fix(tags): preserve acronym casing in tag normalization (DC, NSFW)

normalize_tag_name now only capitalizes the first letter of each word and
leaves the rest of the word untouched (was lowercasing the tail, which turned
DC→Dc / NSFW→Nsfw). This matches ml/tag_name._title_word, so a Camie-suggested
tag keeps the exact casing the suggestion UI showed when it round-trips through
POST /api/tags on Accept — addressing "auto-suggested tags must obey
capitalization" and "don't mangle acronyms" in one rule.

Trade-off (operator-chosen): all-caps input no longer folds to Title Case, so
case-variant merging in #714 still folds the dominant lowercase-vs-Title case
but leaves all-caps stylizations distinct (protecting acronyms wins). Tests
updated + a new test documenting acronym preservation / non-folding.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

backend/app/services/tag_service.py

@@ -17,15 +17,19 @@ log = logging.getLogger(__name__)
 
 
 def normalize_tag_name(name: str) -> str:
-    """Canonical tag form (#701): collapse whitespace + Title Case.
+    """Canonical tag form (#701): collapse whitespace + capitalize the first
+    letter of each word, PRESERVING the rest of the word.
 
     Per-word capitalize (NOT str.title(), which mangles apostrophes:
-    don't → Don'T). Lowercasing the word tail also folds ALL-CAPS input
+    don't → Don'T). The word tail is left untouched so acronyms survive
-    (HATSUNE MIKU → Hatsune Miku); acronym casing is sacrificed, an accepted
+    (DC stays DC, NSFW stays NSFW) — operator-revised 2026-06-06: protecting
-    Title-Case trade-off (operator-chosen 2026-06-06).
+    acronyms matters more than folding ALL-CAPS input. This MATCHES
+    ml/tag_name._title_word, so a tag suggested by the Camie tagger keeps the
+    exact casing the suggestion UI showed when it round-trips through the
+    create endpoint on Accept.
     """
     return " ".join(
-        w[:1].upper() + w[1:].lower() for w in (name or "").split()
+        w[:1].upper() + w[1:] for w in (name or "").split()
     )
 
 

tests/test_api_tags.py

@@ -74,13 +74,14 @@ async def test_create_tag_with_kind_prefix(client):
 @pytest.mark.asyncio
 async def test_explicit_kind_overrides_prefix_parsing(client):
     """If caller passes explicit kind, don't re-parse the name —
-    colon and prefix stay literal. #701: still Title-Cased as a single word."""
+    colon and prefix stay literal. #701: first letter capitalized, tail
+    preserved (acronym-safe), so the 'S' in Saber survives."""
     resp = await client.post(
         "/api/tags", json={"name": "character:Saber", "kind": "general"}
     )
     assert resp.status_code == 201
     body = await resp.get_json()
-    assert body["name"] == "Character:saber"
+    assert body["name"] == "Character:Saber"
     assert body["kind"] == "general"
 
 

tests/test_tag_normalize.py

@@ -87,8 +87,9 @@ async def test_dry_run_projection_counts(db):
 
 @pytest.mark.asyncio
 async def test_live_merges_case_variants_and_repoints_images(db):
+    # lowercase vs sentence-case → both canonicalize to "Hatsune Miku" and fold.
     a = await _raw_tag(db, "hatsune miku", TagKind.character)
-    b = await _raw_tag(db, "HATSUNE MIKU", TagKind.character)
+    b = await _raw_tag(db, "Hatsune miku", TagKind.character)
     i1, i2, i3 = await _img(db), await _img(db), await _img(db)
     await _link(db, i1, a.id)
     await _link(db, i2, a.id)
@@ -125,7 +126,7 @@ async def test_live_merges_case_variants_and_repoints_images(db):
 @pytest.mark.asyncio
 async def test_idempotent_second_run_is_noop(db):
     await _raw_tag(db, "kafka", TagKind.character)
-    await _raw_tag(db, "KAFKA", TagKind.character)
+    await _raw_tag(db, "Kafka", TagKind.character)
 
     first = await normalize_existing_tags(db, dry_run=False)
     assert first["groups_processed"] == 1
@@ -138,6 +139,24 @@ async def test_idempotent_second_run_is_noop(db):
     assert again["renamed"] == 0
 
 
+@pytest.mark.asyncio
+async def test_acronym_casing_is_preserved_not_folded(db):
+    """Acronym-preserving rule (2026-06-06): an already-canonical acronym tag
+    is a no-op, and an all-caps variant does NOT fold into a Title-cased one
+    (protecting DC/NSFW from becoming Dc/Nsfw is the deliberate trade)."""
+    await _raw_tag(db, "DC", TagKind.general)
+    proj = await normalize_existing_tags(db, dry_run=True)
+    assert proj["total_changes"] == 0  # "DC" is already canonical
+
+    # All-caps and Title-case are distinct canonical forms → not merged.
+    await _raw_tag(db, "Nsfw", TagKind.general)
+    await _raw_tag(db, "NSFW", TagKind.general)
+    summary = await normalize_existing_tags(db, dry_run=False)
+    assert summary["merged"] == 0
+    assert await _count_named(db, "DC", TagKind.general) == 1
+    assert await _count_named(db, "NSFW", TagKind.general) == 1
+
+
 @pytest.mark.asyncio
 async def test_same_name_different_fandom_not_merged(db):
     f1 = await _raw_tag(db, "Bleach", TagKind.fandom)

tests/test_tag_service.py

@@ -77,7 +77,12 @@ async def test_normalize_tag_name():
     from backend.app.services.tag_service import normalize_tag_name
     assert normalize_tag_name("hatsune miku") == "Hatsune Miku"
     assert normalize_tag_name("  looking   at  viewer ") == "Looking At Viewer"
-    assert normalize_tag_name("HATSUNE MIKU") == "Hatsune Miku"
+    # Acronym-preserving (2026-06-06): only the first letter of each word is
+    # touched, so already-uppercase tails survive (DC stays DC, not Dc).
+    assert normalize_tag_name("DC comics") == "DC Comics"
+    assert normalize_tag_name("dc") == "Dc"
+    assert normalize_tag_name("NSFW variant") == "NSFW Variant"
+    assert normalize_tag_name("HATSUNE MIKU") == "HATSUNE MIKU"
 
 
 @pytest.mark.asyncio