edward/agenda
1
0
Fork 0
Code Issues 10 Pull requests Projects Releases Packages Wiki Activity

Handle structured conference metadata

Browse source
This commit is contained in:
Edward Betts 2026-08-24 06:59:15 +02:00
parent d1c2c51776
commit 9958632242
2 changed files with 85 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

37
tests/test_add_new_conference.py
View file

@ -174,6 +174,41 @@ def test_validate_generated_conference_reports_missing_dates() -> None:
add_new_conference.validate_generated_conference(conf)
def test_webpage_to_text_includes_hidden_event_metadata() -> None:
"""Schema.org event metadata should be included even when visually hidden."""
root = lxml.html.fromstring("""
<html><body>
<div style="display:none" itemscope itemtype="http://schema.org/Event">
<meta itemprop="name" content="State of the Map Asia 2026 OSAKA">
<meta itemprop="startDate" content="2026-09-06T18:30">
</div>
</body></html>
""")
text = add_new_conference.webpage_to_text(root)
assert "- name: State of the Map Asia 2026 OSAKA" in text
assert "- startDate: 2026-09-06T18:30" in text
def test_apply_event_metadata_dates_fills_missing_generated_dates() -> None:
"""Structured page dates should recover incomplete model output."""
root = lxml.html.fromstring("""
<div itemscope itemtype="http://schema.org/Event">
<meta itemprop="startDate" content="2026-09-06T18:30">
</div>
""")
conf: dict[str, typing.Any] = {"name": "State of the Map Asia 2026"}
add_new_conference.apply_event_metadata_dates(conf, root)
assert conf["dates"] == {
"status": "exact",
"start": datetime(2026, 9, 6, 18, 30),
"end": datetime(2026, 9, 6, 18, 30),
}
def test_build_prompt_includes_nested_dates_and_series() -> None:
"""The prompt should describe nested dates and known series IDs."""
prompt = add_new_conference.build_prompt(
@ -194,6 +229,8 @@ def test_build_prompt_includes_nested_dates_and_series() -> None:
assert "dates.status" in prompt
assert "- pycascades: PyCascades" in prompt
assert "March 2027" in prompt
assert "For an address written in a non-Latin script" in prompt
assert "If no reliable rendering is available, omit" in prompt
def test_validate_country_normalises_name() -> None: