Collect data from glottolog as well as cldr.

This commit is contained in:
2020-10-02 13:07:13 +02:00
parent 0286beb89c
commit 08914685e1
9 changed files with 67669 additions and 45958 deletions

View File

@@ -1,159 +1,31 @@
use std::env;
use std::fs::File;
use std::io::BufReader;
use std::io::{self, Write};
use anyhow::Result;
use isolang::Language;
use quick_xml::{events::Event, Reader};
use thiserror::Error;
use argh::FromArgs;
#[derive(Error, Debug)]
enum Error {
#[error("xml error")]
Xml(#[from] quick_xml::Error),
mod cldr;
mod glottolog;
mod merge;
#[derive(FromArgs, PartialEq, Debug)]
/// txtlang-gen.
struct Opt {
#[argh(subcommand)]
cmd: Command,
}
#[derive(PartialEq, Debug)]
struct LocalDisplayName {
name: String,
lang: Language,
#[derive(FromArgs, PartialEq, Debug)]
#[argh(subcommand)]
enum Command {
Cldr(cldr::Cldr),
Glottolog(glottolog::Glottolog),
Merge(merge::Merge),
}
fn main() -> Result<()> {
let inputs = env::args().skip(1);
let opt: Opt = argh::from_env();
let mut results = Vec::new();
for input in inputs {
let mut reader = File::open(&input)
.map(BufReader::new)
.map(Reader::from_reader)?;
reader.trim_text(true);
let mut buf = Vec::new();
loop {
match reader.read_event(&mut buf) {
Ok(Event::Start(ref e)) if e.name() == b"localeDisplayNames" => {
buf.clear();
results.extend(parse::locale_display_names(&mut reader, &mut buf));
}
Err(e) => panic!("Error at position {}: {:?}", reader.buffer_position(), e),
Ok(Event::Eof) => break,
_ => (),
}
buf.clear();
}
}
let before = results.len();
results.sort_unstable_by(|a, b| a.name.cmp(&b.name));
results.dedup();
let dedup = results.len();
let mut remove = Vec::new();
results.dedup_by(|a, b| {
if a.name == b.name {
remove.push(a.name.clone());
}
a.name == b.name
});
results.retain(|x| !remove.contains(&x.name));
let clean = results.len();
eprintln!("before={}", before);
eprintln!("dedup={}", dedup);
eprintln!("clean={}", clean);
let stdout = io::stdout();
let mut lock = stdout.lock();
for result in results {
assert!(!result.name.contains('\n'));
assert!(!result.name.contains('\t'));
writeln!(lock, "{}\t{}", result.name, result.lang.to_639_3())
.expect("failed to write to stdout");
}
lock.flush().expect("failed to flush to stdout");
Ok(())
}
mod parse {
use std::fs::File;
use std::io::BufReader;
use anyhow::Result;
use isolang::Language;
use quick_xml::{events::Event, Reader};
use crate::LocalDisplayName;
pub(crate) fn locale_display_names(
reader: &mut Reader<BufReader<File>>,
buf: &mut Vec<u8>,
) -> Vec<LocalDisplayName> {
let mut result = Vec::new();
let mut parse = false;
let mut language = None;
loop {
match reader.read_event(buf) {
Ok(Event::End(ref e)) if e.name() == b"localeDisplayNames" => {
break;
}
Ok(Event::Start(ref e)) if e.name() == b"languages" => {
parse = true;
}
Ok(Event::End(ref e)) if e.name() == b"languages" => {
parse = false;
}
Ok(Event::Start(ref e)) if e.name() == b"language" && parse => {
let a = e
.attributes()
.filter_map(Result::ok)
.find(|a| a.key == b"type");
if let Some(a) = a {
if let Ok(lang) = reader.decode_without_bom(&a.value) {
language = Language::from_639_1(lang)
.or_else(|| Language::from_639_3(lang))
.or_else(|| Language::from_locale(lang));
}
}
}
Ok(Event::End(ref e)) if e.name() == b"language" && parse => {
language = None;
}
Ok(Event::Text(e)) if language.is_some() => {
if let Ok(text) = reader.decode_without_bom(&e) {
result.push(LocalDisplayName {
name: text.to_lowercase(),
lang: language.take().unwrap(),
});
}
}
Err(e) => panic!("Error at position {}: {:?}", reader.buffer_position(), e),
Ok(Event::Eof) => break,
_ => (),
}
buf.clear();
}
result
match opt.cmd {
Command::Cldr(cmd) => cmd.run(),
Command::Glottolog(cmd) => cmd.run(),
Command::Merge(cmd) => cmd.run(),
}
}