feat(wordpress): only assign tags that are established

_resolve_wp_tag_ids created a WordPress tag for every keyword the rewriter
invented, up to 12 per post. That is where the 3.055 tags for 955 posts
came from - 1.683 of them used exactly once, 473 attached to no post at
all. The categories are stable now, but the tags would simply grow back.

Three changes, all on the write path:

A proposed tag has to appear for wordpress_new_tag_min_proposals (3)
different articles before it is created. Proposals are counted in the new
tag_proposals table, keyed on (name, article), so re-publishing an article
does not inflate its own count. Tags that already exist in WordPress are
assigned as before - the gate only guards creation.

Only the first wordpress_max_tags_per_post (5) tags reach WordPress. The
full list still feeds the category rules, which were validated against it.

The lookup fallback of reusing the first search hit is gone. It filed
"Camping" under the unrelated existing tag "Campingplatz" whenever the
exact tag was missing, which quietly produced wrong tags rather than none.

If the proposal bookkeeping fails, nothing is creatable that run: existing
tags still get assigned and the taxonomy stays put, rather than falling
back to creating everything.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

backend/app/wordpress.py

@@ -13,6 +13,7 @@ from urllib.parse import quote_plus, urlparse
 from urllib.request import Request, urlopen
 
 from . import categorize
+from . import repositories
 from .config import get_settings
 
 
@@ -91,7 +92,25 @@ def _selected_tags_from_meta(meta_json: str | None) -> list[str]:
     return tags
 
 
-def _resolve_wp_tag_ids(*, base_url: str, auth_header: str, tags: list[str]) -> list[int]:
+def _resolve_wp_tag_ids(
+    *,
+    base_url: str,
+    auth_header: str,
+    tags: list[str],
+    creatable: set[str],
+) -> list[int]:
+    """Map tag names to WordPress tag IDs, creating only established ones.
+
+    A name matches an existing tag only on an exact (case-insensitive) name.
+    The former fallback of taking the first search hit filed "Camping" under
+    the unrelated existing tag "Campingplatz" whenever the exact tag was
+    missing.
+
+    `creatable` holds the casefolded names that cleared the proposal threshold
+    and may be created in WordPress. Everything else is dropped with a log line
+    rather than turned into a new tag - an empty set means this run adds nothing
+    to the taxonomy.
+    """
     ids: list[int] = []
     seen: set[int] = set()
     for tag in tags:
@@ -113,12 +132,12 @@ def _resolve_wp_tag_ids(*, base_url: str, auth_header: str, tags: list[str]) ->
                     if row_name.casefold() == name.casefold():
                         tag_id = rid
                         break
-                if tag_id is None:
-                    for row in result:
-                        if isinstance(row, dict) and int(row.get("id", 0) or 0) > 0:
-                            tag_id = int(row.get("id", 0))
-                            break
             if tag_id is None:
+                if name.casefold() not in creatable:
+                    _logger.info(
+                        "Schlagwort '%s' noch nicht etabliert - wird nicht angelegt", name
+                    )
+                    continue
                 created = _wp_request(
                     base_url=base_url,
                     auth_header=auth_header,
@@ -130,6 +149,7 @@ def _resolve_wp_tag_ids(*, base_url: str, auth_header: str, tags: list[str]) ->
                     rid = int(created.get("id", 0) or 0)
                     if rid > 0:
                         tag_id = rid
+                        _logger.info("Schlagwort '%s' in WordPress neu angelegt (#%s)", name, rid)
             if tag_id is not None and tag_id > 0 and tag_id not in seen:
                 seen.add(tag_id)
                 ids.append(tag_id)
@@ -138,6 +158,32 @@ def _resolve_wp_tag_ids(*, base_url: str, auth_header: str, tags: list[str]) ->
     return ids
 
 
+def _creatable_tag_names(tags: list[str], article_id: Any) -> set[str]:
+    """Casefolded names that may be created as new WordPress tags.
+
+    A proposed tag has to show up for `wordpress_new_tag_min_proposals`
+    different articles before it earns a taxonomy entry. Until then it is
+    counted and dropped, so one-off inventions stop accumulating.
+
+    If the bookkeeping itself fails, nothing is creatable: existing tags are
+    still assigned, and the taxonomy simply does not grow that run.
+    """
+    if not tags:
+        return set()
+    settings = get_settings()
+    threshold = max(1, settings.wordpress_new_tag_min_proposals)
+    try:
+        aid = int(article_id) if article_id is not None else None
+    except (TypeError, ValueError):
+        aid = None
+    try:
+        counts = repositories.record_tag_proposals(tags, aid)
+    except Exception as exc:
+        _logger.warning("Schlagwort-Zähler nicht verfügbar, lege keine neuen an: %s", exc)
+        return set()
+    return {name for name, seen in counts.items() if seen >= threshold}
+
+
 _category_id_cache: dict[str, int | None] = {}
 
 
@@ -552,10 +598,14 @@ def publish_article_draft(article: dict[str, Any]) -> tuple[int, str | None]:
 
     wp_post_id = article.get("wp_post_id")
     tag_names = _selected_tags_from_meta(article.get("meta_json"))
+    # Only the leading few tags reach WordPress; the full list still feeds the
+    # category rules below, which were validated against it.
+    wp_tag_names = tag_names[: max(0, settings.wordpress_max_tags_per_post)]
     tag_ids = _resolve_wp_tag_ids(
         base_url=settings.wordpress_base_url,
         auth_header=auth,
-        tags=tag_names,
+        tags=wp_tag_names,
+        creatable=_creatable_tag_names(wp_tag_names, article.get("id")),
     )
     if tag_ids:
         payload["tags"] = tag_ids