Skip to main content

sqlite_graphrag/
i18n.rs

1//! Bilingual human-readable message layer.
2//!
3//! The CLI uses `--lang en|pt` (global flag) or `SQLITE_GRAPHRAG_LANG` (env var) to choose
4//! the language of stderr progress messages. JSON stdout is deterministic and identical
5//! across languages — only strings intended for humans pass through this module.
6//!
7//! Detection (highest to lowest priority):
8//! 1. Explicit `--lang` flag
9//! 2. Env var `SQLITE_GRAPHRAG_LANG`
10//! 3. OS locale (`LANG`, `LC_ALL`) with `pt` prefix
11//! 4. Fallback `English`
12
13use std::sync::OnceLock;
14
15#[derive(Copy, Clone, Debug, PartialEq, Eq, clap::ValueEnum)]
16pub enum Language {
17    #[value(name = "en", aliases = ["english", "EN"])]
18    English,
19    #[value(name = "pt", aliases = ["portugues", "portuguese", "pt-BR", "pt-br", "PT"])]
20    Portuguese,
21}
22
23impl Language {
24    /// Parses a command-line string into a `Language` without relying on clap.
25    /// Accepts the same aliases defined in `#[value(...)]`: "en", "pt", etc.
26    pub fn from_str_opt(s: &str) -> Option<Self> {
27        match s.to_lowercase().as_str() {
28            "en" | "english" => Some(Language::English),
29            "pt" | "pt-br" | "portugues" | "portuguese" => Some(Language::Portuguese),
30            _ => None,
31        }
32    }
33
34    pub fn from_env_or_locale() -> Self {
35        // Priority 1: explicit SQLITE_GRAPHRAG_LANG env var (highest precedence).
36        // Empty string treated as unset per POSIX convention.
37        if let Ok(v) = std::env::var("SQLITE_GRAPHRAG_LANG") {
38            if !v.is_empty() {
39                let lower = v.to_lowercase();
40                if lower.starts_with("pt") {
41                    return Language::Portuguese;
42                }
43                if lower.starts_with("en") {
44                    return Language::English;
45                }
46                tracing::warn!(target: "i18n",
47                    value = %v,
48                    "SQLITE_GRAPHRAG_LANG value not recognized, falling back to locale detection"
49                );
50            }
51        }
52        // Priority 2: POSIX locale precedence LC_ALL > LC_MESSAGES > LANG.
53        // We read these via std::env (not via sys_locale) because:
54        // (a) `sys_locale::get_locale()` calls into native OS APIs (CFLocaleCopyCurrent
55        //     on macOS, GetUserDefaultLocaleName on Windows) which cache the
56        //     system locale and IGNORE env vars set at runtime by tests;
57        // (b) POSIX specifies LC_ALL > LC_MESSAGES > LANG ordering and an
58        //     unrecognised LC_ALL value must stop iteration (fall back to
59        //     English default).
60        for var in ["LC_ALL", "LC_MESSAGES", "LANG"] {
61            if let Ok(v) = std::env::var(var) {
62                if v.is_empty() {
63                    continue;
64                }
65                let lower = v.to_lowercase();
66                if lower.starts_with("pt") {
67                    return Language::Portuguese;
68                }
69                if lower.starts_with("en") {
70                    return Language::English;
71                }
72                // Unrecognised value in a higher-precedence variable stops
73                // iteration per POSIX.1-2017 §8.2.
74                if var == "LC_ALL" {
75                    return Language::English;
76                }
77            }
78        }
79        // Priority 3: cross-platform locale detection via native OS APIs.
80        // Only reached when no POSIX env var is set.
81        if let Some(locale) = sys_locale::get_locale() {
82            let lower = locale.to_lowercase();
83            if lower.starts_with("pt") {
84                return Language::Portuguese;
85            }
86            if lower.starts_with("en") {
87                return Language::English;
88            }
89        }
90        Language::English
91    }
92}
93
94static GLOBAL_LANGUAGE: OnceLock<Language> = OnceLock::new();
95
96/// Initializes the global language. Subsequent calls are silently ignored
97/// (OnceLock semantics) — guaranteeing thread-safety and determinism.
98///
99/// v1.0.36 (L6): early-return when already initialized so the env-fallback
100/// resolver (`from_env_or_locale`) does not run a second time. Without this
101/// guard, calling `init(None)` after `current()` already populated the
102/// OnceLock causes `from_env_or_locale` to fire its `tracing::warn!` twice
103/// for unrecognized `SQLITE_GRAPHRAG_LANG` values.
104pub fn init(explicit: Option<Language>) {
105    if GLOBAL_LANGUAGE.get().is_some() {
106        return;
107    }
108    let resolved = explicit.unwrap_or_else(Language::from_env_or_locale);
109    let _ = GLOBAL_LANGUAGE.set(resolved);
110}
111
112/// Returns the active language, or fallback English if `init` was never called.
113pub fn current() -> Language {
114    *GLOBAL_LANGUAGE.get_or_init(Language::from_env_or_locale)
115}
116
117/// Translates a bilingual message by selecting the active variant.
118///
119/// v1.0.36 (M4): inputs are constrained to `&'static str` so the function
120/// can return one of them directly without `Box::leak`. The previous
121/// implementation leaked one allocation per call which accumulated in
122/// long-running pipelines; this version is allocation-free. All in-tree
123/// callers already pass string literals, which are `&'static str`.
124pub fn tr(en: &'static str, pt: &'static str) -> &'static str {
125    match current() {
126        Language::English => en,
127        Language::Portuguese => pt,
128    }
129}
130
131/// Progress message emitted after pruning relationships.
132///
133/// English-only: this string is emitted to stderr as a progress notice and
134/// does not vary by language because the prune-relations command targets
135/// agent-first pipelines where deterministic output matters.
136pub fn relations_pruned(count: usize, relation: &str, namespace: &str) -> String {
137    format!("pruned {count} '{relation}' relationships in namespace '{namespace}'")
138}
139
140/// Progress message for dry-run preview of prune-relations.
141///
142/// English-only: emitted to stderr as a progress notice.
143pub fn prune_dry_run(count: usize, relation: &str) -> String {
144    format!("dry run: {count} '{relation}' relationships would be removed")
145}
146
147/// Warning message when --yes is not passed for destructive prune-relations.
148///
149/// English-only: emitted to stderr as a progress notice.
150pub fn prune_requires_yes() -> String {
151    "destructive operation requires --yes flag; use --dry-run to preview".to_string()
152}
153
154/// Localized prefix for error messages displayed to the end user.
155pub fn error_prefix() -> &'static str {
156    match current() {
157        Language::English => "Error",
158        Language::Portuguese => "Erro",
159    }
160}
161
162/// Error messages for `AppError` variants — always English.
163///
164/// These strings end up inside `AppError` inner fields and may appear in
165/// deterministic JSON stdout (e.g. ingest NDJSON). Portuguese translations
166/// for stderr live in `pub mod app_error_pt` and are applied by
167/// `localized_message_for(Language::Portuguese)`.
168pub mod errors_msg {
169    pub fn memory_not_found(nome: &str, namespace: &str) -> String {
170        format!("memory '{nome}' not found in namespace '{namespace}'")
171    }
172
173    pub fn memory_or_entity_not_found(name: &str, namespace: &str) -> String {
174        format!("memory or entity '{name}' not found in namespace '{namespace}'")
175    }
176
177    pub fn database_not_found(path: &str) -> String {
178        format!("database not found at {path}. Run 'sqlite-graphrag init' first.")
179    }
180
181    pub fn entity_not_found(nome: &str, namespace: &str) -> String {
182        format!("entity \"{nome}\" does not exist in namespace \"{namespace}\"")
183    }
184
185    pub fn relationship_not_found(de: &str, rel: &str, para: &str, namespace: &str) -> String {
186        format!(
187            "relationship \"{de}\" --[{rel}]--> \"{para}\" does not exist in namespace \"{namespace}\""
188        )
189    }
190
191    pub fn duplicate_memory(nome: &str, namespace: &str) -> String {
192        format!(
193            "memory '{nome}' already exists in namespace '{namespace}'. Use --force-merge to update."
194        )
195    }
196
197    pub fn duplicate_memory_soft_deleted(name: &str, namespace: &str) -> String {
198        format!(
199            "memory '{name}' exists but is soft-deleted in namespace '{namespace}'; \
200             use --force-merge to restore and update, or `restore` to revive it"
201        )
202    }
203
204    pub fn optimistic_lock_conflict(expected: i64, current_ts: i64) -> String {
205        format!(
206            "optimistic lock conflict: expected updated_at={expected}, but current is {current_ts}"
207        )
208    }
209
210    pub fn version_not_found(versao: i64, nome: &str) -> String {
211        format!("version {versao} not found for memory '{nome}'")
212    }
213
214    pub fn no_recall_results(max_distance: f32, query: &str, namespace: &str) -> String {
215        format!(
216            "no results within --max-distance {max_distance} for query '{query}' in namespace '{namespace}'"
217        )
218    }
219
220    pub fn soft_deleted_memory_not_found(nome: &str, namespace: &str) -> String {
221        format!("soft-deleted memory '{nome}' not found in namespace '{namespace}'")
222    }
223
224    pub fn concurrent_process_conflict() -> String {
225        "optimistic lock conflict: memory was modified by another process".to_string()
226    }
227
228    pub fn entity_limit_exceeded(max: usize) -> String {
229        format!("entities exceed limit of {max}")
230    }
231
232    pub fn relationship_limit_exceeded(max: usize) -> String {
233        format!("relationships exceed limit of {max}")
234    }
235}
236
237/// Localized validation messages for memory fields.
238pub mod validation {
239    use super::current;
240    use crate::i18n::Language;
241
242    pub fn name_length(max: usize) -> String {
243        match current() {
244            Language::English => format!("name must be 1-{max} chars"),
245            Language::Portuguese => format!("nome deve ter entre 1 e {max} caracteres"),
246        }
247    }
248
249    pub fn reserved_name() -> String {
250        match current() {
251            Language::English => {
252                "names and namespaces starting with __ are reserved for internal use".to_string()
253            }
254            Language::Portuguese => {
255                "nomes e namespaces iniciados com __ são reservados para uso interno".to_string()
256            }
257        }
258    }
259
260    pub fn name_kebab(nome: &str) -> String {
261        match current() {
262            Language::English => format!(
263                "name must be kebab-case slug (lowercase letters, digits, hyphens): '{nome}'"
264            ),
265            Language::Portuguese => {
266                format!("nome deve estar em kebab-case (minúsculas, dígitos, hífens): '{nome}'")
267            }
268        }
269    }
270
271    pub fn description_exceeds(max: usize) -> String {
272        match current() {
273            Language::English => format!("description must be <= {max} chars"),
274            Language::Portuguese => format!("descrição deve ter no máximo {max} caracteres"),
275        }
276    }
277
278    pub fn body_exceeds(max: usize) -> String {
279        match current() {
280            Language::English => format!("body exceeds {max} bytes"),
281            Language::Portuguese => format!("corpo excede {max} bytes"),
282        }
283    }
284
285    pub fn new_name_length(max: usize) -> String {
286        match current() {
287            Language::English => format!("new-name must be 1-{max} chars"),
288            Language::Portuguese => format!("novo nome deve ter entre 1 e {max} caracteres"),
289        }
290    }
291
292    pub fn new_name_kebab(nome: &str) -> String {
293        match current() {
294            Language::English => format!(
295                "new-name must be kebab-case slug (lowercase letters, digits, hyphens): '{nome}'"
296            ),
297            Language::Portuguese => format!(
298                "novo nome deve estar em kebab-case (minúsculas, dígitos, hífens): '{nome}'"
299            ),
300        }
301    }
302
303    pub fn namespace_length() -> String {
304        match current() {
305            Language::English => "namespace must be 1-80 chars".to_string(),
306            Language::Portuguese => "namespace deve ter entre 1 e 80 caracteres".to_string(),
307        }
308    }
309
310    pub fn namespace_format() -> String {
311        match current() {
312            Language::English => "namespace must be alphanumeric + hyphens/underscores".to_string(),
313            Language::Portuguese => {
314                "namespace deve ser alfanumérico com hífens/sublinhados".to_string()
315            }
316        }
317    }
318
319    pub fn path_traversal(p: &str) -> String {
320        match current() {
321            Language::English => format!("path traversal rejected: {p}"),
322            Language::Portuguese => format!("traversal de caminho rejeitado: {p}"),
323        }
324    }
325
326    pub fn invalid_tz(v: &str) -> String {
327        match current() {
328            Language::English => format!(
329                "SQLITE_GRAPHRAG_DISPLAY_TZ invalid: '{v}'; use an IANA name like 'America/Sao_Paulo'"
330            ),
331            Language::Portuguese => format!(
332                "SQLITE_GRAPHRAG_DISPLAY_TZ inválido: '{v}'; use um nome IANA como 'America/Sao_Paulo'"
333            ),
334        }
335    }
336
337    pub fn empty_query() -> String {
338        match current() {
339            Language::English => "query cannot be empty".to_string(),
340            Language::Portuguese => "a consulta não pode estar vazia".to_string(),
341        }
342    }
343
344    pub fn empty_body() -> String {
345        match current() {
346            Language::English => "body cannot be empty: provide --body, --body-file, or --body-stdin with content, or supply a graph via --entities-file/--graph-stdin".to_string(),
347            Language::Portuguese => "o corpo não pode estar vazio: forneça --body, --body-file ou --body-stdin com conteúdo, ou um grafo via --entities-file/--graph-stdin".to_string(),
348        }
349    }
350
351    pub fn invalid_namespace_config(path: &str, err: &str) -> String {
352        match current() {
353            Language::English => {
354                format!("invalid project namespace config '{path}': {err}")
355            }
356            Language::Portuguese => {
357                format!("configuração de namespace de projeto inválida '{path}': {err}")
358            }
359        }
360    }
361
362    pub fn invalid_projects_mapping(path: &str, err: &str) -> String {
363        match current() {
364            Language::English => format!("invalid projects mapping '{path}': {err}"),
365            Language::Portuguese => format!("mapeamento de projetos inválido '{path}': {err}"),
366        }
367    }
368
369    pub fn self_referential_link() -> String {
370        match current() {
371            Language::English => "--from and --to must be different entities — self-referential relationships are not supported".to_string(),
372            Language::Portuguese => "--from e --to devem ser entidades diferentes — relacionamentos auto-referenciais não são suportados".to_string(),
373        }
374    }
375
376    pub fn invalid_link_weight(weight: f64) -> String {
377        match current() {
378            Language::English => {
379                format!("--weight: must be between 0.0 and 1.0 (actual: {weight})")
380            }
381            Language::Portuguese => {
382                format!("--weight: deve estar entre 0.0 e 1.0 (atual: {weight})")
383            }
384        }
385    }
386
387    pub fn sync_destination_equals_source() -> String {
388        match current() {
389            Language::English => {
390                "destination path must differ from the source database path".to_string()
391            }
392            Language::Portuguese => {
393                "caminho de destino deve ser diferente do caminho do banco de dados fonte"
394                    .to_string()
395            }
396        }
397    }
398
399    /// Portuguese translations for `AppError` Display messages.
400    ///
401    /// Each helper mirrors a single `AppError` variant's `#[error(...)]` text in
402    /// Portuguese, keeping the language barrier confined to this module. The
403    /// English source of truth lives in `src/errors.rs` via `thiserror`.
404    pub mod app_error_pt {
405        pub fn validation(msg: &str) -> String {
406            format!("erro de validação: {msg}")
407        }
408
409        pub fn duplicate(msg: &str) -> String {
410            let translated = msg
411                .replace("already exists in namespace", "já existe no namespace")
412                .replace(
413                    "exists but is soft-deleted in namespace",
414                    "existe mas está excluída temporariamente no namespace",
415                )
416                .replace(
417                    "Use --force-merge to update.",
418                    "Use --force-merge para atualizar.",
419                )
420                .replace(
421                    "use --force-merge to restore and update, or `restore` to revive it",
422                    "use --force-merge para restaurar e atualizar, ou `restore` para revivê-la",
423                )
424                .replace("memory", "memória");
425            format!("duplicata detectada: {translated}")
426        }
427
428        pub fn conflict(msg: &str) -> String {
429            let translated = msg
430                .replace("optimistic lock conflict", "conflito de lock otimista")
431                .replace("but current is", "mas atual é")
432                .replace(
433                    "was modified by another process",
434                    "foi modificada por outro processo",
435                );
436            format!("conflito: {translated}")
437        }
438
439        pub fn not_found(msg: &str) -> String {
440            // G55 T3: add replacements for the read.rs format produced by the
441            // T1 fix: `memory not found: name='X' in namespace 'Y'`.
442            // The existing chain did not catch ` in namespace '` when broken
443            // by the name label, leaving a bilingual hybrid. New patterns
444            // must run BEFORE the catch-all `memory` → `memória` to avoid
445            // being shadowed.
446            let translated = msg
447                .replace("memory not found:", "memória não encontrada:")
448                .replace("not found in namespace", "não encontrada no namespace")
449                .replace("not found for memory", "não encontrada para memória")
450                .replace("does not exist in namespace", "não existe no namespace")
451                .replace("memory or entity", "memória ou entidade")
452                .replace("name='", "nome='")
453                .replace("memory", "memória")
454                .replace("entity", "entidade")
455                .replace(" in namespace '", " no namespace '")
456                .replace("version", "versão")
457                .replace("soft-deleted", "excluída temporariamente");
458            format!("não encontrado: {translated}")
459        }
460
461        // G55 S2 (v1.0.80): structured variant helpers. They synthesize the
462        // canonical English message and feed it through the `not_found`
463        // replace-chain so the pt-BR translation stays in one place.
464        pub fn memory_not_found(name: &str, namespace: &str) -> String {
465            not_found(&format!(
466                "memory not found: name='{name}' in namespace '{namespace}'"
467            ))
468        }
469
470        pub fn memory_not_found_by_id(id: i64) -> String {
471            not_found(&format!("memory not found: id={id}"))
472        }
473
474        pub fn namespace_error(msg: &str) -> String {
475            format!("namespace não resolvido: {msg}")
476        }
477
478        pub fn limit_exceeded(msg: &str) -> String {
479            let translated = msg
480                .replace("exceeds limit of", "excede limite de")
481                .replace("body exceeds", "corpo excede")
482                .replace("entities exceed limit", "entidades excedem limite")
483                .replace(
484                    "relationships exceed limit",
485                    "relacionamentos excedem limite",
486                );
487            format!("limite excedido: {translated}")
488        }
489
490        pub fn database(err: &str) -> String {
491            format!("erro de banco de dados: {err}")
492        }
493
494        pub fn embedding(msg: &str) -> String {
495            format!("erro de embedding: {msg}")
496        }
497
498        pub fn vec_extension(msg: &str) -> String {
499            format!("extensão sqlite-vec falhou: {msg}")
500        }
501
502        pub fn db_busy(msg: &str) -> String {
503            format!("banco ocupado: {msg}")
504        }
505
506        pub fn batch_partial_failure(total: usize, failed: usize) -> String {
507            format!("falha parcial em batch: {failed} de {total} itens falharam")
508        }
509
510        pub fn io(err: &str) -> String {
511            format!("erro de I/O: {err}")
512        }
513
514        pub fn internal(err: &str) -> String {
515            format!("erro interno: {err}")
516        }
517
518        pub fn json(err: &str) -> String {
519            format!("erro de JSON: {err}")
520        }
521
522        pub fn lock_busy(msg: &str) -> String {
523            format!("lock ocupado: {msg}")
524        }
525
526        pub fn all_slots_full(max: usize, waited_secs: u64) -> String {
527            format!(
528                "todos os {max} slots de concorrência ocupados após aguardar {waited_secs}s \
529                 (exit 75); use --max-concurrency ou aguarde outras invocações terminarem"
530            )
531        }
532
533        pub fn job_singleton_locked(job_type: &str, namespace: &str) -> String {
534            format!(
535                "job {job_type} para o namespace '{namespace}' já está em execução (exit 75); \
536                 aguarde a conclusão ou passe --wait-job-singleton <SEGUNDOS>"
537            )
538        }
539
540        pub fn embedding_singleton_locked(namespace: &str) -> String {
541            format!(
542                "singleton de embedding para o namespace '{namespace}' já está retido (exit 75); \
543                 outra CLI está chamando o LLM neste banco; passe --wait-embed-singleton <SEGUNDOS> para aguardar"
544            )
545        }
546
547        pub fn low_memory(available_mb: u64, required_mb: u64) -> String {
548            format!(
549                "memória disponível ({available_mb}MB) abaixo do mínimo requerido ({required_mb}MB) \
550                 para carregar o modelo; aborte outras cargas ou use --skip-memory-guard (exit 77)"
551            )
552        }
553
554        pub fn shutdown(signal: &str) -> String {
555            format!(
556                "sinal de desligamento recebido: {signal}; operação cancelada pelo usuário (exit 19)"
557            )
558        }
559
560        pub fn binary_not_found(name: &str) -> String {
561            format!("binário não encontrado: {name} — instale e adicione ao PATH")
562        }
563
564        pub fn rate_limited(detail: &str) -> String {
565            format!("taxa de requisição excedida: {detail}")
566        }
567
568        pub fn timeout(operation: &str, secs: u64) -> String {
569            format!("timeout após {secs}s: {operation}")
570        }
571    }
572
573    /// Portuguese translations for runtime startup messages emitted from `main.rs`.
574    ///
575    /// These mirror the English text supplied alongside each call to
576    /// `output::emit_progress_i18n` / `output::emit_error_i18n`, keeping the PT
577    /// strings confined to this module per the language policy.
578    pub mod runtime_pt {
579        pub fn embedding_heavy_must_measure_ram() -> String {
580            "comando intensivo em embedding precisa medir RAM disponível".to_string()
581        }
582
583        pub fn heavy_command_detected(available_mb: u64, safe_concurrency: usize) -> String {
584            format!(
585                "Comando pesado detectado; memória disponível: {available_mb} MB; \
586                 concorrência segura: {safe_concurrency}"
587            )
588        }
589
590        pub fn reducing_concurrency(
591            requested_concurrency: usize,
592            effective_concurrency: usize,
593        ) -> String {
594            format!(
595                "Reduzindo a concorrência solicitada de {requested_concurrency} para \
596                 {effective_concurrency} para evitar oversubscription de memória"
597            )
598        }
599
600        pub fn initializing_embedding_model() -> &'static str {
601            "Inicializando modelo de embedding (pode baixar na primeira execução)..."
602        }
603
604        pub fn embedding_chunks_serially(count: usize) -> String {
605            format!("Embedando {count} chunks serialmente para manter memória limitada...")
606        }
607
608        pub fn remember_step_input_validated(available_mb: u64) -> String {
609            format!("Etapa remember: entrada validada; memória disponível {available_mb} MB")
610        }
611
612        pub fn remember_step_chunking_completed(
613            total_passage_tokens: usize,
614            model_max_length: usize,
615            chunks_count: usize,
616            rss_mb: u64,
617        ) -> String {
618            format!(
619                "Etapa remember: tokenizer contou {total_passage_tokens} tokens de passagem \
620                 (máximo do modelo {model_max_length}); chunking gerou {chunks_count} chunks; \
621                 RSS do processo {rss_mb} MB"
622            )
623        }
624
625        pub fn remember_step_embeddings_completed(rss_mb: u64) -> String {
626            format!("Etapa remember: embeddings dos chunks concluídos; RSS do processo {rss_mb} MB")
627        }
628
629        pub fn restore_recomputing_embedding() -> &'static str {
630            "Recalculando embedding da memória restaurada..."
631        }
632
633        pub fn edit_recomputing_embedding() -> &'static str {
634            "Recalculando embedding da memória editada..."
635        }
636    }
637}
638
639#[cfg(test)]
640mod tests {
641    use super::*;
642    use serial_test::serial;
643
644    #[test]
645    #[serial]
646    fn fallback_english_when_env_absent() {
647        std::env::remove_var("SQLITE_GRAPHRAG_LANG");
648        std::env::set_var("LC_ALL", "C");
649        std::env::set_var("LANG", "C");
650        assert_eq!(Language::from_env_or_locale(), Language::English);
651        std::env::remove_var("LC_ALL");
652        std::env::remove_var("LANG");
653    }
654
655    #[test]
656    #[serial]
657    fn env_pt_selects_portuguese() {
658        std::env::remove_var("LC_ALL");
659        std::env::remove_var("LANG");
660        std::env::set_var("SQLITE_GRAPHRAG_LANG", "pt");
661        assert_eq!(Language::from_env_or_locale(), Language::Portuguese);
662        std::env::remove_var("SQLITE_GRAPHRAG_LANG");
663    }
664
665    #[test]
666    #[serial]
667    fn env_pt_br_selects_portuguese() {
668        std::env::remove_var("LC_ALL");
669        std::env::remove_var("LANG");
670        std::env::set_var("SQLITE_GRAPHRAG_LANG", "pt-BR");
671        assert_eq!(Language::from_env_or_locale(), Language::Portuguese);
672        std::env::remove_var("SQLITE_GRAPHRAG_LANG");
673    }
674
675    #[test]
676    #[serial]
677    fn locale_ptbr_utf8_selects_portuguese() {
678        std::env::remove_var("SQLITE_GRAPHRAG_LANG");
679        std::env::set_var("LC_ALL", "pt_BR.UTF-8");
680        assert_eq!(Language::from_env_or_locale(), Language::Portuguese);
681        std::env::remove_var("LC_ALL");
682    }
683
684    #[test]
685    #[serial]
686    fn posix_precedence_lc_all_overrides_lang() {
687        std::env::remove_var("SQLITE_GRAPHRAG_LANG");
688        std::env::remove_var("LC_MESSAGES");
689        std::env::set_var("LC_ALL", "en_US.UTF-8");
690        std::env::set_var("LANG", "pt_BR.UTF-8");
691        assert_eq!(
692            Language::from_env_or_locale(),
693            Language::English,
694            "LC_ALL=en_US must override LANG=pt_BR per POSIX"
695        );
696        std::env::remove_var("LC_ALL");
697        std::env::remove_var("LANG");
698    }
699
700    #[test]
701    #[serial]
702    fn posix_precedence_lc_all_unrecognized_stops_iteration() {
703        std::env::remove_var("SQLITE_GRAPHRAG_LANG");
704        std::env::remove_var("LC_MESSAGES");
705        std::env::set_var("LC_ALL", "ja_JP.UTF-8");
706        std::env::set_var("LANG", "pt_BR.UTF-8");
707        assert_eq!(
708            Language::from_env_or_locale(),
709            Language::English,
710            "LC_ALL=ja_JP set must stop iteration; falls back to English default"
711        );
712        std::env::remove_var("LC_ALL");
713        std::env::remove_var("LANG");
714    }
715
716    #[test]
717    #[serial]
718    fn lang_pt_selects_portuguese_when_lc_all_unset() {
719        std::env::remove_var("SQLITE_GRAPHRAG_LANG");
720        std::env::remove_var("LC_ALL");
721        std::env::remove_var("LC_MESSAGES");
722        std::env::set_var("LANG", "pt_BR.UTF-8");
723        assert_eq!(Language::from_env_or_locale(), Language::Portuguese);
724        std::env::remove_var("LANG");
725    }
726
727    mod validation_tests {
728        use super::*;
729
730        #[test]
731        fn name_length_en() {
732            let msg = match Language::English {
733                Language::English => format!("name must be 1-{} chars", 80),
734                Language::Portuguese => format!("nome deve ter entre 1 e {} caracteres", 80),
735            };
736            assert!(msg.contains("name must be 1-80 chars"), "obtido: {msg}");
737        }
738
739        #[test]
740        fn name_length_pt() {
741            let msg = match Language::Portuguese {
742                Language::English => format!("name must be 1-{} chars", 80),
743                Language::Portuguese => format!("nome deve ter entre 1 e {} caracteres", 80),
744            };
745            assert!(
746                msg.contains("nome deve ter entre 1 e 80 caracteres"),
747                "obtido: {msg}"
748            );
749        }
750
751        #[test]
752        fn name_kebab_en() {
753            let nome = "Invalid_Name";
754            let msg = match Language::English {
755                Language::English => format!(
756                    "name must be kebab-case slug (lowercase letters, digits, hyphens): '{nome}'"
757                ),
758                Language::Portuguese => {
759                    format!("nome deve estar em kebab-case (minúsculas, dígitos, hífens): '{nome}'")
760                }
761            };
762            assert!(msg.contains("kebab-case slug"), "obtido: {msg}");
763            assert!(msg.contains("Invalid_Name"), "obtido: {msg}");
764        }
765
766        #[test]
767        fn name_kebab_pt() {
768            let nome = "Invalid_Name";
769            let msg = match Language::Portuguese {
770                Language::English => format!(
771                    "name must be kebab-case slug (lowercase letters, digits, hyphens): '{nome}'"
772                ),
773                Language::Portuguese => {
774                    format!("nome deve estar em kebab-case (minúsculas, dígitos, hífens): '{nome}'")
775                }
776            };
777            assert!(msg.contains("kebab-case"), "obtido: {msg}");
778            assert!(msg.contains("minúsculas"), "obtido: {msg}");
779            assert!(msg.contains("Invalid_Name"), "obtido: {msg}");
780        }
781
782        #[test]
783        fn description_exceeds_en() {
784            let msg = match Language::English {
785                Language::English => format!("description must be <= {} chars", 500),
786                Language::Portuguese => format!("descrição deve ter no máximo {} caracteres", 500),
787            };
788            assert!(msg.contains("description must be <= 500"), "obtido: {msg}");
789        }
790
791        #[test]
792        fn description_exceeds_pt() {
793            let msg = match Language::Portuguese {
794                Language::English => format!("description must be <= {} chars", 500),
795                Language::Portuguese => format!("descrição deve ter no máximo {} caracteres", 500),
796            };
797            assert!(
798                msg.contains("descrição deve ter no máximo 500"),
799                "obtido: {msg}"
800            );
801        }
802
803        #[test]
804        fn body_exceeds_en() {
805            let limite = crate::constants::MAX_MEMORY_BODY_LEN;
806            let msg = match Language::English {
807                Language::English => format!("body exceeds {limite} bytes"),
808                Language::Portuguese => format!("corpo excede {limite} bytes"),
809            };
810            assert!(msg.contains("body exceeds 512000"), "obtido: {msg}");
811        }
812
813        #[test]
814        fn body_exceeds_pt() {
815            let limite = crate::constants::MAX_MEMORY_BODY_LEN;
816            let msg = match Language::Portuguese {
817                Language::English => format!("body exceeds {limite} bytes"),
818                Language::Portuguese => format!("corpo excede {limite} bytes"),
819            };
820            assert!(msg.contains("corpo excede 512000"), "obtido: {msg}");
821        }
822
823        #[test]
824        fn new_name_length_en() {
825            let msg = match Language::English {
826                Language::English => format!("new-name must be 1-{} chars", 80),
827                Language::Portuguese => format!("novo nome deve ter entre 1 e {} caracteres", 80),
828            };
829            assert!(msg.contains("new-name must be 1-80"), "obtido: {msg}");
830        }
831
832        #[test]
833        fn new_name_length_pt() {
834            let msg = match Language::Portuguese {
835                Language::English => format!("new-name must be 1-{} chars", 80),
836                Language::Portuguese => format!("novo nome deve ter entre 1 e {} caracteres", 80),
837            };
838            assert!(
839                msg.contains("novo nome deve ter entre 1 e 80"),
840                "obtido: {msg}"
841            );
842        }
843
844        #[test]
845        fn new_name_kebab_en() {
846            let nome = "Bad Name";
847            let msg = match Language::English {
848                Language::English => format!(
849                    "new-name must be kebab-case slug (lowercase letters, digits, hyphens): '{nome}'"
850                ),
851                Language::Portuguese => format!(
852                    "novo nome deve estar em kebab-case (minúsculas, dígitos, hífens): '{nome}'"
853                ),
854            };
855            assert!(msg.contains("new-name must be kebab-case"), "obtido: {msg}");
856        }
857
858        #[test]
859        fn new_name_kebab_pt() {
860            let nome = "Bad Name";
861            let msg = match Language::Portuguese {
862                Language::English => format!(
863                    "new-name must be kebab-case slug (lowercase letters, digits, hyphens): '{nome}'"
864                ),
865                Language::Portuguese => format!(
866                    "novo nome deve estar em kebab-case (minúsculas, dígitos, hífens): '{nome}'"
867                ),
868            };
869            assert!(
870                msg.contains("novo nome deve estar em kebab-case"),
871                "obtido: {msg}"
872            );
873        }
874
875        #[test]
876        fn reserved_name_en() {
877            let msg = match Language::English {
878                Language::English => {
879                    "names and namespaces starting with __ are reserved for internal use"
880                        .to_string()
881                }
882                Language::Portuguese => {
883                    "nomes e namespaces iniciados com __ são reservados para uso interno"
884                        .to_string()
885                }
886            };
887            assert!(msg.contains("reserved for internal use"), "obtido: {msg}");
888        }
889
890        #[test]
891        fn reserved_name_pt() {
892            let msg = match Language::Portuguese {
893                Language::English => {
894                    "names and namespaces starting with __ are reserved for internal use"
895                        .to_string()
896                }
897                Language::Portuguese => {
898                    "nomes e namespaces iniciados com __ são reservados para uso interno"
899                        .to_string()
900                }
901            };
902            assert!(msg.contains("reservados para uso interno"), "obtido: {msg}");
903        }
904    }
905
906    mod app_error_pt_translation_tests {
907        use crate::errors::AppError;
908
909        #[test]
910        fn localized_message_pt_not_found_fully_translated() {
911            let err =
912                AppError::NotFound("memory 'test-mem' not found in namespace 'global'".into());
913            let pt = err.localized_message_for(crate::i18n::Language::Portuguese);
914            assert!(
915                pt.contains("memória"),
916                "PT must translate 'memory' to 'memória': {pt}"
917            );
918            assert!(
919                pt.contains("não encontrada no namespace"),
920                "PT must translate full phrase: {pt}"
921            );
922            assert!(
923                !pt.contains("not found in namespace"),
924                "PT must not contain English phrase: {pt}"
925            );
926        }
927
928        #[test]
929        fn localized_message_pt_duplicate_fully_translated() {
930            let err = AppError::Duplicate(
931                "memory 'x' already exists in namespace 'global'. Use --force-merge to update."
932                    .into(),
933            );
934            let pt = err.localized_message_for(crate::i18n::Language::Portuguese);
935            assert!(pt.contains("memória"), "PT must translate 'memory': {pt}");
936            assert!(
937                pt.contains("já existe no namespace"),
938                "PT must translate 'already exists': {pt}"
939            );
940        }
941    }
942}