kodit 0.4.2__py3-none-any.whl → 0.5.0__py3-none-any.whl

kodit/application/services/code_indexing_application_service.py

@@ -1,393 +0,0 @@
-"""Unified application service for code indexing operations."""
-
-from dataclasses import replace
-from datetime import UTC, datetime
-
-import structlog
-from sqlalchemy.ext.asyncio import AsyncSession
-
-from kodit.application.services.reporting import (
-    OperationType,
-    ProgressTracker,
-)
-from kodit.domain.entities import Index, Snippet
-from kodit.domain.protocols import IndexRepository
-from kodit.domain.services.bm25_service import BM25DomainService
-from kodit.domain.services.embedding_service import EmbeddingDomainService
-from kodit.domain.services.enrichment_service import EnrichmentDomainService
-from kodit.domain.services.index_query_service import IndexQueryService
-from kodit.domain.services.index_service import IndexDomainService
-from kodit.domain.value_objects import (
-    Document,
-    FusionRequest,
-    IndexRequest,
-    MultiSearchRequest,
-    MultiSearchResult,
-    SearchRequest,
-    SearchResult,
-    SnippetSearchFilters,
-)
-from kodit.log import log_event
-
-
-class CodeIndexingApplicationService:
-    """Unified application service for all code indexing operations."""
-
-    def __init__(  # noqa: PLR0913
-        self,
-        indexing_domain_service: IndexDomainService,
-        index_repository: IndexRepository,
-        index_query_service: IndexQueryService,
-        bm25_service: BM25DomainService,
-        code_search_service: EmbeddingDomainService,
-        text_search_service: EmbeddingDomainService,
-        enrichment_service: EnrichmentDomainService,
-        session: AsyncSession,
-        operation: ProgressTracker,
-    ) -> None:
-        """Initialize the code indexing application service."""
-        self.index_domain_service = indexing_domain_service
-        self.index_repository = index_repository
-        self.index_query_service = index_query_service
-        self.bm25_service = bm25_service
-        self.code_search_service = code_search_service
-        self.text_search_service = text_search_service
-        self.enrichment_service = enrichment_service
-        self.session = session
-        self.operation = operation
-        self.log = structlog.get_logger(__name__)
-
-    async def does_index_exist(self, uri: str) -> bool:
-        """Check if an index exists for a source."""
-        # Check if index already exists
-        sanitized_uri, _ = self.index_domain_service.sanitize_uri(uri)
-        existing_index = await self.index_repository.get_by_uri(sanitized_uri)
-        return existing_index is not None
-
-    async def create_index_from_uri(self, uri: str) -> Index:
-        """Create a new index for a source."""
-        log_event("kodit.index.create")
-        with self.operation.create_child(OperationType.CREATE_INDEX.value) as operation:
-            # Check if index already exists
-            sanitized_uri, _ = self.index_domain_service.sanitize_uri(uri)
-            self.log.info("Creating index from URI", uri=str(sanitized_uri))
-            existing_index = await self.index_repository.get_by_uri(sanitized_uri)
-            if existing_index:
-                self.log.debug(
-                    "Index already exists",
-                    uri=str(sanitized_uri),
-                    index_id=existing_index.id,
-                )
-                return existing_index
-
-            # Only prepare working copy if we need to create a new index
-            self.log.info("Preparing working copy", uri=str(sanitized_uri))
-            working_copy = await self.index_domain_service.prepare_index(uri, operation)
-
-            # Create new index
-            self.log.info("Creating index", uri=str(sanitized_uri))
-            index = await self.index_repository.create(sanitized_uri, working_copy)
-            await self.session.commit()
-            return index
-
-    async def run_index(self, index: Index) -> None:
-        """Run the complete indexing process for a specific index."""
-        # Create a new operation
-        with self.operation.create_child(OperationType.RUN_INDEX.value) as operation:
-            # TODO(philwinder): Move this into a reporter # noqa: TD003, FIX002
-            log_event("kodit.index.run")
-
-            if not index or not index.id:
-                msg = f"Index has no ID: {index}"
-                raise ValueError(msg)
-
-            # Refresh working copy
-            with operation.create_child("Refresh working copy") as step:
-                index.source.working_copy = (
-                    await self.index_domain_service.refresh_working_copy(
-                        index.source.working_copy, step
-                    )
-                )
-                if len(index.source.working_copy.changed_files()) == 0:
-                    self.log.info("No new changes to index", index_id=index.id)
-                    step.skip("No new changes to index")
-                    return
-
-            # Delete the old snippets from the files that have changed
-            with operation.create_child("Delete old snippets") as step:
-                await self.index_repository.delete_snippets_by_file_ids(
-                    [
-                        file.id
-                        for file in index.source.working_copy.changed_files()
-                        if file.id
-                    ]
-                )
-
-            # Extract and create snippets (domain service handles progress)
-            with operation.create_child("Extract snippets") as step:
-                index = await self.index_domain_service.extract_snippets_from_index(
-                    index=index, step=step
-                )
-                await self.index_repository.update(index)
-
-                # Refresh index to get snippets with IDs, required for subsequent steps
-                flushed_index = await self.index_repository.get(index.id)
-                if not flushed_index:
-                    msg = f"Index {index.id} not found after snippet extraction"
-                    raise ValueError(msg)
-                index = flushed_index
-                if len(index.snippets) == 0:
-                    self.log.info(
-                        "No snippets to index after extraction", index_id=index.id
-                    )
-                    step.skip("No snippets to index after extraction")
-                    return
-
-            # Create BM25 index
-            self.log.info("Creating keyword index")
-            with operation.create_child("Create BM25 index") as step:
-                await self._create_bm25_index(index.snippets)
-
-            # Create code embeddings
-            with operation.create_child("Create code embeddings") as step:
-                await self._create_code_embeddings(index.snippets, step)
-
-            # Enrich snippets
-            with operation.create_child("Enrich snippets") as step:
-                enriched_snippets = (
-                    await self.index_domain_service.enrich_snippets_in_index(
-                        snippets=index.snippets,
-                        reporting_step=step,
-                    )
-                )
-                # Update snippets in repository
-                await self.index_repository.update_snippets(index.id, enriched_snippets)
-
-            # Create text embeddings (on enriched content)
-            with operation.create_child("Create text embeddings") as step:
-                await self._create_text_embeddings(enriched_snippets, step)
-
-            # Update index timestamp
-            with operation.create_child("Update index timestamp") as step:
-                await self.index_repository.update_index_timestamp(index.id)
-
-            # After indexing, clear the file processing statuses
-            with operation.create_child("Clear file processing statuses") as step:
-                index.source.working_copy.clear_file_processing_statuses()
-                await self.index_repository.update(index)
-
-    async def search(self, request: MultiSearchRequest) -> list[MultiSearchResult]:
-        """Search for relevant snippets across all indexes."""
-        log_event("kodit.index.search")
-
-        # Apply filters if provided
-        filtered_snippet_ids: list[int] | None = None
-        if request.filters:
-            # Use domain service for filtering (use large top_k for pre-filtering)
-            prefilter_request = replace(request, top_k=10000)
-            snippet_results = await self.index_query_service.search_snippets(
-                prefilter_request
-            )
-            filtered_snippet_ids = [
-                snippet.snippet.id for snippet in snippet_results if snippet.snippet.id
-            ]
-
-        # Gather results from different search modes
-        fusion_list: list[list[FusionRequest]] = []
-
-        # Keyword search
-        if request.keywords:
-            result_ids: list[SearchResult] = []
-            for keyword in request.keywords:
-                results = await self.bm25_service.search(
-                    SearchRequest(
-                        query=keyword,
-                        top_k=request.top_k,
-                        snippet_ids=filtered_snippet_ids,
-                    )
-                )
-                result_ids.extend(results)
-
-            fusion_list.append(
-                [FusionRequest(id=x.snippet_id, score=x.score) for x in result_ids]
-            )
-
-        # Semantic code search
-        if request.code_query:
-            query_results = await self.code_search_service.search(
-                SearchRequest(
-                    query=request.code_query,
-                    top_k=request.top_k,
-                    snippet_ids=filtered_snippet_ids,
-                )
-            )
-            fusion_list.append(
-                [FusionRequest(id=x.snippet_id, score=x.score) for x in query_results]
-            )
-
-        # Semantic text search
-        if request.text_query:
-            query_results = await self.text_search_service.search(
-                SearchRequest(
-                    query=request.text_query,
-                    top_k=request.top_k,
-                    snippet_ids=filtered_snippet_ids,
-                )
-            )
-            fusion_list.append(
-                [FusionRequest(id=x.snippet_id, score=x.score) for x in query_results]
-            )
-
-        if len(fusion_list) == 0:
-            return []
-
-        # Fusion ranking
-        final_results = await self.index_query_service.perform_fusion(
-            rankings=fusion_list,
-            k=60,  # This is a parameter in the RRF algorithm, not top_k
-        )
-
-        # Keep only top_k results
-        final_results = final_results[: request.top_k]
-
-        # Get snippet details
-        search_results = await self.index_query_service.get_snippets_by_ids(
-            [x.id for x in final_results]
-        )
-
-        # Create a mapping from snippet ID to search result to handle cases where
-        # some snippet IDs don't exist (e.g., with vectorchord inconsistencies)
-        snippet_map = {
-            result.snippet.id: result
-            for result in search_results
-            if result.snippet.id is not None
-        }
-
-        # Filter final_results to only include IDs that we actually found snippets for
-        valid_final_results = [fr for fr in final_results if fr.id in snippet_map]
-
-        return [
-            MultiSearchResult(
-                id=snippet_map[fr.id].snippet.id or 0,
-                content=snippet_map[fr.id].snippet.original_text(),
-                original_scores=fr.original_scores,
-                # Enhanced fields
-                source_uri=str(snippet_map[fr.id].source.working_copy.remote_uri),
-                relative_path=str(
-                    snippet_map[fr.id]
-                    .file.as_path()
-                    .relative_to(snippet_map[fr.id].source.working_copy.cloned_path)
-                ),
-                language=MultiSearchResult.detect_language_from_extension(
-                    snippet_map[fr.id].file.extension()
-                ),
-                authors=[author.name for author in snippet_map[fr.id].authors],
-                created_at=snippet_map[fr.id].snippet.created_at or datetime.now(UTC),
-                # Summary from snippet entity
-                summary=snippet_map[fr.id].snippet.summary_text(),
-            )
-            for fr in valid_final_results
-        ]
-
-    async def list_snippets(
-        self, file_path: str | None = None, source_uri: str | None = None
-    ) -> list[MultiSearchResult]:
-        """List snippets with optional filtering."""
-        log_event("kodit.index.list_snippets")
-        snippet_results = await self.index_query_service.search_snippets(
-            request=MultiSearchRequest(
-                filters=SnippetSearchFilters(
-                    file_path=file_path,
-                    source_repo=source_uri,
-                )
-            ),
-        )
-        return [
-            MultiSearchResult(
-                id=result.snippet.id or 0,
-                content=result.snippet.original_text(),
-                original_scores=[0.0],
-                # Enhanced fields
-                source_uri=str(result.source.working_copy.remote_uri),
-                relative_path=str(
-                    result.file.as_path().relative_to(
-                        result.source.working_copy.cloned_path
-                    )
-                ),
-                language=MultiSearchResult.detect_language_from_extension(
-                    result.file.extension()
-                ),
-                authors=[author.name for author in result.authors],
-                created_at=result.snippet.created_at or datetime.now(UTC),
-                # Summary from snippet entity
-                summary=result.snippet.summary_text(),
-            )
-            for result in snippet_results
-        ]
-
-    # FUTURE: BM25 index enriched content too
-    async def _create_bm25_index(self, snippets: list[Snippet]) -> None:
-        await self.bm25_service.index_documents(
-            IndexRequest(
-                documents=[
-                    Document(snippet_id=snippet.id, text=snippet.original_text())
-                    for snippet in snippets
-                    if snippet.id
-                ]
-            )
-        )
-
-    async def _create_code_embeddings(
-        self, snippets: list[Snippet], reporting_step: ProgressTracker
-    ) -> None:
-        reporting_step.set_total(len(snippets))
-        processed = 0
-        async for result in self.code_search_service.index_documents(
-            IndexRequest(
-                documents=[
-                    Document(snippet_id=snippet.id, text=snippet.original_text())
-                    for snippet in snippets
-                    if snippet.id
-                ]
-            )
-        ):
-            processed += len(result)
-            reporting_step.set_current(processed)
-
-    async def _create_text_embeddings(
-        self, snippets: list[Snippet], reporting_step: ProgressTracker
-    ) -> None:
-        # Only create text embeddings for snippets that have summary content
-        documents_with_summaries = []
-        for snippet in snippets:
-            if snippet.id:
-                try:
-                    summary_text = snippet.summary_text()
-                    if summary_text.strip():  # Only add if summary is not empty
-                        documents_with_summaries.append(
-                            Document(snippet_id=snippet.id, text=summary_text)
-                        )
-                except ValueError:
-                    # Skip snippets without summary content
-                    continue
-
-        if not documents_with_summaries:
-            reporting_step.skip("No snippets with summaries to create text embeddings")
-            return
-
-        reporting_step.set_total(len(documents_with_summaries))
-        processed = 0
-        async for result in self.text_search_service.index_documents(
-            IndexRequest(documents=documents_with_summaries)
-        ):
-            processed += len(result)
-            reporting_step.set_current(processed)
-
-    async def delete_index(self, index: Index) -> None:
-        """Delete an index."""
-        # Delete the index from the domain
-        await self.index_domain_service.delete_index(index)
-
-        # Delete index from the database
-        await self.index_repository.delete(index)
-        await self.session.commit()