add_hostname_column

Author: chapmanjacobd

library/playback/media_printer.py

@@ -102,6 +102,7 @@ def media_printer(args, data, units: str | None = "media", media_len=None) -> No
     print_args = getattr(args, "print", "")
     cols = getattr(args, "cols", [])
     m_columns = db_utils.columns(args, "media")
+    print(action)
 
     if (args.limit or args.timeout_size) and "path" in data[0].keys():
         new_data = []

library/utils/path_utils.py

@@ -230,8 +230,9 @@ def folder_utime(folder_path, times: tuple[int, int] | tuple[float, float]):
 
 def domain_from_url(tracker):
     url = urlparse(tracker)
-    domain = url.netloc.rsplit(":")[0].lower()
-    return domain
+    if url.hostname:
+        return url.hostname.lower()
+    return url.netloc.rsplit(":")[0].lower()
 
 
 def tld_from_url(url):

library/utils/sqlgroups.py

@@ -6,6 +6,21 @@
 perf_randomize_using_ids = filter_engine.perf_randomize_using_ids
 
 
+def add_hostname_column(args):
+    m_columns = db_utils.columns(args, "media")
+    from library.utils.path_utils import domain_from_url
+
+    args.db.register_function(domain_from_url, deterministic=True)
+    if m_columns and "hostname" not in m_columns:
+        args.db.execute("ALTER TABLE media ADD COLUMN hostname TEXT")
+        m_columns.add("hostname")
+
+    with args.db.conn:
+        args.db.execute("UPDATE media SET hostname = domain_from_url(path) WHERE hostname IS NULL")
+        args.db.execute("CREATE INDEX IF NOT EXISTS media_hostname_idx ON media (hostname)")
+    return m_columns
+
+
 def fs_sql(args, limit) -> tuple[str, dict]:
     m_columns = db_utils.columns(args, "media")
     args.table, m_columns = sql_utils.search_filter(args, m_columns)
@@ -122,7 +137,7 @@ def historical_media(args):
 
 
 def construct_links_query(args, limit) -> tuple[str, dict]:
-    m_columns = db_utils.columns(args, "media")
+    m_columns = add_hostname_column(args)
     args.table, m_columns = sql_utils.search_filter(args, m_columns)
 
     if getattr(args, "category", None) and "category" in m_columns:
@@ -167,7 +182,7 @@ def construct_links_query(args, limit) -> tuple[str, dict]:
 
 
 def construct_tabs_query(args) -> tuple[str, dict]:
-    m_columns = db_utils.columns(args, "media")
+    m_columns = add_hostname_column(args)
     args.table, m_columns = sql_utils.search_filter(args, m_columns)
 
     query = f"""WITH media_history as (

tests/utils/test_path_utils.py

@@ -104,6 +104,23 @@ def test_safe_join(user_path, expected):
     assert utils.p(result) == utils.p(expected)
 
 
+@pytest.mark.parametrize(
+    ("url", "expected"),
+    [
+        ("http://example.com/path/to/file.txt", "example.com"),
+        ("https://www.example.org/another/file.jpg", "www.example.org"),
+        ("ftp://fileserver.net/pub/document.pdf", "fileserver.net"),
+        ("http://example.com:8080/path", "example.com"),
+        ("https://user:password@example.com/path", "example.com"),
+        ("http://[::1]/path", "::1"),
+        ("http://127.0.0.1/path", "127.0.0.1"),
+        ("https://sub.domain.example.com/path", "sub.domain.example.com"),
+    ],
+)
+def test_domain_from_url(url, expected):
+    assert path_utils.domain_from_url(url) == expected
+
+
 @pytest.mark.parametrize(
     ("url", "expected_parent_path", "expected_filename"),
     [