mirror of
https://github.com/storytold/wordcraft.git
synced 2026-11-09 20:00:11 -06:00
Martin BallandClaude Opus 5.5
03c50f8af2
- craft-proof now names ja, zh and th in missing_dictionary: packs::missing_dictionary hides it for a pack language when no pack for it is listed (status bar and Options). - de-CH and pt-PT are pack languages in craft-proof (de-ch, pt-pt): names, hyphenation and grammar stay German and Portuguese (tests); pack names fall back to the region-aware name. - About › Language data shows craft_proof::scowl::notice() when SCOWL is embedded (test). - Flatpak manifest's craft-i18n commit follows craft-i18n.rev. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FypgCX8LsYYE8fsj4byBm2
731 lines
28 KiB
Rust
731 lines
28 KiB
Rust
//! Language packs (CP4.3): spelling dictionaries and hyphenation patterns for languages beyond
|
||
//! the embedded English, one pack per locale, downloaded and verified by the font store
|
||
//! (craft-libs `craft-font-store`, item kind `language-pack`) and read by `craft-proof` from the
|
||
//! store's `language-packs/` folder.
|
||
//!
|
||
//! The commands (`proof.packs`, `proof.installPack`, `proof.removePack`), Options › Language,
|
||
//! the missing-pack bar and About › Language data all go through this module. It never asks
|
||
//! anything: an install needs the font store's single consent setting (downloading on), which only
|
||
//! a person in the app can turn on ([`Backend::enable_network`], called from an Install click).
|
||
//! Agents (MCP, the CLI, the control channel) install only while it is on.
|
||
//!
|
||
//! The [`Backend`] is set once at start-up by the desktop app and the CLI ([`use_font_store`]).
|
||
//! Without one (tests, the web build, `ui_shot`) nothing is available and nothing is downloaded.
|
||
|
||
use std::collections::BTreeSet;
|
||
use std::path::PathBuf;
|
||
use std::sync::atomic::{AtomicBool, Ordering};
|
||
use std::sync::{Arc, Mutex, PoisonError, RwLock};
|
||
|
||
use serde::Serialize;
|
||
|
||
use crate::{CmdError, Session};
|
||
|
||
#[cfg(not(target_arch = "wasm32"))]
|
||
mod store;
|
||
|
||
/// Longest locale tag taken, in bytes (BCP 47 tags in practice stay under 20).
|
||
pub const MAX_LOCALE_LEN: usize = 35;
|
||
/// Most packs an index may list that are read.
|
||
const MAX_PACKS: usize = 1000;
|
||
/// Paragraphs of a document looked at for its languages (the missing-pack bar).
|
||
const MAX_PARAGRAPHS_SCANNED: usize = 20_000;
|
||
/// Distinct languages of a document collected.
|
||
const MAX_DOC_LANGUAGES: usize = 64;
|
||
|
||
/// A pack the pinned index lists.
|
||
#[derive(Clone, Debug, PartialEq, Eq, Serialize)]
|
||
pub struct Pack {
|
||
/// Lowercase BCP 47 (`nl`, `pt-pt`).
|
||
pub locale: String,
|
||
/// The language's own name (`Nederlands`).
|
||
pub name: String,
|
||
/// The craft-i18n commit the pack was built from.
|
||
pub version: String,
|
||
/// The archive.
|
||
pub url: String,
|
||
/// The upstream folder the data comes from, at its pinned commit (the index's optional
|
||
/// `source`), shown before install; the archive when the index gives none.
|
||
pub source: Option<String>,
|
||
/// Bytes of the archive.
|
||
pub size: u64,
|
||
/// SPDX expression.
|
||
pub licence: String,
|
||
/// `Embeddable` or `Download-only`.
|
||
pub class: String,
|
||
/// Has a spelling dictionary.
|
||
pub spelling: bool,
|
||
/// Has hyphenation patterns.
|
||
pub hyphenation: bool,
|
||
}
|
||
|
||
/// An installed pack.
|
||
#[derive(Clone, Debug, PartialEq, Eq, Serialize)]
|
||
pub struct Installed {
|
||
pub locale: String,
|
||
pub name: String,
|
||
pub licence: String,
|
||
pub path: PathBuf,
|
||
}
|
||
|
||
/// Stops a download. On the desktop it is the font store's own flag, which a download watches
|
||
/// even while the server sends nothing.
|
||
#[derive(Clone, Debug, Default)]
|
||
pub struct Cancel(#[cfg(not(target_arch = "wasm32"))] pub(crate) craft_font_store::Cancel, #[cfg(target_arch = "wasm32")] Arc<AtomicBool>);
|
||
|
||
impl Cancel {
|
||
pub fn new() -> Self {
|
||
Self::default()
|
||
}
|
||
#[cfg(not(target_arch = "wasm32"))]
|
||
pub fn cancel(&self) {
|
||
self.0.cancel();
|
||
}
|
||
#[cfg(not(target_arch = "wasm32"))]
|
||
pub fn is_cancelled(&self) -> bool {
|
||
self.0.is_cancelled()
|
||
}
|
||
#[cfg(target_arch = "wasm32")]
|
||
pub fn cancel(&self) {
|
||
self.0.store(true, Ordering::Relaxed);
|
||
}
|
||
#[cfg(target_arch = "wasm32")]
|
||
pub fn is_cancelled(&self) -> bool {
|
||
self.0.load(Ordering::Relaxed)
|
||
}
|
||
}
|
||
|
||
/// Where packs come from: the font store on the desktop, a fixed list in tests and `ui_shot`.
|
||
pub trait Backend: Send + Sync {
|
||
/// The pinned index as last fetched (no request); empty when it never was.
|
||
fn index(&self) -> Result<Vec<Pack>, String>;
|
||
/// Fetch the pinned index (a request; only while downloading is on).
|
||
fn fetch_index(&self, cancel: &Cancel) -> Result<Vec<Pack>, String>;
|
||
/// Whether the font store may download.
|
||
fn network_on(&self) -> bool;
|
||
/// Turn downloading on. Only from the person's own action in the app (an Install click):
|
||
/// the commands never call it, and a backend for an agent refuses.
|
||
fn enable_network(&self) -> Result<(), String>;
|
||
/// Download, verify and unpack `pack` into the packs folder. `progress` gets (done, total) bytes.
|
||
fn install(&self, pack: &Pack, cancel: &Cancel, progress: &mut dyn FnMut(u64, u64)) -> Result<(), String>;
|
||
/// Remove the installed pack for `locale`.
|
||
fn remove(&self, locale: &str) -> Result<(), String>;
|
||
/// The installed-packs folder (`<font store>/language-packs`).
|
||
fn packs_dir(&self) -> Option<PathBuf>;
|
||
/// The packs installed now.
|
||
fn installed(&self) -> Vec<Installed> {
|
||
installed_from_proof()
|
||
}
|
||
/// The test index in use instead of the pinned one (`CRAFT_LANGUAGE_PACK_INDEX`), if any.
|
||
fn index_override(&self) -> Option<String> {
|
||
None
|
||
}
|
||
}
|
||
|
||
static BACKEND: RwLock<Option<Arc<dyn Backend>>> = RwLock::new(None);
|
||
/// The index last read, so the bar and the list don't read the store's cache every frame.
|
||
static INDEX: Mutex<Option<Vec<Pack>>> = Mutex::new(None);
|
||
/// Bumped whenever [`INDEX`] changes.
|
||
static INDEX_GENERATION: std::sync::atomic::AtomicU64 = std::sync::atomic::AtomicU64::new(0);
|
||
|
||
fn set_index(index: Option<Vec<Pack>>) {
|
||
*INDEX.lock().unwrap_or_else(PoisonError::into_inner) = index;
|
||
INDEX_GENERATION.fetch_add(1, Ordering::Relaxed);
|
||
}
|
||
|
||
/// Changes whenever the list of available packs does (a fetch, another backend).
|
||
pub fn index_generation() -> u64 {
|
||
INDEX_GENERATION.load(Ordering::Relaxed)
|
||
}
|
||
|
||
/// Use `backend` for packs from now on (`None`: no packs), and point proofing at its folder.
|
||
pub fn set_backend(backend: Option<Arc<dyn Backend>>) {
|
||
let dir = backend.as_ref().and_then(|b| b.packs_dir());
|
||
*BACKEND.write().unwrap_or_else(PoisonError::into_inner) = backend;
|
||
set_index(None);
|
||
jobs().errors.clear();
|
||
wordcraft_proof::packs::set_packs_dir(dir);
|
||
}
|
||
|
||
/// Use the suite's font store for packs (`<font store>/language-packs`), as the person in the app
|
||
/// (`user`: an Install click may turn downloading on) or as an agent (the CLI, headless MCP: it
|
||
/// never can). The pack index is the pinned one, or `CRAFT_LANGUAGE_PACK_INDEX` (a URL or a path,
|
||
/// for testing packs before they are published), which the font store reads itself. Returns
|
||
/// whether there is a store folder.
|
||
#[cfg(not(target_arch = "wasm32"))]
|
||
pub fn use_font_store(user: bool) -> bool {
|
||
match store::FontStoreBackend::new(user) {
|
||
Some(b) => {
|
||
set_backend(Some(Arc::new(b)));
|
||
true
|
||
}
|
||
None => false,
|
||
}
|
||
}
|
||
|
||
/// [`use_font_store`] with the store in `dir` instead of the person's (`ui_shot`, tests).
|
||
#[cfg(not(target_arch = "wasm32"))]
|
||
pub fn use_font_store_in(dir: PathBuf, user: bool) {
|
||
set_backend(Some(Arc::new(store::FontStoreBackend::in_dir(dir, user))));
|
||
}
|
||
|
||
/// The backend in use.
|
||
pub fn backend() -> Option<Arc<dyn Backend>> {
|
||
BACKEND.read().unwrap_or_else(PoisonError::into_inner).clone()
|
||
}
|
||
|
||
fn installed_from_proof() -> Vec<Installed> {
|
||
wordcraft_proof::packs::installed().into_iter().map(|p| Installed { locale: p.locale, name: p.name, licence: p.licence, path: p.path }).collect()
|
||
}
|
||
|
||
/// The test source packs come from instead of the pinned index (`CRAFT_LANGUAGE_PACK_INDEX`):
|
||
/// Options › Language says so.
|
||
pub fn index_override() -> Option<String> {
|
||
backend().and_then(|b| b.index_override())
|
||
}
|
||
|
||
/// The packs installed now (none without a backend).
|
||
pub fn installed() -> Vec<Installed> {
|
||
backend().map(|b| b.installed()).unwrap_or_default()
|
||
}
|
||
|
||
/// The available packs: the index as last read, read from the store's cache the first time.
|
||
pub fn available() -> Vec<Pack> {
|
||
let Some(b) = backend() else { return Vec::new() };
|
||
let mut cached = INDEX.lock().unwrap_or_else(PoisonError::into_inner);
|
||
if let Some(i) = cached.as_ref() {
|
||
return i.clone();
|
||
}
|
||
let index = clean_index(b.index().unwrap_or_default());
|
||
*cached = Some(index.clone());
|
||
drop(cached);
|
||
INDEX_GENERATION.fetch_add(1, Ordering::Relaxed);
|
||
index
|
||
}
|
||
|
||
/// How many packs [`available`] lists, without copying them (the window asks every frame).
|
||
pub fn available_len() -> usize {
|
||
if let Some(i) = INDEX.lock().unwrap_or_else(PoisonError::into_inner).as_ref() {
|
||
return i.len();
|
||
}
|
||
available().len()
|
||
}
|
||
|
||
/// Fetch the index again (downloading must be on).
|
||
pub fn refresh(cancel: &Cancel) -> Result<Vec<Pack>, String> {
|
||
let b = backend().ok_or_else(no_store)?;
|
||
if !b.network_on() {
|
||
return Err(network_off());
|
||
}
|
||
let index = clean_index(b.fetch_index(cancel)?);
|
||
set_index(Some(index.clone()));
|
||
Ok(index)
|
||
}
|
||
|
||
/// Drop packs with a malformed locale, duplicates and English (embedded), and cap the list.
|
||
fn clean_index(mut packs: Vec<Pack>) -> Vec<Pack> {
|
||
packs.truncate(MAX_PACKS);
|
||
let mut seen = BTreeSet::new();
|
||
packs.retain(|p| valid_locale(&p.locale) && !is_embedded(&p.locale) && seen.insert(p.locale.clone()));
|
||
packs.sort_by(|a, b| a.locale.cmp(&b.locale));
|
||
packs
|
||
}
|
||
|
||
/// English is embedded in every build; it never comes as a pack.
|
||
fn is_embedded(locale: &str) -> bool {
|
||
locale == "en" || locale.starts_with("en-")
|
||
}
|
||
|
||
/// A lowercase BCP 47-shaped tag: letters, digits and `-`, starting with a letter.
|
||
pub fn valid_locale(locale: &str) -> bool {
|
||
!locale.is_empty()
|
||
&& locale.len() <= MAX_LOCALE_LEN
|
||
&& locale.starts_with(|c: char| c.is_ascii_lowercase())
|
||
&& locale.chars().all(|c| c.is_ascii_lowercase() || c.is_ascii_digit() || c == '-')
|
||
&& !locale.contains("--")
|
||
&& !locale.ends_with('-')
|
||
}
|
||
|
||
/// A locale from a command's params: trimmed, lowercased, `_` as `-`.
|
||
pub fn parse_locale(raw: &str) -> Result<String, CmdError> {
|
||
let l = raw.trim().replace('_', "-").to_ascii_lowercase();
|
||
if valid_locale(&l) { Ok(l) } else { Err(CmdError::Params(format!("`locale` must be a language tag like nl or pt-PT, not {raw:?}"))) }
|
||
}
|
||
|
||
fn no_store() -> String {
|
||
if cfg!(target_arch = "wasm32") {
|
||
"Language packs aren't available in the browser".into()
|
||
} else {
|
||
"there is no font store folder for language packs".into()
|
||
}
|
||
}
|
||
|
||
fn network_off() -> String {
|
||
"downloading is turned off: turn it on in WordCraft (File › Options › Language, or an Install button there)".into()
|
||
}
|
||
|
||
/// The pack for `locale` in the index.
|
||
pub fn find(locale: &str) -> Option<Pack> {
|
||
available().into_iter().find(|p| p.locale == locale)
|
||
}
|
||
|
||
/// Whether a pack is installed for `locale`.
|
||
pub fn is_installed(locale: &str) -> bool {
|
||
installed().iter().any(|p| p.locale == locale)
|
||
}
|
||
|
||
/// Install the pack for `locale` (downloading must be on): the commands and the Install button.
|
||
/// Proofing picks it up at once.
|
||
pub fn install(locale: &str, cancel: &Cancel, progress: &mut dyn FnMut(u64, u64)) -> Result<Installed, String> {
|
||
let b = backend().ok_or_else(no_store)?;
|
||
let pack = find(locale).ok_or_else(|| format!("there is no language pack for {locale}"))?;
|
||
if !b.network_on() {
|
||
return Err(network_off());
|
||
}
|
||
if let Some(dir) = b.packs_dir() {
|
||
crate::file_access::check_write(&dir.to_string_lossy())?;
|
||
}
|
||
b.install(&pack, cancel, progress)?;
|
||
reload(&b);
|
||
b.installed().into_iter().find(|p| p.locale == locale).ok_or_else(|| format!("the {locale} pack didn't install"))
|
||
}
|
||
|
||
/// Remove the installed pack for `locale`.
|
||
pub fn remove(locale: &str) -> Result<(), String> {
|
||
let b = backend().ok_or_else(no_store)?;
|
||
if jobs().installs.iter().any(|(l, ..)| l == locale) {
|
||
return Err(format!("the {locale} pack is being installed: cancel it first"));
|
||
}
|
||
if !b.installed().iter().any(|p| p.locale == locale) {
|
||
return Err(format!("no language pack is installed for {locale}"));
|
||
}
|
||
if let Some(dir) = b.packs_dir() {
|
||
crate::file_access::check_write(&dir.to_string_lossy())?;
|
||
}
|
||
b.remove(locale)?;
|
||
reload(&b);
|
||
Ok(())
|
||
}
|
||
|
||
/// What the app does with a long pack request (an install or a fetch of the list): `None` runs it
|
||
/// in the caller (the CLI, headless MCP, tests); `Some(notify)` runs it on a thread, so the window
|
||
/// keeps drawing whoever asked (the person, the control channel, MCP), and calls `notify` on each
|
||
/// change (the window repaints). Set once by the desktop app.
|
||
pub type Notify = Arc<dyn Fn() + Send + Sync>;
|
||
|
||
static BACKGROUND: RwLock<Option<Notify>> = RwLock::new(None);
|
||
|
||
/// Installs and fetches running on a thread, and how the finished ones ended.
|
||
#[derive(Default)]
|
||
struct Jobs {
|
||
/// Running installs by locale: (bytes done, bytes in all, its Cancel).
|
||
installs: Vec<(String, u64, u64, Cancel)>,
|
||
fetching: bool,
|
||
/// The last failure of each locale's install ("" for the list), until it is tried again.
|
||
errors: Vec<(String, String)>,
|
||
/// Bumped as each job ends: the window relays out and shows the result.
|
||
finished: u64,
|
||
}
|
||
|
||
static JOBS: Mutex<Jobs> = Mutex::new(Jobs { installs: Vec::new(), fetching: false, errors: Vec::new(), finished: 0 });
|
||
|
||
fn jobs() -> std::sync::MutexGuard<'static, Jobs> {
|
||
JOBS.lock().unwrap_or_else(PoisonError::into_inner)
|
||
}
|
||
|
||
/// Run long pack requests on a thread (`Some`: the desktop app), or in the caller (`None`).
|
||
pub fn set_background(notify: Option<Notify>) {
|
||
*BACKGROUND.write().unwrap_or_else(PoisonError::into_inner) = notify;
|
||
}
|
||
|
||
fn background() -> Option<Notify> {
|
||
if cfg!(target_arch = "wasm32") {
|
||
return None;
|
||
}
|
||
BACKGROUND.read().unwrap_or_else(PoisonError::into_inner).clone()
|
||
}
|
||
|
||
/// Whether installs and fetches run on a thread ([`set_background`]).
|
||
pub fn runs_in_background() -> bool {
|
||
background().is_some()
|
||
}
|
||
|
||
/// A running install: (locale, bytes done, bytes in all).
|
||
#[derive(Clone, Debug, PartialEq, Eq, Serialize)]
|
||
pub struct Progress {
|
||
pub locale: String,
|
||
pub done: u64,
|
||
pub total: u64,
|
||
}
|
||
|
||
/// The installs running now.
|
||
pub fn installing() -> Vec<Progress> {
|
||
jobs().installs.iter().map(|(l, d, t, _)| Progress { locale: l.clone(), done: *d, total: *t }).collect()
|
||
}
|
||
|
||
/// Whether the list is being fetched.
|
||
pub fn fetching() -> bool {
|
||
jobs().fetching
|
||
}
|
||
|
||
/// The last failure of `locale`'s install ("" for the list), until it is tried again.
|
||
pub fn error(locale: &str) -> Option<String> {
|
||
jobs().errors.iter().find(|(l, _)| l == locale).map(|(_, e)| e.clone())
|
||
}
|
||
|
||
/// Every recorded failure, (locale or "" for the list, message).
|
||
pub fn errors() -> Vec<(String, String)> {
|
||
jobs().errors.clone()
|
||
}
|
||
|
||
/// Counts finished jobs: when it changes, the window lays the documents out again.
|
||
pub fn finished() -> u64 {
|
||
jobs().finished
|
||
}
|
||
|
||
/// Stop `locale`'s running install.
|
||
pub fn cancel(locale: &str) {
|
||
if let Some((_, _, _, c)) = jobs().installs.iter().find(|(l, ..)| l == locale) {
|
||
c.cancel();
|
||
}
|
||
}
|
||
|
||
fn set_error(locale: &str, e: Option<String>) {
|
||
let mut j = jobs();
|
||
j.errors.retain(|(l, _)| l != locale);
|
||
if let Some(e) = e {
|
||
j.errors.push((locale.to_string(), e));
|
||
}
|
||
}
|
||
|
||
/// Start installing `locale` ([`install`]): on a thread when [`runs_in_background`], else now.
|
||
/// What's wrong up front (no pack, downloading off, outside the write root) is an error at once.
|
||
/// Returns whether it is still running (`false`: done, or it already was running).
|
||
pub fn start_install(locale: &str) -> Result<bool, String> {
|
||
let b = backend().ok_or_else(no_store)?;
|
||
let pack = find(locale).ok_or_else(|| format!("there is no language pack for {locale}"))?;
|
||
if !b.network_on() {
|
||
return Err(network_off());
|
||
}
|
||
if let Some(dir) = b.packs_dir() {
|
||
crate::file_access::check_write(&dir.to_string_lossy())?;
|
||
}
|
||
let Some(notify) = background() else {
|
||
let r = install(locale, &Cancel::new(), &mut |_, _| {});
|
||
set_error(locale, r.as_ref().err().cloned());
|
||
return r.map(|_| false);
|
||
};
|
||
let cancel = Cancel::new();
|
||
{
|
||
let mut j = jobs();
|
||
if j.installs.iter().any(|(l, ..)| l == locale) {
|
||
return Ok(true);
|
||
}
|
||
j.installs.push((locale.to_string(), 0, pack.size, cancel.clone()));
|
||
j.errors.retain(|(l, _)| l != locale);
|
||
}
|
||
let key = locale.to_string();
|
||
let locale = locale.to_string();
|
||
let started = spawn(move || {
|
||
let progress_notify = notify.clone();
|
||
let r = install(&locale, &cancel, &mut |done, total| {
|
||
if let Some(job) = jobs().installs.iter_mut().find(|(l, ..)| *l == locale) {
|
||
job.1 = done;
|
||
job.2 = total;
|
||
}
|
||
progress_notify();
|
||
});
|
||
let mut j = jobs();
|
||
j.installs.retain(|(l, ..)| *l != locale);
|
||
if let Err(e) = r {
|
||
j.errors.push((locale, e));
|
||
}
|
||
j.finished = j.finished.wrapping_add(1);
|
||
drop(j);
|
||
notify();
|
||
});
|
||
if let Err(e) = started {
|
||
jobs().installs.retain(|(l, ..)| *l != key);
|
||
return Err(e);
|
||
}
|
||
Ok(true)
|
||
}
|
||
|
||
/// Start fetching the list again ([`refresh`]): on a thread when [`runs_in_background`], else now.
|
||
/// Returns whether it is still running.
|
||
pub fn start_refresh() -> Result<bool, String> {
|
||
let b = backend().ok_or_else(no_store)?;
|
||
if !b.network_on() {
|
||
return Err(network_off());
|
||
}
|
||
let Some(notify) = background() else {
|
||
let r = refresh(&Cancel::new());
|
||
set_error("", r.as_ref().err().cloned());
|
||
return r.map(|_| false);
|
||
};
|
||
{
|
||
let mut j = jobs();
|
||
if j.fetching {
|
||
return Ok(true);
|
||
}
|
||
j.fetching = true;
|
||
j.errors.retain(|(l, _)| !l.is_empty());
|
||
}
|
||
let started = spawn(move || {
|
||
let r = refresh(&Cancel::new());
|
||
let mut j = jobs();
|
||
j.fetching = false;
|
||
if let Err(e) = r {
|
||
j.errors.push((String::new(), e));
|
||
}
|
||
j.finished = j.finished.wrapping_add(1);
|
||
drop(j);
|
||
notify();
|
||
});
|
||
if let Err(e) = started {
|
||
jobs().fetching = false;
|
||
return Err(e);
|
||
}
|
||
Ok(true)
|
||
}
|
||
|
||
#[cfg(not(target_arch = "wasm32"))]
|
||
fn spawn(f: impl FnOnce() + Send + 'static) -> Result<(), String> {
|
||
std::thread::Builder::new().name("language-pack".into()).spawn(f).map(|_| ()).map_err(|e| format!("couldn't start the download: {e}"))
|
||
}
|
||
|
||
#[cfg(target_arch = "wasm32")]
|
||
fn spawn(_: impl FnOnce() + Send + 'static) -> Result<(), String> {
|
||
Err(no_store())
|
||
}
|
||
|
||
/// Have proofing look at the packs folder again (its dictionaries are reloaded).
|
||
fn reload(b: &Arc<dyn Backend>) {
|
||
if wordcraft_proof::packs::packs_dir() == b.packs_dir() {
|
||
wordcraft_proof::packs::reload();
|
||
} else {
|
||
wordcraft_proof::packs::set_packs_dir(b.packs_dir());
|
||
}
|
||
}
|
||
|
||
/// The full licence texts for About › Language data: the embedded English data, then each
|
||
/// installed pack.
|
||
pub fn licence_texts() -> Vec<(String, String)> {
|
||
let mut texts = wordcraft_proof::packs::licence_texts();
|
||
// SCOWL's notices (with the Ispell BSD licence of its affix files), when English spells from
|
||
// the embedded SCOWL data; after the Moby and supplement entries.
|
||
if let Some(notice) = wordcraft_proof::scowl::notice() {
|
||
let at = texts.len().min(2);
|
||
texts.insert(at, (SCOWL_TITLE.into(), notice.into()));
|
||
}
|
||
texts
|
||
}
|
||
|
||
/// About › Language data's title for SCOWL's notices.
|
||
pub const SCOWL_TITLE: &str = "English spelling dictionaries (SCOWL)";
|
||
|
||
/// The pack locale for a document language tag: the whole tag (`pt-pt`), else its language
|
||
/// (`nl` for `nl-NL`), whichever the index has. `None` for English and tags with no pack.
|
||
pub fn pack_for_tag<'a>(tag: &str, packs: &'a [Pack]) -> Option<&'a Pack> {
|
||
let tag = tag.trim().replace('_', "-").to_ascii_lowercase();
|
||
let base = tag.split('-').next().unwrap_or("");
|
||
if base.is_empty() || base == "en" {
|
||
return None;
|
||
}
|
||
packs.iter().find(|p| p.locale == tag).or_else(|| packs.iter().find(|p| p.locale == base))
|
||
}
|
||
|
||
/// The languages (`w:lang`, resolved through the styles) of the body text, in first-seen order.
|
||
pub fn document_languages(s: &Session) -> Vec<String> {
|
||
let mut out: Vec<String> = Vec::new();
|
||
for path in s.doc.para_paths(wordcraft_doc::StoryRef::Body).into_iter().take(MAX_PARAGRAPHS_SCANNED) {
|
||
let Some(p) = s.doc.para(wordcraft_doc::StoryRef::Body, &path) else { continue };
|
||
if p.text.trim().is_empty() {
|
||
continue;
|
||
}
|
||
let style = p.props.style.as_deref();
|
||
for (range, c) in p.run_ranges() {
|
||
let Some(text) = p.text.get(range) else { continue };
|
||
if !text.chars().any(char::is_alphabetic) {
|
||
continue;
|
||
}
|
||
let rc = s.doc.styles.resolve_char_in(style, None, c);
|
||
if rc.no_proof {
|
||
continue;
|
||
}
|
||
let tag = rc.lang.unwrap_or_else(|| "en-US".into());
|
||
if !out.contains(&tag) {
|
||
out.push(tag);
|
||
if out.len() >= MAX_DOC_LANGUAGES {
|
||
return out;
|
||
}
|
||
}
|
||
}
|
||
}
|
||
out
|
||
}
|
||
|
||
/// Whether proofing reads a pack for text in `tag`: every language craft-proof can name, its own
|
||
/// (Polish…) and any other (`ProofLang::Pack`, read from the pack alone); not `und`, `zxx` or a
|
||
/// malformed tag. The one place the bar's language rule lives.
|
||
pub fn proofable(tag: &str) -> bool {
|
||
wordcraft_proof::ProofLang::from_tag(Some(tag)) != wordcraft_proof::ProofLang::Other
|
||
}
|
||
|
||
/// The Hunspell dictionary `tag`'s language needs and doesn't have, for the status bar's
|
||
/// "(no dictionary)": craft-proof's [`wordcraft_proof::missing_dictionary`], except for a
|
||
/// language craft-proof has no rules of its own for (`ProofLang::Pack`: Japanese, Chinese,
|
||
/// Thai…) when no pack for it is available either, as no Hunspell dictionary may exist at all.
|
||
pub fn missing_dictionary(tag: Option<&str>) -> Option<&'static str> {
|
||
let name = wordcraft_proof::missing_dictionary(tag)?;
|
||
let lang = wordcraft_proof::ProofLang::from_tag(tag);
|
||
if lang.pack_locale().is_some() && pack_for_tag(tag.unwrap_or(""), &available()).is_none() {
|
||
return None;
|
||
}
|
||
Some(name)
|
||
}
|
||
|
||
/// The packs the document's text asks for that are available and not installed: what the
|
||
/// missing-pack bar offers. Never a language without a pack, nor one proofing can't use.
|
||
pub fn missing(s: &Session) -> Vec<Pack> {
|
||
let packs = available();
|
||
if packs.is_empty() {
|
||
return Vec::new();
|
||
}
|
||
let installed = installed();
|
||
let mut out: Vec<Pack> = Vec::new();
|
||
for tag in document_languages(s) {
|
||
if proofable(&tag)
|
||
&& let Some(p) = pack_for_tag(&tag, &packs)
|
||
&& !installed.iter().any(|i| i.locale == p.locale)
|
||
&& !out.iter().any(|o| o.locale == p.locale)
|
||
{
|
||
out.push(p.clone());
|
||
}
|
||
}
|
||
out
|
||
}
|
||
|
||
/// A backend with a fixed list and installs that only record themselves: tests and `ui_shot`.
|
||
#[derive(Default)]
|
||
pub struct FixedBackend {
|
||
pub packs: Vec<Pack>,
|
||
pub network: AtomicBool,
|
||
/// Agents can't turn the network on.
|
||
pub agent: bool,
|
||
pub installed: Mutex<Vec<Installed>>,
|
||
}
|
||
|
||
impl Pack {
|
||
/// Where the data comes from: the upstream source when the index names it, else the archive.
|
||
pub fn source_link(&self) -> &str {
|
||
self.source.as_deref().filter(|s| !s.is_empty()).unwrap_or(&self.url)
|
||
}
|
||
}
|
||
|
||
impl FixedBackend {
|
||
/// A few packs shaped like the published ones, for screenshots and tests.
|
||
pub fn sample() -> Self {
|
||
let pack = |locale: &str, name: &str, size: u64, licence: &str, class: &str| Pack {
|
||
locale: locale.into(),
|
||
name: name.into(),
|
||
version: "0000000".into(),
|
||
url: format!("https://github.com/storytold/craft-i18n/releases/download/language-packs-1/{locale}.tar.zst"),
|
||
source: Some("https://github.com/LibreOffice/dictionaries".into()),
|
||
size,
|
||
licence: licence.into(),
|
||
class: class.into(),
|
||
spelling: true,
|
||
hyphenation: true,
|
||
};
|
||
FixedBackend {
|
||
packs: vec![
|
||
pack("ca", "Català", 1_480_000, "GPL-2.0-or-later OR LGPL-2.1-or-later", "Download-only"),
|
||
pack("el", "Ελληνικά", 2_310_000, "MPL-1.1 OR GPL-2.0-or-later OR LGPL-2.1-or-later", "Embeddable"),
|
||
pack("fr", "Français", 1_120_000, "MPL-2.0", "Embeddable"),
|
||
pack("nl", "Nederlands", 1_950_000, "BSD-3-Clause", "Embeddable"),
|
||
pack("pl", "Polski", 4_620_000, "GPL-2.0-or-later OR LGPL-2.1-or-later OR MPL-1.1 OR Apache-2.0 OR CC-BY-4.0", "Embeddable"),
|
||
pack("sk", "Slovenčina", 2_040_000, "GPL-2.0-only OR LGPL-2.1-only OR MPL-1.1", "Download-only"),
|
||
],
|
||
..Default::default()
|
||
}
|
||
}
|
||
}
|
||
|
||
impl Backend for FixedBackend {
|
||
fn index(&self) -> Result<Vec<Pack>, String> {
|
||
Ok(self.packs.clone())
|
||
}
|
||
fn fetch_index(&self, _: &Cancel) -> Result<Vec<Pack>, String> {
|
||
Ok(self.packs.clone())
|
||
}
|
||
fn network_on(&self) -> bool {
|
||
self.network.load(Ordering::Relaxed)
|
||
}
|
||
fn enable_network(&self) -> Result<(), String> {
|
||
if self.agent {
|
||
return Err("only the user can turn on downloading".into());
|
||
}
|
||
self.network.store(true, Ordering::Relaxed);
|
||
Ok(())
|
||
}
|
||
fn install(&self, pack: &Pack, cancel: &Cancel, progress: &mut dyn FnMut(u64, u64)) -> Result<(), String> {
|
||
if cancel.is_cancelled() {
|
||
return Err("cancelled".into());
|
||
}
|
||
progress(pack.size, pack.size);
|
||
let mut i = self.installed.lock().unwrap_or_else(PoisonError::into_inner);
|
||
i.retain(|p| p.locale != pack.locale);
|
||
i.push(Installed { locale: pack.locale.clone(), name: pack.name.clone(), licence: pack.licence.clone(), path: PathBuf::new() });
|
||
Ok(())
|
||
}
|
||
fn remove(&self, locale: &str) -> Result<(), String> {
|
||
self.installed.lock().unwrap_or_else(PoisonError::into_inner).retain(|p| p.locale != locale);
|
||
Ok(())
|
||
}
|
||
fn packs_dir(&self) -> Option<PathBuf> {
|
||
None
|
||
}
|
||
fn installed(&self) -> Vec<Installed> {
|
||
self.installed.lock().unwrap_or_else(PoisonError::into_inner).clone()
|
||
}
|
||
}
|
||
|
||
#[cfg(test)]
|
||
mod tests {
|
||
/// CP4.4 (`cargo xtask language`): a build embeds English Language data and nothing it
|
||
/// wasn't asked to. A Language build names the locales it embeds in `CRAFT_PROOF_EMBED`
|
||
/// (`en` alone: English only); the default embeds the carry-over locales too, until Language
|
||
/// packs are published and pinned (craft-proof's `DEFAULT_EMBED`; TODO(CP4.4): English only
|
||
/// then).
|
||
/// About › Language data shows SCOWL's notices (the Ispell BSD licence among them) whenever
|
||
/// English spells from the embedded SCOWL data.
|
||
#[test]
|
||
fn about_shows_the_scowl_notice_when_scowl_is_embedded() {
|
||
let texts = super::licence_texts();
|
||
match wordcraft_proof::scowl::notice() {
|
||
Some(notice) => {
|
||
let entry = texts.iter().find(|(t, _)| t == super::SCOWL_TITLE).map(|(_, x)| x.as_str());
|
||
assert_eq!(entry, Some(notice));
|
||
assert!(notice.contains("Ispell") || notice.contains("ispell"), "the Ispell notice is part of it");
|
||
}
|
||
None => assert!(texts.iter().all(|(t, _)| t != super::SCOWL_TITLE), "no SCOWL data: no SCOWL entry"),
|
||
}
|
||
}
|
||
|
||
#[test]
|
||
fn the_default_build_embeds_only_english() {
|
||
const CARRY_OVER: [&str; 5] = ["de", "pl", "pt", "it", "ru"];
|
||
let named: Vec<String> = std::env::var("CRAFT_PROOF_EMBED")
|
||
.unwrap_or_default()
|
||
.split(',')
|
||
.map(|l| l.trim().to_ascii_lowercase())
|
||
.filter(|l| !l.is_empty())
|
||
.collect();
|
||
let allowed = |base: &str| base == "en" || if named.is_empty() { CARRY_OVER.contains(&base) } else { named.iter().any(|n| n == base) };
|
||
let embedded = wordcraft_proof::packs::embedded_locales();
|
||
let extra: Vec<&&str> = embedded.iter().filter(|l| !allowed(l.split('-').next().unwrap_or(l))).collect();
|
||
assert!(extra.is_empty(), "embedded beyond English and {named:?} (or the carry-over by default): {extra:?}");
|
||
}
|
||
}
|