removal of search segmenter

Author: gdetari

background_tasks.py

@@ -55,99 +55,4 @@ Paginated list endpoints return:
 
 ## Active Route Inventory
 
-Applications:
-
-- `POST /v2/applications`
-
-Languages:
-
-- `GET /v2/languages`
-- `POST /v2/languages`
-
-Categories:
-
-- `GET /v2/categories`
-- `GET /v2/categories/{category_id}`
-- `POST /v2/categories`
-
-Tags:
-
-- `GET /v2/tags`
-- `POST /v2/tags`
-- `DELETE /v2/tags/{tag_id}`
-- `POST /v2/texts/{text_id}/tags/{tag_id}`
-- `DELETE /v2/texts/{text_id}/tags/{tag_id}`
-- `POST /v2/segments/{segment_id}/tags/{tag_id}`
-- `DELETE /v2/segments/{segment_id}/tags/{tag_id}`
-
-Persons:
-
-- `GET /v2/persons`
-- `GET /v2/persons/{person_id}`
-- `POST /v2/persons`
-- `PATCH /v2/persons/{person_id}`
-
-Texts:
-
-- `GET /v2/texts`
-- `GET /v2/texts/{text_id}`
-- `POST /v2/texts`
-- `PATCH /v2/texts/{text_id}`
-- `GET /v2/texts/{text_id}/editions`
-- `POST /v2/texts/{text_id}/editions`
-
-Content Search:
-
-- `GET /v2/content-search`
-
-Editions:
-
-- `GET /v2/editions/{edition_id}`
-- `GET /v2/editions/{edition_id}/content`
-- `PATCH /v2/editions/{edition_id}/content`
-- `DELETE /v2/editions/{edition_id}`
-- `GET /v2/editions/{edition_id}/related`
-- `GET /v2/editions/{edition_id}/segments/related`
-- `GET /v2/editions/{edition_id}/segmentations`
-- `POST /v2/editions/{edition_id}/segmentations`
-- `GET /v2/editions/{edition_id}/alignments`
-- `POST /v2/editions/{edition_id}/alignments`
-- `GET /v2/editions/{edition_id}/pagination`
-- `POST /v2/editions/{edition_id}/pagination`
-- `GET /v2/editions/{edition_id}/table-of-contents`
-- `POST /v2/editions/{edition_id}/table-of-contents`
-- `GET /v2/editions/{edition_id}/bibliographic`
-- `POST /v2/editions/{edition_id}/bibliographic`
-- `GET /v2/editions/{edition_id}/durchens`
-- `POST /v2/editions/{edition_id}/durchens`
-
-Annotation objects by ID:
-
-- `GET /v2/segmentations/{segmentation_id}`
-- `DELETE /v2/segmentations/{segmentation_id}`
-- `GET /v2/alignments/{alignment_id}`
-- `DELETE /v2/alignments/{alignment_id}`
-- `GET /v2/paginations/{pagination_id}`
-- `DELETE /v2/paginations/{pagination_id}`
-- `GET /v2/table-of-contents/{toc_id}`
-- `DELETE /v2/table-of-contents/{toc_id}`
-- `GET /v2/bibliographic/{bibliographic_id}`
-- `DELETE /v2/bibliographic/{bibliographic_id}`
-- `GET /v2/durchens/{durchen_id}`
-- `DELETE /v2/durchens/{durchen_id}`
-
-Segments:
-
-- `GET /v2/segments/{segment_id}`
-- `GET /v2/segments/{segment_id}/content`
-- `GET /v2/segments/{segment_id}/related`
-- `GET /v2/segments/search`
-
-Operational:
-
-- `GET /__/health`
-- `GET /openapi.json`
-- `GET /docs`
-- `GET /redoc`
-
-There is currently no registered `/v2/relations/...` route and no registered `/v2/schema/openapi` route. Text relations are exposed on text records (`translation_of`, `commentary_of`, `translations`, `commentaries`) and related edition lookup is exposed through `GET /v2/editions/{edition_id}/related`.
+Use `GET /docs`

docs/api-documentation/README.md

@@ -1,16 +1,13 @@
 import logging
 from typing import TYPE_CHECKING, Annotated
 
-import httpx
 from fastapi import APIRouter, Depends, Path, Query, status
 
-from config import settings
 from dependencies import OptionalAppHeader, get_api_key, get_db, get_storage
-from exceptions import DataNotFoundError, InvalidRequestError
+from exceptions import DataNotFoundError
 from models.annotation import SegmentWithContextOutput
-from models.requests import DirectRelatedSegmentsQueryParams, SegmentsQueryParams
+from models.requests import DirectRelatedSegmentsQueryParams
 from models.responses import PaginatedResponse
-from models.search import SearchFilter, SearchResponse, SearchResult
 
 if TYPE_CHECKING:
     from database import Database
@@ -102,101 +99,6 @@ async def untag_segment(
     await db.tag.untag_segment(segment_id, tag_id)
 
 
-@router.get(
-    "/search",
-    summary="Search segments",
-    description="Search segments using the external search API.",
-)
-async def search_segments(
-    query: Annotated[str, Query(description="Search query")],
-    params: Annotated[SegmentsQueryParams, Query()],
-    _api_key: Annotated[str, Depends(get_api_key)],
-    db: Annotated[Database, Depends(get_db)],
-) -> SearchResponse:
-    """Search segments."""
-    filter_obj = SearchFilter(title=params.title) if params.title else None
-
-    try:
-        logger.info("Forwarding search request to %s/search", settings.search_api_url)
-
-        request_params = {
-            "query": query,
-            "search_type": params.search_type,
-            "limit": params.limit,
-            "return_text": params.return_text,
-        }
-        if filter_obj and filter_obj.title:
-            request_params["title"] = filter_obj.title
-
-        async with httpx.AsyncClient(timeout=60) as client:
-            response = await client.get(f"{settings.search_api_url}/search", params=request_params)
-            response.raise_for_status()
-            search_response_data = response.json()
-
-    except httpx.RequestError:
-        logger.exception("Error calling search API")
-        raise InvalidRequestError("Failed to call search API") from None
-
-    enriched_results = []
-
-    for result_item in search_response_data.get("results", []):
-        segment_id = result_item.get("id")
-        if not segment_id:
-            enriched_results.append(
-                SearchResult(
-                    id=result_item.get("id", ""),
-                    distance=result_item.get("distance", 0.0),
-                    entity=result_item.get("entity", {}),
-                    segmentation_ids=[],
-                )
-            )
-            continue
-
-        try:
-            segment = await db.segment.get(segment_id)
-            segmentation_ids = await db.segment.find_by_span(
-                edition_id=segment.edition_id,
-                start=segment.span.start,
-                end=segment.span.end,
-            )
-            enriched_results.append(
-                SearchResult(
-                    id=result_item.get("id", ""),
-                    distance=result_item.get("distance", 0.0),
-                    entity=result_item.get("entity", {}),
-                    segmentation_ids=segmentation_ids,
-                )
-            )
-        except DataNotFoundError:
-            logger.warning("Segment %s not found, skipping segmentation mapping", segment_id)
-            enriched_results.append(
-                SearchResult(
-                    id=result_item.get("id", ""),
-                    distance=result_item.get("distance", 0.0),
-                    entity=result_item.get("entity", {}),
-                    segmentation_ids=[],
-                )
-            )
-        except Exception:
-            logger.exception("Error processing segment %s", segment_id)
-            enriched_results.append(
-                SearchResult(
-                    id=result_item.get("id", ""),
-                    distance=result_item.get("distance", 0.0),
-                    entity=result_item.get("entity", {}),
-                    segmentation_ids=[],
-                )
-            )
-
-    return SearchResponse.model_validate(
-        {
-            "query": search_response_data.get("query", query),
-            "results": enriched_results,
-            "count": len(enriched_results),
-        }
-    )
-
-
 @router.get(
     "/{segment_id}",
     summary="Get segment",

routers/segments.py

@@ -3,7 +3,6 @@
 
 from fastapi import APIRouter, BackgroundTasks, Depends, Path, Query, status
 
-from background_tasks import trigger_search_segmenter
 from content_search import ContentSearchService
 from dependencies import OptionalAppHeader, get_api_key, get_content_search, get_db, get_storage
 from identifier import generate_id
@@ -125,7 +124,6 @@ async def create_edition(
         segmentation=data.segmentation,
     )
 
-    background_tasks.add_task(trigger_search_segmenter, edition_id)
     background_tasks.add_task(content_search.index_edition, edition_id, db, storage)
 
     return IdResponse(id=edition_id)

routers/texts.py

@@ -7,7 +7,6 @@
 from contextlib import suppress
 from pathlib import Path
 from typing import LiteralString, cast
-from unittest.mock import patch
 
 import httpx
 from neo4j import GraphDatabase
@@ -260,20 +259,6 @@ async def content_search(_opensearch_endpoint: str) -> AsyncGenerator[ContentSea
         await service.close()
 
 
-@pytest.fixture(autouse=True)
-def mock_search_segmenter():
-    """
-    Prevent background threads / network calls during tests.
-
-    These helpers are "fire-and-forget" and call external services; tests should never
-    hit the network or spawn those background threads.
-    """
-    with (
-        patch("background_tasks.trigger_search_segmenter"),
-        patch("background_tasks.trigger_delete_search_segments"),
-    ):
-        yield
-
 
 class MockS3Storage:
     """In-memory S3 storage mock for tests."""