Browse Source

text_processing: implement Unicode line breaking algorithm

Not conforming to the unicode standard yet
tags/pre-alpha-0.2.0
Manos Pitsidianakis 4 weeks ago
parent
commit
5b679be782
No known key found for this signature in database

+ 4
- 0
testing/Cargo.toml View File

@@ -8,7 +8,11 @@ edition = "2018"
[[bin]]
name = "emailparse"
path = "src/email_parse.rs"
[[bin]]
name = "linebreak"
path = "src/linebreak.rs"


[dependencies]
melib = { path = "../melib", version = "*" }
text_processing = { path = "../text_processing", version = "*" }

+ 162
- 0
testing/src/linebreak.rs View File

@@ -0,0 +1,162 @@
extern crate melib;
use melib::Result;
use melib::StackVec;

extern crate text_processing;
use text_processing::line_break::*;

fn cost(i: usize, j: usize, width: usize, minima: &Vec<usize>, offsets: &Vec<usize>) -> usize {
let w = offsets[j] - offsets[i] + j - i - 1;
if w > width {
return 65536 * (w - width);
}
minima[i] + (width - w) * (width - w)
}

fn smawk(
rows: &mut StackVec<usize>,
columns: &mut StackVec<usize>,
minima: &mut Vec<usize>,
breaks: &mut Vec<usize>,
width: usize,
offsets: &Vec<usize>,
) {
let mut stack = StackVec::new();
let mut i = 0;
while i < rows.len() {
if stack.len() > 0 {
let c = columns[stack.len() - 1];
if cost(*stack.iter().last().unwrap(), c, width, minima, offsets)
< cost(rows[i], c, width, minima, offsets)
{
if stack.len() < columns.len() {
stack.push(rows[i]);
}
i += 1;
} else {
stack.pop();
}
} else {
stack.push(rows[i]);
i += 1;
}
}
let rows = &mut stack;
if columns.len() > 1 {
let mut odd_columns = columns.iter().skip(1).step_by(2).cloned().collect();
smawk(rows, &mut odd_columns, minima, breaks, width, offsets);
for (i, o) in odd_columns.into_iter().enumerate() {
columns.set(2 * i + 1, o);
}
}
let mut i = 0;
let mut j = 0;
while j < columns.len() {
let end = if j + 1 < columns.len() {
breaks[columns[j + 1]]
} else {
*rows.iter().last().unwrap()
};
let c = cost(rows[i], columns[j], width, minima, offsets);
if c < minima[columns[j]] {
minima[columns[j]] = c;
breaks[columns[j]] = rows[i];
}
if rows[i] < end {
i += 1;
} else {
j += 2;
}
}
}

fn linear(text: &str, width: usize) -> Vec<String> {
let mut words = Vec::new();
let breaks = LineBreakCandidateIter::new(text).collect::<Vec<(usize, LineBreakCandidate)>>();
{
let mut prev = 0;
for b in breaks {
if &text[prev..b.0] != "\n" {
words.push(text[prev..b.0].trim_end_matches("\n"));
if text[prev..b.0].ends_with("\n") {
words.push(" ");
}
}
prev = b.0;
}
if &text[prev..] != "\n" {
words.push(text[prev..].trim_end_matches("\n"));
}
}
let count = words.len();
let mut minima = vec![std::usize::MAX - 1; count + 1];
minima[0] = 0;
let mut offsets = Vec::with_capacity(words.len());
offsets.push(0);
for w in words.iter() {
offsets.push(offsets.iter().last().unwrap() + w.len());
}

let mut breaks = vec![0; count + 1];

let mut n = count + 1;
let mut i = 1;
let mut offset = 0;
loop {
let r = std::cmp::min(n, 2 * i);
let edge = i + offset;
smawk(
&mut (offset..edge).collect(),
&mut (edge..(r + offset)).collect(),
&mut minima,
&mut breaks,
width,
&offsets,
);
let x = minima[r - 1 + offset];
let mut for_was_broken = false;
for j in i..(r - 1) {
let y = cost(j + offset, r - 1 + offset, width, &minima, &offsets);
if y <= x {
n -= j;
i = 1;
offset += j;
for_was_broken = true;
break;
}
}

if !for_was_broken || i >= (r - 1) {
if r == n {
break;
}
i *= 2;
}
}
let mut lines = Vec::new();
let mut j = count;
while j > 0 {
let mut line = String::new();
for i in breaks[j]..j {
line.push_str(words[i]);
}
lines.push(line);
j = breaks[j];
}
lines.reverse();
lines
}

fn main() -> Result<()> {
let text = std::fs::read_to_string(std::env::args().nth(1).unwrap())?;
let paragraphs = text.split("\n\n").collect::<Vec<&str>>();
for (i, p) in paragraphs.iter().enumerate() {
for l in linear(&p, 72) {
println!("{}", l.trim());
}
if i + 1 < paragraphs.len() {
println!("");
}
}
Ok(())
}

+ 1
- 0
text_processing/Cargo.toml View File

@@ -4,6 +4,7 @@ version = "0.0.1" #:version
authors = ["Manos Pitsidianakis <el13635@mail.ntua.gr>"]
workspace = ".."
edition = "2018"
build = "build.rs"

[dependencies]
unicode-segmentation = "1.2.1"

+ 73
- 0
text_processing/build.rs View File

@@ -0,0 +1,73 @@
const LINE_BREAK_TABLE_URL: &str = "http://www.unicode.org/Public/UCD/latest/ucd/LineBreak.txt";
use std::fs::File;
use std::io::prelude::*;
use std::io::BufReader;
use std::path::PathBuf;
use std::process::Command;

include!("src/types.rs");

fn main() -> Result<(), std::io::Error> {
let mod_path = PathBuf::from("src/tables.rs");
if mod_path.exists() {
eprintln!(
"{} already exists, delete it if you want to replace it.",
mod_path.display()
);
std::process::exit(0);
}
let mut tmpdir_path = PathBuf::from(
std::str::from_utf8(&Command::new("mktemp").arg("-d").output()?.stdout)
.unwrap()
.trim(),
);
tmpdir_path.push("LineBreak.txt");
Command::new("curl")
.args(&["-o", tmpdir_path.to_str().unwrap(), LINE_BREAK_TABLE_URL])
.output()?;

let file = File::open(&tmpdir_path)?;
let buf_reader = BufReader::new(file);

let mut line_break_table: Vec<(u32, u32, LineBreakClass)> = Vec::with_capacity(3800);
for line in buf_reader.lines() {
let line = line.unwrap();
if line.starts_with('#') || line.starts_with(' ') || line.is_empty() {
continue;
}
let tokens: &str = line.split_whitespace().next().unwrap();

let semicolon_idx: usize = tokens.chars().position(|c| c == ';').unwrap();
/* LineBreak.txt list is ascii encoded so we can assume each char takes one byte: */
let chars_str: &str = &tokens[..semicolon_idx];

let mut codepoint_iter = chars_str.split("..");

let first_codepoint: u32 =
u32::from_str_radix(std::dbg!(codepoint_iter.next().unwrap()), 16).unwrap();

let sec_codepoint: u32 = codepoint_iter
.next()
.map(|v| u32::from_str_radix(std::dbg!(v), 16).unwrap())
.unwrap_or(first_codepoint);
let class = &tokens[semicolon_idx + 1..semicolon_idx + 1 + 2];
line_break_table.push((first_codepoint, sec_codepoint, LineBreakClass::from(class)));
}

let mut file = File::create(&mod_path)?;
file.write_all(b"use crate::types::LineBreakClass::*;\n")
.unwrap();
file.write_all(b"use crate::types::LineBreakClass;\n\n")
.unwrap();
file.write_all(b"const line_break_rules: &'static [(u32, u32, LineBreakClass)] = &[\n")
.unwrap();
for l in &line_break_table {
file.write_all(format!(" (0x{:X}, 0x{:X}, {:?}),\n", l.0, l.1, l.2).as_bytes())
.unwrap();
}
file.write_all(b"];").unwrap();
std::fs::remove_file(&tmpdir_path).unwrap();
tmpdir_path.pop();
std::fs::remove_dir(&tmpdir_path).unwrap();
Ok(())
}

+ 4
- 0
text_processing/src/lib.rs View File

@@ -1,4 +1,8 @@
pub mod grapheme_clusters;
pub mod line_break;
mod tables;
mod types;
pub mod wcwidth;
pub use grapheme_clusters::*;
pub use line_break::*;
pub use wcwidth::*;

+ 703
- 0
text_processing/src/line_break.rs View File

@@ -0,0 +1,703 @@
extern crate unicode_segmentation;
use self::unicode_segmentation::UnicodeSegmentation;
use crate::tables::LINE_BREAK_RULES;
use crate::types::LineBreakClass;
use core::cmp::Ordering;
use core::iter::Peekable;
use core::str::FromStr;
use LineBreakClass::*;

#[derive(Debug, PartialEq)]
pub enum LineBreakCandidate {
MandatoryBreak,
BreakAllowed,
// NoBreak, Not used.
}

use LineBreakCandidate::*;

pub struct LineBreakCandidateIter<'a> {
text: &'a str,
iter: Peekable<unicode_segmentation::GraphemeIndices<'a>>,
pos: usize,
/* Needed for rule LB30a */
reg_ind_streak: u32,
}

impl<'a> LineBreakCandidateIter<'a> {
pub fn new(text: &'a str) -> Self {
LineBreakCandidateIter {
text,
pos: 0,
iter: UnicodeSegmentation::grapheme_indices(text, true).peekable(),
reg_ind_streak: 0,
}
}
}

macro_rules! get_base_character {
($grapheme:ident) => {{
char::from_str($grapheme.get(0..1).unwrap_or_else(|| {
$grapheme.get(0..2).unwrap_or_else(|| {
$grapheme
.get(0..3)
.unwrap_or_else(|| $grapheme.get(0..4).unwrap())
})
}))
}};
($grapheme:expr) => {{
char::from_str($grapheme.get(0..1).unwrap_or_else(|| {
$grapheme.get(0..2).unwrap_or_else(|| {
$grapheme
.get(0..3)
.unwrap_or_else(|| $grapheme.get(0..4).unwrap())
})
}))
}};
}

/// Side effects: none
macro_rules! get_class {
($grapheme:ident) => {{
get_base_character!($grapheme)
.map(|char| search_table(char as u32, LINE_BREAK_RULES))
.unwrap_or(XX)
}};
($grapheme:expr) => {{
get_base_character!($grapheme)
.map(|char| search_table(char as u32, LINE_BREAK_RULES))
.unwrap_or(XX)
}};
}

/// Side effects: Updates $graph_iter and potentially $idx and $grapheme
macro_rules! next_grapheme_class {
($graph_iter:ident, $grapheme:ident) => ({
if let Some((_, g)) = $graph_iter.next() {
$grapheme = g;
Some(get_class!(g))
} else { None }
});
(($next_char:ident is $class:expr)) => ({
$next_char.is_some() && get_class!(($next_char.unwrap().1)) == $class
});
(($next_char:ident is $($class:ident),+)) => ({
$next_char.is_some() && ($(get_class!(($next_char.unwrap().1)) == $class)||+)
});
}

/// Returns positions where breaks can happen
/// Examples:
/// ```
/// use text_processing::{self, LineBreakCandidate::{self, *}};
/// use text_processing::line_break::LineBreakCandidateIter;
///
/// assert!(LineBreakCandidateIter::new("").collect::<Vec<(usize, LineBreakCandidate)>>().is_empty());
/// assert_eq!(&[(7, BreakAllowed), (12, MandatoryBreak)],
/// LineBreakCandidateIter::new("Sample Text.").collect::<Vec<(usize, LineBreakCandidate)>>().as_slice());
/// assert_eq!(&[(3, MandatoryBreak), (7, MandatoryBreak), (10, BreakAllowed), (17, MandatoryBreak)],
/// LineBreakCandidateIter::new("Sa\nmp\r\nle T(e)xt.").collect::<Vec<(usize, LineBreakCandidate)>>().as_slice());
/// ```
impl<'a> Iterator for LineBreakCandidateIter<'a> {
type Item = (usize, LineBreakCandidate);
fn next(&mut self) -> Option<Self::Item> {
// After end of text, there are no breaks.
if self.pos >= self.text.len() {
return None;
}
// LB3 Always break at the end of text
if self.pos + 1 == self.text.len() {
self.pos += 1;
return Some((self.pos, MandatoryBreak));
}

let (idx, mut grapheme) = self.iter.next().unwrap();
let LineBreakCandidateIter {
ref mut iter,
ref text,
ref mut reg_ind_streak,
ref mut pos,
} = self;
let iter = iter.by_ref();

debug_assert_eq!(idx, *pos);

// LB2 Never break at the start of text
if idx == 0 {
*pos += grapheme.len();
return self.next();
}

let class = get_class!(grapheme);

if class != RI {
*reg_ind_streak = 0;
}

/* LB1 Assign a line breaking class to each code point of the input. Resolve AI, CB, CJ,
* SA, SG, and XX into other line breaking classes depending on criteria outside the scope
* of this algorithm.
*
* In the absence of such criteria all characters with a specific combination of original
* class and General_Category property value are resolved as follows:
* Resolved Original General_Category
* AL AI, SG, XX Any
* CM SA Only Mn or Mc
* AL SA Any except Mn and Mc
* NS SJ Any
*/

// TODO: LB1

/* Check if next character class allows breaks before it */
let next_char: Option<&(usize, &str)> = iter.peek();

match class {
BK => {
// LB4 Always Break after hard line breaks.
*pos += grapheme.len();
return Some((*pos, MandatoryBreak));
}
// LB5 Treat CR followed by LF, as well as CR, LF, and NL as hard line breaks
CR if next_grapheme_class!((next_char is LF)) => {
*pos += grapheme.len();
assert!(Some(LF) == next_grapheme_class!(iter, grapheme));
*pos += grapheme.len();
return Some((*pos, MandatoryBreak));
}
CR | LF | NL => {
*pos += grapheme.len();
return Some((*pos, MandatoryBreak));
}
_ => {}
}
if let Some((_, next_grapheme)) = next_char {
let next_class = get_class!(next_grapheme);
match next_class {
/* LB6 Do not break before hard line breaks. × ( BK | CR | LF | NL ) */
BK | CR | LF | NL => {
*pos += grapheme.len();
return self.next();
}
/* LB7 Do not break before spaces or zero width
* space. × SP × ZW */
SP | ZW => {
*pos += grapheme.len();
return self.next();
}
_ => {}
}
}
match class {
ZW => {
// LB8 Break before any character following a zero-width space, even if one or more
// spaces intervene
// ZW SP* ÷
*pos += grapheme.len();
while Some(SP) == next_grapheme_class!(iter, grapheme) {
*pos += grapheme.len();
}
return Some((*pos, MandatoryBreak));
}
ZWJ => {
// LB8a Do not break after a zero width joiner.
*pos += grapheme.len();
return self.next();
}

CM => {
// LB9 Do not break a combining character sequence; treat it as if it has the line
// breaking class of the base character in all of the following rules. Treat ZWJ as
// if it were CM.
// Treat X (CM | ZWJ)* as if it were X.
// where X is any line break class except BK, CR, LF, NL, SP, or ZW.

/* Unreachable since we break lines based on graphemes, not characters */
unreachable!();
}
WJ => {
/*: LB11 Do not break before or after Word joiner and related characters.*/
*pos += grapheme.len();
/* Get next grapheme */
if next_grapheme_class!(iter, grapheme).is_some() {
*pos += grapheme.len();
}
return self.next();
}
GL => {
/*LB12 Non-breaking characters: LB12 Do not break after NBSP and related characters.*/
*pos += grapheme.len();
return self.next();
}
_ => {}
}
if let Some((next_idx, next_grapheme)) = next_char {
let next_class = get_class!(next_grapheme);
match next_class {
GL if ![SP, BA, HY].contains(&class) => {
/* LB12a Do not break before NBSP and related characters, except after spaces and
* hyphens. [^SP BA HY] × GL
* Also LB12 Do not break after NBSP and related characters */
*pos += grapheme.len();
return self.next();
}
/* LB13 Do not break before ‘]’ or ‘!’ or ‘;’ or ‘/’, even after spaces. */
CL | CP | EX | IS | SY => {
*pos = *next_idx;
return self.next();
}
_ => {}
}
}

match class {
/* LB13 Do not break before ‘]’ or ‘!’ or ‘;’ or ‘/’, even after spaces. */
SP if [CL, CP, EX, IS, SY].contains(&get_class!(text[idx..].trim_start())) => {
*pos += grapheme.len();
while ![CL, CP, EX, IS, SY].contains(&next_grapheme_class!(iter, grapheme).unwrap())
{
*pos += grapheme.len();
}
*pos += grapheme.len();
return self.next();
}
OP => {
/* LB14 Do not break after ‘[’, even after spaces.
* OP SP* ×
*/
while let Some((idx, grapheme)) = self.iter.next() {
*pos = idx + grapheme.len();
if !(get_class!(grapheme) == SP) {
break;
}
}
return self.next();
}
QU if get_class!(text[idx..].trim_start()) == OP => {
/* LB15 Do not break within ‘”[’, even with intervening spaces.
* QU SP* × OP */
*pos += grapheme.len();
while Some(SP) == next_grapheme_class!(iter, grapheme) {
*pos += grapheme.len();
}
*pos = idx;
return self.next();
}
QU => {
/* LB19 Do not break before or after quotation marks, such as ‘ ” ’. */
*pos += grapheme.len();
if let Some((_, g)) = self.iter.next() {
*pos += g.len();
}
return self.next();
}
LineBreakClass::CL | LineBreakClass::CP
if get_class!(text[idx..].trim_start()) == NS =>
{
/* LB16 Do not break between closing punctuation and a nonstarter (lb=NS), even with
* intervening spaces.
* (CL | CP) SP* × NS */
*pos += grapheme.len();
while Some(SP) == next_grapheme_class!(iter, grapheme) {
*pos += grapheme.len();
}
return self.next();
}
B2 if get_class!(text[idx..].trim_start()) == B2 => {
*pos += grapheme.len();
while Some(SP) == next_grapheme_class!(iter, grapheme) {
*pos += grapheme.len();
}
return self.next();
}
SP => {
/* LB18 Break after spaces. SP ÷ */
// Space 0x20 is 1 byte long.
*pos += 1;
return Some((*pos, BreakAllowed));
}
_ => {}
}
if let Some((next_idx, next_grapheme)) = next_char {
let next_class = get_class!(next_grapheme);
match next_class {
QU if class != SP => {
/* LB19 Do not break before or after quotation marks, such as ‘ ” ’. */
*pos = *next_idx + next_grapheme.len();
self.iter.next();
return self.next();
}
_ => {}
}
}
match class {
CB => {
/* LB20 Break before and after unresolved CB. */
*pos += grapheme.len();
return Some((*pos - 1, BreakAllowed));
}
/* LB21 Do not break before hyphen-minus, other hyphens, fixed-width spaces, small
* kana, and other non-starters, or after acute accents. × BA, × HY, × NS, BB × */
BB => {
*pos += grapheme.len();
return self.next();
}
_ => {}
}

if let Some((_, next_grapheme)) = next_char {
let next_class = get_class!(next_grapheme);
match next_class {
BA | HY | NS => {
/* LB21 Do not break before hyphen-minus, other hyphens, fixed-width spaces, small
* kana, and other non-starters, or after acute accents. × BA, × HY, × NS, BB × */
*pos += grapheme.len();
return self.next();
}
_ => {}
}
}
match class {
HL if next_grapheme_class!((next_char is HY, BA)) => {
/* LB21a Don’t break after Hebrew + Hyphen. HL (HY | BA) × */
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* LB21b Don’t break between ,Solidus and Hebrew letters. SY × HL */
SY if next_grapheme_class!((next_char is HL)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
/* bypass next_char */
self.iter.next().unwrap();
if let Some((idx, next_grapheme)) = self.iter.next() {
*pos = idx + next_grapheme.len();
}
return self.next();
}
/* LB22 Do not break between two ellipses, or between letters, numbers or excla-
* mations and ellipsis.
* Examples: ‘9...’, ‘a...’, ‘H...’
* (AL | HL) × IN */
AL | HL if next_grapheme_class!((next_char is IN)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* EX × IN */
EX if next_grapheme_class!((next_char is IN)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
EX => {
// LB13
*pos += grapheme.len();
return self.next();
}
/* (ID | EB | EM) × IN */
ID | EB | EM if next_grapheme_class!((next_char is IN)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* IN × IN */
IN if next_grapheme_class!((next_char is IN)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* NU × IN */
NU if next_grapheme_class!((next_char is IN)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* LB23 Do not break between digits and letters.
* (AL | HL) × NU */
AL | HL if next_grapheme_class!((next_char is NU)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* NU × (AL | HL) */
NU if next_grapheme_class!((next_char is AL, HL)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* LB23a Do not break between numeric prefixes and ideographs, or between ideographs
* and numeric postfixes.
* PR × (ID | EB | EM) */
PR if next_grapheme_class!((next_char is ID, EB, EM)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* (ID | EB | EM) × PO */
ID | EB | EM if next_grapheme_class!((next_char is PO)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* B24 Do not break between numeric prefix/postfix and letters, or between
letters and prefix/postfix.
(PR | PO) × (AL | HL)*/
PR | PO if next_grapheme_class!((next_char is AL, HL)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/*(AL | HL) × (PR | PO) */
AL | HL if next_grapheme_class!((next_char is PR, PO)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* LB25 Do not break between the following pairs of classes relevant to numbers:
* CL × PO */
CL if next_grapheme_class!((next_char is PO)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* CP × PO */
CP if next_grapheme_class!((next_char is PO)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* CL × PR */
CL if next_grapheme_class!((next_char is PR)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* CP × PR */
CP if next_grapheme_class!((next_char is PR)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* NU × PO */
NU if next_grapheme_class!((next_char is PO)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* NU × PR */
NU if next_grapheme_class!((next_char is PR)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* PO × OP */
PO if next_grapheme_class!((next_char is OP)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* PO × NU */
PO if next_grapheme_class!((next_char is NU)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* PR × OP */
PR if next_grapheme_class!((next_char is OP)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* PR × NU */
PR if next_grapheme_class!((next_char is NU)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* HY × NU */
HY if next_grapheme_class!((next_char is NU)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* IS × NU */
IS if next_grapheme_class!((next_char is NU)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* NU × NU */
NU if next_grapheme_class!((next_char is NU)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* SY × NU */
SY if next_grapheme_class!((next_char is NU)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* LB26 Do not break a Korean syllable.
* JL × (JL | JV | H2 | H3) */
JL if next_grapheme_class!((next_char is JL, JV, H2, H3)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* (JV | H2) × (JV | JT) */
JV | H2 if next_grapheme_class!((next_char is JV, JT)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* (JT | H3) × JT */
JT | H3 if next_grapheme_class!((next_char is JT)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* LB27 Treat a Korean Syllable Block the same as ID.
* (JL | JV | JT | H2 | H3) × IN */
JL | JV | JT | H2 | H3 if next_grapheme_class!((next_char is IN)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* (JL | JV | JT | H2 | H3) × PO */
JL | JV | JT | H2 | H3 if next_grapheme_class!((next_char is PO)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* PR × (JL | JV | JT | H2 | H3) */
PR if next_grapheme_class!((next_char is JL, JV, JT, H2, H3)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* LB28 Do not break between alphabetics (“at”).
(AL | HL) × (AL | HL) */
AL | HL if next_grapheme_class!((next_char is AL, HL)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* LB29 Do not break between numeric punctuation and alphabetics (“e.g.”).
IS × (AL | HL) */
IS if next_grapheme_class!((next_char is AL, HL)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* LB30 Do not break between letters, numbers, or ordinary symbols and opening
or closing parentheses.
(AL | HL | NU) × OP */
AL | HL | NU if next_grapheme_class!((next_char is OP)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/* CP × (AL | HL | NU) */
CP if next_grapheme_class!((next_char is AL, HL , NU)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
/*LB30b Do not break between an emoji base and an emoji modifier.
* EB × EM */
EB if next_grapheme_class!((next_char is EM)) => {
let (idx, next_grapheme) = next_char.unwrap();
*pos = idx + next_grapheme.len();
self.iter.next();
return self.next();
}
RI => {
/* LB30a Break between two regional indicator symbols if and only if there are an
* even number of regional indicators preceding the position of the break.
* sot (RI RI)* RI × RI
* [^RI] (RI RI)* RI × RI */
*reg_ind_streak += 1;
*pos += grapheme.len();
if *reg_ind_streak % 2 == 1 {
return Some((*pos - grapheme.len(), BreakAllowed));
}
self.iter.next();
return self.next();
}
_ => {
*pos += grapheme.len();
return Some((*pos - grapheme.len(), BreakAllowed));
}
}
}
}

fn search_table(c: u32, t: &'static [(u32, u32, LineBreakClass)]) -> LineBreakClass {
match t.binary_search_by(|&(lo, hi, _)| {
if lo <= c && c <= hi {
Ordering::Equal
} else if hi < c {
Ordering::Less
} else {
Ordering::Greater
}
}) {
Ok(idx) => t[idx].2,
Err(_) => XX,
}
}

#[cfg(test)]
mod tests {
use super::*;

#[test]
fn test_line_breaks() {
let s = "Fell past it.\n\n‘Well!’ thought Alice to herself.";
let breaks = LineBreakCandidateIter::new(s).collect::<Vec<(usize, LineBreakCandidate)>>();
let mut prev = 0;
for b in breaks {
println!("{:?}", &s[prev..b.0]);
prev = b.0;
}
println!("{:?}", &s[prev..]);
}
}

+ 3389
- 0
text_processing/src/tables.rs
File diff suppressed because it is too large
View File


+ 102
- 0
text_processing/src/types.rs View File

@@ -0,0 +1,102 @@
#[derive(Debug, Copy, Clone, PartialEq)]
pub enum LineBreakClass {
BK,
CM,
CR,
GL,
LF,
NL,
SP,
WJ,
ZW,
ZWJ,
AI,
AL,
B2,
BA,
BB,
CB,
CJ,
CL,
CP,
EB,
EM,
EX,
H2,
H3,
HL,
HY,
ID,
IN,
IS,
JL,
JT,
JV,
NS,
NU,
OP,
PO,
PR,
QU,
RI,
SA,
SG,
SY,
XX,
}

use LineBreakClass::*;

impl From<&str> for LineBreakClass {
fn from(val: &str) -> Self {
match val {
stringify!(BK) => BK,
stringify!(CM) => CM,
stringify!(CR) => CR,
stringify!(GL) => GL,
stringify!(LF) => LF,
stringify!(NL) => NL,
stringify!(SP) => SP,
stringify!(WJ) => WJ,
stringify!(ZW) => ZW,
stringify!(ZWJ) => ZWJ,
stringify!(AI) => AI,
stringify!(AL) => AL,
stringify!(B2) => B2,
stringify!(BA) => BA,
stringify!(BB) => BB,
stringify!(CB) => CB,
stringify!(CJ) => CJ,
stringify!(CL) => CL,
stringify!(CP) => CP,
stringify!(EB) => EB,

stringify!(EM) => EM,
stringify!(EX) => EX,
stringify!(H2) => H2,
stringify!(H3) => H3,
stringify!(HL) => HL,
stringify!(HY) => HY,
stringify!(ID) => ID,
stringify!(IN) => IN,
stringify!(IS) => IS,
stringify!(JL) => JL,

stringify!(JT) => JT,
stringify!(JV) => JV,
stringify!(NS) => NS,
stringify!(NU) => NU,
stringify!(OP) => OP,
stringify!(PO) => PO,
stringify!(PR) => PR,
stringify!(QU) => QU,
stringify!(RI) => RI,
stringify!(SA) => SA,

stringify!(SG) => SG,
stringify!(SY) => SY,
stringify!(XX) => XX,
_ => unreachable!(),
}
}
}

Loading…
Cancel
Save