chunk before that is requested. This will |
| 222 |
/// only be returned if the `chunk_start` parameter is nonzero. |
| 223 |
PrevChunk, |
| 224 |
|
| 225 |
/// When requesting `next_boundary`, the cursor is moving past the end of the |
| 226 |
/// current chunk, so the chunk after that is requested. This will only be |
| 227 |
/// returned if the chunk ends before the `len` parameter provided on |
| 228 |
/// creation of the cursor. |
| 229 |
NextChunk, // requesting chunk following the one given |
| 230 |
|
| 231 |
/// An error returned when the chunk given does not contain the cursor position. |
| 232 |
InvalidOffset, |
| 233 |
} |
| 234 |
|
| 235 |
// An enum describing the result from lookup of a pair of categories. |
| 236 |
#[derive(PartialEq, Eq)] |
| 237 |
enum PairResult { |
| 238 |
NotBreak, // definitely not a break |
| 239 |
Break, // definitely a break |
| 240 |
Extended, // a break iff not in extended mode |
| 241 |
Regional, // a break if preceded by an even number of RIS |
| 242 |
Emoji, // a break if preceded by emoji base and (Extend)* |
| 243 |
} |
| 244 |
|
| 245 |
#[inline] |
| 246 |
fn check_pair(before: GraphemeCat, after: GraphemeCat) -> PairResult { |
| 247 |
use self::PairResult::*; |
| 248 |
use crate::tables::grapheme::GraphemeCat::*; |
| 249 |
match (before, after) { |
| 250 |
(GC_CR, GC_LF) => NotBreak, // GB3 |
| 251 |
(GC_Control, _) => Break, // GB4 |
| 252 |
(GC_CR, _) => Break, // GB4 |
| 253 |
(GC_LF, _) => Break, // GB4 |
| 254 |
(_, GC_Control) => Break, // GB5 |
| 255 |
(_, GC_CR) => Break, // GB5 |
| 256 |
(_, GC_LF) => Break, // GB5 |
| 257 |
(GC_L, GC_L) => NotBreak, // GB6 |
| 258 |
(GC_L, GC_V) => NotBreak, // GB6 |
| 259 |
(GC_L, GC_LV) => NotBreak, // GB6 |
| 260 |
(GC_L, GC_LVT) => NotBreak, // GB6 |
| 261 |
(GC_LV, GC_V) => NotBreak, // GB7 |
| 262 |
(GC_LV, GC_T) => NotBreak, // GB7 |
| 263 |
(GC_V, GC_V) => NotBreak, // GB7 |
| 264 |
(GC_V, GC_T) => NotBreak, // GB7 |
| 265 |
(GC_LVT, GC_T) => NotBreak, // GB8 |
| 266 |
(GC_T, GC_T) => NotBreak, // GB8 |
| 267 |
(_, GC_Extend) => NotBreak, // GB9 |
| 268 |
(_, GC_ZWJ) => NotBreak, // GB9 |
| 269 |
(_, GC_SpacingMark) => Extended, // GB9a |
| 270 |
(GC_Prepend, _) => Extended, // GB9b |
| 271 |
(GC_ZWJ, GC_Extended_Pictographic) => Emoji, // GB11 |
| 272 |
(GC_Regional_Indicator, GC_Regional_Indicator) => Regional, // GB12, GB13 |
| 273 |
(_, _) => Break, // GB999 |
| 274 |
} |
| 275 |
} |
| 276 |
|
| 277 |
impl GraphemeCursor { |
| 278 |
/// Create a new cursor. The string and initial offset are given at creation |
| 279 |
/// time, but the contents of the string are not. The `is_extended` parameter |
| 280 |
/// controls whether extended grapheme clusters are selected. |
| 281 |
/// |
| 282 |
/// The `offset` parameter must be on a codepoint boundary. |
| 283 |
/// |
| 284 |
/// ```rust |
| 285 |
/// # use unicode_segmentation::GraphemeCursor; |
| 286 |
/// let s = "हिन्दी"; |
| 287 |
/// let mut legacy = GraphemeCursor::new(0, s.len(), false); |
| 288 |
/// assert_eq!(legacy.next_boundary(s, 0), Ok(Some("ह".len()))); |
| 289 |
/// let mut extended = GraphemeCursor::new(0, s.len(), true); |
| 290 |
/// assert_eq!(extended.next_boundary(s, 0), Ok(Some("हि".len()))); |
| 291 |
/// ``` |
| 292 |
pub fn new(offset: usize, len: usize, is_extended: bool) -> GraphemeCursor { |
| 293 |
let state = if offset == 0 || offset == len { |
| 294 |
GraphemeState::Break |
| 295 |
} else { |
| 296 |
GraphemeState::Unknown |
| 297 |
}; |
| 298 |
GraphemeCursor { |
| 299 |
offset: offset, |
| 300 |
len: len, |
| 301 |
state: state, |
| 302 |
is_extended: is_extended, |
| 303 |
cat_before: None, |
| 304 |
cat_after: None, |
| 305 |
pre_context_offset: None, |
| 306 |
ris_count: None, |
| 307 |
resuming: false, |
| 308 |
grapheme_cat_cache: (0, 0, GraphemeCat::GC_Control), |
| 309 |
} |
| 310 |
} |
| 311 |
|
| 312 |
fn grapheme_category(&mut self, ch: char) -> GraphemeCat { |
| 313 |
use crate::tables::grapheme as gr; |
| 314 |
use crate::tables::grapheme::GraphemeCat::*; |
| 315 |
|
| 316 |
if ch <= '\u{7e}' { |
| 317 |
// Special-case optimization for ascii, except U+007F. This |
| 318 |
// improves performance even for many primarily non-ascii texts, |
| 319 |
// due to use of punctuation and white space characters from the |
| 320 |
// ascii range. |
| 321 |
if ch >= '\u{20}' { |
| 322 |
GC_Any |
| 323 |
} else if ch == '\n' { |
| 324 |
GC_LF |
| 325 |
} else if ch == '\r' { |
| 326 |
GC_CR |
| 327 |
} else { |
| 328 |
GC_Control |
| 329 |
} |
| 330 |
} else { |
| 331 |
// If this char isn't within the cached range, update the cache to the |
| 332 |
// range that includes it. |
| 333 |
if (ch as u32) < self.grapheme_cat_cache.0 || (ch as u32) > self.grapheme_cat_cache.1 { |
| 334 |
self.grapheme_cat_cache = gr::grapheme_category(ch); |
| 335 |
} |
| 336 |
self.grapheme_cat_cache.2 |
| 337 |
} |
| 338 |
} |
| 339 |
|
| 340 |
// Not sure I'm gonna keep this, the advantage over new() seems thin. |
| 341 |
|
| 342 |
/// Set the cursor to a new location in the same string. |
| 343 |
/// |
| 344 |
/// ```rust |
| 345 |
/// # use unicode_segmentation::GraphemeCursor; |
| 346 |
/// let s = "abcd"; |
| 347 |
/// let mut cursor = GraphemeCursor::new(0, s.len(), false); |
| 348 |
/// assert_eq!(cursor.cur_cursor(), 0); |
| 349 |
/// cursor.set_cursor(2); |
| 350 |
/// assert_eq!(cursor.cur_cursor(), 2); |
| 351 |
/// ``` |
| 352 |
pub fn set_cursor(&mut self, offset: usize) { |
| 353 |
if offset != self.offset { |
| 354 |
self.offset = offset; |
| 355 |
self.state = if offset == 0 || offset == self.len { |
| 356 |
GraphemeState::Break |
| 357 |
} else { |
| 358 |
GraphemeState::Unknown |
| 359 |
}; |
| 360 |
// reset state derived from text around cursor |
| 361 |
self.cat_before = None; |
| 362 |
self.cat_after = None; |
| 363 |
self.ris_count = None; |
| 364 |
} |
| 365 |
} |
| 366 |
|
| 367 |
#[inline] |
| 368 |
/// The current offset of the cursor. Equal to the last value provided to |
| 369 |
/// `new()` or `set_cursor()`, or returned from `next_boundary()` or |
| 370 |
/// `prev_boundary()`. |
| 371 |
/// |
| 372 |
/// ```rust |
| 373 |
/// # use unicode_segmentation::GraphemeCursor; |
| 374 |
/// // Two flags (🇷🇸🇮🇴), each flag is two RIS codepoints, each RIS is 4 bytes. |
| 375 |
/// let flags = "\u{1F1F7}\u{1F1F8}\u{1F1EE}\u{1F1F4}"; |
| 376 |
/// let mut cursor = GraphemeCursor::new(4, flags.len(), false); |
| 377 |
/// assert_eq!(cursor.cur_cursor(), 4); |
| 378 |
/// assert_eq!(cursor.next_boundary(flags, 0), Ok(Some(8))); |
| 379 |
/// assert_eq!(cursor.cur_cursor(), 8); |
| 380 |
/// ``` |
| 381 |
pub fn cur_cursor(&self) -> usize { |
| 382 |
self.offset |
| 383 |
} |
| 384 |
|
| 385 |
/// Provide additional pre-context when it is needed to decide a boundary. |
| 386 |
/// The end of the chunk must coincide with the value given in the |
| 387 |
/// `GraphemeIncomplete::PreContext` request. |
| 388 |
/// |
| 389 |
/// ```rust |
| 390 |
/// # use unicode_segmentation::{GraphemeCursor, GraphemeIncomplete}; |
| 391 |
/// let flags = "\u{1F1F7}\u{1F1F8}\u{1F1EE}\u{1F1F4}"; |
| 392 |
/// let mut cursor = GraphemeCursor::new(8, flags.len(), false); |
| 393 |
/// // Not enough pre-context to decide if there's a boundary between the two flags. |
| 394 |
/// assert_eq!(cursor.is_boundary(&flags[8..], 8), Err(GraphemeIncomplete::PreContext(8))); |
| 395 |
/// // Provide one more Regional Indicator Symbol of pre-context |
| 396 |
/// cursor.provide_context(&flags[4..8], 4); |
| 397 |
/// // Still not enough context to decide. |
| 398 |
/// assert_eq!(cursor.is_boundary(&flags[8..], 8), Err(GraphemeIncomplete::PreContext(4))); |
| 399 |
/// // Provide additional requested context. |
| 400 |
/// cursor.provide_context(&flags[0..4], 0); |
| 401 |
/// // That's enough to decide (it always is when context goes to the start of the string) |
| 402 |
/// assert_eq!(cursor.is_boundary(&flags[8..], 8), Ok(true)); |
| 403 |
/// ``` |
| 404 |
pub fn provide_context(&mut self, chunk: &str, chunk_start: usize) { |
| 405 |
use crate::tables::grapheme as gr; |
| 406 |
assert!(chunk_start + chunk.len() == self.pre_context_offset.unwrap()); |
| 407 |
self.pre_context_offset = None; |
| 408 |
if self.is_extended && chunk_start + chunk.len() == self.offset { |
| 409 |
let ch = chunk.chars().rev().next().unwrap(); |
| 410 |
if self.grapheme_category(ch) == gr::GC_Prepend { |
| 411 |
self.decide(false); // GB9b |
| 412 |
return; |
| 413 |
} |
| 414 |
} |
| 415 |
match self.state { |
| 416 |
GraphemeState::Regional => self.handle_regional(chunk, chunk_start), |
| 417 |
GraphemeState::Emoji => self.handle_emoji(chunk, chunk_start), |
| 418 |
_ => { |
| 419 |
if self.cat_before.is_none() && self.offset == chunk.len() + chunk_start { |
| 420 |
let ch = chunk.chars().rev().next().unwrap(); |
| 421 |
self.cat_before = Some(self.grapheme_category(ch)); |
| 422 |
} |
| 423 |
} |
| 424 |
} |
| 425 |
} |
| 426 |
|
| 427 |
#[inline] |
| 428 |
fn decide(&mut self, is_break: bool) { |
| 429 |
self.state = if is_break { |
| 430 |
GraphemeState::Break |
| 431 |
} else { |
| 432 |
GraphemeState::NotBreak |
| 433 |
}; |
| 434 |
} |
| 435 |
|
| 436 |
#[inline] |
| 437 |
fn decision(&mut self, is_break: bool) -> Result<bool, GraphemeIncomplete> { |
| 438 |
self.decide(is_break); |
| 439 |
Ok(is_break) |
| 440 |
} |
| 441 |
|
| 442 |
#[inline] |
| 443 |
fn is_boundary_result(&self) -> Result<bool, GraphemeIncomplete> { |
| 444 |
if self.state == GraphemeState::Break { |
| 445 |
Ok(true) |
| 446 |
} else if self.state == GraphemeState::NotBreak { |
| 447 |
Ok(false) |
| 448 |
} else if let Some(pre_context_offset) = self.pre_context_offset { |
| 449 |
Err(GraphemeIncomplete::PreContext(pre_context_offset)) |
| 450 |
} else { |
| 451 |
unreachable!("inconsistent state"); |
| 452 |
} |
| 453 |
} |
| 454 |
|
| 455 |
#[inline] |
| 456 |
fn handle_regional(&mut self, chunk: &str, chunk_start: usize) { |
| 457 |
use crate::tables::grapheme as gr; |
| 458 |
let mut ris_count = self.ris_count.unwrap_or(0); |
| 459 |
for ch in chunk.chars().rev() { |
| 460 |
if self.grapheme_category(ch) != gr::GC_Regional_Indicator { |
| 461 |
self.ris_count = Some(ris_count); |
| 462 |
self.decide((ris_count % 2) == 0); |
| 463 |
return; |
| 464 |
} |
| 465 |
ris_count += 1; |
| 466 |
} |
| 467 |
self.ris_count = Some(ris_count); |
| 468 |
if chunk_start == 0 { |
| 469 |
self.decide((ris_count % 2) == 0); |
| 470 |
return; |
| 471 |
} |
| 472 |
self.pre_context_offset = Some(chunk_start); |
| 473 |
self.state = GraphemeState::Regional; |
| 474 |
} |
| 475 |
|
| 476 |
#[inline] |
| 477 |
fn handle_emoji(&mut self, chunk: &str, chunk_start: usize) { |
| 478 |
use crate::tables::grapheme as gr; |
| 479 |
let mut iter = chunk.chars().rev(); |
| 480 |
if let Some(ch) = iter.next() { |
| 481 |
if self.grapheme_category(ch) != gr::GC_ZWJ { |
| 482 |
self.decide(true); |
| 483 |
return; |
| 484 |
} |
| 485 |
} |
| 486 |
for ch in iter { |
| 487 |
match self.grapheme_category(ch) { |
| 488 |
gr::GC_Extend => (), |
| 489 |
gr::GC_Extended_Pictographic => { |
| 490 |
self.decide(false); |
| 491 |
return; |
| 492 |
} |
| 493 |
_ => { |
| 494 |
self.decide(true); |
| 495 |
return; |
| 496 |
} |
| 497 |
} |
| 498 |
} |
| 499 |
if chunk_start == 0 { |
| 500 |
self.decide(true); |
| 501 |
return; |
| 502 |
} |
| 503 |
self.pre_context_offset = Some(chunk_start); |
| 504 |
self.state = GraphemeState::Emoji; |
| 505 |
} |
| 506 |
|
| 507 |
#[inline] |
| 508 |
/// Determine whether the current cursor location is a grapheme cluster boundary. |
| 509 |
/// Only a part of the string need be supplied. If `chunk_start` is nonzero or |
| 510 |
/// the length of `chunk` is not equal to `len` on creation, then this method |
| 511 |
/// may return `GraphemeIncomplete::PreContext`. The caller should then |
| 512 |
/// call `provide_context` with the requested chunk, then retry calling this |
| 513 |
/// method. |
| 514 |
/// |
| 515 |
/// For partial chunks, if the cursor is not at the beginning or end of the |
| 516 |
/// string, the chunk should contain at least the codepoint following the cursor. |
| 517 |
/// If the string is nonempty, the chunk must be nonempty. |
| 518 |
/// |
| 519 |
/// All calls should have consistent chunk contents (ie, if a chunk provides |
| 520 |
/// content for a given slice, all further chunks covering that slice must have |
| 521 |
/// the same content for it). |
| 522 |
/// |
| 523 |
/// ```rust |
| 524 |
/// # use unicode_segmentation::GraphemeCursor; |
| 525 |
/// let flags = "\u{1F1F7}\u{1F1F8}\u{1F1EE}\u{1F1F4}"; |
| 526 |
/// let mut cursor = GraphemeCursor::new(8, flags.len(), false); |
| 527 |
/// assert_eq!(cursor.is_boundary(flags, 0), Ok(true)); |
| 528 |
/// cursor.set_cursor(12); |
| 529 |
/// assert_eq!(cursor.is_boundary(flags, 0), Ok(false)); |
| 530 |
/// ``` |
| 531 |
pub fn is_boundary( |
| 532 |
&mut self, |
| 533 |
chunk: &str, |
| 534 |
chunk_start: usize, |
| 535 |
) -> Result<bool, GraphemeIncomplete> { |
| 536 |
use crate::tables::grapheme as gr; |
| 537 |
if self.state == GraphemeState::Break { |
| 538 |
return Ok(true); |
| 539 |
} |
| 540 |
if self.state == GraphemeState::NotBreak { |
| 541 |
return Ok(false); |
| 542 |
} |
| 543 |
if self.offset < chunk_start || self.offset >= chunk_start + chunk.len() { |
| 544 |
if self.offset > chunk_start + chunk.len() || self.cat_after.is_none() { |
| 545 |
return Err(GraphemeIncomplete::InvalidOffset); |
| 546 |
} |
| 547 |
} |
| 548 |
if let Some(pre_context_offset) = self.pre_context_offset { |
| 549 |
return Err(GraphemeIncomplete::PreContext(pre_context_offset)); |
| 550 |
} |
| 551 |
let offset_in_chunk = self.offset - chunk_start; |
| 552 |
if self.cat_after.is_none() { |
| 553 |
let ch = chunk[offset_in_chunk..].chars().next().unwrap(); |
| 554 |
self.cat_after = Some(self.grapheme_category(ch)); |
| 555 |
} |
| 556 |
if self.offset == chunk_start { |
| 557 |
let mut need_pre_context = true; |
| 558 |
match self.cat_after.unwrap() { |
| 559 |
gr::GC_Regional_Indicator => self.state = GraphemeState::Regional, |
| 560 |
gr::GC_Extended_Pictographic => self.state = GraphemeState::Emoji, |
| 561 |
_ => need_pre_context = self.cat_before.is_none(), |
| 562 |
} |
| 563 |
if need_pre_context { |
| 564 |
self.pre_context_offset = Some(chunk_start); |
| 565 |
return Err(GraphemeIncomplete::PreContext(chunk_start)); |
| 566 |
} |
| 567 |
} |
| 568 |
if self.cat_before.is_none() { |
| 569 |
let ch = chunk[..offset_in_chunk].chars().rev().next().unwrap(); |
| 570 |
self.cat_before = Some(self.grapheme_category(ch)); |
| 571 |
} |
| 572 |
match check_pair(self.cat_before.unwrap(), self.cat_after.unwrap()) { |
| 573 |
PairResult::NotBreak => return self.decision(false), |
| 574 |
PairResult::Break => return self.decision(true), |
| 575 |
PairResult::Extended => { |
| 576 |
let is_extended = self.is_extended; |
| 577 |
return self.decision(!is_extended); |
| 578 |
} |
| 579 |
PairResult::Regional => { |
| 580 |
if let Some(ris_count) = self.ris_count { |
| 581 |
return self.decision((ris_count % 2) == 0); |
| 582 |
} |
| 583 |
self.handle_regional(&chunk[..offset_in_chunk], chunk_start); |
| 584 |
self.is_boundary_result() |
| 585 |
} |
| 586 |
PairResult::Emoji => { |
| 587 |
self.handle_emoji(&chunk[..offset_in_chunk], chunk_start); |
| 588 |
self.is_boundary_result() |
| 589 |
} |
| 590 |
} |
| 591 |
} |
| 592 |
|
| 593 |
#[inline] |
| 594 |
/// Find the next boundary after the current cursor position. Only a part of |
| 595 |
/// the string need be supplied. If the chunk is incomplete, then this |
| 596 |
/// method might return `GraphemeIncomplete::PreContext` or |
| 597 |
/// `GraphemeIncomplete::NextChunk`. In the former case, the caller should |
| 598 |
/// call `provide_context` with the requested chunk, then retry. In the |
| 599 |
/// latter case, the caller should provide the chunk following the one |
| 600 |
/// given, then retry. |
| 601 |
/// |
| 602 |
/// See `is_boundary` for expectations on the provided chunk. |
| 603 |
/// |
| 604 |
/// ```rust |
| 605 |
/// # use unicode_segmentation::GraphemeCursor; |
| 606 |
/// let flags = "\u{1F1F7}\u{1F1F8}\u{1F1EE}\u{1F1F4}"; |
| 607 |
/// let mut cursor = GraphemeCursor::new(4, flags.len(), false); |
| 608 |
/// assert_eq!(cursor.next_boundary(flags, 0), Ok(Some(8))); |
| 609 |
/// assert_eq!(cursor.next_boundary(flags, 0), Ok(Some(16))); |
| 610 |
/// assert_eq!(cursor.next_boundary(flags, 0), Ok(None)); |
| 611 |
/// ``` |
| 612 |
/// |
| 613 |
/// And an example that uses partial strings: |
| 614 |
/// |
| 615 |
/// ```rust |
| 616 |
/// # use unicode_segmentation::{GraphemeCursor, GraphemeIncomplete}; |
| 617 |
/// let s = "abcd"; |
| 618 |
/// let mut cursor = GraphemeCursor::new(0, s.len(), false); |
| 619 |
/// assert_eq!(cursor.next_boundary(&s[..2], 0), Ok(Some(1))); |
| 620 |
/// assert_eq!(cursor.next_boundary(&s[..2], 0), Err(GraphemeIncomplete::NextChunk)); |
| 621 |
/// assert_eq!(cursor.next_boundary(&s[2..4], 2), Ok(Some(2))); |
| 622 |
/// assert_eq!(cursor.next_boundary(&s[2..4], 2), Ok(Some(3))); |
| 623 |
/// assert_eq!(cursor.next_boundary(&s[2..4], 2), Ok(Some(4))); |
| 624 |
/// assert_eq!(cursor.next_boundary(&s[2..4], 2), Ok(None)); |
| 625 |
/// ``` |
| 626 |
pub fn next_boundary( |
| 627 |
&mut self, |
| 628 |
chunk: &str, |
| 629 |
chunk_start: usize, |
| 630 |
) -> Result<Option<usize>, GraphemeIncomplete> { |
| 631 |
if self.offset == self.len { |
| 632 |
return Ok(None); |
| 633 |
} |
| 634 |
let mut iter = chunk[self.offset - chunk_start..].chars(); |
| 635 |
let mut ch = iter.next().unwrap(); |
| 636 |
loop { |
| 637 |
if self.resuming { |
| 638 |
if self.cat_after.is_none() { |
| 639 |
self.cat_after = Some(self.grapheme_category(ch)); |
| 640 |
} |
| 641 |
} else { |
| 642 |
self.offset += ch.len_utf8(); |
| 643 |
self.state = GraphemeState::Unknown; |
| 644 |
self.cat_before = self.cat_after.take(); |
| 645 |
if self.cat_before.is_none() { |
| 646 |
self.cat_before = Some(self.grapheme_category(ch)); |
| 647 |
} |
| 648 |
if self.cat_before.unwrap() == GraphemeCat::GC_Regional_Indicator { |
| 649 |
self.ris_count = self.ris_count.map(|c| c + 1); |
| 650 |
} else { |
| 651 |
self.ris_count = Some(0); |
| 652 |
} |
| 653 |
if let Some(next_ch) = iter.next() { |
| 654 |
ch = next_ch; |
| 655 |
self.cat_after = Some(self.grapheme_category(ch)); |
| 656 |
} else if self.offset == self.len { |
| 657 |
self.decide(true); |
| 658 |
} else { |
| 659 |
self.resuming = true; |
| 660 |
return Err(GraphemeIncomplete::NextChunk); |
| 661 |
} |
| 662 |
} |
| 663 |
self.resuming = true; |
| 664 |
if self.is_boundary(chunk, chunk_start)? { |
| 665 |
self.resuming = false; |
| 666 |
return Ok(Some(self.offset)); |
| 667 |
} |
| 668 |
self.resuming = false; |
| 669 |
} |
| 670 |
} |
| 671 |
|
| 672 |
/// Find the previous boundary after the current cursor position. Only a part |
| 673 |
/// of the string need be supplied. If the chunk is incomplete, then this |
| 674 |
/// method might return `GraphemeIncomplete::PreContext` or |
| 675 |
/// `GraphemeIncomplete::PrevChunk`. In the former case, the caller should |
| 676 |
/// call `provide_context` with the requested chunk, then retry. In the |
| 677 |
/// latter case, the caller should provide the chunk preceding the one |
| 678 |
/// given, then retry. |
| 679 |
/// |
| 680 |
/// See `is_boundary` for expectations on the provided chunk. |
| 681 |
/// |
| 682 |
/// ```rust |
| 683 |
/// # use unicode_segmentation::GraphemeCursor; |
| 684 |
/// let flags = "\u{1F1F7}\u{1F1F8}\u{1F1EE}\u{1F1F4}"; |
| 685 |
/// let mut cursor = GraphemeCursor::new(12, flags.len(), false); |
| 686 |
/// assert_eq!(cursor.prev_boundary(flags, 0), Ok(Some(8))); |
| 687 |
/// assert_eq!(cursor.prev_boundary(flags, 0), Ok(Some(0))); |
| 688 |
/// assert_eq!(cursor.prev_boundary(flags, 0), Ok(None)); |
| 689 |
/// ``` |
| 690 |
/// |
| 691 |
/// And an example that uses partial strings (note the exact return is not |
| 692 |
/// guaranteed, and may be `PrevChunk` or `PreContext` arbitrarily): |
| 693 |
/// |
| 694 |
/// ```rust |
| 695 |
/// # use unicode_segmentation::{GraphemeCursor, GraphemeIncomplete}; |
| 696 |
/// let s = "abcd"; |
| 697 |
/// let mut cursor = GraphemeCursor::new(4, s.len(), false); |
| 698 |
/// assert_eq!(cursor.prev_boundary(&s[2..4], 2), Ok(Some(3))); |
| 699 |
/// assert_eq!(cursor.prev_boundary(&s[2..4], 2), Err(GraphemeIncomplete::PrevChunk)); |
| 700 |
/// assert_eq!(cursor.prev_boundary(&s[0..2], 0), Ok(Some(2))); |
| 701 |
/// assert_eq!(cursor.prev_boundary(&s[0..2], 0), Ok(Some(1))); |
| 702 |
/// assert_eq!(cursor.prev_boundary(&s[0..2], 0), Ok(Some(0))); |
| 703 |
/// assert_eq!(cursor.prev_boundary(&s[0..2], 0), Ok(None)); |
| 704 |
/// ``` |
| 705 |
pub fn prev_boundary( |
| 706 |
&mut self, |
| 707 |
chunk: &str, |
| 708 |
chunk_start: usize, |
| 709 |
) -> Result<Option<usize>, GraphemeIncomplete> { |
| 710 |
if self.offset == 0 { |
| 711 |
return Ok(None); |
| 712 |
} |
| 713 |
if self.offset == chunk_start { |
| 714 |
return Err(GraphemeIncomplete::PrevChunk); |
| 715 |
} |
| 716 |
let mut iter = chunk[..self.offset - chunk_start].chars().rev(); |
| 717 |
let mut ch = iter.next().unwrap(); |
| 718 |
loop { |
| 719 |
if self.offset == chunk_start { |
| 720 |
self.resuming = true; |
| 721 |
return Err(GraphemeIncomplete::PrevChunk); |
| 722 |
} |
| 723 |
if self.resuming { |
| 724 |
self.cat_before = Some(self.grapheme_category(ch)); |
| 725 |
} else { |
| 726 |
self.offset -= ch.len_utf8(); |
| 727 |
self.cat_after = self.cat_before.take(); |
| 728 |
self.state = GraphemeState::Unknown; |
| 729 |
if let Some(ris_count) = self.ris_count { |
| 730 |
self.ris_count = if ris_count > 0 { |
| 731 |
Some(ris_count - 1) |
| 732 |
} else { |
| 733 |
None |
| 734 |
}; |
| 735 |
} |
| 736 |
if let Some(prev_ch) = iter.next() { |
| 737 |
ch = prev_ch; |
| 738 |
self.cat_before = Some(self.grapheme_category(ch)); |
| 739 |
} else if self.offset == 0 { |
| 740 |
self.decide(true); |
| 741 |
} else { |
| 742 |
self.resuming = true; |
| 743 |
self.cat_after = Some(self.grapheme_category(ch)); |
| 744 |
return Err(GraphemeIncomplete::PrevChunk); |
| 745 |
} |
| 746 |
} |
| 747 |
self.resuming = true; |
| 748 |
if self.is_boundary(chunk, chunk_start)? { |
| 749 |
self.resuming = false; |
| 750 |
return Ok(Some(self.offset)); |
| 751 |
} |
| 752 |
self.resuming = false; |
| 753 |
} |
| 754 |
} |
| 755 |
} |
| 756 |
|
| 757 |
#[test] |
| 758 |
fn test_grapheme_cursor_ris_precontext() { |
| 759 |
let s = "\u{1f1fa}\u{1f1f8}\u{1f1fa}\u{1f1f8}\u{1f1fa}\u{1f1f8}"; |
| 760 |
let mut c = GraphemeCursor::new(8, s.len(), true); |
| 761 |
assert_eq!( |
| 762 |
c.is_boundary(&s[4..], 4), |
| 763 |
Err(GraphemeIncomplete::PreContext(4)) |
| 764 |
); |
| 765 |
c.provide_context(&s[..4], 0); |
| 766 |
assert_eq!(c.is_boundary(&s[4..], 4), Ok(true)); |
| 767 |
} |
| 768 |
|
| 769 |
#[test] |
| 770 |
fn test_grapheme_cursor_chunk_start_require_precontext() { |
| 771 |
let s = "\r\n"; |
| 772 |
let mut c = GraphemeCursor::new(1, s.len(), true); |
| 773 |
assert_eq!( |
| 774 |
c.is_boundary(&s[1..], 1), |
| 775 |
Err(GraphemeIncomplete::PreContext(1)) |
| 776 |
); |
| 777 |
c.provide_context(&s[..1], 0); |
| 778 |
assert_eq!(c.is_boundary(&s[1..], 1), Ok(false)); |
| 779 |
} |
| 780 |
|
| 781 |
#[test] |
| 782 |
fn test_grapheme_cursor_prev_boundary() { |
| 783 |
let s = "abcd"; |
| 784 |
let mut c = GraphemeCursor::new(3, s.len(), true); |
| 785 |
assert_eq!( |
| 786 |
c.prev_boundary(&s[2..], 2), |
| 787 |
Err(GraphemeIncomplete::PrevChunk) |
| 788 |
); |
| 789 |
assert_eq!(c.prev_boundary(&s[..2], 0), Ok(Some(2))); |
| 790 |
} |
| 791 |
|
| 792 |
#[test] |
| 793 |
fn test_grapheme_cursor_prev_boundary_chunk_start() { |
| 794 |
let s = "abcd"; |
| 795 |
let mut c = GraphemeCursor::new(2, s.len(), true); |
| 796 |
assert_eq!( |
| 797 |
c.prev_boundary(&s[2..], 2), |
| 798 |
Err(GraphemeIncomplete::PrevChunk) |
| 799 |
); |
| 800 |
assert_eq!(c.prev_boundary(&s[..2], 0), Ok(Some(1))); |
| 801 |
} |