Ontology rendering

[jnu]

Add a new paint module (differentiated from the general render module that operates on RedactedText) to annotate the input pdf with the results of the ontology extraction.

Refactors some existing modules to pass through the appropriate data in the pipeline context.

[sentry]

Bug: pymupdf.Quad(*scaled_points) crashes if region.points has anything other than exactly 4 points.
_{Severity: MEDIUM}

Suggested Fix

Clamp scaled_points to exactly 4 points (e.g., use scaled_points[:4]) or construct the rect directly via pymupdf.Rect(min_x, min_y, max_x, max_y) computed from all points, avoiding the Quad constraint entirely.

Prompt for AI Agent

Review the code at the location below. A potential bug has been identified by an AI
agent.
Verify if this is a real issue. If it is, propose a fix; if not, explain why it's not
valid.

Location: bc2/core/common/ontopainter.py#L127

Potential issue: In `_paint_rect` (line 127), `pymupdf.Quad(*scaled_points)` unpacks all
points from `region.points` as positional arguments. `pymupdf.Quad` accepts exactly 4
corner points (ul, ur, ll, lr). The polygon from Azure DI is converted in `openai.py` by
iterating `range(0, len(polygon), 2)`, which typically yields 4 points (8-float
polygon). However, `SourceChunkBoundingRegion.points` is typed as an unconstrained
`list[tuple[float, float]]`, so if Azure DI returns a polygon with more or fewer than 4
points (e.g., for irregular regions or future API changes),
`pymupdf.Quad(*scaled_points)` will raise a `TypeError` about incorrect number of
arguments, crashing the painting pipeline for that document.

[sentry]

Bug: Subject DOB field is registered twice in painter, causing it to be painted twice on every document.
_{Severity: LOW}

Suggested Fix

Remove the duplicate entry (lines 113-120) or replace it with the intended field configuration (e.g., subject.seq).

Prompt for AI Agent

Review the code at the location below. A potential bug has been identified by an AI
agent.
Verify if this is a real issue. If it is, propose a fix; if not, explain why it's not
valid.

Location: bc2/core/paint/ontology.py#L113-L120

Potential issue: In `bc2/core/paint/ontology.py`, lines 105-120, the
`OntoPainterFieldConfig` for `Subject DOB` is defined twice with identical configuration
(`accessor=lambda report: [subject.dob for subject in report.subjects]`, label `"Subject
DOB"`, same stroke color). This causes the painter to iterate over DOB fields twice per
document: once at lines 105-112 and again at lines 113-120. Each DOB annotation will be
drawn twice on top of itself, wasting rendering time and potentially confusing
downstream users who try to match labels to fields. This is clearly a copy-paste
mistake—one entry should likely be for a different field (e.g., subject `seq`).