a353e1c683
All checks were successful
CI / Lint, PostgreSQL tests, and package (push) Successful in 12s
203 lines
9.7 KiB
Python
203 lines
9.7 KiB
Python
"""Opt-in headed Chromium tests against a disposable *_test PostgreSQL database."""
|
|
|
|
from contextlib import redirect_stderr, redirect_stdout
|
|
from decimal import Decimal
|
|
import gzip
|
|
import io
|
|
import json
|
|
import os
|
|
from pathlib import Path
|
|
from tempfile import TemporaryDirectory
|
|
import unittest
|
|
from unittest.mock import patch
|
|
|
|
from psycopg import sql
|
|
|
|
from car_tracker import database, pipeline
|
|
from car_tracker.collect import search_url
|
|
import test_database_integration as database_tests
|
|
|
|
|
|
@unittest.skipUnless(
|
|
os.environ.get("RUN_BROWSER_TESTS") == "1" and os.environ.get("TEST_DATABASE_URL"),
|
|
"RUN_BROWSER_TESTS=1 and TEST_DATABASE_URL are required",
|
|
)
|
|
class BrowserPipelineTests(unittest.TestCase):
|
|
tables = database_tests.DatabaseIntegrationTests.tables
|
|
|
|
@classmethod
|
|
def setUpClass(cls):
|
|
cls.conn = database_tests.DatabaseIntegrationTests.connect()
|
|
cls.addClassCleanup(cls.conn.close)
|
|
database.migrate(cls.conn)
|
|
|
|
def setUp(self):
|
|
self.conn.execute(sql.SQL("TRUNCATE {} RESTART IDENTITY").format(
|
|
sql.SQL(", ").join(sql.Identifier("silver", table) for table in self.tables)
|
|
))
|
|
temporary = TemporaryDirectory()
|
|
self.addCleanup(temporary.cleanup)
|
|
self.directory = Path(temporary.name)
|
|
self.config = self.directory / "searches.toml"
|
|
self.config.write_text(
|
|
'[[searches]]\nfamily = "synthetic"\ntext = "synthetic car"\n',
|
|
encoding="utf-8",
|
|
)
|
|
self.output = self.directory / "raw"
|
|
self.search = {"list_id": 123, "price": ["12000"]}
|
|
self.detail = {
|
|
"list_id": 123, "price": ["12500.50"],
|
|
"body": "Synthetic private description.\nRetain this raw text exactly.",
|
|
"location": {"lat": 48.125, "lng": 2.375},
|
|
}
|
|
|
|
def run_browser_pipeline(self, *, detail_status=200, import_status="succeeded"):
|
|
# Import only when opted in; never replace the browser, collector or importer.
|
|
from playwright.sync_api import BrowserContext
|
|
|
|
original_new_page = BrowserContext.new_page
|
|
search = search_url("synthetic car", 1)
|
|
detail = "https://www.leboncoin.fr/ad/voitures/123"
|
|
requests, responses, unexpected = [], [], []
|
|
|
|
def handle(route):
|
|
url = route.request.url
|
|
requests.append(url)
|
|
if url == search:
|
|
status = 200
|
|
props = {"searchData": {"ads": [self.search], "total": 1, "max_pages": 1}}
|
|
elif url == detail:
|
|
status = detail_status
|
|
# Even the 403 has valid data: ignoring HTTP status must fail this test.
|
|
props = {"ad": self.detail}
|
|
else:
|
|
unexpected.append(url)
|
|
route.abort()
|
|
return
|
|
data = json.dumps({"props": {"pageProps": props}})
|
|
route.fulfill(
|
|
status=status, content_type="text/html",
|
|
body=f'<!doctype html><script id="__NEXT_DATA__" type="application/json">{data}</script>',
|
|
)
|
|
|
|
def new_page(context, *args, **kwargs):
|
|
page = original_new_page(context, *args, **kwargs)
|
|
page.route("**/*", handle)
|
|
page.on("response", lambda response: responses.append((response.url, response.status)))
|
|
return page
|
|
|
|
stdout, stderr = io.StringIO(), io.StringIO()
|
|
# Preserve DISPLAY/browser paths, but discard inherited libpq/homelab settings.
|
|
environment = {key: value for key, value in os.environ.items() if not key.startswith("PG")}
|
|
environment["DATABASE_URL"] = os.environ["TEST_DATABASE_URL"]
|
|
with patch.dict(os.environ, environment, clear=True), patch.object(
|
|
BrowserContext, "new_page", new_page,
|
|
), patch("car_tracker.collect.time.sleep"), redirect_stdout(stdout), redirect_stderr(stderr):
|
|
code = pipeline.main([
|
|
"--headed", "--config", str(self.config), "--output", str(self.output),
|
|
"--delay", "1", "--max-pages", "1", "--max-details", "1",
|
|
])
|
|
|
|
self.assertEqual(unexpected, [])
|
|
self.assertEqual(requests, [search, detail])
|
|
self.assertEqual(responses, [(search, 200), (detail, detail_status)])
|
|
self.assertNotIn(self.detail["body"], stdout.getvalue() + stderr.getvalue())
|
|
self.assertNotIn("48.125", stdout.getvalue() + stderr.getvalue())
|
|
self.assertNotIn("2.375", stdout.getvalue() + stderr.getvalue())
|
|
paths = list(self.output.rglob("manifest.json"))
|
|
self.assertEqual(len(paths), 1)
|
|
path = paths[0]
|
|
manifest = json.loads(path.read_text(encoding="utf-8"))
|
|
state = json.loads((path.parent / "pipeline.json").read_text(encoding="utf-8"))
|
|
self.assertEqual(state["collection_status"], manifest["status"])
|
|
self.assertEqual(state["import_status"], import_status)
|
|
if import_status == "succeeded":
|
|
self.assertEqual(state["import_summary"]["status"], "imported")
|
|
self.assertEqual(state["import_summary"]["source_status"], manifest["status"])
|
|
self.assertEqual(state["import_summary"]["rejected"], 0)
|
|
self.assertEqual(self.conn.execute(
|
|
"SELECT run_id::text, source_status, manifest FROM silver.collection_runs"
|
|
).fetchall(), [(manifest["run_id"], manifest["status"], manifest)])
|
|
self.assertEqual(self.conn.execute(
|
|
"SELECT family, status, pages_observed, records_reported FROM silver.search_coverage"
|
|
).fetchall(), [("synthetic", "complete", 1, 1)])
|
|
with gzip.open(path.parent / manifest["searches"][0]["pages"][0]["file"], "rt") as stream:
|
|
records = [json.loads(line) for line in stream]
|
|
self.assertEqual(len(records), 1)
|
|
self.assertEqual(records[0]["record_type"], "search")
|
|
self.assertEqual(records[0]["payload"], self.search)
|
|
return code, path, manifest
|
|
|
|
def test_complete_run_preserves_raw_and_typed_detail_and_replays(self):
|
|
code, path, manifest = self.run_browser_pipeline()
|
|
self.assertEqual(code, 0)
|
|
self.assertEqual(manifest["status"], "complete")
|
|
with gzip.open(path.parent / manifest["details"]["listings"]["123"]["file"], "rt") as stream:
|
|
records = [json.loads(line) for line in stream]
|
|
self.assertEqual(len(records), 1)
|
|
self.assertEqual(records[0]["record_type"], "detail")
|
|
self.assertEqual(records[0]["payload"], self.detail)
|
|
self.assertEqual(self.conn.execute(
|
|
"SELECT source_listing_id, record_type, price_eur, currency, description, "
|
|
"latitude, longitude, raw_payload FROM silver.observations ORDER BY record_type"
|
|
).fetchall(), [
|
|
("123", "detail", Decimal("12500.50"), "EUR", self.detail["body"],
|
|
48.125, 2.375, self.detail),
|
|
("123", "search", Decimal("12000"), "EUR", None, None, None, self.search),
|
|
])
|
|
self.assertEqual(self.conn.execute(
|
|
"SELECT source_listing_id, status FROM silver.detail_coverage"
|
|
).fetchall(), [("123", "complete")])
|
|
before = self.state()
|
|
self.assertEqual({table: len(rows) for table, rows in before.items()}, {
|
|
"collection_runs": 1, "raw_files": 2, "search_coverage": 1,
|
|
"detail_coverage": 1, "listings": 1, "observations": 2, "rejected_records": 0,
|
|
})
|
|
with redirect_stdout(io.StringIO()):
|
|
replay = database.import_run(self.conn, path)
|
|
self.assertEqual(replay, {
|
|
"run_id": manifest["run_id"], "status": "already_imported", "rejected": 0,
|
|
})
|
|
self.assertEqual(self.state(), before)
|
|
|
|
def test_blocked_detail_imports_partial_run_and_retains_search(self):
|
|
code, path, manifest = self.run_browser_pipeline(detail_status=403)
|
|
self.assertEqual(code, 1)
|
|
self.assertEqual(manifest["status"], "partial")
|
|
self.assertEqual(manifest["error_type"], "SourceAccessError")
|
|
self.assertEqual(manifest["details"]["status"], "partial")
|
|
self.assertEqual(manifest["details"]["listings"]["123"]["http_status"], 403)
|
|
self.assertFalse((path.parent / "detail-123.jsonl.gz").exists())
|
|
self.assertEqual(self.conn.execute(
|
|
"SELECT source_listing_id, status, reason, http_status FROM silver.detail_coverage"
|
|
).fetchall(), [("123", "failed", "SourceAccessError", 403)])
|
|
self.assertEqual(self.conn.execute(
|
|
"SELECT source_listing_id, record_type, price_eur, raw_payload FROM silver.observations"
|
|
).fetchall(), [("123", "search", Decimal("12000"), self.search)])
|
|
self.assertEqual(self.conn.execute(
|
|
"SELECT record_type, expected_records, accepted_records, rejected_records FROM silver.raw_files"
|
|
).fetchall(), [("search", 1, 1, 0)])
|
|
self.assertEqual(self.conn.execute("SELECT count(*) FROM silver.rejected_records").fetchone(), (0,))
|
|
|
|
def state(self):
|
|
return {table: self.conn.execute(sql.SQL(
|
|
"SELECT to_jsonb(t) FROM silver.{} t ORDER BY to_jsonb(t)::text"
|
|
).format(sql.Identifier(table))).fetchall() for table in self.tables}
|
|
|
|
def test_interrupt_after_commit_is_unknown_and_replay_is_safe(self):
|
|
def committed_then_interrupted(conn, path):
|
|
database.import_run(conn, path)
|
|
raise KeyboardInterrupt
|
|
|
|
with patch.object(pipeline, "import_run", side_effect=committed_then_interrupted):
|
|
code, path, _ = self.run_browser_pipeline(import_status="unknown")
|
|
self.assertEqual(code, 130)
|
|
before = self.state()
|
|
self.assertEqual(len(before["observations"]), 2)
|
|
self.assertEqual(database.import_run(self.conn, path)["status"], "already_imported")
|
|
self.assertEqual(self.state(), before)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
unittest.main()
|