feat: add language support for wiki generation

packages/cli/src/repowise/cli/commands/init_cmd.py

@@ -977,6 +977,7 @@ def init_command(
 
     # Merge exclude_patterns from config.yaml and --exclude/-x flags
     config = load_config(repo_path)
+    language = config.get("language", "en")
     exclude_patterns: list[str] = list(config.get("exclude_patterns") or []) + list(exclude)
 
     # Resolve commit limit: CLI flag → config.yaml → default (500)
@@ -1038,6 +1039,8 @@ def init_command(
             f"  Provider: [cyan]{provider.provider_name}[/cyan] / Model: [cyan]{provider.model_name}[/cyan]"
         )
         console.print(f"  Embedder: [cyan]{embedder_name_resolved}[/cyan]")
+        if language != "en":
+            console.print(f"  Language: [cyan]{language}[/cyan]")
 
         # Validate provider connection
         from repowise.core.providers.llm.base import ProviderError
@@ -1150,7 +1153,9 @@ def init_command(
         # Cost estimation
         from repowise.core.generation import GenerationConfig
 
-        gen_config = GenerationConfig(max_concurrency=concurrency)
+        cfg = load_config(repo_path)
+        language = cfg.get("language", "en")
+        gen_config = GenerationConfig(max_concurrency=concurrency, language=language)
         plans = build_generation_plan(
             result.parsed_files, result.graph_builder, gen_config, skip_tests, skip_infra
         )
@@ -1291,6 +1296,7 @@ async def _make_cost_tracker() -> CostTracker:
                     progress=gen_callback,
                     resume=resume,
                     cost_tracker=cost_tracker,
+                    generation_config=gen_config,
                 )
             )
 

packages/cli/src/repowise/cli/commands/update_cmd.py

@@ -209,7 +209,9 @@ def update_command(
     from repowise.core.generation import ContextAssembler, GenerationConfig, PageGenerator
     from repowise.core.ingestion import ASTParser, FileTraverser, GraphBuilder
 
-    config = GenerationConfig()
+    cfg = load_config(repo_path)
+    language = cfg.get("language", "en")
+    config = GenerationConfig(max_concurrency=concurrency, language=language)()
 
     # Read exclude patterns from config (set during init or via web UI)
     repo_config = load_config(repo_path)

packages/core/src/repowise/core/generation/models.py

@@ -82,7 +82,7 @@ class GenerationConfig:
     max_pages_pct: float = 0.10  # hard cap: total pages ≤ max(50, N_files * this)
     jobs_dir: str = ".repowise/jobs"
     large_file_source_pct: float = 0.4  # use structural summary when source tokens > budget * this
-
+    language: str = "en"
 
 # ---------------------------------------------------------------------------
 # GeneratedPage

packages/core/src/repowise/core/generation/page_generator.py

@@ -154,12 +154,14 @@ def __init__(
         assembler: ContextAssembler,
         config: GenerationConfig,
         jinja_env: jinja2.Environment | None = None,
-        vector_store: Any | None = None,  # VectorStore | None
+        vector_store: Any | None = None,
+        language: str = "en",
     ) -> None:
         self._provider = provider
         self._assembler = assembler
         self._config = config
         self._vector_store = vector_store
+        self._language = language
         self._cache: dict[str, GeneratedResponse] = {}
 
         if jinja_env is None:
@@ -171,7 +173,6 @@ def __init__(
                 autoescape=False,
             )
         self._jinja_env = jinja_env
-
     # ------------------------------------------------------------------
     # Per-type generation methods
     # ------------------------------------------------------------------
@@ -929,21 +930,31 @@ async def _generate_file_page_from_ctx(
             )
             page.metadata["hallucination_warnings"] = hal_warnings
         return page
-
+    
     async def _call_provider(
         self,
         page_type: str,
         user_prompt: str,
         request_id: str,
     ) -> GeneratedResponse:
-        """Call the provider with caching."""
         key = self._compute_cache_key(page_type, user_prompt)
         if self._config.cache_enabled and key in self._cache:
             log.debug("Cache hit", page_type=page_type, key=key[:8])
             return self._cache[key]
 
+        base_system = SYSTEM_PROMPTS[page_type]
+        if self._language != "en":
+            language_instruction = (
+                f"Generate all documentation content in {self._language}. "
+                "Keep all code, file paths, and symbol names in their original form. "
+                "Do not translate them.\n\n"
+            )
+            system_prompt = language_instruction + base_system
+        else:
+            system_prompt = base_system
+
         response = await self._provider.generate(
-            SYSTEM_PROMPTS[page_type],
+            system_prompt,
             user_prompt,
             max_tokens=self._config.max_tokens,
             temperature=self._config.temperature,
@@ -954,7 +965,7 @@ async def _call_provider(
             self._cache[key] = response
 
         return response
-
+   
     def _compute_cache_key(self, page_type: str, user_prompt: str) -> str:
         """Return SHA256(model + page_type + user_prompt) as cache key."""
         raw = f"{self._provider.model_name}:{page_type}:{user_prompt}"

packages/core/src/repowise/core/pipeline/orchestrator.py

@@ -723,6 +723,7 @@ async def run_generation(
     progress: ProgressCallback | None,
     resume: bool = False,
     cost_tracker: Any | None = None,
+    generation_config: Any,   # <-- ДОБАВЛЕННЫЙ ПАРАМЕТР
 ) -> list[Any]:
     """Run LLM-powered page generation.
 
@@ -741,7 +742,23 @@ async def run_generation(
     if cost_tracker is not None and llm_client is not None and hasattr(llm_client, "_cost_tracker"):
         llm_client._cost_tracker = cost_tracker
 
-    config = GenerationConfig(max_concurrency=concurrency)
+    # Создаём новый конфиг на основе переданного generation_config, но с нужным max_concurrency
+    config = GenerationConfig(
+        max_tokens=generation_config.max_tokens,
+        temperature=generation_config.temperature,
+        token_budget=generation_config.token_budget,
+        max_concurrency=concurrency,
+        cache_enabled=generation_config.cache_enabled,
+        staleness_threshold_days=generation_config.staleness_threshold_days,
+        expiry_threshold_days=generation_config.expiry_threshold_days,
+        top_symbol_percentile=generation_config.top_symbol_percentile,
+        file_page_top_percentile=generation_config.file_page_top_percentile,
+        file_page_min_symbols=generation_config.file_page_min_symbols,
+        max_pages_pct=generation_config.max_pages_pct,
+        jobs_dir=generation_config.jobs_dir,
+        large_file_source_pct=generation_config.large_file_source_pct,
+        language=generation_config.language,
+    )
     assembler = ContextAssembler(config)
 
     # Resolve embedder and vector store
@@ -750,8 +767,6 @@ async def run_generation(
     if vector_store is None:
         vector_store = InMemoryVectorStore(embedder_impl)
 
-    generator = PageGenerator(llm_client, assembler, config, vector_store=vector_store)
-
     # Job system — use a temp-like dir under repo_path for checkpoints
     jobs_dir = repo_path / ".repowise" / "jobs"
     jobs_dir.mkdir(parents=True, exist_ok=True)
@@ -778,6 +793,14 @@ def on_total_known(total: int) -> None:
         if progress:
             progress.on_phase_start("generation", total)
 
+    generator = PageGenerator(
+        llm_client,
+        assembler,
+        config,
+        vector_store=vector_store,
+        language=config.language,
+    )
+
     generated_pages = await generator.generate_all(
         parsed_files,
         source_map,
@@ -795,4 +818,4 @@ def on_total_known(total: int) -> None:
     if progress:
         progress.on_message("info", f"Generated {len(generated_pages)} pages")
 
-    return generated_pages
+    return generated_pages