zng_ext_l10n/
types.rs

1use std::{borrow::Cow, collections::HashMap, fmt, mem, ops, path::PathBuf, sync::Arc};
2
3use fluent::types::FluentNumber;
4use once_cell::sync::Lazy;
5use semver::Version;
6use zng_ext_fs_watcher::WatcherReadStatus;
7use zng_layout::context::LayoutDirection;
8use zng_txt::{ToTxt, Txt};
9use zng_var::{ArcEq, IntoVar, Var, VarValue, const_var, context_var, impl_from_and_into_var};
10
11use crate::{L10N, lang, service::L10N_SV};
12
13/// Handle to multiple localization resources.
14#[derive(Clone, Debug)]
15pub struct LangResources(pub Vec<LangResource>);
16impl ops::Deref for LangResources {
17    type Target = Vec<LangResource>;
18
19    fn deref(&self) -> &Self::Target {
20        &self.0
21    }
22}
23impl ops::DerefMut for LangResources {
24    fn deref_mut(&mut self) -> &mut Self::Target {
25        &mut self.0
26    }
27}
28impl LangResources {
29    /// Wait for all the resources to load.
30    pub async fn wait(&self) {
31        for res in &self.0 {
32            res.wait().await;
33        }
34    }
35
36    /// Drop all handles without dropping the resource.
37    pub fn perm(self) {
38        for res in self.0 {
39            res.perm()
40        }
41    }
42}
43
44/// Handle to a localization resource.
45#[derive(Clone)]
46#[must_use = "resource can unload if dropped"]
47pub struct LangResource {
48    pub(super) res: Var<Option<ArcEq<fluent::FluentResource>>>,
49    pub(super) status: Var<LangResourceStatus>,
50}
51
52impl fmt::Debug for LangResource {
53    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
54        f.debug_struct("LangResource")
55            .field("status", &self.status.get())
56            .finish_non_exhaustive()
57    }
58}
59impl LangResource {
60    /// Read-only variable with the resource.
61    pub fn resource(&self) -> &Var<Option<ArcEq<fluent::FluentResource>>> {
62        &self.res
63    }
64
65    /// Read-only variable with the resource status.
66    pub fn status(&self) -> &Var<LangResourceStatus> {
67        &self.status
68    }
69
70    /// Drop the handle without unloading the resource.
71    pub fn perm(self) {
72        L10N_SV.write().push_perm_resource(self);
73    }
74
75    /// Await resource status to not be loading.
76    pub async fn wait(&self) {
77        while matches!(self.status.get(), LangResourceStatus::Loading) {
78            self.status.wait_update().await;
79        }
80    }
81}
82
83/// Status of a localization resource.
84#[derive(Clone, Debug)]
85pub enum LangResourceStatus {
86    /// Resource not available.
87    ///
88    /// This can change if the localization directory changes, or the file is created.
89    NotAvailable,
90    /// Resource is loading.
91    Loading,
92    /// Resource loaded ok.
93    Loaded,
94    /// Resource failed to load.
95    ///
96    /// This can be any IO or parse errors. If the resource if *not found* the status is set to
97    /// `NotAvailable`, not an error. Localization messages fallback on error just like they do
98    /// for not available.
99    Errors(StatusError),
100}
101impl fmt::Display for LangResourceStatus {
102    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
103        match self {
104            LangResourceStatus::NotAvailable => write!(f, "not available"),
105            LangResourceStatus::Loading => write!(f, "loading…"),
106            LangResourceStatus::Loaded => write!(f, "loaded"),
107            LangResourceStatus::Errors(e) => {
108                writeln!(f, "errors:")?;
109                for e in e {
110                    writeln!(f, "   {e}")?;
111                }
112                Ok(())
113            }
114        }
115    }
116}
117impl PartialEq for LangResourceStatus {
118    fn eq(&self, other: &Self) -> bool {
119        match (self, other) {
120            (Self::Errors(a), Self::Errors(b)) => a.is_empty() && b.is_empty(),
121            _ => core::mem::discriminant(self) == core::mem::discriminant(other),
122        }
123    }
124}
125impl Eq for LangResourceStatus {}
126impl WatcherReadStatus<StatusError> for LangResourceStatus {
127    fn idle() -> Self {
128        Self::Loaded
129    }
130
131    fn reading() -> Self {
132        Self::Loading
133    }
134
135    fn read_error(e: StatusError) -> Self {
136        Self::Errors(e)
137    }
138}
139impl WatcherReadStatus<LangResourceStatus> for LangResourceStatus {
140    fn idle() -> Self {
141        Self::Loaded
142    }
143
144    fn reading() -> Self {
145        Self::Loading
146    }
147
148    fn read_error(e: LangResourceStatus) -> Self {
149        e
150    }
151}
152
153type StatusError = Vec<Arc<dyn std::error::Error + Send + Sync>>;
154
155/// Localized message variable builder.
156///
157/// See [`L10N.message`] for more details.
158///
159/// [`L10N.message`]: L10N::message
160pub struct L10nMessageBuilder {
161    pub(super) file: LangFilePath,
162    pub(super) id: Txt,
163    pub(super) attribute: Txt,
164    pub(super) fallback: Txt,
165    pub(super) args: Vec<(Txt, Var<L10nArgument>)>,
166}
167impl L10nMessageBuilder {
168    /// Add a format arg variable.
169    pub fn arg(mut self, name: Txt, value: impl IntoVar<L10nArgument>) -> Self {
170        self.args.push((name, value.into_var()));
171        self
172    }
173    #[doc(hidden)]
174    pub fn l10n_arg(self, name: &'static str, value: Var<L10nArgument>) -> Self {
175        self.arg(Txt::from_static(name), value)
176    }
177
178    /// Build the message var for the given languages.
179    pub fn build_for(self, lang: impl Into<Langs>) -> Var<Txt> {
180        L10N_SV
181            .write()
182            .localized_message(lang.into(), self.file, self.id, self.attribute, self.fallback, self.args)
183    }
184
185    /// Build the message var for the contextual language.
186    pub fn build(self) -> Var<Txt> {
187        let Self {
188            file,
189            id,
190            attribute,
191            fallback,
192            args,
193        } = self;
194        LANG_VAR.flat_map(move |l| {
195            L10N_SV.write().localized_message(
196                l.clone(),
197                file.clone(),
198                id.clone(),
199                attribute.clone(),
200                fallback.clone(),
201                args.clone(),
202            )
203        })
204    }
205}
206
207/// Represents an argument value for a localization message.
208///
209/// See [`L10nMessageBuilder::arg`] for more details.
210#[derive(Clone, Debug, PartialEq)]
211pub enum L10nArgument {
212    /// String.
213    Txt(Txt),
214    /// Number, with optional style details.
215    Number(FluentNumber),
216}
217impl_from_and_into_var! {
218    fn from(txt: Txt) -> L10nArgument {
219        L10nArgument::Txt(txt)
220    }
221    fn from(txt: &'static str) -> L10nArgument {
222        L10nArgument::Txt(Txt::from_static(txt))
223    }
224    fn from(txt: String) -> L10nArgument {
225        L10nArgument::Txt(Txt::from(txt))
226    }
227    fn from(t: char) -> L10nArgument {
228        L10nArgument::Txt(Txt::from_char(t))
229    }
230    fn from(number: FluentNumber) -> L10nArgument {
231        L10nArgument::Number(number)
232    }
233    fn from(b: bool) -> L10nArgument {
234        b.to_txt().into()
235    }
236}
237macro_rules! impl_from_and_into_var_number {
238    ($($literal:tt),+) => {
239        impl_from_and_into_var! {
240            $(
241                fn from(number: $literal) -> L10nArgument {
242                    FluentNumber::from(number).into()
243                }
244            )+
245        }
246    }
247}
248impl_from_and_into_var_number! { u8, i8, u16, i16, u32, i32, u64, i64, u128, i128, usize, isize, f32, f64 }
249impl L10nArgument {
250    /// Borrow argument as a fluent value.
251    pub fn fluent_value(&self) -> fluent::FluentValue<'_> {
252        match self {
253            L10nArgument::Txt(t) => fluent::FluentValue::String(Cow::Borrowed(t.as_str())),
254            L10nArgument::Number(n) => fluent::FluentValue::Number(n.clone()),
255        }
256    }
257    /// Clone argument as a fluent value.
258    pub fn to_fluent_value(&self) -> fluent::FluentValue<'static> {
259        match self {
260            L10nArgument::Txt(t) => fluent::FluentValue::String(Cow::Owned(t.to_string())),
261            L10nArgument::Number(n) => fluent::FluentValue::Number(n.clone()),
262        }
263    }
264}
265
266#[doc(hidden)]
267pub struct L10nSpecialize<T>(pub Option<T>);
268#[doc(hidden)]
269pub trait IntoL10nVar {
270    fn to_l10n_var(&mut self) -> Var<L10nArgument>;
271}
272
273impl<T: Into<L10nArgument>> IntoL10nVar for L10nSpecialize<T> {
274    fn to_l10n_var(&mut self) -> Var<L10nArgument> {
275        const_var(self.0.take().unwrap().into())
276    }
277}
278impl<T: VarValue + Into<L10nArgument>> IntoL10nVar for &mut L10nSpecialize<Var<T>> {
279    fn to_l10n_var(&mut self) -> Var<L10nArgument> {
280        self.0.take().unwrap().map_into()
281    }
282}
283impl IntoL10nVar for &mut &mut L10nSpecialize<Var<L10nArgument>> {
284    fn to_l10n_var(&mut self) -> Var<L10nArgument> {
285        self.0.take().unwrap()
286    }
287}
288
289context_var! {
290    /// Language of text in a widget context.
291    ///
292    /// Is [`L10N.app_lang`] by default.
293    ///
294    /// [`L10N.app_lang`]: L10N::app_lang
295    pub static LANG_VAR: Langs = L10N.app_lang();
296}
297
298/// Identifies the language, region and script of text.
299///
300/// Use the [`lang!`] macro to construct one, it does compile-time validation.
301///
302/// Use the [`unic_langid`] crate for more advanced operations such as runtime parsing and editing identifiers, this
303/// type is just an alias for the core struct of that crate.
304///
305/// [`unic_langid`]: https://docs.rs/unic-langid
306#[derive(PartialEq, Eq, Hash, Clone, Default, PartialOrd, Ord, serde::Serialize, serde::Deserialize)]
307#[serde(transparent)]
308pub struct Lang(pub unic_langid::LanguageIdentifier);
309impl Lang {
310    /// Returns character direction of the language.
311    pub fn direction(&self) -> LayoutDirection {
312        crate::from_unic_char_direction(self.0.character_direction())
313    }
314
315    /// Compares a language to another allowing for either side to use the missing fields as wildcards.
316    ///
317    /// This allows for matching between `en` (treated as `en-*-*-*`) and `en-US`.
318    ///
319    /// # Examples
320    ///
321    /// ```
322    /// # use zng_ext_l10n::*;
323    /// #
324    /// let l1 = lang!("en");
325    /// let l2 = lang!("en-US");
326    ///
327    /// assert_ne!(l1, l2); // "en" != "en-US"
328    /// assert!(!l1.matches(&l2, false, false)); // "en" != "en-US"
329    ///
330    /// assert!(l1.matches(&l2, true, false)); // "en-*-*-*" == "en-US"
331    /// assert!(!l1.matches(&l2, false, true)); // "en" != "en-*-US-*"
332    /// assert!(l1.matches(&l2, true, true)); // "en-*-*-*" == "en-*-US-*"
333    /// ```
334    ///
335    /// Note that `machine` variant is ignored when matching.
336    ///
337    /// ```
338    /// # use zng_ext_l10n::*;
339    ///
340    /// let pt_br = lang!("pt-BR");
341    /// let pt_pt = lang!("pt-PT");
342    /// let pt = lang!("pt");
343    /// let pt_m = lang!("pt-machine");
344    ///
345    /// assert!(!pt_br.matches(&pt_pt, true, false)); // "pt-BR-*-*" != "pt-PT"
346    /// assert!(pt.matches(&pt_pt, true, false)); // "pt-*-*-*" == "pt-PT"
347    /// assert!(pt_m.matches(&pt_pt, true, false)); // "pt-machine-*-*" == "pt-PT"
348    ///
349    /// assert!(pt_m.matches(&pt, false, false)); // "pt-machine" == "pt"
350    /// assert_ne!(pt_m, pt); // but "pt-machine" != "pt" in `PartialEq`
351    /// ```
352    pub fn matches(&self, other: &Self, self_as_range: bool, other_as_range: bool) -> bool {
353        if !self.0.language.matches(other.0.language, self_as_range, other_as_range) {
354            return false;
355        }
356
357        fn tag_matches<P: PartialEq>(tag1: &Option<P>, tag2: &Option<P>, as_range1: bool, as_range2: bool) -> bool {
358            (as_range1 && tag1.is_none()) || (as_range2 && tag2.is_none()) || tag1 == tag2
359        }
360        if !tag_matches(&self.0.script, &other.0.script, self_as_range, other_as_range)
361            || !tag_matches(&self.region, &other.region, self_as_range, other_as_range)
362        {
363            return false;
364        }
365
366        (self_as_range && self.variants_no_machine().next().is_none())
367            || (other_as_range && other.variants_no_machine().next().is_none())
368            || self.variants_no_machine().eq(other.variants_no_machine())
369    }
370    fn variants_no_machine(&self) -> impl Iterator<Item = &unic_langid::subtags::Variant> {
371        self.0.variants().filter(|v| v.as_str() != "machine")
372    }
373
374    /// If the language has a `machine` variant.
375    ///
376    /// Machine translated resources are expected to be of lesser quality, but better than nothing.
377    /// This variant indicates that the resource should be given lower priority. If a machine generated
378    /// resource if reviewed by a human translator the `machine` variant must be removed.
379    pub fn is_machine_translation(&self) -> bool {
380        self.0.variants().any(|v| v == "machine")
381    }
382
383    /// Returns the name of the language and region in own language.
384    ///
385    /// The names are in localized title case, when the name is the same in different scripts the script name
386    /// is part of the language name (currently Hans/Hant).
387    ///
388    /// Returns `None` if the locale is not one of the 90 most common locales embedded. Language must match,
389    /// script must match exactly (omitted different from explicit), region is optional (if not matched returns only language name),
390    /// variants are ignored.
391    #[cfg(feature = "lang_autonym")]
392    pub fn autonym(&self) -> Option<LangAutonym> {
393        let lang = self.0.language.as_str();
394        let script = self.0.script.as_ref().map(|s| s.as_str()).unwrap_or("");
395        let region = self.0.region.as_ref().map(|r| r.as_str()).unwrap_or("");
396        let (name, region) = match (lang, script, region) {
397            // generated with cargo run --manifest-path tools/lang-autonym-gen/Cargo.toml
398            // please implement any manual corrections in the tool, not here
399            ("af", "", "") => ("Afrikaans", ""),
400            ("am", "", "") => ("አማርኛ", ""),
401            ("ar", "", "") => ("العربية", ""),
402            ("as", "", "") => ("অসমীয়া", ""),
403            ("az", "", "") => ("Azərbaycan", ""),
404            ("be", "", "") => ("Беларуская", ""),
405            ("bg", "", "") => ("Български", ""),
406            ("bn", "", "") => ("বাংলা", ""),
407            ("bs", "", "") => ("Bosanski", ""),
408            ("ca", "", "") => ("Català", ""),
409            ("cs", "", "") => ("Čeština", ""),
410            ("cy", "", "") => ("Cymraeg", ""),
411            ("da", "", "") => ("Dansk", ""),
412            ("de", "", "") => ("Deutsch", ""),
413            ("el", "", "") => ("Ελληνικά", ""),
414            ("en", "", "") => ("English", ""),
415            ("en", "", "GB") => ("English", "United Kingdom"),
416            ("en", "", "US") => ("English", "United States"),
417            ("es", "", "") => ("Español", ""),
418            ("es", "", "419") => ("Español", "Latinoamérica"),
419            ("es", "", "ES") => ("Español", "España"),
420            ("et", "", "") => ("Eesti", ""),
421            ("eu", "", "") => ("Euskara", ""),
422            ("fa", "", "") => ("فارسی", ""),
423            ("fi", "", "") => ("Suomi", ""),
424            ("fil", "", "") => ("Filipino", ""),
425            ("fr", "", "") => ("Français", ""),
426            ("fr", "", "CA") => ("Français", "Canada"),
427            ("ga", "", "") => ("Gaeilge", ""),
428            ("gd", "", "") => ("Gàidhlig", ""),
429            ("gl", "", "") => ("Galego", ""),
430            ("gu", "", "") => ("ગુજરાતી", ""),
431            ("he", "", "") => ("עברית", ""),
432            ("hi", "", "") => ("हिन्दी", ""),
433            ("hr", "", "") => ("Hrvatski", ""),
434            ("hu", "", "") => ("Magyar", ""),
435            ("hy", "", "") => ("Հայերեն", ""),
436            ("id", "", "") => ("Indonesia", ""),
437            ("is", "", "") => ("Íslenska", ""),
438            ("it", "", "") => ("Italiano", ""),
439            ("ja", "", "") => ("日本語", ""),
440            ("ka", "", "") => ("ქართული", ""),
441            ("kk", "", "") => ("Қазақ тілі", ""),
442            ("km", "", "") => ("ខ្មែរ", ""),
443            ("kn", "", "") => ("ಕನ್ನಡ", ""),
444            ("ko", "", "") => ("한국어", ""),
445            ("ky", "", "") => ("Кыргызча", ""),
446            ("lo", "", "") => ("ລາວ", ""),
447            ("lt", "", "") => ("Lietuvių", ""),
448            ("lv", "", "") => ("Latviešu", ""),
449            ("mk", "", "") => ("Македонски", ""),
450            ("ml", "", "") => ("മലയാളം", ""),
451            ("mn", "", "") => ("Монгол", ""),
452            ("mr", "", "") => ("मराठी", ""),
453            ("ms", "", "") => ("Melayu", ""),
454            ("my", "", "") => ("မြန်မာ", ""),
455            ("nb", "", "") => ("Norsk bokmål", ""),
456            ("ne", "", "") => ("नेपाली", ""),
457            ("nl", "", "") => ("Nederlands", ""),
458            ("nn", "", "") => ("Norsk nynorsk", ""),
459            ("or", "", "") => ("ଓଡ଼ିଆ", ""),
460            ("pa", "", "") => ("ਪੰਜਾਬੀ", ""),
461            ("pl", "", "") => ("Polski", ""),
462            ("ps", "", "") => ("پښتو", ""),
463            ("pt", "", "") => ("Português", ""),
464            ("pt", "", "BR") => ("Português", "Brasil"),
465            ("pt", "", "PT") => ("Português", "Portugal"),
466            ("ro", "", "") => ("Română", ""),
467            ("ru", "", "") => ("Русский", ""),
468            ("si", "", "") => ("සිංහල", ""),
469            ("sk", "", "") => ("Slovenčina", ""),
470            ("sl", "", "") => ("Slovenščina", ""),
471            ("sq", "", "") => ("Shqip", ""),
472            ("sr", "", "") => ("Српски", ""),
473            ("sr", "Latn", "") => ("Srpski", ""),
474            ("sv", "", "") => ("Svenska", ""),
475            ("sw", "", "") => ("Kiswahili", ""),
476            ("ta", "", "") => ("தமிழ்", ""),
477            ("te", "", "") => ("తెలుగు", ""),
478            ("th", "", "") => ("ไทย", ""),
479            ("tr", "", "") => ("Türkçe", ""),
480            ("uk", "", "") => ("Українська", ""),
481            ("ur", "", "") => ("اردو", ""),
482            ("uz", "", "") => ("O‘zbek", ""),
483            ("vi", "", "") => ("Tiếng việt", ""),
484            ("zh", "", "") => ("中文", ""),
485            ("zh", "Hans", "") => ("简体中文", ""),
486            ("zh", "Hant", "") => ("繁體中文", ""),
487            ("zh", "", "TW") => ("繁體中文", "台灣"),
488            ("zu", "", "") => ("Isizulu", ""),
489            _ => {
490                if self.0.region.is_some() {
491                    let q = Lang(unic_langid::LanguageIdentifier::from_parts(
492                        self.0.language,
493                        self.0.script,
494                        None,
495                        &[],
496                    ));
497                    return q.autonym();
498                } else {
499                    return None;
500                }
501            }
502        };
503        Some(LangAutonym {
504            language: name,
505            region: if region.is_empty() { None } else { Some(region) },
506        })
507    }
508
509    /// Get ordering of this and `other` as [`fmt::Display`] text.
510    ///
511    /// This method avoids actually allocating a string.
512    pub fn cmp_display(&self, other: &Self) -> std::cmp::Ordering {
513        let (name, region) = self.name_region_str();
514        let (other_name, other_region) = other.name_region_str();
515        name.cmp(other_name).then_with(|| region.cmp(&other_region))
516    }
517
518    fn name_region_str(&self) -> (&str, Option<&str>) {
519        #[cfg(feature = "lang_autonym")]
520        if let Some(a) = self.autonym() {
521            let region = a.region.or_else(|| self.0.region.as_ref().map(|r| r.as_str()));
522            return (a.language, region);
523        }
524        (self.0.language.as_str(), self.0.region.as_ref().map(|r| r.as_str()))
525    }
526}
527impl ops::Deref for Lang {
528    type Target = unic_langid::LanguageIdentifier;
529
530    fn deref(&self) -> &Self::Target {
531        &self.0
532    }
533}
534impl fmt::Debug for Lang {
535    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
536        write!(f, "{}", self.0)
537    }
538}
539/// Prints the autonym if built with `"lang_autonym"` and has one, otherwise prints the tags.
540///
541/// `{}` prints language and region (if set), `{:#}` only prints language.
542impl fmt::Display for Lang {
543    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
544        let (name, region) = self.name_region_str();
545        if f.alternate()
546            && let Some(r) = region
547        {
548            write!(f, "{name} ({r})")
549        } else {
550            write!(f, "{name}")
551        }
552    }
553}
554impl std::str::FromStr for Lang {
555    type Err = unic_langid::LanguageIdentifierError;
556
557    fn from_str(s: &str) -> Result<Self, Self::Err> {
558        let s = s.trim();
559        if s.is_empty() {
560            return Ok(lang!(und));
561        }
562        unic_langid::LanguageIdentifier::from_str(s).map(Lang)
563    }
564}
565
566/// Represents the translated name of a language and region in own language, using the selected script.
567#[derive(Clone, PartialEq, Eq, Hash, PartialOrd, Ord)]
568#[cfg(feature = "lang_autonym")]
569pub struct LangAutonym {
570    /// Language name in its own language.
571    pub language: &'static str,
572    /// Region name in the language.
573    ///
574    /// Can be `None` when a region was requested and only the language name was available.
575    pub region: Option<&'static str>,
576}
577#[cfg(feature = "lang_autonym")]
578impl fmt::Debug for LangAutonym {
579    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
580        if f.alternate() {
581            f.debug_struct("LangAutonym")
582                .field("language", &self.language)
583                .field("region", &self.region)
584                .finish()
585        } else {
586            write!(f, "{self}")
587        }
588    }
589}
590/// `{}` prints language and region (if set), `{:#}` only prints language.
591#[cfg(feature = "lang_autonym")]
592impl fmt::Display for LangAutonym {
593    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
594        write!(f, "{}", self.language)?;
595        if let Some(r) = self.region
596            && !f.alternate()
597        {
598            write!(f, " ({r})")?;
599        }
600        Ok(())
601    }
602}
603
604/// List of languages, in priority order.
605#[derive(Clone, PartialEq, Eq, Default, Hash, serde::Serialize, serde::Deserialize)]
606#[serde(transparent)]
607pub struct Langs(pub Vec<Lang>);
608impl fmt::Debug for Langs {
609    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
610        struct DisplayLangs<'a>(&'a [Lang]);
611        impl fmt::Debug for DisplayLangs<'_> {
612            fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
613                f.debug_list().entries(self.0.iter()).finish()
614            }
615        }
616        if f.alternate() {
617            f.debug_tuple("Langs").field(&DisplayLangs(&self.0)).finish()
618        } else {
619            fmt::Debug::fmt(&DisplayLangs(&self.0), f)
620        }
621    }
622}
623impl Langs {
624    /// The first lang on the list or `und` if the list is empty.
625    pub fn best(&self) -> &Lang {
626        static NONE: Lazy<Lang> = Lazy::new(|| lang!(und));
627        self.first().unwrap_or(&NONE)
628    }
629}
630impl ops::Deref for Langs {
631    type Target = Vec<Lang>;
632
633    fn deref(&self) -> &Self::Target {
634        &self.0
635    }
636}
637impl ops::DerefMut for Langs {
638    fn deref_mut(&mut self) -> &mut Self::Target {
639        &mut self.0
640    }
641}
642impl_from_and_into_var! {
643    fn from(lang: Lang) -> Langs {
644        Langs(vec![lang])
645    }
646    fn from(lang: Option<Lang>) -> Langs {
647        Langs(lang.into_iter().collect())
648    }
649}
650impl fmt::Display for Langs {
651    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
652        let mut sep = "";
653        for l in self.iter() {
654            write!(f, "{sep}{l}")?;
655            sep = ", ";
656        }
657        Ok(())
658    }
659}
660impl std::str::FromStr for Langs {
661    type Err = unic_langid::LanguageIdentifierError;
662
663    fn from_str(s: &str) -> Result<Self, Self::Err> {
664        if s.trim().is_empty() {
665            return Ok(Langs(vec![]));
666        }
667        let mut r = Self(vec![]);
668        for lang in s.split(',') {
669            r.0.push(lang.trim().parse()?)
670        }
671        Ok(r)
672    }
673}
674
675/// Represents a map of [`Lang`] keys that can be partially matched.
676#[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
677#[serde(transparent)]
678pub struct LangMap<V> {
679    inner: Vec<(Lang, V)>,
680}
681impl<V> Default for LangMap<V> {
682    fn default() -> Self {
683        Self { inner: Default::default() }
684    }
685}
686impl<V> LangMap<V> {
687    /// New empty default.
688    pub fn new() -> Self {
689        LangMap::default()
690    }
691
692    /// New empty with pre-allocated capacity.
693    pub fn with_capacity(capacity: usize) -> Self {
694        LangMap {
695            inner: Vec::with_capacity(capacity),
696        }
697    }
698
699    fn exact_i(&self, lang: &Lang) -> Option<usize> {
700        for (i, (key, _)) in self.inner.iter().enumerate() {
701            if key == lang {
702                return Some(i);
703            }
704        }
705        None
706    }
707
708    fn best_i(&self, lang: &Lang) -> Option<usize> {
709        let mut best = None;
710        let mut best_weight = 0;
711
712        for (i, (key, _)) in self.inner.iter().enumerate() {
713            if lang.matches(key, true, true) {
714                let mut weight = 1;
715                let mut eq = 0;
716
717                // must match language
718                if key.language == lang.language {
719                    weight += 200;
720                    eq += 1;
721                }
722                // must match script
723                if key.script == lang.script {
724                    weight += 100;
725                    eq += 1;
726                }
727                // can fallback to different region
728                if key.region == lang.region {
729                    weight += 30;
730                    eq += 1;
731                }
732
733                if eq == 3 && lang.variants().eq(key.variants()) {
734                    // exact match
735                    return Some(i);
736                }
737
738                // human translated of different regions has priority over machine translated
739                if !key.is_machine_translation() {
740                    weight += 60;
741                }
742
743                if best_weight < weight {
744                    best_weight = weight;
745                    best = Some(i);
746                }
747            }
748        }
749
750        best
751    }
752
753    /// Returns the best match to `lang` currently in the map.
754    pub fn best_match(&self, lang: &Lang) -> Option<&Lang> {
755        if let Some(i) = self.best_i(lang) {
756            Some(&self.inner[i].0)
757        } else {
758            None
759        }
760    }
761
762    /// Returns the best match for `lang`.
763    pub fn get(&self, lang: &Lang) -> Option<&V> {
764        if let Some(i) = self.best_i(lang) {
765            Some(&self.inner[i].1)
766        } else {
767            None
768        }
769    }
770
771    /// Returns the exact match for `lang`.
772    pub fn get_exact(&self, lang: &Lang) -> Option<&V> {
773        if let Some(i) = self.exact_i(lang) {
774            Some(&self.inner[i].1)
775        } else {
776            None
777        }
778    }
779
780    /// Returns the best match for `lang`.
781    pub fn get_mut(&mut self, lang: &Lang) -> Option<&mut V> {
782        if let Some(i) = self.best_i(lang) {
783            Some(&mut self.inner[i].1)
784        } else {
785            None
786        }
787    }
788
789    /// Returns the exact match for `lang`.
790    pub fn get_exact_mut(&mut self, lang: &Lang) -> Option<&mut V> {
791        if let Some(i) = self.exact_i(lang) {
792            Some(&mut self.inner[i].1)
793        } else {
794            None
795        }
796    }
797
798    /// Returns the current value or insert `new` and return a reference to it.
799    pub fn get_exact_or_insert(&mut self, lang: Lang, new: impl FnOnce() -> V) -> &mut V {
800        if let Some(i) = self.exact_i(&lang) {
801            return &mut self.inner[i].1;
802        }
803        let i = self.inner.len();
804        self.inner.push((lang, new()));
805        &mut self.inner[i].1
806    }
807
808    /// Insert the value with the exact match of `lang`.
809    ///
810    /// Returns the previous exact match.
811    pub fn insert(&mut self, lang: Lang, value: V) -> Option<V> {
812        if let Some(i) = self.exact_i(&lang) {
813            Some(mem::replace(&mut self.inner[i].1, value))
814        } else {
815            self.inner.push((lang, value));
816            None
817        }
818    }
819
820    /// Remove the exact match of `lang`.
821    pub fn remove(&mut self, lang: &Lang) -> Option<V> {
822        if let Some(i) = self.exact_i(lang) {
823            Some(self.inner.swap_remove(i).1)
824        } else {
825            None
826        }
827    }
828
829    /// Remove all exact and partial matches of `lang`.
830    ///
831    /// Returns a count of items removed.
832    pub fn remove_all(&mut self, lang: &Lang) -> usize {
833        let mut count = 0;
834        self.inner.retain(|(key, _)| {
835            let rmv = lang.matches(key, true, false);
836            if rmv {
837                count += 1
838            }
839            !rmv
840        });
841        count
842    }
843
844    /// Remove the last inserted lang.
845    pub fn pop(&mut self) -> Option<(Lang, V)> {
846        self.inner.pop()
847    }
848
849    /// Returns if the map is empty.
850    pub fn is_empty(&self) -> bool {
851        self.inner.is_empty()
852    }
853
854    /// Returns the number of languages in the map.
855    pub fn len(&self) -> usize {
856        self.inner.len()
857    }
858
859    /// Remove all entries.
860    pub fn clear(&mut self) {
861        self.inner.clear()
862    }
863
864    /// Iterator over lang keys.
865    pub fn keys(&self) -> impl std::iter::ExactSizeIterator<Item = &Lang> {
866        self.inner.iter().map(|(k, _)| k)
867    }
868
869    /// Iterator over values.
870    pub fn values(&self) -> impl std::iter::ExactSizeIterator<Item = &V> {
871        self.inner.iter().map(|(_, v)| v)
872    }
873
874    /// Iterator over values.
875    pub fn values_mut(&mut self) -> impl std::iter::ExactSizeIterator<Item = &mut V> {
876        self.inner.iter_mut().map(|(_, v)| v)
877    }
878
879    /// Into iterator of values.
880    pub fn into_values(self) -> impl std::iter::ExactSizeIterator<Item = V> {
881        self.inner.into_iter().map(|(_, v)| v)
882    }
883
884    /// Iterate over key-value pairs.
885    pub fn iter(&self) -> impl std::iter::ExactSizeIterator<Item = (&Lang, &V)> {
886        self.inner.iter().map(|(k, v)| (k, v))
887    }
888
889    /// Iterate over key-value pairs with mutable values.
890    pub fn iter_mut(&mut self) -> impl std::iter::ExactSizeIterator<Item = (&Lang, &mut V)> {
891        self.inner.iter_mut().map(|(k, v)| (&*k, v))
892    }
893}
894impl<V> LangMap<HashMap<LangFilePath, V>> {
895    /// Returns the match for `lang` and `file`.
896    pub fn get_file(&self, lang: &Lang, file: &LangFilePath) -> Option<&V> {
897        let files = self.get(lang)?;
898        if let Some(exact) = files.get(file) {
899            return Some(exact);
900        }
901        Self::best_file(files, file).map(|(_, v)| v)
902    }
903
904    /// Returns the best match to `lang` and `file` currently in the map.
905    pub fn best_file_match(&self, lang: &Lang, file: &LangFilePath) -> Option<&LangFilePath> {
906        let files = self.get(lang)?;
907        if let Some((exact, _)) = files.get_key_value(file) {
908            return Some(exact);
909        }
910        Self::best_file(files, file).map(|(k, _)| k)
911    }
912
913    fn best_file<'a>(files: &'a HashMap<LangFilePath, V>, file: &LangFilePath) -> Option<(&'a LangFilePath, &'a V)> {
914        let mut best = None;
915        let mut best_dist = u64::MAX;
916        for (k, v) in files {
917            if let Some(d) = k.matches(file)
918                && d < best_dist
919            {
920                best = Some((k, v));
921                best_dist = d;
922            }
923        }
924        best
925    }
926}
927impl<V> IntoIterator for LangMap<V> {
928    type Item = (Lang, V);
929
930    type IntoIter = std::vec::IntoIter<(Lang, V)>;
931
932    fn into_iter(self) -> Self::IntoIter {
933        self.inner.into_iter()
934    }
935}
936impl<V: PartialEq> PartialEq for LangMap<V> {
937    fn eq(&self, other: &Self) -> bool {
938        if self.len() != other.len() {
939            return false;
940        }
941        for (k, v) in &self.inner {
942            if other.get_exact(k) != Some(v) {
943                return false;
944            }
945        }
946        true
947    }
948}
949impl<V: Eq> Eq for LangMap<V> {}
950
951/// Errors found parsing a fluent resource file.
952#[derive(Clone, Debug)]
953pub struct FluentParserErrors(pub Vec<fluent_syntax::parser::ParserError>);
954impl fmt::Display for FluentParserErrors {
955    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
956        let mut sep = "";
957        for e in &self.0 {
958            write!(f, "{sep}{e}")?;
959            sep = "\n";
960        }
961        Ok(())
962    }
963}
964impl std::error::Error for FluentParserErrors {
965    fn source(&self) -> Option<&(dyn std::error::Error + 'static)> {
966        if self.0.len() == 1 { Some(&self.0[0]) } else { None }
967    }
968}
969
970/// Localization resource file path in the localization directory.
971///
972/// In the default directory layout, localization dependencies are collected using `cargo zng l10n`
973/// and copied to `l10n/{lang}/deps/{name}/{version}/`, and localization for the app ([`is_current_app`])
974/// is placed in `l10n/{lang}/`.
975///
976/// [`is_current_app`]: Self::is_current_app
977#[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
978#[non_exhaustive]
979pub struct LangFilePath {
980    /// Package name.
981    pub pkg_name: Txt,
982    /// Package version.
983    pub pkg_version: Version,
984    /// The localization file name, without extension.
985    pub file: Txt,
986}
987impl Ord for LangFilePath {
988    fn cmp(&self, other: &Self) -> std::cmp::Ordering {
989        let self_pkg = self.actual_pkg_data();
990        let other_pkg = other.actual_pkg_data();
991        match self_pkg.0.cmp(other_pkg.0) {
992            core::cmp::Ordering::Equal => {}
993            ord => return ord,
994        }
995        match self_pkg.1.cmp(other_pkg.1) {
996            core::cmp::Ordering::Equal => {}
997            ord => return ord,
998        }
999        self.file().cmp(&other.file())
1000    }
1001}
1002impl PartialOrd for LangFilePath {
1003    fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
1004        Some(self.cmp(other))
1005    }
1006}
1007impl std::hash::Hash for LangFilePath {
1008    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
1009        self.actual_pkg_data().hash(state);
1010        self.file().hash(state);
1011    }
1012}
1013impl Eq for LangFilePath {}
1014impl PartialEq for LangFilePath {
1015    fn eq(&self, other: &Self) -> bool {
1016        self.actual_pkg_data() == other.actual_pkg_data() && self.file() == other.file()
1017    }
1018}
1019impl LangFilePath {
1020    /// New from package name, version and file.
1021    pub fn new(pkg_name: impl Into<Txt>, pkg_version: Version, file: impl Into<Txt>) -> Self {
1022        let r = Self {
1023            pkg_name: pkg_name.into(),
1024            pkg_version,
1025            file: file.into(),
1026        };
1027        // these non-standard names are matched on fallback, but they can cause duplicate caching
1028        debug_assert!(
1029            r.file
1030                .rsplit_once('.')
1031                .map(|(_, ext)| !ext.eq_ignore_ascii_case("ftl"))
1032                .unwrap_or(true),
1033            "file `{}` must not have extension",
1034            r.file
1035        );
1036        debug_assert!(r.file != "_", "file `_` should be an empty string");
1037        r
1038    }
1039
1040    /// Gets a file in the current app.
1041    ///
1042    /// This value indicates that the localization resources are directly on the `l10n/{lang?}/` directories, not
1043    /// in the dependencies directories.
1044    ///
1045    /// See [`zng_env::about()`] for more details.
1046    pub fn current_app(file: impl Into<Txt>) -> LangFilePath {
1047        let about = zng_env::about();
1048        Self::new(about.pkg_name.clone(), about.version.clone(), file.into())
1049    }
1050
1051    /// Gets if this file is in the [`current_app`] resources, or the `pkg_name` is empty or the `pkg_version.pre` is `#.#.#-local`.
1052    ///
1053    /// [`current_app`]: Self::current_app
1054    pub fn is_current_app(&self) -> bool {
1055        self.is_current_app_no_check() || {
1056            let about = zng_env::about();
1057            self.pkg_name == about.pkg_name && self.pkg_version == about.version
1058        }
1059    }
1060
1061    fn is_current_app_no_check(&self) -> bool {
1062        self.pkg_name.is_empty() || self.pkg_version.pre.as_str() == "local"
1063    }
1064
1065    fn actual_pkg_data(&self) -> (&Txt, &Version) {
1066        if self.is_current_app_no_check() {
1067            let about = zng_env::about();
1068            (&about.pkg_name, &about.version)
1069        } else {
1070            (&self.pkg_name, &self.pkg_version)
1071        }
1072    }
1073
1074    /// Gets the normalized package name.
1075    ///
1076    /// This is the app package name if [`is_current_app`], otherwise is just the `pkg_name` value.
1077    ///
1078    /// [`is_current_app`]: Self::is_current_app
1079    pub fn pkg_name(&self) -> Txt {
1080        self.actual_pkg_data().0.clone()
1081    }
1082
1083    /// Gets the normalized package version.
1084    ///
1085    /// This is the app version if [`is_current_app`], otherwise is just the `pkg_version` value.
1086    ///
1087    /// [`is_current_app`]: Self::is_current_app
1088    pub fn pkg_version(&self) -> Version {
1089        self.actual_pkg_data().1.clone()
1090    }
1091
1092    /// Gets the normalized file name.
1093    ///
1094    /// This `"_"` for empty file or the file.
1095    pub fn file(&self) -> Txt {
1096        if self.file.is_empty() {
1097            Txt::from_char('_')
1098        } else {
1099            self.file.clone()
1100        }
1101    }
1102
1103    /// Get the file path, relative to the localization dir.
1104    ///
1105    /// * Empty file name is the same as `_`.
1106    /// * If package [`is_current_app`] gets `{lang}/{file}.ftl`.
1107    /// * Else if is another package gets `{lang}/deps/{pkg_name}/{pkg_version}/{file}.ftl`.
1108    ///
1109    /// [`is_current_app`]: Self::is_current_app
1110    pub fn to_path(&self, lang: &Lang) -> PathBuf {
1111        let mut file = self.file.as_str();
1112        if file.is_empty() {
1113            file = "_";
1114        }
1115        if self.is_current_app() {
1116            format!("{lang}/{file}.ftl")
1117        } else {
1118            format!("{lang}/deps/{}/{}/{file}.ftl", self.pkg_name, self.pkg_version)
1119        }
1120        .into()
1121    }
1122
1123    /// Gets a value that indicates if the resources represented by `self` can be used for `search`.
1124    ///
1125    /// The number indicates the quality of the match:
1126    ///
1127    /// * `0` is an exact match.
1128    /// * `b1` is a match with only version `build` differences.
1129    /// * `b10` is a match with only version `pre` differences.
1130    /// * `(0..u16::MAX) << 16` is a match with only `patch` differences and the absolute distance.
1131    /// * `(0..u16::MAX) << 16 * 2` is a match with `minor` differences and the absolute distance.
1132    /// * `(0..u16::MAX) << 16 * 3` is a match with `major` differences and the absolute distance.
1133    /// * `None`` is a `pkg_name` mismatch.
1134    pub fn matches(&self, search: &Self) -> Option<u64> {
1135        let (self_name, self_version) = self.actual_pkg_data();
1136        let (search_name, search_version) = search.actual_pkg_data();
1137
1138        if self_name != search_name {
1139            return None;
1140        }
1141
1142        if self.file != search.file {
1143            let file_a = self.file.rsplit_once('.').map(|t| t.0).unwrap_or(self.file.as_str());
1144            let file_b = search.file.rsplit_once('.').map(|t| t.0).unwrap_or(search.file.as_str());
1145            if file_a != file_b {
1146                let is_empty_a = file_a == "_" || file_a.is_empty();
1147                let is_empty_b = file_b == "_" || file_b.is_empty();
1148                if !(is_empty_a && is_empty_b) {
1149                    return None;
1150                }
1151            }
1152            tracing::warn!(
1153                "fallback matching `{}` with `{}`, file was not expected to have extension",
1154                self.file,
1155                search.file
1156            )
1157        }
1158
1159        fn dist(a: u64, b: u64, shift: u64) -> u64 {
1160            let (l, s) = match a.cmp(&b) {
1161                std::cmp::Ordering::Equal => return 0,
1162                std::cmp::Ordering::Less => (b, a),
1163                std::cmp::Ordering::Greater => (a, b),
1164            };
1165
1166            (l - s).min(u16::MAX as u64) << (16 * shift)
1167        }
1168
1169        let mut d = 0;
1170        if self_version.build != search_version.build {
1171            d = 1;
1172        }
1173        if self_version.pre != search_version.pre {
1174            d |= 0b10;
1175        }
1176
1177        d |= dist(self_version.patch, search_version.patch, 1);
1178        d |= dist(self_version.minor, search_version.minor, 2);
1179        d |= dist(self_version.major, search_version.major, 3);
1180
1181        Some(d)
1182    }
1183}
1184impl_from_and_into_var! {
1185    fn from(file: Txt) -> LangFilePath {
1186        LangFilePath::current_app(file)
1187    }
1188
1189    fn from(file: &'static str) -> LangFilePath {
1190        LangFilePath::current_app(file)
1191    }
1192
1193    fn from(file: String) -> LangFilePath {
1194        LangFilePath::current_app(file)
1195    }
1196}
1197
1198#[cfg(test)]
1199mod tests {
1200    use super::*;
1201
1202    #[test]
1203    fn file_matches() {
1204        fn check(a: &str, b: &str, c: &str) {
1205            let ap = LangFilePath::new("name", a.parse().unwrap(), "file");
1206            let bp = LangFilePath::new("name", b.parse().unwrap(), "file");
1207            let cp = LangFilePath::new("name", c.parse().unwrap(), "file");
1208
1209            let ab = ap.matches(&bp);
1210            let ac = ap.matches(&cp);
1211
1212            assert!(ab < ac, "expected {a}.matches({b}) < {a}.matches({c})")
1213        }
1214
1215        check("0.0.0", "0.0.1", "0.1.0");
1216        check("0.0.1", "0.1.0", "1.0.0");
1217        check("0.0.0-pre", "0.0.0-pre+build", "0.0.0-other+build");
1218        check("0.0.0+build", "0.0.0+build", "0.0.0+other");
1219        check("0.0.1", "0.0.2", "0.0.3");
1220        check("0.1.0", "0.2.0", "0.3.0");
1221        check("1.0.0", "2.0.0", "3.0.0");
1222        check("1.0.0", "1.1.0", "2.0.0");
1223    }
1224
1225    #[test]
1226    fn file_name_mismatches() {
1227        let ap = LangFilePath::new("name", "1.0.0".parse().unwrap(), "file-a");
1228        let bp = LangFilePath::new("name", "1.0.0".parse().unwrap(), "file-b");
1229
1230        assert!(ap.matches(&bp).is_none());
1231    }
1232}
zng_ext_l10n/types.rs

zng_ext_l10n/
types.rs
