Merge branch 'master' into archive-result

tests/test_add.py

@@ -81,4 +81,13 @@ def test_add_updates_history_json_index(tmp_path, process, disable_extractors_di
 
     with open(archived_item_path / "index.json", "r") as f:
         output_json = json.load(f)
-    assert output_json["history"] != {}
+    assert output_json["history"] != {}
+
+def test_extract_input_uses_only_passed_extractors(tmp_path, process):
+    subprocess.run(["archivebox", "add", "http://127.0.0.1:8080/static/example.com.html", "--extract", "wget"],
+                    capture_output=True)
+    
+    archived_item_path = list(tmp_path.glob('archive/**/*'))[0]
+
+    assert (archived_item_path / "warc").exists()
+    assert not (archived_item_path / "singlefile.html").exists()

tests/test_remove.py

@@ -70,6 +70,29 @@ def test_remove_domain(tmp_path, process, disable_extractors_dict):
 
     assert count == 0
 
+
+def test_remove_tag(tmp_path, process, disable_extractors_dict):
+    subprocess.run(['archivebox', 'add', 'http://127.0.0.1:8080/static/example.com.html'], capture_output=True, env=disable_extractors_dict)
+    subprocess.run(['archivebox', 'add', 'http://127.0.0.1:8080/static/iana.org.html'], capture_output=True, env=disable_extractors_dict)
+    assert list((tmp_path / "archive").iterdir()) != []
+    
+    conn = sqlite3.connect("index.sqlite3")
+    c = conn.cursor()
+    c.execute("INSERT INTO core_tag (id, name, slug) VALUES (2, 'test-tag', 'test-tag')")
+    snapshot_ids = c.execute("SELECT id from core_snapshot")
+    c.executemany('INSERT INTO core_snapshot_tags (snapshot_id, tag_id) VALUES (?, 2)', list(snapshot_ids))
+    conn.commit()
+
+    remove_process = subprocess.run(['archivebox', 'remove', '--filter-type=tag', 'test-tag', '--yes', '--delete'], capture_output=True)
+
+    assert len(list((tmp_path / "archive").iterdir())) == 0
+
+    count = c.execute("SELECT COUNT() from core_snapshot").fetchone()[0]
+    conn.commit()
+    conn.close()
+
+    assert count == 0
+
 def test_remove_before(tmp_path, process, disable_extractors_dict):
     subprocess.run(['archivebox', 'add', 'http://127.0.0.1:8080/static/example.com.html'], capture_output=True, env=disable_extractors_dict)
     subprocess.run(['archivebox', 'add', 'http://127.0.0.1:8080/static/iana.org.html'], capture_output=True, env=disable_extractors_dict)

tests/test_title.py

@@ -5,7 +5,7 @@ from .fixtures import *
 
 def test_title_is_htmlencoded_in_index_html(tmp_path, process, disable_extractors_dict):
     """
-    https://github.com/pirate/ArchiveBox/issues/330
+    https://github.com/ArchiveBox/ArchiveBox/issues/330
     Unencoded content should not be rendered as it facilitates xss injections
     and breaks the layout.
     """