Audit category selector identities

tools/prompt_map_audit.py

@@ -118,6 +118,10 @@ AUDIT_DOC_SNIPPETS: tuple[tuple[str, str], ...] = (
         "docs/prompt-pool-routing-map.md",
         "multi-seed route sweeps",
     ),
+    (
+        "docs/prompt-pool-routing-map.md",
+        "category identity validation",
+    ),
     (
         "docs/prompt-pool-routing-map.md",
         "node documentation validation",
@@ -141,6 +145,15 @@ ALLOWED_PROMPT_ROW_READS: set[tuple[str, str]] = {
 
 ROUTE_POLICY_ACTION_EXCLUSIONS = {"default"}
 ROUTE_POLICY_POSITION_EXCLUSIONS = {"any"}
+RESERVED_CATEGORY_CHOICES = {
+    "auto_weighted",
+    "auto_full",
+    "woman",
+    "man",
+    "couple",
+    "group_or_layout",
+    "custom_random",
+}
 NODE_DOC_PATHS = (
     "docs/prompt-pool-routing-map.md",
     "README.md",
@@ -418,6 +431,117 @@ def _json_reference_errors(paths: list[Path]) -> list[tuple[str, str, str]]:
     return errors
 
 
+def _identity_key(value: Any) -> str:
+    return re.sub(r"\s+", " ", str(value or "").strip()).lower()
+
+
+def _identity_location(category: dict[str, Any], subcategory: dict[str, Any] | None = None) -> str:
+    category_slug = str(category.get("slug") or category.get("name") or "category").strip()
+    location = f"categories.{category_slug}"
+    if subcategory is not None:
+        sub_slug = str(subcategory.get("slug") or subcategory.get("name") or "subcategory").strip()
+        location = f"{location}.subcategories.{sub_slug}"
+    return location
+
+
+def _add_duplicate_identity_error(
+    seen: dict[str, str],
+    *,
+    key: str,
+    location: str,
+    label: str,
+    errors: list[tuple[str, str, str]],
+) -> None:
+    if not key:
+        return
+    previous = seen.get(key)
+    if previous is not None:
+        errors.append(("(category library)", location, f"duplicate {label}: {key!r} also used at {previous}"))
+        return
+    seen[key] = location
+
+
+def _category_identity_errors(paths: list[Path]) -> list[tuple[str, str, str]]:
+    # Use the normalized library because that is the selector surface exposed by
+    # category/preset nodes and exact subcategory routing.
+    _ = paths
+    try:
+        categories = category_policy.load_category_library()
+    except Exception as exc:
+        return [("(category library)", "categories", f"cannot load categories: {exc}")]
+
+    errors: list[tuple[str, str, str]] = []
+    category_names: dict[str, str] = {}
+    category_slugs: dict[str, str] = {}
+    exact_selectors: dict[str, str] = {}
+    exact_slug_selectors: dict[str, str] = {}
+
+    for category in categories:
+        location = _identity_location(category)
+        name = _identity_key(category.get("name"))
+        slug = _identity_key(category.get("slug"))
+        if not name:
+            errors.append(("(category library)", f"{location}.name", "missing normalized category name"))
+        if not slug:
+            errors.append(("(category library)", f"{location}.slug", "missing normalized category slug"))
+        if name in RESERVED_CATEGORY_CHOICES or slug in RESERVED_CATEGORY_CHOICES:
+            errors.append(
+                (
+                    "(category library)",
+                    location,
+                    "category identity collides with reserved built-in selector",
+                )
+            )
+        _add_duplicate_identity_error(category_names, key=name, location=location, label="category name", errors=errors)
+        _add_duplicate_identity_error(category_slugs, key=slug, location=location, label="category slug", errors=errors)
+
+        subcategory_names: dict[str, str] = {}
+        subcategory_slugs: dict[str, str] = {}
+        for subcategory in category.get("subcategories") or []:
+            if not isinstance(subcategory, dict):
+                continue
+            sub_location = _identity_location(category, subcategory)
+            sub_name = _identity_key(subcategory.get("name"))
+            sub_slug = _identity_key(subcategory.get("slug"))
+            if not sub_name:
+                errors.append(("(category library)", f"{sub_location}.name", "missing normalized subcategory name"))
+            if not sub_slug:
+                errors.append(("(category library)", f"{sub_location}.slug", "missing normalized subcategory slug"))
+            _add_duplicate_identity_error(
+                subcategory_names,
+                key=sub_name,
+                location=sub_location,
+                label=f"subcategory name in {category.get('name')}",
+                errors=errors,
+            )
+            _add_duplicate_identity_error(
+                subcategory_slugs,
+                key=sub_slug,
+                location=sub_location,
+                label=f"subcategory slug in {category.get('name')}",
+                errors=errors,
+            )
+
+            exact_selector = _identity_key(category_policy.exact_subcategory_selector(category, subcategory))
+            slug_selector = f"{slug} / {sub_slug}" if slug and sub_slug else ""
+            _add_duplicate_identity_error(
+                exact_selectors,
+                key=exact_selector,
+                location=sub_location,
+                label="exact subcategory selector",
+                errors=errors,
+            )
+            _add_duplicate_identity_error(
+                exact_slug_selectors,
+                key=slug_selector,
+                location=sub_location,
+                label="category/subcategory slug selector",
+                errors=errors,
+            )
+
+    return errors
+
+
 def _hardcore_template_metadata_errors(paths: list[Path]) -> list[tuple[str, str, str]]:
     errors: list[tuple[str, str, str]] = []
     for path in paths:
@@ -1064,6 +1188,13 @@ def main() -> int:
         return 1
     print("OK: all JSON pool references and item template axes resolve.")
 
+    print("\n# Category Identity Validation")
+    category_identity_errors = _category_identity_errors(category_paths)
+    if category_identity_errors:
+        print_table(("Source", "Path", "Issue"), category_identity_errors)
+        return 1
+    print("OK: category and subcategory identities are unique and selector-safe.")
+
     print("\n# Hardcore Template Metadata Validation")
     hardcore_metadata_errors = _hardcore_template_metadata_errors(category_paths)
     if hardcore_metadata_errors: