playback: add -rr for re-ranking

Author: chapmanjacobd

library/playback/play_actions.py

@@ -92,6 +92,21 @@ def parse_args(action, default_chromecast=None) -> argparse.Namespace:
 --fetch-siblings each -O          # get the first result per directory
 --fetch-siblings if-audiobook -O  # get the first result per directory if 'audiobook' is in the path
 --fetch-siblings always -O        # get 2,000 results per directory
+""",
+    )
+    ordering.add_argument(
+        "--re-rank",
+        "--rerank",
+        "-rr",
+        action=argparse_utils.ArgparseDict,
+        default={},
+        metavar="COLUMN=WEIGHT",
+        help="""Add key/value pairs re-rank sorting by multiple attributes (similar to MCDA)
+-rr 'regex_sort=1 cluster_sort=1 -size=3'
+will sort the playlist by taking into account size, regex, and clusters prioritizing size over the other two.
+
+-u size desc -L 500 --cols path,title,size,time_modified -rs -rr 'regex_sort=1 time_modified=1'
+will get the 500 largest and then sort by regex_sort and time_modified
 """,
     )
 
@@ -407,7 +422,37 @@ def process_playqueue(args) -> None:
     if args.play_in_order:
         media = db_media.natsort_media(args, media)
 
-    if args.regex_sort:
+    if args.re_rank:
+        import pandas
+
+        from library.utils import pd_utils
+
+        df = pandas.DataFrame(media)
+
+        if args.regex_sort:
+            from library.text import regex_sort
+
+            sorted_media = regex_sort.sort_dicts(args, media)
+            df = pd_utils.from_dict_add_path_rank(df, sorted_media, "regex_sort")
+            log.debug("regex-sort: %s", t.elapsed())
+        elif args.cluster_sort:
+            from library.text import cluster_sort
+
+            sorted_media = cluster_sort.sort_dicts(args, media)
+            df = pd_utils.from_dict_add_path_rank(df, sorted_media, "cluster_sort")
+            log.debug("cluster-sort: %s", t.elapsed())
+
+        column_weights = {
+            k.lstrip("-"): {
+                "direction": "desc" if k.startswith("-") else "asc",
+                "weight": v or 1,
+            }
+            for k, v in args.re_rank.items()
+        }
+        df = pd_utils.rank_dataframe(df, column_weights)
+        media = df.to_dict(orient="records")
+        log.debug("re-rank: %s", t.elapsed())
+    elif args.regex_sort:
         from library.text import regex_sort
 
         media = regex_sort.sort_dicts(args, media)

library/utils/pd_utils.py

@@ -85,3 +85,9 @@ def rank_dataframe(df, column_weights):
 def count_category(df, key_name):
     df[f"{key_name}_count"] = df.groupby(key_name)[key_name].transform("size")
     return df
+
+
+def from_dict_add_path_rank(df, sorted_media, new_rank_column_name):
+    rank_dict = {item["path"]: rank + 1 for rank, item in enumerate(sorted_media)}
+    df[new_rank_column_name] = df["path"].map(rank_dict)
+    return df

tests/utils/test_web.py

@@ -251,6 +251,7 @@ def test_parent_property():
     assert str(web_path.parent.parent.parent.parent.parent.parent.parent) == "https://<netloc>"
     assert str(web_path.parent.parent.parent.parent.parent.parent.parent.parent) == "https://<netloc>"
 
+
 @pytest.mark.parametrize(
     "base_url, href, expected",
     [
@@ -264,7 +265,7 @@ def test_parent_property():
         ("https://unli.xyz/", "//example.com/ch/", "https://example.com/ch/"),
         ("https://unli.xyz/diskprices", "ftp://example.com/ch/", "ftp://example.com/ch/"),
         ("https://unli.xyz/diskprices", "ssh://example.com/ch/", "ssh://example.com/ch/"),
-    ]
+    ],
 )
 def test_construct_absolute_url(base_url, href, expected):
     result = construct_absolute_url(base_url, href)