fix(suggestions): resolve bare WD14 character names to existing "(Fandom)" tags

WD14 emits bare names (e.g. "Ruby Rose") while curated character/fandom
tags carry a fandom suffix ("Ruby Rose (RWBY)"). Two gaps caused a
duplicate, fandom-less tag to be created when users accepted a WD14
suggestion for an image that already had the curated version:

- Suggestion compute: bare character/copyright suggestions are now
  suppressed when the image already carries a `Name (Fandom)` variant,
  so the duplicate chip no longer surfaces.
- Accept endpoint: if the exact-name lookup misses for character/fandom
  kinds, we try a `Name (%)`-scoped lookup and attach the existing tag
  only when exactly one candidate matches. Zero or multiple matches
  fall through to prior behavior to avoid silently guessing across
  ambiguous fandoms.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
This commit is contained in:
2026-04-21 20:30:49 -04:00
parent 03184e7f1c
commit 0257a79a15
2 changed files with 44 additions and 3 deletions
+23
View File
@@ -8,6 +8,7 @@ The result is grouped by category and annotated with whether each tag already
exists in the Tag table (the modal dims disallowed auto-creations).
"""
from __future__ import annotations
import re
from typing import Iterable
from sqlalchemy import select
@@ -106,6 +107,21 @@ def _existing_tag_names() -> set[str]:
return {row[0] for row in db.session.query(Tag.name).all()}
# Matches `Name (Fandom)` — used to recognize that an applied character/copyright
# tag covers a bare WD14 suggestion like `Name`.
_FANDOM_SUFFIX_RE = re.compile(r'^(.+?) \([^()]+\)$')
def _fandom_suffix_prefixes(names: Iterable[str]) -> set[str]:
"""Extract the `Name` portion from any `Name (Fandom)` entries in `names`."""
out: set[str] = set()
for n in names:
m = _FANDOM_SUFFIX_RE.match(n)
if m:
out.add(m.group(1))
return out
def _wd14_suggestions(image_id: int, cfg: dict, already: set[str]) -> list[dict]:
wd14_ver = cfg.get('wd14_model_version', _DEFAULTS['wd14_model_version'])
preds = (
@@ -113,6 +129,11 @@ def _wd14_suggestions(image_id: int, cfg: dict, already: set[str]) -> list[dict]
.filter_by(image_id=image_id, model_version=wd14_ver)
.all()
)
# Bare names covered by an already-applied `Name (Fandom)` tag. WD14 emits
# bare names (e.g. "Ruby Rose") even when the curated tag is fandom-scoped
# ("Ruby Rose (RWBY)"); without this, the same character resurfaces as a
# duplicate suggestion and accepting it creates a second, fandom-less tag.
applied_fandom_prefixes = _fandom_suffix_prefixes(already)
out: list[dict] = []
for p in preds:
threshold = _category_threshold(p.tag_category, cfg)
@@ -121,6 +142,8 @@ def _wd14_suggestions(image_id: int, cfg: dict, already: set[str]) -> list[dict]
canonical = _canonicalize_wd14_name(p.tag_name, p.tag_category)
if canonical in already:
continue
if p.tag_category in ('character', 'copyright') and canonical in applied_fandom_prefixes:
continue
out.append({
'name': canonical,
'category': p.tag_category,