🔴 Blocking finding from PR #773 latest review:
`get_chunk_bbox_and_page_from_qdrant` (`search/context.py:199`) still
declared `doc_id: int | str` and passed the raw value into
`MatchValue(value=doc_id)` at lines 239 and 256 without `str()`
coercion. After this branch's startup backfill normalises every Qdrant
`doc_id` payload to a string, an `int` filter would silently match zero
points — the function would return `(None, None)` instead of the
chunk bbox / page, and PDF highlight overlays would fail in production.
Take option 2 from the reviewer's two suggestions (annotation
tightening over inline coercion): the producer side of this PR has
already narrowed every other `doc_id` annotation to `str`, so this
function is the last hold-out. Pushing the contract into the type
system means `ty` will catch any future regression at the call site.
Production callers in `api/visualization.py` and `auth/viz_routes.py`
already pass `doc_id` (str) verbatim after the recent merge with
master's chunk_index-first refactor, so no caller-side changes needed.
Update the 9 calls in `tests/unit/test_chunk_bbox_helper.py` to use
string literals (`"42"` / `"99"` / `"1"`) instead of integers. The
mock doesn't validate `MatchValue` value types, so the tests passed
with stale int doc_ids today — but they were exercising a path
production no longer takes.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Resolves both 🟡 important issues from the latest review:
1. `page_number` was unconditionally overwritten in `viz_routes.py:696` even
when Qdrant's payload lacked the field, clobbering the value resolved
from `chunk_context.page_number`. The new helper returns each field
independently and both call sites only overwrite via `is not None`
guards, matching the existing logic in `visualization.py`.
2. The ~60-line `if chunk_index is not None: ... else: ...` Qdrant scroll
block was duplicated between `api/visualization.py` and
`auth/viz_routes.py`. Extracted into `get_chunk_bbox_and_page_from_qdrant`
in `search/context.py` alongside the existing private `_get_chunk_*_from_qdrant`
helpers; both routes now share ~12 lines of caller code.
New unit tests at `tests/unit/test_chunk_bbox_helper.py` cover the indexed
and offset paths, the `(bbox, None)` regression case, and graceful
degradation on Qdrant strict-mode 400 (which also closes nit #4).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
