refactor(ebook-search): simplify search and phrase matching

This commit is contained in:
2026-07-24 11:38:51 -04:00
parent 0028237579
commit e010756e09
16 changed files with 490 additions and 557 deletions
+6 -11
View File
@@ -13,7 +13,7 @@ from python.ebook_search.api.dependencies import ( # noqa: TC001 FastAPI resolv
AppEngine,
AppHttpClient,
)
from python.ebook_search.api.web import templates
from python.ebook_search.api.web import error_response, templates
from python.ebook_search.embeddings import embed_missing_chunks, embedding_model_stats
from python.ebook_search.ingest import ingest_configured_paths
from python.ebook_search.protected_phrases.generate_ngrams import generate_candidate_phrases_for_books
@@ -50,7 +50,7 @@ async def scan_library(request: Request, config: AppConfig, session: AsyncDbSess
await session.commit()
except Exception as error:
logger.exception("ebook_admin_scan_failed")
return templates.TemplateResponse(request, "partials/error.html", {"message": str(error)}, status_code=500)
return error_response(request, error)
logger.info(f"ebook_admin_scan_complete {count=}")
if count > 0:
@@ -65,7 +65,7 @@ async def generate_all_phrases(request: Request, config: AppConfig, engine: AppE
result = await generate_candidate_phrases_for_books(engine, config)
except Exception as error:
logger.exception("ebook_admin_generate_phrases_failed")
return templates.TemplateResponse(request, "partials/error.html", {"message": str(error)}, status_code=500)
return error_response(request, error)
logger.info(
f"ebook_admin_generate_phrases_complete {result.books_seen=} {result.books_built=} {result.candidate_phrases=}"
@@ -128,7 +128,7 @@ async def run_phrase_judgment(
result = await judge_candidate_phrases_for_books(engine, config, source_ids=source_ids)
except Exception as error:
logger.exception("ebook_admin_judge_phrases_failed")
return templates.TemplateResponse(request, "partials/error.html", {"message": str(error)}, status_code=500)
return error_response(request, error)
logger.info(
f"ebook_admin_judge_phrases_complete {result.books_seen=} {result.books_judged=} {result.books_failed=} "
@@ -161,7 +161,7 @@ async def embed_missing(
await session.commit()
except Exception as error:
logger.exception("ebook_admin_embed_missing_failed")
return templates.TemplateResponse(request, "partials/error.html", {"message": str(error)}, status_code=500)
return error_response(request, error)
logger.info(f"ebook_admin_embed_missing_complete {count=}")
return templates.TemplateResponse(
@@ -192,12 +192,7 @@ async def embed_all(
logger.info(f"ebook_admin_embed_all_batch_complete {batches=} {count=} {total=}")
except Exception as error:
logger.exception(f"ebook_admin_embed_all_failed {batches=} {total=}")
return templates.TemplateResponse(
request,
"partials/error.html",
{"message": f"Embed all failed after {total} chunks in {batches} batches: {error}"},
status_code=500,
)
return error_response(request, f"Embed all failed after {total} chunks in {batches} batches: {error}")
logger.info(f"ebook_admin_embed_all_complete {batches=} {total=}")
return templates.TemplateResponse(
+2 -9
View File
@@ -15,6 +15,7 @@ from python.ebook_search.api.dependencies import (
from python.ebook_search.api.judge_tasks import is_judging_book, pop_book_judgment_outcome, start_book_phrase_judgment
from python.ebook_search.api.web import templates
from python.ebook_search.protected_phrases.generate_ngrams import recalculate_candidate_phrases_for_book
from python.ebook_search.protected_phrases.store import count_protected_phrases
from python.fastapi_tools import AsyncDbSession # noqa: TC001 FastAPI resolves this annotated dependency at runtime
from python.orm.richie import EbookCandidatePhrase, EbookChapter, EbookChunk, EbookProtectedPhrase, EbookSource
@@ -71,14 +72,6 @@ async def get_judged_candidate_count(session: AsyncSession, book_id: int) -> int
)
async def get_protected_count(session: AsyncSession, book_id: int) -> int:
"""Return the number of protected phrases for one book."""
return (
await session.scalar(select(func.count(EbookProtectedPhrase.id)).where(EbookProtectedPhrase.book_id == book_id))
or 0
)
async def get_candidates(session: AsyncSession, book_id: int) -> list[EbookCandidatePhrase]:
"""Return the indexed candidates for one book."""
return list(
@@ -122,7 +115,7 @@ async def book_detail(source_id: int, request: Request, session: AsyncDbSession)
chunk_count = await get_chunk_count(session, source.id)
candidate_count = await get_candidate_count(session, source.id)
judged_candidate_count = await get_judged_candidate_count(session, source.id)
protected_count = await get_protected_count(session, source.id)
protected_count = await count_protected_phrases(session, source.id)
candidates = await get_candidates(session, source.id)
protected_phrases = await get_protected_phrases(session, source.id)
else:
+2 -2
View File
@@ -16,7 +16,7 @@ from python.ebook_search.api.dependencies import ( # noqa: TC001 FastAPI resolv
AppEngine,
AppHttpClient,
)
from python.ebook_search.api.web import templates
from python.ebook_search.api.web import error_response, templates
from python.ebook_search.guardrails import (
CitationReport,
is_confident,
@@ -95,7 +95,7 @@ async def search(
)
except Exception as error:
logger.exception("ebook_search_request_failed")
return templates.TemplateResponse(request, "partials/error.html", {"message": str(error)}, status_code=500)
return error_response(request, error)
answer_start = perf_counter()
answer, low_confidence, citation_report = await build_answer(client, query, response, config)