Add gettext po/pot support

This commit is contained in:
2025-09-06 17:05:29 +08:00
parent 65b22ccddf
commit ca3cd41c0a
4 changed files with 843 additions and 0 deletions

View File

@@ -1,2 +1,3 @@
//! Module for output scripts.
pub mod m3t;
pub mod po;

734
src/output_scripts/po.rs Normal file
View File

@@ -0,0 +1,734 @@
//! tools to process gettext po/pot files
//!
//! See [spec](https://www.gnu.org/software/gettext/manual/html_node/PO-Files.html)
use crate::types::*;
use anyhow::{Result, anyhow};
use std::collections::HashMap;
use unicode_segmentation::UnicodeSegmentation;
#[derive(Debug)]
/// A comment line
pub enum Comment {
/// Translator comment, starting with `# `
Translator(String),
/// Extracted comment, starting with `#.`.
Extracted(String),
/// Reference, starting with `#:`.
Reference(String),
/// Flag, starting with `#,`.
Flag(Vec<String>),
/// Previous untranslated string, starting with `#|`.
Previous(String),
/// Previous message block, starting with `#~`.
PreviousStr(String),
}
#[derive(Debug)]
/// A line in a .po file.
pub enum PoLine {
/// A comment line, starting with `#`.
Comment(Comment),
/// A msgid line
MsgId(String),
/// A msgstr line
MsgStr(String),
/// A msgctxt line
MsgCtxt(String),
/// A msgid_plural line
MsgIdPlural(String),
/// A msgstr[n] line
MsgStrN(usize, String),
/// Empty line
EmptyLine,
}
fn dump_text_in_multi_lines(s: &str) -> Result<String> {
if s.contains("\n") {
let mut result = Vec::new();
result.push("\"\"".to_string());
let mut s = s;
while let Some(pos) = s.find('\n') {
let line = &s[..pos + 1];
result.push(format!("\"{}\"", escape_c_str(line)?));
s = &s[pos + 1..];
}
if !s.is_empty() {
result.push(format!("\"{}\"", escape_c_str(s)?));
}
Ok(result.join("\n"))
} else {
Ok(format!("\"{}\"", escape_c_str(s)?))
}
}
impl PoLine {
fn dump(&self) -> Result<String> {
Ok(match self {
PoLine::Comment(c) => match c {
Comment::Translator(s) => format!("# {}", s),
Comment::Extracted(s) => format!("#. {}", s),
Comment::Reference(s) => format!("#: {}", s),
Comment::Flag(flags) => format!("#, {}", flags.join(", ")),
Comment::Previous(s) => format!("#| {}", s),
Comment::PreviousStr(s) => format!("#~ {}", s),
},
PoLine::MsgId(s) => format!("msgid {}", dump_text_in_multi_lines(s)?),
PoLine::MsgStr(s) => format!("msgstr {}", dump_text_in_multi_lines(s)?),
PoLine::MsgCtxt(s) => format!("msgctxt {}", dump_text_in_multi_lines(s)?),
PoLine::MsgIdPlural(s) => format!("msgid_plural {}", dump_text_in_multi_lines(s)?),
PoLine::MsgStrN(n, s) => format!("msgstr[{}] {}", n, dump_text_in_multi_lines(s)?),
PoLine::EmptyLine => String::new(),
})
}
}
#[derive(Debug)]
pub enum MsgStr {
Single(String),
Plural(Vec<(usize, String)>),
}
#[derive(Debug)]
pub struct PoEntry {
comments: Vec<Comment>,
msgctxt: Option<String>,
msgid: String,
msgid_plural: Option<String>,
msgstr: MsgStr,
}
/// Escapes a string according to C-style rules.
///
/// This function handles common escape sequences like \n, \t, \", \\, etc.
/// For other ASCII control characters or non-printable characters, it uses octal notation (e.g., \0, \177).
///
/// # Arguments
/// * `s`: The string slice to be escaped.
///
/// # Returns
/// A `Result<String>` containing the new escaped string.
pub fn escape_c_str(s: &str) -> Result<String> {
let mut escaped = String::with_capacity(s.len());
for c in s.chars() {
match c {
'\n' => escaped.push_str("\\n"),
'\r' => escaped.push_str("\\r"),
'\t' => escaped.push_str("\\t"),
'\\' => escaped.push_str("\\\\"),
'\"' => escaped.push_str("\\\""),
'\0' => escaped.push_str("\\0"),
'\x08' => escaped.push_str("\\b"),
'\x0c' => escaped.push_str("\\f"),
'\x0b' => escaped.push_str("\\v"),
'\x07' => escaped.push_str("\\a"),
c if c.is_ascii_control() && c != '\n' && c != '\r' && c != '\t' => {
escaped.push_str(&format!("\\{:03o}", c as u8));
}
_ => escaped.push(c),
}
}
Ok(escaped)
}
/// Unescapes a string that has been escaped C-style.
///
/// This function parses common escape sequences (like \n, \t, \", \\) as well as
/// octal (\ooo) and hexadecimal (\xHH) escape notations.
///
/// # Arguments
/// * `s`: The string slice containing C-style escape sequences.
///
/// # Returns
/// A `Result<String>` containing the new unescaped string.
/// If an invalid escape sequence is encountered, an error is returned.
pub fn unescape_c_str(s: &str) -> Result<String> {
let mut unescaped = String::with_capacity(s.len());
let mut chars = s.chars().peekable();
while let Some(c) = chars.next() {
if c == '\\' {
match chars.next() {
Some('n') => unescaped.push('\n'),
Some('r') => unescaped.push('\r'),
Some('t') => unescaped.push('\t'),
Some('b') => unescaped.push('\x08'),
Some('f') => unescaped.push('\x0c'),
Some('v') => unescaped.push('\x0b'),
Some('a') => unescaped.push('\x07'),
Some('\\') => unescaped.push('\\'),
Some('\'') => unescaped.push('\''),
Some('\"') => unescaped.push('\"'),
Some('?') => unescaped.push('?'),
Some(o @ '0'..='7') => {
let mut octal = String::new();
octal.push(o);
while let Some(peek_char) = chars.peek() {
if peek_char.is_digit(8) && octal.len() < 3 {
octal.push(chars.next().unwrap());
} else {
break;
}
}
let value = u8::from_str_radix(&octal, 8).map_err(|e| {
anyhow!("Invalid octal escape sequence: \\{}: {}", octal, e)
})?;
unescaped.push(value as char);
}
// --- FIX START: Reworked hexadecimal parsing logic ---
Some('x') => {
let mut hex = String::new();
// Read the first character, which must be a hex digit
if let Some(c1) = chars.peek() {
if c1.is_ascii_hexdigit() {
hex.push(chars.next().unwrap());
} else {
// Handle cases like \xG
return Err(anyhow!(
"Invalid hex escape sequence: \\x followed by non-hex character '{}'",
c1
));
}
} else {
// Handle cases where \x is at the end of the string
return Err(anyhow!(
"Invalid hex escape sequence: \\x must be followed by a hex digit"
));
}
// Try to read the second character, which must also be a hex digit
if let Some(c2) = chars.peek() {
if c2.is_ascii_hexdigit() {
hex.push(chars.next().unwrap());
} else {
// Handle cases like \xFG
// We have successfully parsed one digit (like F), but it's followed by an invalid hex character (like G)
// As per the test requirements, this should be an error
return Err(anyhow!(
"Invalid hex escape sequence: \\x{} followed by non-hex character '{}'",
hex,
c2
));
}
}
let value =
u8::from_str_radix(&hex, 16).expect("Hex parsing should be valid here");
unescaped.push(value as char);
}
// --- FIX END ---
Some(other) => {
return Err(anyhow!("Unknown escape sequence: \\{}", other));
}
None => {
return Err(anyhow!("String cannot end with a single backslash"));
}
}
} else {
unescaped.push(c);
}
}
Ok(unescaped)
}
pub struct PoDumper {
entries: Vec<PoLine>,
}
impl PoDumper {
pub fn new() -> Self {
Self {
entries: Vec::new(),
}
}
fn gen_start_str(encoding: Encoding) -> String {
let mut map = HashMap::new();
let content_type = match encoding.charset() {
Some(e) => format!("text/plain; charset={}", e),
None => String::from("text/plain"),
};
map.insert("Content-Type", content_type);
map.insert("X-Generator", String::from("msg-tool"));
map.insert("MIME-Version", String::from("1.0"));
let mut result = String::new();
for (k, v) in map {
result.push_str(&format!("{}: {}\n", k, v));
}
result
}
pub fn dump(mut self, entries: &[Message], encoding: Encoding) -> Result<String> {
self.add_entry(PoEntry {
comments: vec![
Comment::Translator(String::from("Generated by msg-tool")),
Comment::Flag(vec![String::from("fuzzy")]),
],
msgctxt: None,
msgid: String::new(),
msgid_plural: None,
msgstr: MsgStr::Single(Self::gen_start_str(encoding)),
});
let mut added_messages: HashMap<&String, usize> = HashMap::new();
for entry in entries {
let count = added_messages.get(&entry.message).map(|&s| s).unwrap_or(0);
self.add_entry(PoEntry {
comments: entry
.name
.as_ref()
.map(|name| vec![Comment::Translator(format!("NAME: {}", name))])
.unwrap_or_default(),
msgctxt: if count > 0 {
Some(format!("{}", count))
} else {
None
},
msgid: entry.message.clone(),
msgid_plural: None,
msgstr: MsgStr::Single(String::new()),
});
added_messages.insert(&entry.message, count + 1);
}
let mut result = String::new();
for line in &self.entries {
result.push_str(&line.dump()?);
result.push('\n');
}
Ok(result)
}
fn add_entry(&mut self, entry: PoEntry) {
for comment in entry.comments {
self.entries.push(PoLine::Comment(comment));
}
if let Some(ctx) = entry.msgctxt {
self.entries.push(PoLine::MsgCtxt(ctx));
}
self.entries.push(PoLine::MsgId(entry.msgid));
let is_plural = entry.msgid_plural.is_some();
if let Some(plural) = entry.msgid_plural {
self.entries.push(PoLine::MsgIdPlural(plural));
}
match entry.msgstr {
MsgStr::Single(s) => {
if is_plural {
self.entries.push(PoLine::MsgStrN(0, s));
} else {
self.entries.push(PoLine::MsgStr(s));
}
}
MsgStr::Plural(v) => {
for (n, s) in v {
self.entries.push(PoLine::MsgStrN(n, s));
}
}
}
self.entries.push(PoLine::EmptyLine);
}
}
pub struct PoParser<'a> {
texts: Vec<&'a str>,
pos: usize,
llm_mark: Option<&'a str>,
}
impl<'a> PoParser<'a> {
pub fn new(s: &'a str, llm_mark: Option<&'a str>) -> Self {
Self {
texts: s.graphemes(true).collect(),
pos: 0,
llm_mark,
}
}
pub fn parse_lines(&mut self) -> Result<Vec<PoLine>> {
let mut lines = Vec::new();
while let Some(f) = self.next_line() {
let f = f.trim();
if f.starts_with("#") {
if f.len() < 2 {
return Err(anyhow!("Invalid comment line: {}", f));
}
let c = &f[1..];
if c.starts_with(' ') {
lines.push(PoLine::Comment(Comment::Translator(c[1..].to_string())));
} else if c.starts_with('.') {
lines.push(PoLine::Comment(Comment::Extracted(
c[1..].trim_start().to_string(),
)));
} else if c.starts_with(':') {
lines.push(PoLine::Comment(Comment::Reference(
c[1..].trim_start().to_string(),
)));
} else if c.starts_with(',') {
let flags = c[1..].split(',').map(|s| s.trim().to_string()).collect();
lines.push(PoLine::Comment(Comment::Flag(flags)));
} else if c.starts_with('|') {
lines.push(PoLine::Comment(Comment::Previous(
c[1..].trim_start().to_string(),
)));
} else if c.starts_with('~') {
lines.push(PoLine::Comment(Comment::PreviousStr(
c[1..].trim_start().to_string(),
)));
} else {
return Err(anyhow!("Unknown comment type: {}", f));
}
} else if f.starts_with("msgid ") {
let content = self.read_string_literal(&f[6..])?;
lines.push(PoLine::MsgId(content));
} else if f.starts_with("msgstr ") {
let content = self.read_string_literal(&f[7..])?;
lines.push(PoLine::MsgStr(content));
} else if f.starts_with("msgctxt ") {
let content = self.read_string_literal(&f[8..])?;
lines.push(PoLine::MsgCtxt(content));
} else if f.starts_with("msgid_plural ") {
let content = self.read_string_literal(&f[13..])?;
lines.push(PoLine::MsgIdPlural(content));
} else if f.starts_with("msgstr[") {
let end_bracket = f
.find(']')
.ok_or_else(|| anyhow!("Invalid msgstr[n] line: {}", f))?;
let index_str = &f[7..end_bracket];
let index: usize = index_str
.parse()
.map_err(|_| anyhow!("Invalid index in msgstr[n]: {}", index_str))?;
let content = self.read_string_literal(&f[end_bracket + 1..])?;
lines.push(PoLine::MsgStrN(index, content));
} else if f.trim().is_empty() {
lines.push(PoLine::EmptyLine);
} else if f.starts_with('"') {
// This is a continuation of a previous string.
// According to GNU gettext manual, a string literal cannot appear on its own.
// It must follow a keyword. However, some tools generate this.
// We will append it to the last string-like element.
let content = self.read_string_literal(&f)?;
if let Some(last_line) = lines.last_mut() {
match last_line {
PoLine::MsgId(s) => s.push_str(&content),
PoLine::MsgStr(s) => s.push_str(&content),
PoLine::MsgCtxt(s) => s.push_str(&content),
PoLine::MsgIdPlural(s) => s.push_str(&content),
PoLine::MsgStrN(_, s) => s.push_str(&content),
_ => return Err(anyhow!("Orphan string literal continuation: {}", f)),
}
} else {
return Err(anyhow!("Orphan string literal continuation: {}", f));
}
} else {
return Err(anyhow!("Unknown line type: {}", f));
}
}
Ok(lines)
}
fn read_string_literal(&mut self, s: &str) -> Result<String> {
let mut content = String::new();
let current_line_str = s.trim();
if current_line_str.starts_with('"') && current_line_str.ends_with('"') {
content.push_str(&unescape_c_str(
&current_line_str[1..current_line_str.len() - 1],
)?);
} else {
return Err(anyhow!("Invalid string literal: {}", s));
}
while let Some(peeked_line) = self.peek_line() {
if peeked_line.trim().starts_with('"') {
self.next_line(); // consume it
let trimmed_line = peeked_line.trim();
if trimmed_line.starts_with('"') && trimmed_line.ends_with('"') {
content.push_str(&unescape_c_str(&trimmed_line[1..trimmed_line.len() - 1])?);
} else {
return Err(anyhow!(
"Invalid string literal continuation: {}",
peeked_line
));
}
} else {
break;
}
}
Ok(content)
}
pub fn parse_entries(&mut self) -> Result<Vec<PoEntry>> {
let lines = self.parse_lines()?;
let mut entries = Vec::new();
let mut current_entry_lines: Vec<PoLine> = Vec::new();
for line in lines {
if let PoLine::EmptyLine = line {
if !current_entry_lines.is_empty() {
entries.push(self.build_entry_from_lines(current_entry_lines)?);
current_entry_lines = Vec::new();
}
} else {
current_entry_lines.push(line);
}
}
if !current_entry_lines.is_empty() {
entries.push(self.build_entry_from_lines(current_entry_lines)?);
}
Ok(entries)
}
fn build_entry_from_lines(&self, lines: Vec<PoLine>) -> Result<PoEntry> {
let mut comments = Vec::new();
let mut msgctxt: Option<String> = None;
let mut msgid: Option<String> = None;
let mut msgid_plural: Option<String> = None;
let mut msgstr: Option<String> = None;
let mut msgstr_plural: Vec<(usize, String)> = Vec::new();
for line in lines {
match line {
PoLine::Comment(c) => comments.push(c),
PoLine::MsgCtxt(s) => {
if msgctxt.is_some() {
return Err(anyhow!("Duplicate msgctxt in entry"));
}
msgctxt = Some(s);
}
PoLine::MsgId(s) => {
if msgid.is_some() {
return Err(anyhow!("Duplicate msgid in entry"));
}
msgid = Some(s);
}
PoLine::MsgIdPlural(s) => {
if msgid_plural.is_some() {
return Err(anyhow!("Duplicate msgid_plural in entry"));
}
msgid_plural = Some(s);
}
PoLine::MsgStr(s) => {
if msgstr.is_some() {
return Err(anyhow!("Duplicate msgstr in entry"));
}
msgstr = Some(s);
}
PoLine::MsgStrN(n, s) => {
if msgstr_plural.iter().any(|(i, _)| *i == n) {
return Err(anyhow!("Duplicate msgstr[{}] in entry", n));
}
msgstr_plural.push((n, s));
}
PoLine::EmptyLine => {
// This should not be reached if called from parse_entries
return Err(anyhow!("Unexpected empty line in build_entry_from_lines"));
}
}
}
let final_msgstr = if !msgstr_plural.is_empty() {
if msgstr.is_some() {
return Err(anyhow!(
"Mixing msgstr and msgstr[n] in the same entry is not allowed"
));
}
MsgStr::Plural(msgstr_plural)
} else {
MsgStr::Single(msgstr.unwrap_or_default())
};
Ok(PoEntry {
comments,
msgctxt,
msgid: msgid.ok_or_else(|| anyhow!("Entry is missing msgid"))?,
msgid_plural,
msgstr: final_msgstr,
})
}
fn peek(&self) -> Option<&'a str> {
self.texts.get(self.pos).copied()
}
fn peek_line(&self) -> Option<String> {
let mut line = String::new();
let mut current_pos = self.pos;
while let Some(c) = self.texts.get(current_pos).copied() {
current_pos += 1;
if c == "\n" || c == "\r\n" {
break;
}
line.push_str(c);
}
if line.is_empty() && current_pos >= self.texts.len() {
None
} else {
Some(line)
}
}
fn next_line(&mut self) -> Option<String> {
let mut line = String::new();
while let Some(c) = self.next() {
if c == "\n" || c == "\r\n" {
break;
}
line.push_str(c);
}
if line.is_empty() && self.peek().is_none() {
None
} else {
Some(line)
}
}
fn next(&mut self) -> Option<&'a str> {
let r = self.texts.get(self.pos).copied();
if r.is_some() {
self.pos += 1;
}
r
}
pub fn parse(&mut self) -> Result<Vec<Message>> {
let mut messages = Vec::new();
let mut llm = None;
let mut name = None;
for entry in self.parse_entries()? {
if entry.msgid.is_empty() {
// This is the header entry, skip it
continue;
}
for comment in &entry.comments {
if let Comment::Translator(s) = comment {
let s = s.trim();
if s.starts_with("NAME:") {
name = Some(s[5..].trim().to_string());
} else if s.starts_with("LLM:") {
llm = Some(s[4..].trim().to_string());
}
}
}
let message = match entry.msgstr {
MsgStr::Single(s) => {
let s = s.trim();
if s.is_empty() {
llm.take()
.map(|mut llm| {
if let Some(mark) = self.llm_mark {
llm.push_str(mark);
}
llm
})
.unwrap_or_else(|| {
String::from(if entry.msgid.is_empty() { "" } else { "" })
})
} else {
let mut tmp = s.to_string();
if let Some(llm) = llm.take() {
if tmp == llm {
if let Some(mark) = self.llm_mark {
tmp.push_str(mark);
}
}
}
tmp
}
}
MsgStr::Plural(_) => {
return Err(anyhow!("Plural msgstr not supported in this context"));
}
};
let m = Message::new(message, name.take());
messages.push(m);
}
Ok(messages)
}
}
// --- Unit Tests ---
#[cfg(test)]
mod c_escape_tests {
use super::*;
#[test]
fn test_escape_basic() {
assert_eq!(escape_c_str("hello world").unwrap(), "hello world");
}
#[test]
fn test_escape_quotes_and_slashes() {
assert_eq!(
escape_c_str(r#"he"llo\world"#).unwrap(),
r#"he\"llo\\world"#
);
}
#[test]
fn test_escape_control_chars() {
assert_eq!(escape_c_str("a\nb\tc\rd").unwrap(), r#"a\nb\tc\rd"#);
assert_eq!(escape_c_str("\0").unwrap(), r#"\0"#);
}
#[test]
fn test_escape_other_control_chars_as_octal() {
assert_eq!(escape_c_str("\x07").unwrap(), r#"\a"#);
assert_eq!(escape_c_str("\x7f").unwrap(), r#"\177"#);
}
#[test]
fn test_unescape_basic() {
assert_eq!(unescape_c_str("hello world").unwrap(), "hello world");
}
#[test]
fn test_unescape_quotes_and_slashes() {
assert_eq!(
unescape_c_str(r#"he\"llo\\world"#).unwrap(),
r#"he"llo\world"#
);
}
#[test]
fn test_unescape_control_chars() {
assert_eq!(unescape_c_str(r#"a\nb\tc\rd"#).unwrap(), "a\nb\tc\rd");
}
#[test]
fn test_unescape_octal() {
assert_eq!(unescape_c_str(r#"\101"#).unwrap(), "A");
assert_eq!(unescape_c_str(r#"\60"#).unwrap(), "0");
assert_eq!(unescape_c_str(r#"\0"#).unwrap(), "\0");
assert_eq!(unescape_c_str(r#"\177"#).unwrap(), "\x7f");
assert_eq!(unescape_c_str(r#"hello\101world"#).unwrap(), "helloAworld");
}
#[test]
fn test_unescape_hex() {
assert_eq!(unescape_c_str(r#"\x41"#).unwrap(), "A");
assert_eq!(unescape_c_str(r#"\x30"#).unwrap(), "0");
assert_eq!(unescape_c_str(r#"\x7F"#).unwrap(), "\x7f");
assert_eq!(unescape_c_str(r#"\x7f"#).unwrap(), "\x7f");
assert_eq!(unescape_c_str(r#"hello\x41world"#).unwrap(), "helloAworld");
// A single hex digit is also valid
assert_eq!(unescape_c_str(r#"\xF"#).unwrap(), "\x0f");
}
#[test]
fn test_unescape_mixed() {
let original = "A\tB\"C\\D\0E";
let escaped = r#"A\tB\"C\\D\0E"#;
assert_eq!(unescape_c_str(escaped).unwrap(), original);
}
#[test]
fn test_unescape_invalid_sequence() {
assert!(unescape_c_str(r#"\q"#).is_err());
assert!(unescape_c_str(r#"hello\"#).is_err());
assert!(unescape_c_str(r#"\x"#).is_err());
// New: test \x followed immediately by an invalid character
assert!(unescape_c_str(r#"\xG"#).is_err());
// This should now pass
assert!(unescape_c_str(r#"\xFG"#).is_err());
assert!(unescape_c_str(r#"\8"#).is_err());
}
}