test: make extract cli setup deterministic

archivebox/tests/test_cli_extract.py

@@ -1,15 +1,10 @@
 #!/usr/bin/env python3
-"""
-Tests for archivebox extract command.
-Verify extract re-runs extractors on existing snapshots.
-"""
-
-import json
+"""Tests for archivebox extract command."""
 
 import pytest
 
-from archivebox.core.models import Snapshot
-from archivebox.tests.conftest import run_archivebox_cmd, cli_env
+from archivebox.core.models import ArchiveResult, Snapshot
+from archivebox.tests.conftest import cli_env, parse_jsonl_output, run_archivebox_cmd
 
 from archivebox.tests.test_orm_helpers import use_archivebox_db
 
@@ -17,50 +12,56 @@ pytestmark = pytest.mark.django_db(transaction=True)
 
 
 def _create_snapshot(data_dir, env, url="https://example.com"):
-    record = json.dumps({"type": "Snapshot", "url": url})
-    run_archivebox_cmd(
-        ["snapshot", "create"],
+    result = run_archivebox_cmd(
+        ["snapshot", "create", url],
         cwd=data_dir,
-        stdin=f"{record}\n",
         env=env,
         check=True,
     )
+    snapshot = next(record for record in parse_jsonl_output(result.stdout) if record.get("type") == "Snapshot")
+    return snapshot
 
 
 def test_extract_runs_on_existing_snapshots(initialized_archive):
-    """Test that extract command runs on existing snapshots."""
+    """Extract queues a requested plugin for an existing snapshot."""
     env = cli_env(disable_extractors=True)
 
-    _create_snapshot(initialized_archive, env)
+    snapshot = _create_snapshot(initialized_archive, env)
+    snapshot_id = snapshot["id"]
 
-    # Run extract
     result = run_archivebox_cmd(
-        ["extract"],
+        ["extract", "--plugin=title", "--no-wait", snapshot_id],
         cwd=initialized_archive,
         env=env,
         timeout=30,
     )
 
-    # Should complete
-    assert result.returncode in [0, 1]
+    assert result.returncode == 0, result.stderr or result.stdout
+
+    with use_archivebox_db(initialized_archive):
+        archiveresult = ArchiveResult.objects.get(snapshot_id=snapshot_id, plugin="title")
+        extracted_snapshot = Snapshot.objects.get(id=snapshot_id)
+
+    assert archiveresult.status == ArchiveResult.StatusChoices.QUEUED
+    assert extracted_snapshot.retry_at is not None
 
 
 def test_extract_preserves_snapshot_count(initialized_archive):
-    """Test that extract doesn't change snapshot count."""
+    """Extract queues work without creating duplicate snapshots."""
     env = cli_env(disable_extractors=True)
 
-    _create_snapshot(initialized_archive, env)
+    snapshot = _create_snapshot(initialized_archive, env)
 
     with use_archivebox_db(initialized_archive):
         count_before = Snapshot.objects.count()
 
-    # Run extract
-    run_archivebox_cmd(
-        ["extract", "--overwrite"],
+    result = run_archivebox_cmd(
+        ["extract", "--plugin=title", "--no-wait", snapshot["id"]],
         cwd=initialized_archive,
         env=env,
         timeout=30,
     )
+    assert result.returncode == 0, result.stderr or result.stdout
 
     with use_archivebox_db(initialized_archive):
         count_after = Snapshot.objects.count()