First draft of tag_sort.py

samuellieberman · samuellieberman · commit b9316fa1f488 · 2024-07-26T15:57:30.000-06:00
Adds functionality to "group" tags based on an automatically generated "lineage" from a root grandparent tag to most given grandchild tags.

Needs major improvements:
* Better naming of variables.
* Better encapsulation of sort information
* Support for circular lineages
* Comments
* More?
diff --git a/tagstudio/src/core/tag_sort.py b/tagstudio/src/core/tag_sort.py
@@ -0,0 +1,243 @@
+from enum import Enum, Flag, auto
+from typing import Any, Callable
+
+from src.core.constants import TAG_COLORS
+from src.core.library import Library
+
+
+class TagSortProperty(int, Enum):
+    MATCHES_PREFIX = auto()
+    CATEGORY = auto()
+    LINEAGE_DEPTH = auto()
+    DESCENDENT_COUNT = auto()
+    NAME = auto()
+    COLOR = auto()
+    ID = auto()
+
+
+class TagSortDirection(Flag):
+    ASC = False
+    DESC = True
+
+
+Sort = list[tuple[TagSortProperty, TagSortDirection]]
+
+# _default_sort must contain LINEAGE_DEPTH and ID
+_default_sort: Sort = [
+    (TagSortProperty.LINEAGE_DEPTH, TagSortDirection.DESC),
+    (TagSortProperty.DESCENDENT_COUNT, TagSortDirection.DESC),
+    (TagSortProperty.NAME, TagSortDirection.ASC),
+    (TagSortProperty.COLOR, TagSortDirection.ASC),
+    (TagSortProperty.ID, TagSortDirection.ASC),
+]
+
+def normalize_sort(old_sort: Sort = None) -> Sort:
+    if old_sort is None:
+        old_sort = []
+    old_sort.extend(_default_sort)
+
+    sorted_properties: set[TagSortProperty] = set()
+    new_sort = []
+    for sort_property, sort_direction in old_sort:
+        if sort_property not in sorted_properties:
+            sorted_properties.add(sort_property)
+            new_sort.append((sort_property, sort_direction))
+
+    return new_sort
+
+
+def add_sort_property(
+    new_property: TagSortProperty,
+    new_direction: TagSortDirection,
+    old_sort: Sort = None,
+) -> Sort:
+    new_sort = [(new_property, new_direction)]
+
+    if old_sort is not None:
+        new_sort.extend(old_sort)
+
+    return normalize_sort(new_sort)
+
+
+def reverse_sort(old_sort: Sort) -> Sort:
+    new_sort: Sort = []
+    for sort_property, sort_direction in old_sort:
+        if sort_direction is TagSortDirection.ASC:
+            new_direction = TagSortDirection.DESC
+        else:
+            new_direction = TagSortDirection.ASC
+
+        new_sort.append((sort_property, new_direction))
+    return new_sort
+
+
+def get_key(
+    lib: Library, tag_id_list, sort: Sort = _default_sort
+) -> Callable[[int], list[Any]]:
+    sort = normalize_sort(sort)
+
+    outer_sort: Sort = []
+    lineage_direction: TagSortDirection
+    inner_sort: Sort = []
+
+    outer = True
+    for sort_property, sort_direction in sort:
+        if sort_property is TagSortProperty.LINEAGE_DEPTH:
+            outer = False
+            lineage_direction = sort_direction
+            continue
+
+        if outer:
+            outer_sort.append((sort_property, sort_direction))
+        else:
+            inner_sort.append((sort_property, sort_direction))
+
+    if lineage_direction is TagSortDirection.DESC:
+        inner_sort = reverse_sort(inner_sort)
+    def key(tag_id: int) -> list[Any]:
+        nonlocal lineage_direction
+        canonical_lineage: Any = _get_canonical_lineage(
+            lib, outer_sort, inner_sort, tag_id, tag_id_list
+        )
+        if lineage_direction is TagSortDirection.DESC:
+            canonical_lineage = _ReverseComparison(canonical_lineage)
+
+        key_items = _get_basic_key_items(lib, tag_id, outer_sort)
+        key_items.append(canonical_lineage)
+
+        print(key_items)
+        return key_items
+
+    return key
+
+
+def _get_basic_key_items(
+    lib: Library, tag_id: int, sort: Sort, tag_id_set: set[int] = None, prefix: str = None
+) -> list[Any]:
+    key_items = []
+    for sort_property, sort_direction in sort:
+        key_item: Any = None
+        match sort_property:
+            case TagSortProperty.MATCHES_PREFIX:
+                key_item = lib.get_tag(tag_id).name.lower().startswith(prefix.lower())
+            # case TagSortProperty.CATEGORY:
+            # case TagSortProperty.LINEAGE_DEPTH:
+            case TagSortProperty.DESCENDENT_COUNT:
+                tag_cluster = lib.get_tag_cluster(tag_id)
+                if tag_id_set is not None:
+                    key_item = len(tag_id_set.intersection(set(tag_cluster)))
+                else:
+                    key_item = len(tag_cluster)
+            case TagSortProperty.NAME:
+                key_item = lib.get_tag(tag_id).display_name(lib)
+            case TagSortProperty.COLOR:
+                key_item = TAG_COLORS.index(lib.get_tag(tag_id).color.lower())
+            case TagSortProperty.ID:
+                key_item = tag_id
+
+        if sort_direction is TagSortDirection.DESC:
+            key_item = _ReverseComparison(key_item)
+
+        key_items.append(key_item)
+
+    return key_items
+
+
+class _ReverseComparison:
+    def __init__(self, inner: Any):
+        self.inner = inner
+
+    def __lt__(self, other):
+        return other.inner.__lt__(self.inner)
+
+    def __le__(self, other):
+        return other.inner.__le__(self.inner)
+
+    def __eq__(self, other):
+        return other.inner.__eq__(self.inner)
+
+    def __ne__(self, other):
+        return other.inner.__ne__(self.inner)
+
+    def __gt__(self, other):
+        return other.inner.__gt__(self.inner)
+
+    def __ge__(self, other):
+        return other.inner.__ge__(self.inner)
+    
+    def __str__(self) -> str:
+        return f"rev:{self.inner}"
+    
+    def __repr__(self) -> str:
+        return str(self)
+
+def _get_canonical_lineage(
+    lib: Library,
+    outer_sort: Sort,
+    inner_sort: Sort,
+    tag_id: int,
+    tag_id_list: list[int],
+    last_generation_ids=set([-1]),
+    first_gen=True,
+) -> list[list[Any]]:
+    ancestor_id_queue: list[int] = [tag_id]
+    encountered_tag_ids: set[int] = set(last_generation_ids)
+    encountered_tag_ids.add(tag_id)
+
+    this_generation_ids: set[int] = set()
+
+    while ancestor_id_queue:
+        next_ancestor_id = ancestor_id_queue.pop()
+        parent_ids: set[int] = set(lib.get_tag(next_ancestor_id).subtag_ids)
+
+        if first_gen and not parent_ids:
+            this_generation_ids.add(next_ancestor_id)
+
+        if not first_gen and last_generation_ids.intersection(parent_ids):
+            this_generation_ids.add(next_ancestor_id)
+
+        #TODO: make this work for looping relationships
+        for parent_id in parent_ids:
+            if parent_id not in encountered_tag_ids:
+                encountered_tag_ids.add(parent_id)
+                ancestor_id_queue.append(parent_id)
+
+    outer_key = _get_basic_key_items(lib, tag_id, outer_sort)
+    first_in_generation_id = None
+    first_in_generation_inner_key_item = None
+    for challenger_id in this_generation_ids:
+        if challenger_id not in tag_id_list:
+            continue
+
+        challenger_outer_key_item = _get_basic_key_items(lib, challenger_id, outer_sort)
+        if challenger_outer_key_item != outer_key:
+            continue
+
+        challenger_inner_key_item = _get_basic_key_items(lib, challenger_id, inner_sort)
+        if (
+            first_in_generation_id is None
+            or challenger_inner_key_item < first_in_generation_inner_key_item
+        ):
+            first_in_generation_id = challenger_id
+            first_in_generation_inner_key_item = challenger_inner_key_item
+    
+    lineage: list[list[Any]] = []
+    if first_in_generation_id is not None:
+        lineage = [first_in_generation_inner_key_item]
+        lineage.extend(
+            _get_canonical_lineage(
+                lib,
+                outer_sort,
+                inner_sort,
+                tag_id,
+                tag_id_list,
+                set([first_in_generation_id]),
+                False,
+            )
+        )
+    elif this_generation_ids:
+        lineage = _get_canonical_lineage(
+            lib, outer_sort, inner_sort, tag_id, tag_id_list, this_generation_ids, False
+        )
+    
+    return lineage
diff --git a/tagstudio/src/qt/modals/tag_database.py b/tagstudio/src/qt/modals/tag_database.py
@@ -13,6 +13,7 @@
 )
 
 from src.core.library import Library
+from src.core.tag_sort import get_key
 from src.qt.widgets.panel import PanelWidget, PanelModal
 from src.qt.widgets.tag import TagWidget
 from src.qt.modals.build_tag import BuildTagPanel
@@ -103,8 +104,10 @@ def update_tags(self, query: str):
             # Get tag ids to keep this behaviorally identical
             tags = [t.id for t in self.lib.tags]
 
+        sorted_tags = sorted(tags, key=get_key(self.lib, tags))
+
         first_id_set = False
-        for tag_id in tags:
+        for tag_id in sorted_tags:
             if not first_id_set:
                 self.first_tag_id = tag_id
                 first_id_set = True
diff --git a/tagstudio/src/qt/modals/tag_search.py b/tagstudio/src/qt/modals/tag_search.py
@@ -18,6 +18,7 @@
 )
 
 from src.core.library import Library
+from src.core.tag_sort import get_key
 from src.core.palette import ColorType, get_tag_color
 from src.qt.widgets.panel import PanelWidget
 from src.qt.widgets.tag import TagWidget
@@ -111,7 +112,9 @@ def update_tags(self, query: str = ""):
         found_tags = self.lib.search_tags(query, include_cluster=True)[: self.tag_limit]
         self.first_tag_id = found_tags[0] if found_tags else None
 
-        for tag_id in found_tags:
+        sorted_tags = sorted(found_tags, key=get_key(self.lib, found_tags))
+
+        for tag_id in sorted_tags:
             c = QWidget()
             l = QHBoxLayout(c)
             l.setContentsMargins(0, 0, 0, 0)
