77 lines
2.9 KiB
Python
77 lines
2.9 KiB
Python
|
"""Unit tests for per-source /search result shaping."""
|
||
|
|
from __future__ import annotations
|
||
|
|
|
||
|
|
import pathlib
|
||
|
|
import sys
|
||
|
|
|
||
|
|
sys.path.insert(0, str(pathlib.Path(__file__).resolve().parents[1]))
|
||
|
|
|
||
|
|
from app.links import build_result # noqa: E402
|
||
|
|
|
||
|
|
|
||
|
|
def _chunk(envelope_id="paperless:119", dist=0.34, chunk_index=2, text="hit text"):
|
||
|
|
return {"envelope_id": envelope_id, "dist": dist, "chunk_index": chunk_index, "text": text}
|
||
|
|
|
||
|
|
|
||
|
|
class TestBuildResultPaperless:
|
||
|
|
def test_shape_and_link(self):
|
||
|
|
envelope = {"source": "paperless", "entities": []}
|
||
|
|
result = build_result(_chunk(envelope_id="paperless:119"), envelope)
|
||
|
|
assert result["source"] == "paperless"
|
||
|
|
assert result["link"] == "https://paper.kapala.org/documents/119/details"
|
||
|
|
assert "subject" not in result
|
||
|
|
assert "mail_ui_url" not in result
|
||
|
|
|
||
|
|
|
||
|
|
class TestBuildResultGmail:
|
||
|
|
def test_shape_and_headers(self):
|
||
|
|
envelope = {
|
||
|
|
"source": "gmail",
|
||
|
|
"entities": [
|
||
|
|
{
|
||
|
|
"type": "headers",
|
||
|
|
"from": {"name": "Promocja Lunchroom", "address": "promocja@lunchroom.pl"},
|
||
|
|
"subject": "-20% w wybranych restauracjach",
|
||
|
|
"date_raw": "Wed, 9 May 2018 10:03:32 +0200",
|
||
|
|
}
|
||
|
|
],
|
||
|
|
}
|
||
|
|
result = build_result(_chunk(envelope_id="<msgid@example.com>"), envelope)
|
||
|
|
assert result["source"] == "gmail"
|
||
|
|
assert result["subject"] == "-20% w wybranych restauracjach"
|
||
|
|
assert result["from"] == "Promocja Lunchroom <promocja@lunchroom.pl>"
|
||
|
|
assert result["date"] == "Wed, 9 May 2018 10:03:32 +0200"
|
||
|
|
assert result["link"] is None
|
||
|
|
assert result["mail_ui_url"] is None
|
||
|
|
|
||
|
|
def test_from_without_name_falls_back_to_address(self):
|
||
|
|
envelope = {
|
||
|
|
"source": "gmail",
|
||
|
|
"entities": [
|
||
|
|
{"type": "headers", "from": {"name": None, "address": "a@b.com"}, "subject": "s", "date_raw": "d"}
|
||
|
|
],
|
||
|
|
}
|
||
|
|
result = build_result(_chunk(), envelope)
|
||
|
|
assert result["from"] == "a@b.com"
|
||
|
|
|
||
|
|
def test_missing_headers_entity_yields_none_fields(self):
|
||
|
|
envelope = {"source": "gmail", "entities": []}
|
||
|
|
result = build_result(_chunk(), envelope)
|
||
|
|
assert result["subject"] is None
|
||
|
|
assert result["from"] is None
|
||
|
|
assert result["date"] is None
|
||
|
|
|
||
|
|
|
||
|
|
class TestBuildResultEdgeCases:
|
||
|
|
def test_unknown_source_gets_null_link_no_extra_fields(self):
|
||
|
|
result = build_result(_chunk(), {"source": "mystery", "entities": []})
|
||
|
|
assert result["source"] == "mystery"
|
||
|
|
assert result["link"] is None
|
||
|
|
assert "subject" not in result
|
||
|
|
|
||
|
|
def test_missing_envelope_does_not_crash(self):
|
||
|
|
# e.g. a chunk whose envelope vanished between embedding and query time.
|
||
|
|
result = build_result(_chunk(), None)
|
||
|
|
assert result["source"] == "unknown"
|
||
|
|
assert result["link"] is None
|