fix(tagging): skip a damaged WD14 label list in the preview (#431)

py/api/autotag_routes.py

@@ -44,6 +44,8 @@ BATCH_JOB_BODY_MAX = 32 * 1024
 SKIPPED_PREVIEW_BODY_MAX = 64 * 1024
 SKIPPED_PREVIEW_LEASE = "Skipped tags preview"
 SKIPPED_PREVIEW_SAMPLE = 12
+# WD14 label lists the preview found missing or unreadable (logged once each).
+_MISSING_LABEL_LISTS = set()
 REVIEW_REQUIRED = (
     "Replacing tags needs a review of each image. Use Preview tagging review, "
     "or choose Skip or Merge for a batch run."
@@ -760,8 +762,11 @@ class AutotagRoutesMixin:
             try:
                 names = wd14_label_names(path)
             except (OSError, ValueError) as exc:
-                self.logger.debug(f"No WD14 label list for {model_type}: {exc}")
+                if path not in _MISSING_LABEL_LISTS:  # logged once per list
+                    _MISSING_LABEL_LISTS.add(path)
+                    self.logger.debug(f"No WD14 label list for {model_type}: {exc}")
                 continue
+            _MISSING_LABEL_LISTS.discard(path)
             lists.append(MODELS[model_type]["name"])
             for name in names:
                 known.setdefault(name.casefold(), (1, name))

py/tagging_defaults.py

@@ -204,8 +204,8 @@ def read_wd14_label_rows(path) -> List[Dict[str, str]]:
 
 def wd14_label_names(path) -> Tuple[str, ...]:
     """The names a WD14 label list can suggest (categories 0 and 4), cached
-    by ``(mtime_ns, size)``. OSError and ValueError propagate (the caller
-    skips a missing or unreadable list)."""
+    by ``(mtime_ns, size)``. OSError (missing) and ValueError (unreadable)
+    propagate; the caller skips that list."""
     key = os.fspath(path)
     stat = os.stat(key)
     stamp = (stat.st_mtime_ns, stat.st_size)
@@ -213,9 +213,13 @@ def wd14_label_names(path) -> Tuple[str, ...]:
         cached = _label_cache.get(key)
     if cached is not None and cached[0] == stamp:
         return cached[1]
+    try:
+        rows = read_wd14_label_rows(key)
+    except csv.Error as exc:  # a damaged list (a NUL byte, an oversized field)
+        raise ValueError(f"unreadable label list: {exc}") from None
     names = tuple(
         name
-        for row in read_wd14_label_rows(key)
+        for row in rows
         if row.get("category", "0") in WD14_SUGGESTED_CATEGORIES
         and (name := (row.get("name") or "").strip())
     )

tests/test_skipped_tags.py

@@ -5,6 +5,7 @@ reader and the filter in ``AutoTagService.generate_tag_suggestions``. No model
 is loaded: ``_scored_tags`` is replaced by a fixed list.
 """
 
+import csv
 import os
 import tempfile
 import unicodedata
@@ -190,6 +191,15 @@ class TestWd14Labels(unittest.TestCase):
             self.assertEqual(wd14_label_names(self.path), ("blurry", "sky"))
             self.assertEqual(reader.call_count, 2)
 
+    def test_a_damaged_list_is_a_value_error_for_the_caller_to_skip(self):
+        # A field over the csv module's size limit raises csv.Error.
+        big = "x" * (csv.field_size_limit() + 1)
+        self.path.write_text(
+            f"tag_id,name,category,count\n1,{big},0,1\n", encoding="utf-8"
+        )
+        with self.assertRaises(ValueError):
+            wd14_label_names(self.path)
+
     def test_a_missing_list_raises_for_the_caller_to_skip(self):
         with self.assertRaises(OSError):
             wd14_label_names(self.path)

web/js/workspace/system-skipped-tags.mjs

@@ -107,7 +107,6 @@ export function skippedTagsBlock({ api, scope, drafts, delay = PREVIEW_DELAY_MS
     input.id = 'tag-skipped-input';
     input.autocomplete = 'off';
     input.placeholder = SKIPPED_COPY.placeholder;
-    input.maxLength = MAX_ENTRY_CHARS + 10;
     input.setAttribute('aria-describedby', 'tag-skipped-problem');
     const add = createButton(SKIPPED_COPY.add);
     add.id = 'tag-skipped-add';
@@ -116,6 +115,7 @@ export function skippedTagsBlock({ api, scope, drafts, delay = PREVIEW_DELAY_MS
     problem.id = 'tag-skipped-problem';
     problem.setAttribute('role', 'alert');
     const chips = el('div', undefined, 'tags skip-chips');
+    chips.setAttribute('role', 'group');
     chips.setAttribute('aria-label', SKIPPED_COPY.chips);
     const preview = el('p', '', 'small skip-preview');
     preview.setAttribute('role', 'status');