kodit 0.4.3__py3-none-any.whl → 0.5.1__py3-none-any.whl

kodit/infrastructure/api/v1/routers/search.py

@@ -3,15 +3,18 @@
 from fastapi import APIRouter
 
 from kodit.domain.value_objects import MultiSearchRequest, SnippetSearchFilters
-from kodit.infrastructure.api.v1.dependencies import (
-    IndexingAppServiceDep,
-)
+from kodit.infrastructure.api.v1.dependencies import CodeSearchAppServiceDep
 from kodit.infrastructure.api.v1.schemas.search import (
     SearchRequest,
     SearchResponse,
     SnippetAttributes,
     SnippetData,
 )
+from kodit.infrastructure.api.v1.schemas.snippet import (
+    EnrichmentSchema,
+    GitFileSchema,
+    SnippetContentSchema,
+)
 
 router = APIRouter(tags=["search"])
 
@@ -19,7 +22,7 @@ router = APIRouter(tags=["search"])
 @router.post("/api/v1/search")
 async def search_snippets(
     request: SearchRequest,
-    app_service: IndexingAppServiceDep,
+    search_application_service: CodeSearchAppServiceDep,
 ) -> SearchResponse:
     """Search code snippets with filters matching MCP tool."""
     # Convert API request to domain request
@@ -50,23 +53,37 @@ async def search_snippets(
     )
 
     # Execute search using application service
-    results = await app_service.search(domain_request)
+    results = await search_application_service.search(domain_request)
 
     return SearchResponse(
         data=[
             SnippetData(
                 type="snippet",
-                id=result.id,
+                id=result.snippet.id,
                 attributes=SnippetAttributes(
-                    content=result.content,
-                    created_at=result.created_at,
-                    updated_at=result.created_at,  # Use created_at as fallback
+                    created_at=result.snippet.created_at,
+                    updated_at=result.snippet.updated_at,
+                    derives_from=[
+                        GitFileSchema(
+                            blob_sha=file.blob_sha,
+                            path=file.path,
+                            mime_type=file.mime_type,
+                            size=file.size,
+                        )
+                        for file in result.snippet.derives_from
+                    ],
+                    content=SnippetContentSchema(
+                        value=result.snippet.content,
+                        language=result.snippet.extension,
+                    ),
+                    enrichments=[
+                        EnrichmentSchema(
+                            type=enrichment.type.value,
+                            content=enrichment.content,
+                        )
+                        for enrichment in result.snippet.enrichments
+                    ],
                     original_scores=result.original_scores,
-                    source_uri=result.source_uri,
-                    relative_path=result.relative_path,
-                    language=result.language,
-                    authors=result.authors,
-                    summary=result.summary,
                 ),
             )
             for result in results

kodit/infrastructure/api/v1/schemas/__init__.py

@@ -1,25 +1 @@
 """JSON:API schemas for the REST API."""
-
-from .index import (
-    IndexCreateRequest,
-    IndexDetailResponse,
-    IndexListResponse,
-    IndexResponse,
-)
-from .search import (
-    SearchRequest,
-    SearchResponse,
-    SearchResponseWithIncluded,
-    SnippetDetailResponse,
-)
-
-__all__ = [
-    "IndexCreateRequest",
-    "IndexDetailResponse",
-    "IndexListResponse",
-    "IndexResponse",
-    "SearchRequest",
-    "SearchResponse",
-    "SearchResponseWithIncluded",
-    "SnippetDetailResponse",
-]

kodit/infrastructure/api/v1/schemas/commit.py

@@ -0,0 +1,96 @@
+"""Commit JSON-API schemas."""
+
+from datetime import datetime
+
+from pydantic import BaseModel
+
+
+class GitFileData(BaseModel):
+    """Git file data."""
+
+    blob_sha: str
+    path: str
+    mime_type: str
+    size: int
+
+
+class CommitAttributes(BaseModel):
+    """Commit attributes following JSON-API spec."""
+
+    commit_sha: str
+    date: datetime
+    message: str
+    parent_commit_sha: str
+    author: str
+
+
+class CommitData(BaseModel):
+    """Commit data following JSON-API spec."""
+
+    type: str = "commit"
+    id: str
+    attributes: CommitAttributes
+
+
+class CommitResponse(BaseModel):
+    """Single commit response following JSON-API spec."""
+
+    data: CommitData
+
+
+class CommitListResponse(BaseModel):
+    """Commit list response following JSON-API spec."""
+
+    data: list[CommitData]
+
+
+class FileAttributes(BaseModel):
+    """File attributes following JSON-API spec."""
+
+    blob_sha: str
+    path: str
+    mime_type: str
+    size: int
+    extension: str
+
+
+class FileData(BaseModel):
+    """File data following JSON-API spec."""
+
+    type: str = "file"
+    id: str
+    attributes: FileAttributes
+
+
+class FileResponse(BaseModel):
+    """Single file response following JSON-API spec."""
+
+    data: FileData
+
+
+class FileListResponse(BaseModel):
+    """File list response following JSON-API spec."""
+
+    data: list[FileData]
+
+
+class EmbeddingAttributes(BaseModel):
+    """Embedding attributes following JSON-API spec."""
+
+    snippet_sha: str
+    embedding_type: str
+    embedding: list[float]
+
+
+class EmbeddingData(BaseModel):
+    """Embedding data following JSON-API spec."""
+
+    type: str = "embedding"
+    id: str
+    attributes: EmbeddingAttributes
+
+
+class EmbeddingListResponse(BaseModel):
+    """Embedding list response following JSON-API spec."""
+
+    data: list[EmbeddingData]

kodit/infrastructure/api/v1/schemas/context.py

@@ -2,6 +2,7 @@
 
 from typing import TypedDict
 
+from kodit.application.factories.server_factory import ServerFactory
 from kodit.config import AppContext
 
 
@@ -9,3 +10,4 @@ class AppLifespanState(TypedDict):
     """Application lifespan state."""
 
     app_context: AppContext
+    server_factory: ServerFactory

kodit/infrastructure/api/v1/schemas/enrichment.py

@@ -0,0 +1,29 @@
+"""Enrichment JSON-API schemas."""
+
+from datetime import datetime
+
+from pydantic import BaseModel
+
+
+class EnrichmentAttributes(BaseModel):
+    """Enrichment attributes following JSON-API spec."""
+
+    type: str
+    subtype: str | None
+    content: str
+    created_at: datetime | None
+    updated_at: datetime | None
+
+
+class EnrichmentData(BaseModel):
+    """Enrichment data following JSON-API spec."""
+
+    type: str = "enrichment"
+    id: str
+    attributes: EnrichmentAttributes
+
+
+class EnrichmentListResponse(BaseModel):
+    """Enrichment list response following JSON-API spec."""
+
+    data: list[EnrichmentData]

kodit/infrastructure/api/v1/schemas/repository.py

@@ -0,0 +1,128 @@
+"""Repository JSON-API schemas."""
+
+from datetime import datetime
+from pathlib import Path
+
+from pydantic import AnyUrl, BaseModel
+
+from kodit.domain.entities.git import GitRepo
+
+
+class RepositoryAttributes(BaseModel):
+    """Repository attributes following JSON-API spec."""
+
+    remote_uri: AnyUrl
+    created_at: datetime | None = None
+    updated_at: datetime | None = None
+    last_scanned_at: datetime | None = None
+    cloned_path: Path | None = None
+    tracking_branch: str | None = None
+    num_commits: int = 0
+    num_branches: int = 0
+    num_tags: int = 0
+
+    @staticmethod
+    def from_git_repo(repo: GitRepo) -> "RepositoryAttributes":
+        """Create a repository attributes from a Git repository."""
+        return RepositoryAttributes(
+            remote_uri=repo.sanitized_remote_uri,
+            cloned_path=repo.cloned_path,
+            created_at=repo.created_at,
+            updated_at=repo.updated_at,
+            last_scanned_at=repo.last_scanned_at,
+            tracking_branch=repo.tracking_branch.name if repo.tracking_branch else None,
+            num_commits=repo.num_commits,
+            num_branches=repo.num_branches,
+            num_tags=repo.num_tags,
+        )
+
+
+class RepositoryData(BaseModel):
+    """Repository data following JSON-API spec."""
+
+    type: str = "repository"
+    id: str
+    attributes: RepositoryAttributes
+
+    @staticmethod
+    def from_git_repo(repo: GitRepo) -> "RepositoryData":
+        """Create a repository data from a Git repository."""
+        return RepositoryData(
+            id=str(repo.id) or "",
+            attributes=RepositoryAttributes.from_git_repo(repo),
+        )
+
+
+class RepositoryResponse(BaseModel):
+    """Single repository response following JSON-API spec."""
+
+    data: RepositoryData
+
+
+class RepositoryListResponse(BaseModel):
+    """Repository list response following JSON-API spec."""
+
+    data: list[RepositoryData]
+
+
+class RepositoryCreateAttributes(BaseModel):
+    """Repository creation attributes."""
+
+    remote_uri: AnyUrl
+
+
+class RepositoryCreateData(BaseModel):
+    """Repository creation data."""
+
+    type: str = "repository"
+    attributes: RepositoryCreateAttributes
+
+
+class RepositoryCreateRequest(BaseModel):
+    """Repository creation request."""
+
+    data: RepositoryCreateData
+
+
+class RepositoryUpdateAttributes(BaseModel):
+    """Repository update attributes."""
+
+    pull_latest: bool = False
+
+
+class RepositoryUpdateData(BaseModel):
+    """Repository update data."""
+
+    type: str = "repository"
+    attributes: RepositoryUpdateAttributes
+
+
+class RepositoryUpdateRequest(BaseModel):
+    """Repository update request."""
+
+    data: RepositoryUpdateData
+
+
+class RepositoryBranchData(BaseModel):
+    """Repository branch data."""
+
+    name: str
+    is_default: bool
+    commit_count: int
+
+
+class RepositoryCommitData(BaseModel):
+    """Repository commit data for repository details."""
+
+    sha: str
+    message: str
+    author: str
+    timestamp: datetime
+
+
+class RepositoryDetailsResponse(BaseModel):
+    """Repository details response with branches and commits."""
+
+    data: RepositoryData
+    branches: list[RepositoryBranchData]
+    recent_commits: list[RepositoryCommitData]

kodit/infrastructure/api/v1/schemas/search.py

@@ -4,6 +4,12 @@ from datetime import datetime
 
 from pydantic import BaseModel, Field
 
+from kodit.infrastructure.api.v1.schemas.snippet import (
+    EnrichmentSchema,
+    GitFileSchema,
+    SnippetContentSchema,
+)
+
 
 class SearchFilters(BaseModel):
     """Search filters for JSON:API requests."""
@@ -111,22 +117,19 @@ class SearchRequest(BaseModel):
 class SnippetAttributes(BaseModel):
     """Snippet attributes for JSON:API responses."""
 
-    content: str
-    created_at: datetime
-    updated_at: datetime
+    created_at: datetime | None = None
+    updated_at: datetime | None = None
+    derives_from: list[GitFileSchema]
+    content: SnippetContentSchema
+    enrichments: list[EnrichmentSchema]
     original_scores: list[float]
-    source_uri: str
-    relative_path: str
-    language: str
-    authors: list[str]
-    summary: str
 
 
 class SnippetData(BaseModel):
     """Snippet data for JSON:API responses."""
 
     type: str = "snippet"
-    id: int
+    id: str
     attributes: SnippetAttributes
 
 

kodit/infrastructure/api/v1/schemas/snippet.py

@@ -0,0 +1,58 @@
+"""Snippet JSON-API schemas."""
+
+from datetime import datetime
+
+from pydantic import BaseModel
+
+
+class SnippetContentSchema(BaseModel):
+    """Snippet content schema following JSON-API spec."""
+
+    value: str
+    language: str
+
+
+class GitFileSchema(BaseModel):
+    """Git file schema following JSON-API spec."""
+
+    blob_sha: str
+    path: str
+    mime_type: str
+    size: int
+
+
+class EnrichmentSchema(BaseModel):
+    """Enrichment schema following JSON-API spec."""
+
+    type: str
+    content: str
+
+
+class SnippetAttributes(BaseModel):
+    """Snippet attributes following JSON-API spec."""
+
+    created_at: datetime | None = None
+    updated_at: datetime | None = None
+    derives_from: list[GitFileSchema]
+    content: SnippetContentSchema
+    enrichments: list[EnrichmentSchema]
+
+
+class SnippetData(BaseModel):
+    """Snippet data following JSON-API spec."""
+
+    type: str = "snippet"
+    id: str
+    attributes: SnippetAttributes
+
+
+class SnippetResponse(BaseModel):
+    """Single snippet response following JSON-API spec."""
+
+    data: SnippetData
+
+
+class SnippetListResponse(BaseModel):
+    """Snippet list response following JSON-API spec."""
+
+    data: list[SnippetData]

kodit/infrastructure/api/v1/schemas/tag.py

@@ -0,0 +1,31 @@
+"""Tag JSON-API schemas."""
+
+from pydantic import BaseModel
+
+
+class TagAttributes(BaseModel):
+    """Tag attributes following JSON-API spec."""
+
+    name: str
+    target_commit_sha: str
+    is_version_tag: bool
+
+
+class TagData(BaseModel):
+    """Tag data following JSON-API spec."""
+
+    type: str = "tag"
+    id: str  # The tag name
+    attributes: TagAttributes
+
+
+class TagResponse(BaseModel):
+    """Single tag response following JSON-API spec."""
+
+    data: TagData
+
+
+class TagListResponse(BaseModel):
+    """Tag list response following JSON-API spec."""
+
+    data: list[TagData]

kodit/infrastructure/api/v1/schemas/task_status.py

@@ -17,6 +17,8 @@ class TaskStatusAttributes(BaseModel):
     current: int = Field(default=0, description="Current number of items processed")
     created_at: datetime | None = Field(default=None, description="Task start time")
     updated_at: datetime | None = Field(default=None, description="Last update time")
+    error: str = Field(default="", description="Error message")
+    message: str = Field(default="", description="Message")
 
 
 class TaskStatusData(BaseModel):

kodit/infrastructure/bm25/local_bm25_repository.py

@@ -37,7 +37,7 @@ class LocalBM25Repository(BM25Repository):
         """
         self.log = structlog.get_logger(__name__)
         self.index_path = data_dir / "bm25s_index"
-        self.snippet_ids: list[int] = []
+        self.snippet_ids: list[str] = []
         self.stemmer = Stemmer.Stemmer("english")
         self.__retriever: bm25s.BM25 | None = None
 
@@ -76,11 +76,23 @@ class LocalBM25Repository(BM25Repository):
             self.log.warning("Corpus is empty, skipping bm25 index")
             return
 
-        vocab = self._tokenize([doc.text for doc in request.documents])
+        if not self.snippet_ids and (self.index_path / SNIPPET_IDS_FILE).exists():
+            async with aiofiles.open(self.index_path / SNIPPET_IDS_FILE) as f:
+                self.snippet_ids = json.loads(await f.read())
+
+        # Filter out documents that have already been indexed
+        new_documents = [
+            doc for doc in request.documents if doc.snippet_id not in self.snippet_ids
+        ]
+        if not new_documents:
+            self.log.info("No new documents to index")
+            return
+
+        vocab = self._tokenize([doc.text for doc in new_documents])
         self._retriever().index(vocab, show_progress=False)
         self._retriever().save(self.index_path)
         # Replace snippet_ids instead of appending, since the BM25 index is rebuilt
-        self.snippet_ids = [doc.snippet_id for doc in request.documents]
+        self.snippet_ids = [doc.snippet_id for doc in new_documents]
         async with aiofiles.open(self.index_path / SNIPPET_IDS_FILE, "w") as f:
             await f.write(json.dumps(self.snippet_ids))
 
@@ -121,7 +133,7 @@ class LocalBM25Repository(BM25Repository):
         # Filter results by snippet_ids if provided
         filtered_results = []
         for result, score in zip(results[0], scores[0], strict=True):
-            snippet_id = int(result)
+            snippet_id = result
             if score > 0.0 and (
                 request.snippet_ids is None or snippet_id in request.snippet_ids
             ):