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}
272impl<T: fmt::Display> IntoL10nVar for L10nSpecialize<T> {
273    fn to_l10n_var(&mut self) -> Var<L10nArgument> {
274        const_var(self.0.take().unwrap().to_txt().into())
275    }
276}
277impl<T: Into<L10nArgument>> IntoL10nVar for &mut L10nSpecialize<T> {
278    fn to_l10n_var(&mut self) -> Var<L10nArgument> {
279        const_var(self.0.take().unwrap().into())
280    }
281}
282impl<T: VarValue + Into<L10nArgument>> IntoL10nVar for &mut &mut L10nSpecialize<Var<T>> {
283    fn to_l10n_var(&mut self) -> Var<L10nArgument> {
284        self.0.take().unwrap().map_into()
285    }
286}
287impl IntoL10nVar for &mut &mut &mut L10nSpecialize<Var<L10nArgument>> {
288    fn to_l10n_var(&mut self) -> Var<L10nArgument> {
289        self.0.take().unwrap()
290    }
291}
292
293context_var! {
294    /// Language of text in a widget context.
295    ///
296    /// Is [`L10N.app_lang`] by default.
297    ///
298    /// [`L10N.app_lang`]: L10N::app_lang
299    pub static LANG_VAR: Langs = L10N.app_lang();
300}
301
302/// Identifies the language, region and script of text.
303///
304/// Use the [`lang!`] macro to construct one, it does compile-time validation.
305///
306/// Use the [`unic_langid`] crate for more advanced operations such as runtime parsing and editing identifiers, this
307/// type is just an alias for the core struct of that crate.
308///
309/// [`unic_langid`]: https://docs.rs/unic-langid
310#[derive(PartialEq, Eq, Hash, Clone, Default, PartialOrd, Ord, serde::Serialize, serde::Deserialize)]
311#[serde(transparent)]
312pub struct Lang(pub unic_langid::LanguageIdentifier);
313impl Lang {
314    /// Returns character direction of the language.
315    pub fn direction(&self) -> LayoutDirection {
316        if self == &lang!("ms") {
317            // Malay should default to "ms-Latn" that LTR, but unic_langid is matching "ms-Arab"
318            return LayoutDirection::LTR;
319        }
320        if self == &lang!("uz") {
321            // Uzbek should default to "uz-Latn"
322            return LayoutDirection::LTR;
323        }
324        crate::from_unic_char_direction(self.0.character_direction())
325    }
326
327    /// Compares a language to another allowing for either side to use the missing fields as wildcards.
328    ///
329    /// This allows for matching between `en` (treated as `en-*-*-*`) and `en-US`.
330    ///
331    /// # Examples
332    ///
333    /// ```
334    /// # use zng_ext_l10n::*;
335    /// #
336    /// let l1 = lang!("en");
337    /// let l2 = lang!("en-US");
338    ///
339    /// assert_ne!(l1, l2); // "en" != "en-US"
340    /// assert!(!l1.matches(&l2, false, false)); // "en" != "en-US"
341    ///
342    /// assert!(l1.matches(&l2, true, false)); // "en-*-*-*" == "en-US"
343    /// assert!(!l1.matches(&l2, false, true)); // "en" != "en-*-US-*"
344    /// assert!(l1.matches(&l2, true, true)); // "en-*-*-*" == "en-*-US-*"
345    /// ```
346    ///
347    /// Note that `machine` variant is ignored when matching.
348    ///
349    /// ```
350    /// # use zng_ext_l10n::*;
351    ///
352    /// let pt_br = lang!("pt-BR");
353    /// let pt_pt = lang!("pt-PT");
354    /// let pt = lang!("pt");
355    /// let pt_m = lang!("pt-machine");
356    ///
357    /// assert!(!pt_br.matches(&pt_pt, true, false)); // "pt-BR-*-*" != "pt-PT"
358    /// assert!(pt.matches(&pt_pt, true, false)); // "pt-*-*-*" == "pt-PT"
359    /// assert!(pt_m.matches(&pt_pt, true, false)); // "pt-machine-*-*" == "pt-PT"
360    ///
361    /// assert!(pt_m.matches(&pt, false, false)); // "pt-machine" == "pt"
362    /// assert_ne!(pt_m, pt); // but "pt-machine" != "pt" in `PartialEq`
363    /// ```
364    pub fn matches(&self, other: &Self, self_as_range: bool, other_as_range: bool) -> bool {
365        if !self.0.language.matches(other.0.language, self_as_range, other_as_range) {
366            return false;
367        }
368
369        fn tag_matches<P: PartialEq>(tag1: &Option<P>, tag2: &Option<P>, as_range1: bool, as_range2: bool) -> bool {
370            (as_range1 && tag1.is_none()) || (as_range2 && tag2.is_none()) || tag1 == tag2
371        }
372        if !tag_matches(&self.0.script, &other.0.script, self_as_range, other_as_range)
373            || !tag_matches(&self.region, &other.region, self_as_range, other_as_range)
374        {
375            return false;
376        }
377
378        (self_as_range && self.variants_no_machine().next().is_none())
379            || (other_as_range && other.variants_no_machine().next().is_none())
380            || self.variants_no_machine().eq(other.variants_no_machine())
381    }
382    fn variants_no_machine(&self) -> impl Iterator<Item = &unic_langid::subtags::Variant> {
383        self.0.variants().filter(|v| v.as_str() != "machine")
384    }
385
386    /// If the language has a `machine` variant.
387    ///
388    /// Machine translated resources are expected to be of lesser quality, but better than nothing.
389    /// This variant indicates that the resource should be given lower priority. If a machine generated
390    /// resource if reviewed by a human translator the `machine` variant must be removed.
391    pub fn is_machine_translation(&self) -> bool {
392        self.0.variants().any(|v| v == "machine")
393    }
394
395    /// Returns the name of the language and region in own language.
396    ///
397    /// The names are in localized title case, when the name is the same in different scripts the script name
398    /// is part of the language name (currently Hans/Hant).
399    ///
400    /// Returns `None` if the locale is not one of the 90 most common locales embedded. Language must match,
401    /// script must match exactly (omitted different from explicit), region is optional (if not matched returns only language name),
402    /// variants are ignored.
403    #[cfg(feature = "lang_autonym")]
404    pub fn autonym(&self) -> Option<LangAutonym> {
405        let lang = self.0.language.as_str();
406        let script = self.0.script.as_ref().map(|s| s.as_str()).unwrap_or("");
407        let region = self.0.region.as_ref().map(|r| r.as_str()).unwrap_or("");
408        let (name, region) = match (lang, script, region) {
409            // generated with cargo run --manifest-path tools/lang-autonym-gen/Cargo.toml
410            // please implement any manual corrections in the tool, not here
411            ("af", "", "") => ("Afrikaans", ""),
412            ("am", "", "") => ("አማርኛ", ""),
413            ("ar", "", "") => ("العربية", ""),
414            ("as", "", "") => ("অসমীয়া", ""),
415            ("az", "", "") => ("Azərbaycan", ""),
416            ("be", "", "") => ("Беларуская", ""),
417            ("bg", "", "") => ("Български", ""),
418            ("bn", "", "") => ("বাংলা", ""),
419            ("bs", "", "") => ("Bosanski", ""),
420            ("ca", "", "") => ("Català", ""),
421            ("cs", "", "") => ("Čeština", ""),
422            ("cy", "", "") => ("Cymraeg", ""),
423            ("da", "", "") => ("Dansk", ""),
424            ("de", "", "") => ("Deutsch", ""),
425            ("el", "", "") => ("Ελληνικά", ""),
426            ("en", "", "") => ("English", ""),
427            ("en", "", "GB") => ("English", "United Kingdom"),
428            ("en", "", "US") => ("English", "United States"),
429            ("es", "", "") => ("Español", ""),
430            ("es", "", "419") => ("Español", "Latinoamérica"),
431            ("es", "", "ES") => ("Español", "España"),
432            ("et", "", "") => ("Eesti", ""),
433            ("eu", "", "") => ("Euskara", ""),
434            ("fa", "", "") => ("فارسی", ""),
435            ("fi", "", "") => ("Suomi", ""),
436            ("fil", "", "") => ("Filipino", ""),
437            ("fr", "", "") => ("Français", ""),
438            ("fr", "", "FR") => ("Français", "France"),
439            ("fr", "", "CA") => ("Français", "Canada"),
440            ("ga", "", "") => ("Gaeilge", ""),
441            ("gd", "", "") => ("Gàidhlig", ""),
442            ("gl", "", "") => ("Galego", ""),
443            ("gu", "", "") => ("ગુજરાતી", ""),
444            ("he", "", "") => ("עברית", ""),
445            ("hi", "", "") => ("हिन्दी", ""),
446            ("hr", "", "") => ("Hrvatski", ""),
447            ("hu", "", "") => ("Magyar", ""),
448            ("hy", "", "") => ("Հայերեն", ""),
449            ("id", "", "") => ("Indonesia", ""),
450            ("is", "", "") => ("Íslenska", ""),
451            ("it", "", "") => ("Italiano", ""),
452            ("ja", "", "") => ("日本語", ""),
453            ("ka", "", "") => ("ქართული", ""),
454            ("kk", "", "") => ("Қазақ тілі", ""),
455            ("km", "", "") => ("ខ្មែរ", ""),
456            ("kn", "", "") => ("ಕನ್ನಡ", ""),
457            ("ko", "", "") => ("한국어", ""),
458            ("ky", "", "") => ("Кыргызча", ""),
459            ("lo", "", "") => ("ລາວ", ""),
460            ("lt", "", "") => ("Lietuvių", ""),
461            ("lv", "", "") => ("Latviešu", ""),
462            ("mk", "", "") => ("Македонски", ""),
463            ("ml", "", "") => ("മലയാളം", ""),
464            ("mn", "", "") => ("Монгол", ""),
465            ("mr", "", "") => ("मराठी", ""),
466            ("ms", "", "") => ("Melayu", ""),
467            ("my", "", "") => ("မြန်မာ", ""),
468            ("nb", "", "") => ("Norsk bokmål", ""),
469            ("ne", "", "") => ("नेपाली", ""),
470            ("nl", "", "") => ("Nederlands", ""),
471            ("nn", "", "") => ("Norsk nynorsk", ""),
472            ("or", "", "") => ("ଓଡ଼ିଆ", ""),
473            ("pa", "", "") => ("ਪੰਜਾਬੀ", ""),
474            ("pl", "", "") => ("Polski", ""),
475            ("ps", "", "") => ("پښتو", ""),
476            ("pt", "", "") => ("Português", ""),
477            ("pt", "", "BR") => ("Português", "Brasil"),
478            ("pt", "", "PT") => ("Português", "Portugal"),
479            ("ro", "", "") => ("Română", ""),
480            ("ru", "", "") => ("Русский", ""),
481            ("si", "", "") => ("සිංහල", ""),
482            ("sk", "", "") => ("Slovenčina", ""),
483            ("sl", "", "") => ("Slovenščina", ""),
484            ("sq", "", "") => ("Shqip", ""),
485            ("sr", "", "") => ("Српски", ""),
486            ("sr", "Latn", "") => ("Srpski", ""),
487            ("sv", "", "") => ("Svenska", ""),
488            ("sw", "", "") => ("Kiswahili", ""),
489            ("ta", "", "") => ("தமிழ்", ""),
490            ("te", "", "") => ("తెలుగు", ""),
491            ("th", "", "") => ("ไทย", ""),
492            ("tr", "", "") => ("Türkçe", ""),
493            ("uk", "", "") => ("Українська", ""),
494            ("ur", "", "") => ("اردو", ""),
495            ("uz", "", "") => ("O‘zbek", ""),
496            ("vi", "", "") => ("Tiếng việt", ""),
497            ("zh", "", "") => ("中文", ""),
498            ("zh", "Hans", "") => ("简体中文", ""),
499            ("zh", "Hant", "") => ("繁體中文", ""),
500            ("zh", "", "TW") => ("繁體中文", "台灣"),
501            ("zu", "", "") => ("Isizulu", ""),
502            ("pseudo", "", "") => ("Ƥşeuḓo", ""),
503            ("pseudo", "Mirr", "") => ("Ԁsǝnpo-Wıɹɹoɹǝp", ""),
504            ("pseudo", "Wide", "") => ("Ƥşeeuuḓoo-Ẇiḓee", ""),
505            _ => {
506                if self.0.region.is_some() {
507                    let q = Lang(unic_langid::LanguageIdentifier::from_parts(
508                        self.0.language,
509                        self.0.script,
510                        None,
511                        &[],
512                    ));
513                    return q.autonym();
514                } else {
515                    return None;
516                }
517            }
518        };
519        Some(LangAutonym {
520            language: name,
521            region: if region.is_empty() { None } else { Some(region) },
522        })
523    }
524
525    /// Get ordering of this and `other` as [`fmt::Display`] text.
526    ///
527    /// This method avoids actually allocating a string.
528    pub fn cmp_display(&self, other: &Self) -> std::cmp::Ordering {
529        let (name, region) = self.name_region_str();
530        let (other_name, other_region) = other.name_region_str();
531        name.cmp(other_name).then_with(|| region.cmp(&other_region))
532    }
533
534    fn name_region_str(&self) -> (&str, Option<&str>) {
535        #[cfg(feature = "lang_autonym")]
536        if let Some(a) = self.autonym() {
537            let region = a.region.or_else(|| self.0.region.as_ref().map(|r| r.as_str()));
538            return (a.language, region);
539        }
540        (self.0.language.as_str(), self.0.region.as_ref().map(|r| r.as_str()))
541    }
542}
543impl ops::Deref for Lang {
544    type Target = unic_langid::LanguageIdentifier;
545
546    fn deref(&self) -> &Self::Target {
547        &self.0
548    }
549}
550impl fmt::Debug for Lang {
551    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
552        write!(f, "{}", self.0)
553    }
554}
555/// Prints the autonym if built with `"lang_autonym"` and has one, otherwise prints the tags.
556///
557/// `{}` prints language and region (if set), `{:#}` only prints language.
558impl fmt::Display for Lang {
559    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
560        let (name, region) = self.name_region_str();
561        if f.alternate()
562            && let Some(r) = region
563        {
564            write!(f, "{name} ({r})")
565        } else {
566            write!(f, "{name}")
567        }
568    }
569}
570impl std::str::FromStr for Lang {
571    type Err = unic_langid::LanguageIdentifierError;
572
573    fn from_str(s: &str) -> Result<Self, Self::Err> {
574        let s = s.trim();
575        if s.is_empty() {
576            return Ok(lang!(und));
577        }
578        unic_langid::LanguageIdentifier::from_str(s).map(Lang)
579    }
580}
581
582/// Represents the translated name of a language and region in own language, using the selected script.
583#[derive(Clone, PartialEq, Eq, Hash, PartialOrd, Ord)]
584#[cfg(feature = "lang_autonym")]
585pub struct LangAutonym {
586    /// Language name in its own language.
587    pub language: &'static str,
588    /// Region name in the language.
589    ///
590    /// Can be `None` when a region was requested and only the language name was available.
591    pub region: Option<&'static str>,
592}
593#[cfg(feature = "lang_autonym")]
594impl fmt::Debug for LangAutonym {
595    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
596        if f.alternate() {
597            f.debug_struct("LangAutonym")
598                .field("language", &self.language)
599                .field("region", &self.region)
600                .finish()
601        } else {
602            write!(f, "{self}")
603        }
604    }
605}
606/// `{}` prints language and region (if set), `{:#}` only prints language.
607#[cfg(feature = "lang_autonym")]
608impl fmt::Display for LangAutonym {
609    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
610        write!(f, "{}", self.language)?;
611        if let Some(r) = self.region
612            && !f.alternate()
613        {
614            write!(f, " ({r})")?;
615        }
616        Ok(())
617    }
618}
619
620/// List of languages, in priority order.
621#[derive(Clone, PartialEq, Eq, Default, Hash, serde::Serialize, serde::Deserialize)]
622#[serde(transparent)]
623pub struct Langs(pub Vec<Lang>);
624impl fmt::Debug for Langs {
625    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
626        struct DisplayLangs<'a>(&'a [Lang]);
627        impl fmt::Debug for DisplayLangs<'_> {
628            fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
629                f.debug_list().entries(self.0.iter()).finish()
630            }
631        }
632        if f.alternate() {
633            f.debug_tuple("Langs").field(&DisplayLangs(&self.0)).finish()
634        } else {
635            fmt::Debug::fmt(&DisplayLangs(&self.0), f)
636        }
637    }
638}
639impl Langs {
640    /// The first lang on the list or `und` if the list is empty.
641    pub fn best(&self) -> &Lang {
642        static NONE: Lazy<Lang> = Lazy::new(|| lang!(und));
643        self.first().unwrap_or(&NONE)
644    }
645}
646impl ops::Deref for Langs {
647    type Target = Vec<Lang>;
648
649    fn deref(&self) -> &Self::Target {
650        &self.0
651    }
652}
653impl ops::DerefMut for Langs {
654    fn deref_mut(&mut self) -> &mut Self::Target {
655        &mut self.0
656    }
657}
658impl_from_and_into_var! {
659    fn from(lang: Lang) -> Langs {
660        Langs(vec![lang])
661    }
662    fn from(lang: Option<Lang>) -> Langs {
663        Langs(lang.into_iter().collect())
664    }
665}
666impl fmt::Display for Langs {
667    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
668        let mut sep = "";
669        for l in self.iter() {
670            write!(f, "{sep}{l}")?;
671            sep = ", ";
672        }
673        Ok(())
674    }
675}
676impl std::str::FromStr for Langs {
677    type Err = unic_langid::LanguageIdentifierError;
678
679    fn from_str(s: &str) -> Result<Self, Self::Err> {
680        if s.trim().is_empty() {
681            return Ok(Langs(vec![]));
682        }
683        let mut r = Self(vec![]);
684        for lang in s.split(',') {
685            r.0.push(lang.trim().parse()?)
686        }
687        Ok(r)
688    }
689}
690
691/// Represents a map of [`Lang`] keys that can be partially matched.
692#[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
693#[serde(transparent)]
694pub struct LangMap<V> {
695    inner: Vec<(Lang, V)>,
696}
697impl<V> Default for LangMap<V> {
698    fn default() -> Self {
699        Self { inner: Default::default() }
700    }
701}
702impl<V> LangMap<V> {
703    /// New empty default.
704    pub fn new() -> Self {
705        LangMap::default()
706    }
707
708    /// New empty with pre-allocated capacity.
709    pub fn with_capacity(capacity: usize) -> Self {
710        LangMap {
711            inner: Vec::with_capacity(capacity),
712        }
713    }
714
715    fn exact_i(&self, lang: &Lang) -> Option<usize> {
716        for (i, (key, _)) in self.inner.iter().enumerate() {
717            if key == lang {
718                return Some(i);
719            }
720        }
721        None
722    }
723
724    fn best_i(&self, lang: &Lang) -> Option<usize> {
725        let mut best = None;
726        let mut best_weight = 0;
727
728        for (i, (key, _)) in self.inner.iter().enumerate() {
729            if lang.matches(key, true, true) {
730                let mut weight = 1;
731                let mut eq = 0;
732
733                // must match language
734                if key.language == lang.language {
735                    weight += 200;
736                    eq += 1;
737                }
738                // must match script
739                if key.script == lang.script {
740                    weight += 100;
741                    eq += 1;
742                }
743                // can fallback to different region
744                if key.region == lang.region {
745                    weight += 30;
746                    eq += 1;
747                }
748
749                if eq == 3 && lang.variants().eq(key.variants()) {
750                    // exact match
751                    return Some(i);
752                }
753
754                // human translated of different regions has priority over machine translated
755                if !key.is_machine_translation() {
756                    weight += 60;
757                }
758
759                if best_weight < weight {
760                    best_weight = weight;
761                    best = Some(i);
762                }
763            }
764        }
765
766        best
767    }
768
769    /// Returns the best match to `lang` currently in the map.
770    pub fn best_match(&self, lang: &Lang) -> Option<&Lang> {
771        if let Some(i) = self.best_i(lang) {
772            Some(&self.inner[i].0)
773        } else {
774            None
775        }
776    }
777
778    /// Returns the best match for `lang`.
779    pub fn get(&self, lang: &Lang) -> Option<&V> {
780        if let Some(i) = self.best_i(lang) {
781            Some(&self.inner[i].1)
782        } else {
783            None
784        }
785    }
786
787    /// Returns the exact match for `lang`.
788    pub fn get_exact(&self, lang: &Lang) -> Option<&V> {
789        if let Some(i) = self.exact_i(lang) {
790            Some(&self.inner[i].1)
791        } else {
792            None
793        }
794    }
795
796    /// Returns the best match for `lang`.
797    pub fn get_mut(&mut self, lang: &Lang) -> Option<&mut V> {
798        if let Some(i) = self.best_i(lang) {
799            Some(&mut self.inner[i].1)
800        } else {
801            None
802        }
803    }
804
805    /// Returns the exact match for `lang`.
806    pub fn get_exact_mut(&mut self, lang: &Lang) -> Option<&mut V> {
807        if let Some(i) = self.exact_i(lang) {
808            Some(&mut self.inner[i].1)
809        } else {
810            None
811        }
812    }
813
814    /// Returns the current value or insert `new` and return a reference to it.
815    pub fn get_exact_or_insert(&mut self, lang: Lang, new: impl FnOnce() -> V) -> &mut V {
816        if let Some(i) = self.exact_i(&lang) {
817            return &mut self.inner[i].1;
818        }
819        let i = self.inner.len();
820        self.inner.push((lang, new()));
821        &mut self.inner[i].1
822    }
823
824    /// Insert the value with the exact match of `lang`.
825    ///
826    /// Returns the previous exact match.
827    pub fn insert(&mut self, lang: Lang, value: V) -> Option<V> {
828        if let Some(i) = self.exact_i(&lang) {
829            Some(mem::replace(&mut self.inner[i].1, value))
830        } else {
831            self.inner.push((lang, value));
832            None
833        }
834    }
835
836    /// Remove the exact match of `lang`.
837    pub fn remove(&mut self, lang: &Lang) -> Option<V> {
838        if let Some(i) = self.exact_i(lang) {
839            Some(self.inner.swap_remove(i).1)
840        } else {
841            None
842        }
843    }
844
845    /// Remove all exact and partial matches of `lang`.
846    ///
847    /// Returns a count of items removed.
848    pub fn remove_all(&mut self, lang: &Lang) -> usize {
849        let mut count = 0;
850        self.inner.retain(|(key, _)| {
851            let rmv = lang.matches(key, true, false);
852            if rmv {
853                count += 1
854            }
855            !rmv
856        });
857        count
858    }
859
860    /// Remove the last inserted lang.
861    pub fn pop(&mut self) -> Option<(Lang, V)> {
862        self.inner.pop()
863    }
864
865    /// Returns if the map is empty.
866    pub fn is_empty(&self) -> bool {
867        self.inner.is_empty()
868    }
869
870    /// Returns the number of languages in the map.
871    pub fn len(&self) -> usize {
872        self.inner.len()
873    }
874
875    /// Remove all entries.
876    pub fn clear(&mut self) {
877        self.inner.clear()
878    }
879
880    /// Iterator over lang keys.
881    pub fn keys(&self) -> impl std::iter::ExactSizeIterator<Item = &Lang> {
882        self.inner.iter().map(|(k, _)| k)
883    }
884
885    /// Iterator over values.
886    pub fn values(&self) -> impl std::iter::ExactSizeIterator<Item = &V> {
887        self.inner.iter().map(|(_, v)| v)
888    }
889
890    /// Iterator over values.
891    pub fn values_mut(&mut self) -> impl std::iter::ExactSizeIterator<Item = &mut V> {
892        self.inner.iter_mut().map(|(_, v)| v)
893    }
894
895    /// Into iterator of values.
896    pub fn into_values(self) -> impl std::iter::ExactSizeIterator<Item = V> {
897        self.inner.into_iter().map(|(_, v)| v)
898    }
899
900    /// Iterate over key-value pairs.
901    pub fn iter(&self) -> impl std::iter::ExactSizeIterator<Item = (&Lang, &V)> {
902        self.inner.iter().map(|(k, v)| (k, v))
903    }
904
905    /// Iterate over key-value pairs with mutable values.
906    pub fn iter_mut(&mut self) -> impl std::iter::ExactSizeIterator<Item = (&Lang, &mut V)> {
907        self.inner.iter_mut().map(|(k, v)| (&*k, v))
908    }
909
910    /// Shrink capacity as much as possible.
911    pub fn shrink_to_fit(&mut self) {
912        self.inner.shrink_to_fit();
913    }
914}
915impl<V> LangMap<HashMap<LangFilePath, V>> {
916    /// Returns the match for `lang` and `file`.
917    pub fn get_file(&self, lang: &Lang, file: &LangFilePath) -> Option<&V> {
918        let files = self.get(lang)?;
919        if let Some(exact) = files.get(file) {
920            return Some(exact);
921        }
922        Self::best_file(files, file).map(|(_, v)| v)
923    }
924
925    /// Returns the best match to `lang` and `file` currently in the map.
926    pub fn best_file_match(&self, lang: &Lang, file: &LangFilePath) -> Option<&LangFilePath> {
927        let files = self.get(lang)?;
928        if let Some((exact, _)) = files.get_key_value(file) {
929            return Some(exact);
930        }
931        Self::best_file(files, file).map(|(k, _)| k)
932    }
933
934    fn best_file<'a>(files: &'a HashMap<LangFilePath, V>, file: &LangFilePath) -> Option<(&'a LangFilePath, &'a V)> {
935        let mut best = None;
936        let mut best_dist = u64::MAX;
937        for (k, v) in files {
938            if let Some(d) = k.matches(file)
939                && d < best_dist
940            {
941                best = Some((k, v));
942                best_dist = d;
943            }
944        }
945        best
946    }
947}
948impl<V> IntoIterator for LangMap<V> {
949    type Item = (Lang, V);
950
951    type IntoIter = std::vec::IntoIter<(Lang, V)>;
952
953    fn into_iter(self) -> Self::IntoIter {
954        self.inner.into_iter()
955    }
956}
957impl<V: PartialEq> PartialEq for LangMap<V> {
958    fn eq(&self, other: &Self) -> bool {
959        if self.len() != other.len() {
960            return false;
961        }
962        for (k, v) in &self.inner {
963            if other.get_exact(k) != Some(v) {
964                return false;
965            }
966        }
967        true
968    }
969}
970impl<V: Eq> Eq for LangMap<V> {}
971
972/// Errors found parsing a fluent resource file.
973#[derive(Clone, Debug)]
974pub struct FluentParserErrors(pub Vec<fluent_syntax::parser::ParserError>);
975impl fmt::Display for FluentParserErrors {
976    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
977        let mut sep = "";
978        for e in &self.0 {
979            write!(f, "{sep}{e}")?;
980            sep = "\n";
981        }
982        Ok(())
983    }
984}
985impl std::error::Error for FluentParserErrors {
986    fn source(&self) -> Option<&(dyn std::error::Error + 'static)> {
987        if self.0.len() == 1 { Some(&self.0[0]) } else { None }
988    }
989}
990
991/// Localization resource file path in the localization directory.
992///
993/// In the default directory layout, localization dependencies are collected using `cargo zng l10n`
994/// and copied to `l10n/{lang}/deps/{name}/{version}/`, and localization for the app ([`is_current_app`])
995/// is placed in `l10n/{lang}/`.
996///
997/// [`is_current_app`]: Self::is_current_app
998#[derive(Debug, Clone, serde::Serialize, serde::Deserialize)]
999#[non_exhaustive]
1000pub struct LangFilePath {
1001    /// Package name.
1002    pub pkg_name: Txt,
1003    /// Package version.
1004    pub pkg_version: Version,
1005    /// The localization file name, without extension.
1006    pub file: Txt,
1007}
1008impl Ord for LangFilePath {
1009    fn cmp(&self, other: &Self) -> std::cmp::Ordering {
1010        let self_pkg = self.actual_pkg_data();
1011        let other_pkg = other.actual_pkg_data();
1012        match self_pkg.0.cmp(other_pkg.0) {
1013            core::cmp::Ordering::Equal => {}
1014            ord => return ord,
1015        }
1016        match self_pkg.1.cmp(other_pkg.1) {
1017            core::cmp::Ordering::Equal => {}
1018            ord => return ord,
1019        }
1020        self.file().cmp(&other.file())
1021    }
1022}
1023impl PartialOrd for LangFilePath {
1024    fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
1025        Some(self.cmp(other))
1026    }
1027}
1028impl std::hash::Hash for LangFilePath {
1029    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
1030        self.actual_pkg_data().hash(state);
1031        self.file().hash(state);
1032    }
1033}
1034impl Eq for LangFilePath {}
1035impl PartialEq for LangFilePath {
1036    fn eq(&self, other: &Self) -> bool {
1037        self.actual_pkg_data() == other.actual_pkg_data() && self.file() == other.file()
1038    }
1039}
1040impl LangFilePath {
1041    /// New from package name, version and file.
1042    pub fn new(pkg_name: impl Into<Txt>, pkg_version: Version, file: impl Into<Txt>) -> Self {
1043        let r = Self {
1044            pkg_name: pkg_name.into(),
1045            pkg_version,
1046            file: file.into(),
1047        };
1048        // these non-standard names are matched on fallback, but they can cause duplicate caching
1049        debug_assert!(
1050            r.file
1051                .rsplit_once('.')
1052                .map(|(_, ext)| !ext.eq_ignore_ascii_case("ftl"))
1053                .unwrap_or(true),
1054            "file `{}` must not have extension",
1055            r.file
1056        );
1057        debug_assert!(r.file != "_", "file `_` should be an empty string");
1058        r
1059    }
1060
1061    /// Gets a file in the current app.
1062    ///
1063    /// This value indicates that the localization resources are directly on the `l10n/{lang?}/` directories, not
1064    /// in the dependencies directories.
1065    ///
1066    /// See [`zng_env::about()`] for more details.
1067    pub fn current_app(file: impl Into<Txt>) -> LangFilePath {
1068        let about = zng_env::about();
1069        Self::new(about.pkg_name.clone(), about.version.clone(), file.into())
1070    }
1071
1072    /// Gets if this file is in the [`current_app`] resources, or the `pkg_name` is empty or the `pkg_version.pre` is `#.#.#-local`.
1073    ///
1074    /// [`current_app`]: Self::current_app
1075    pub fn is_current_app(&self) -> bool {
1076        self.is_current_app_no_check() || {
1077            let about = zng_env::about();
1078            self.pkg_name == about.pkg_name && self.pkg_version == about.version
1079        }
1080    }
1081
1082    fn is_current_app_no_check(&self) -> bool {
1083        self.pkg_name.is_empty() || self.pkg_version.pre.as_str() == "local"
1084    }
1085
1086    fn actual_pkg_data(&self) -> (&Txt, &Version) {
1087        if self.is_current_app_no_check() {
1088            let about = zng_env::about();
1089            (&about.pkg_name, &about.version)
1090        } else {
1091            (&self.pkg_name, &self.pkg_version)
1092        }
1093    }
1094
1095    /// Gets the normalized package name.
1096    ///
1097    /// This is the app package name if [`is_current_app`], otherwise is just the `pkg_name` value.
1098    ///
1099    /// [`is_current_app`]: Self::is_current_app
1100    pub fn pkg_name(&self) -> Txt {
1101        self.actual_pkg_data().0.clone()
1102    }
1103
1104    /// Gets the normalized package version.
1105    ///
1106    /// This is the app version if [`is_current_app`], otherwise is just the `pkg_version` value.
1107    ///
1108    /// [`is_current_app`]: Self::is_current_app
1109    pub fn pkg_version(&self) -> Version {
1110        self.actual_pkg_data().1.clone()
1111    }
1112
1113    /// Gets the normalized file name.
1114    ///
1115    /// This `"_"` for empty file or the file.
1116    pub fn file(&self) -> Txt {
1117        if self.file.is_empty() {
1118            Txt::from_char('_')
1119        } else {
1120            self.file.clone()
1121        }
1122    }
1123
1124    /// Get the file path, relative to the localization dir.
1125    ///
1126    /// * Empty file name is the same as `_`.
1127    /// * If package [`is_current_app`] gets `{lang}/{file}.ftl`.
1128    /// * Else if is another package gets `{lang}/deps/{pkg_name}/{pkg_version}/{file}.ftl`.
1129    ///
1130    /// [`is_current_app`]: Self::is_current_app
1131    pub fn to_path(&self, lang: &Lang) -> PathBuf {
1132        let mut file = self.file.as_str();
1133        if file.is_empty() {
1134            file = "_";
1135        }
1136        if self.is_current_app() {
1137            format!("{lang}/{file}.ftl")
1138        } else {
1139            format!("{lang}/deps/{}/{}/{file}.ftl", self.pkg_name, self.pkg_version)
1140        }
1141        .into()
1142    }
1143
1144    /// Gets a value that indicates if the resources represented by `self` can be used for `search`.
1145    ///
1146    /// The number indicates the quality of the match:
1147    ///
1148    /// * `0` is an exact match.
1149    /// * `b1` is a match with only version `build` differences.
1150    /// * `b10` is a match with only version `pre` differences.
1151    /// * `(0..u16::MAX) << 16` is a match with only `patch` differences and the absolute distance.
1152    /// * `(0..u16::MAX) << 16 * 2` is a match with `minor` differences and the absolute distance.
1153    /// * `(0..u16::MAX) << 16 * 3` is a match with `major` differences and the absolute distance.
1154    /// * `None`` is a `pkg_name` mismatch.
1155    pub fn matches(&self, search: &Self) -> Option<u64> {
1156        let (self_name, self_version) = self.actual_pkg_data();
1157        let (search_name, search_version) = search.actual_pkg_data();
1158
1159        if self_name != search_name {
1160            return None;
1161        }
1162
1163        if self.file != search.file {
1164            let file_a = self.file.rsplit_once('.').map(|t| t.0).unwrap_or(self.file.as_str());
1165            let file_b = search.file.rsplit_once('.').map(|t| t.0).unwrap_or(search.file.as_str());
1166            if file_a != file_b {
1167                let is_empty_a = file_a == "_" || file_a.is_empty();
1168                let is_empty_b = file_b == "_" || file_b.is_empty();
1169                if !(is_empty_a && is_empty_b) {
1170                    return None;
1171                }
1172            }
1173            tracing::warn!(
1174                "fallback matching `{}` with `{}`, file was not expected to have extension",
1175                self.file,
1176                search.file
1177            )
1178        }
1179
1180        fn dist(a: u64, b: u64, shift: u64) -> u64 {
1181            let (l, s) = match a.cmp(&b) {
1182                std::cmp::Ordering::Equal => return 0,
1183                std::cmp::Ordering::Less => (b, a),
1184                std::cmp::Ordering::Greater => (a, b),
1185            };
1186
1187            (l - s).min(u16::MAX as u64) << (16 * shift)
1188        }
1189
1190        let mut d = 0;
1191        if self_version.build != search_version.build {
1192            d = 1;
1193        }
1194        if self_version.pre != search_version.pre {
1195            d |= 0b10;
1196        }
1197
1198        d |= dist(self_version.patch, search_version.patch, 1);
1199        d |= dist(self_version.minor, search_version.minor, 2);
1200        d |= dist(self_version.major, search_version.major, 3);
1201
1202        Some(d)
1203    }
1204}
1205impl_from_and_into_var! {
1206    fn from(file: Txt) -> LangFilePath {
1207        LangFilePath::current_app(file)
1208    }
1209
1210    fn from(file: &'static str) -> LangFilePath {
1211        LangFilePath::current_app(file)
1212    }
1213
1214    fn from(file: String) -> LangFilePath {
1215        LangFilePath::current_app(file)
1216    }
1217}
1218
1219#[cfg(test)]
1220mod tests {
1221    use super::*;
1222
1223    #[test]
1224    fn file_matches() {
1225        fn check(a: &str, b: &str, c: &str) {
1226            let ap = LangFilePath::new("name", a.parse().unwrap(), "file");
1227            let bp = LangFilePath::new("name", b.parse().unwrap(), "file");
1228            let cp = LangFilePath::new("name", c.parse().unwrap(), "file");
1229
1230            let ab = ap.matches(&bp);
1231            let ac = ap.matches(&cp);
1232
1233            assert!(ab < ac, "expected {a}.matches({b}) < {a}.matches({c})")
1234        }
1235
1236        check("0.0.0", "0.0.1", "0.1.0");
1237        check("0.0.1", "0.1.0", "1.0.0");
1238        check("0.0.0-pre", "0.0.0-pre+build", "0.0.0-other+build");
1239        check("0.0.0+build", "0.0.0+build", "0.0.0+other");
1240        check("0.0.1", "0.0.2", "0.0.3");
1241        check("0.1.0", "0.2.0", "0.3.0");
1242        check("1.0.0", "2.0.0", "3.0.0");
1243        check("1.0.0", "1.1.0", "2.0.0");
1244    }
1245
1246    #[test]
1247    fn file_name_mismatches() {
1248        let ap = LangFilePath::new("name", "1.0.0".parse().unwrap(), "file-a");
1249        let bp = LangFilePath::new("name", "1.0.0".parse().unwrap(), "file-b");
1250
1251        assert!(ap.matches(&bp).is_none());
1252    }
1253}
zng_ext_l10n/types.rs

zng_ext_l10n/
types.rs
