Video2Guide / tests /test_metadata.py
Claude
Third-pass audit fixes: migration setdefaults, robust JS mark builder
7562488
Raw History Blame Contribute Delete
5.62 kB
import json
from pathlib import Path
import pytest
from PIL import Image
from app.pipeline.html_gen import (
Segment,
metadata_to_segment,
render_from_metadata,
render_guide,
)
from app.pipeline.metadata import (
CURRENT_SCHEMA_VERSION,
KeyTerm,
MathEquation,
PageMetadata,
SegmentMetadata,
dump_metadata,
load_metadata,
migrate_v1_to_v2,
page_metadata_from_segments,
segment_to_metadata,
)
def _make_image(path: Path) -> None:
path.parent.mkdir(parents=True, exist_ok=True)
Image.new("RGB", (32, 24), (123, 200, 50)).save(path, "JPEG")
def _sample_segment(idx: int = 0) -> Segment:
return Segment(
index=idx,
start_seconds=float(idx * 5),
end_seconds=float(idx * 5 + 5),
image_filename=f"scene_{idx:04d}.jpg",
image_data_uri=None,
narration=["First paragraph.", "Second paragraph."],
ocr_text="ocr text",
alt_draft="An image showing a slide.",
section_title=f"Section {idx + 1}",
key_terms=[KeyTerm(term="Photon", definition="A particle of light.")],
math_equations=[
MathEquation(latex="E=mc^2", aria_label="E equals m c squared"),
],
include=True,
)
def test_pagemetadata_roundtrips_through_json(tmp_path):
page = page_metadata_from_segments(
title="Round-Trip",
segments=[_sample_segment(0), _sample_segment(1)],
subtitle="A subtitle",
module="Module 7",
meta_lines=["2 segments", "Source: 0:10"],
)
out = tmp_path / "meta.json"
dump_metadata(out, page)
loaded = load_metadata(out)
assert loaded.schema_version == CURRENT_SCHEMA_VERSION
assert loaded.title == "Round-Trip"
assert loaded.subtitle == "A subtitle"
assert loaded.module == "Module 7"
assert len(loaded.segments) == 2
assert loaded.segments[0].key_terms[0].term == "Photon"
assert loaded.segments[0].math_equations[0].latex == "E=mc^2"
assert loaded.segments[0].narration == ["First paragraph.", "Second paragraph."]
def test_segment_to_and_from_metadata(tmp_path):
seg = _sample_segment(3)
seg_meta = segment_to_metadata(seg)
assert isinstance(seg_meta, SegmentMetadata)
assert seg_meta.index == 3
assert seg_meta.alt_text == "An image showing a slide."
assert seg_meta.key_terms[0].definition == "A particle of light."
img = tmp_path / seg.image_filename
_make_image(img)
rebuilt = metadata_to_segment(seg_meta, tmp_path, inline_images=False)
assert isinstance(rebuilt, Segment)
assert rebuilt.section_title == seg.section_title
assert rebuilt.alt_draft == seg.alt_draft
assert rebuilt.narration == seg.narration
def test_migrate_v1_to_v2_fills_defaults():
v1 = {
"title": "Old Save",
"eyebrow": "Legacy",
"segments": [
{
"index": 0,
"section_title": "Intro",
"alt_text": "Slide",
"narration": "Single string narration.",
"ocr_text": "OCR",
"include": True,
}
],
}
v2 = migrate_v1_to_v2(v1)
page = PageMetadata.model_validate(v2)
assert page.schema_version == CURRENT_SCHEMA_VERSION
assert page.subtitle is None
assert page.module is None
assert page.frames_dir == "frames"
seg = page.segments[0]
assert seg.narration == ["Single string narration."]
assert seg.key_terms == []
assert seg.math_equations == []
assert seg.image_filename == "scene_0000.jpg"
# Learner-feature fields (added later than the original v1→v2
# migration spec) must still receive empty defaults so downstream
# consumers of the raw migrated dict aren't surprised.
assert v2["segments"][0]["summary"] == ""
assert v2["segments"][0]["self_check_questions"] == []
assert seg.summary == ""
assert seg.self_check_questions == []
def test_migrate_v1_via_load_metadata_with_no_schema_version_key(tmp_path):
v1_json = tmp_path / "v1.json"
v1_json.write_text(
json.dumps(
{
"title": "Old",
"segments": [
{"index": 0, "narration": ["one", "two"], "section_title": "S", "alt_text": "A"}
],
}
),
encoding="utf-8",
)
page = load_metadata(v1_json)
assert page.schema_version == CURRENT_SCHEMA_VERSION
assert page.segments[0].narration == ["one", "two"]
def test_render_from_metadata_idempotent(tmp_path):
img0 = tmp_path / "scene_0000.jpg"
img1 = tmp_path / "scene_0001.jpg"
_make_image(img0)
_make_image(img1)
page = page_metadata_from_segments(
title="Idempotent",
segments=[_sample_segment(0), _sample_segment(1)],
subtitle="Sub",
module="Mod 3",
)
page.frames_dir = str(tmp_path)
html_a = render_from_metadata(page, frames_dir=tmp_path, mode="single")
html_b = render_from_metadata(page, frames_dir=tmp_path, mode="single")
assert html_a == html_b
assert "Idempotent" in html_a
def test_render_from_metadata_links_images_in_guide_mode(tmp_path):
img0 = tmp_path / "scene_0000.jpg"
_make_image(img0)
page = page_metadata_from_segments(
title="Linked",
segments=[_sample_segment(0)],
)
page.frames_dir = str(tmp_path)
html = render_from_metadata(page, frames_dir=tmp_path, mode="guide")
# In guide mode (non-inline), images are referenced by path.
assert 'src="static/scene_0000.jpg"' in html
assert "data:image/jpeg;base64," not in html