fix base lemma query

app.py

@@ -1998,8 +1998,12 @@ def _wiktionary_build_report_for_entry(entry_id: int, conn: sqlite3.Connection)
 
 def _wiktionary_find_all_entries(word: str, conn: sqlite3.Connection) -> List[Dict[str, Any]]:
     """
-    Finds all entries related to a word, checking both lemmas and 
-    NON-VARIANT, NON-AUXILIARY inflected forms.
+    (FIXED V24)
+    Finds all entries related to a word.
+    1. Finds direct lemma matches (e.g., input "Vertrag" -> finds "Vertrag" entry)
+    2. Finds inflection matches (e.g., input "Häuser" -> finds "Haus" entry via `forms` table)
+    3. Finds declined form matches (e.g., input "Verträge" -> finds "Verträge" entry,
+       then finds "Vertrag" entry via `senses.form_of` table)
     Returns a list of full entry reports.
     """
     log(f"Wiktionary: Querying for '{word}'...")
@@ -2007,12 +2011,39 @@ def _wiktionary_find_all_entries(word: str, conn: sqlite3.Connection) -> List[Di
     
     # 1. Check if the word is a lemma (base form)
     lemma_q = conn.execute(
-        "SELECT id FROM entries WHERE word = ? AND lang = 'Deutsch'", (word,)
+        "SELECT id, pos_title FROM entries WHERE word = ? AND lang = 'Deutsch'", (word,)
     ).fetchall()
+    
+    parent_lemmas_to_find: Set[str] = set()
+    
     for row in lemma_q:
-        found_entry_ids.add(row["id"])
+        entry_id = row["id"]
+        pos_title = row["pos_title"]
+        found_entry_ids.add(entry_id)
         
-    # 2. Check if the word is a true inflected form, but NOT a "variant" or "auxiliary"
+        # --- THIS IS THE NEW LOGIC (STEP 3) ---
+        if pos_title in ("Deklinierte Form", "Konjugierte Form", "Komparativ", "Superlativ"):
+            log(f"Wiktionary: Word '{word}' is an inflected entry (ID {entry_id}). Looking for its parent lemma...")
+            form_of_q = conn.execute(
+                "SELECT form_of FROM senses WHERE entry_id = ?", (entry_id,)
+            ).fetchall()
+            
+            for form_row in form_of_q:
+                form_of_json = form_row["form_of"]
+                if not form_of_json:
+                    continue
+                try:
+                    # Parse the JSON string (e.g., '[{"word": "Vertrag"}]')
+                    form_of_data = json.loads(form_of_json)
+                    if isinstance(form_of_data, list) and form_of_data:
+                        parent_lemma_word = form_of_data[0].get("word")
+                        if parent_lemma_word:
+                            parent_lemmas_to_find.add(parent_lemma_word)
+                except json.JSONDecodeError:
+                    log(f"Wiktionary: Failed to parse form_of JSON: {form_of_json}")
+        # --- END OF NEW LOGIC ---
+
+    # 2. Check if the word is an inflected form (in the `forms` table)
     form_q = conn.execute(
         """
         SELECT DISTINCT e.id
@@ -2031,6 +2062,16 @@ def _wiktionary_find_all_entries(word: str, conn: sqlite3.Connection) -> List[Di
     for row in form_q:
         found_entry_ids.add(row["id"])
         
+    # --- NEW: Add parent lemmas found in step 3 ---
+    if parent_lemmas_to_find:
+        log(f"Wiktionary: Found parent lemmas to add: {parent_lemmas_to_find}")
+        for lemma_word in parent_lemmas_to_find:
+            parent_id_q = conn.execute(
+                "SELECT id FROM entries WHERE word = ? AND lang = 'Deutsch'", (lemma_word,)
+            ).fetchall()
+            for row in parent_id_q:
+                found_entry_ids.add(row["id"])
+    
     log(f"Wiktionary: Found {len(found_entry_ids)} unique matching entries.")
     
     # 3. Build a full report for each unique entry
@@ -2050,26 +2091,33 @@ def _wiktionary_format_semantics_block(
     top_n: int
 ) -> Dict[str, Any]:
     """
+    (FIXED V24)
     Combines Wiktionary senses with OdeNet/ConceptNet senses,
-    using the CORRECT lemma from the pattern.de analysis block.
+    using the *correct* lemma.
+    
+    Priority:
+    1. Wiktionary's lemma (from `wikt_report`)
+    2. Pattern.de's lemma (from `pattern_block`)
     """
     
-    # --- THIS IS THE FIX ---
-    # Determine the true lemma from the pattern.de block, as it's more reliable
-    # for semantic lookup than the wiktionary lemma (which could be an inflected form).
     pos_key = _wiktionary_map_pos_key(wikt_report.get("pos"))
     
-    semantic_lemma = ""
-    if pos_key == "verb":
-        semantic_lemma = pattern_block.get("infinitive")
-    elif pos_key == "noun":
-        semantic_lemma = pattern_block.get("base_form")
-    elif pos_key == "adjective":
-        semantic_lemma = pattern_block.get("predicative")
+    # --- THIS IS THE FIX ---
+    # Prioritize Wiktionary's lemma first, as it's more reliable.
+    semantic_lemma = wikt_report.get("lemma")
     
-    # Fallback if pattern.de fails or it's a non-inflecting POS
+    # If Wiktionary's lemma is missing or bad, try pattern.de's
+    if not semantic_lemma:
+        if pos_key == "verb":
+            semantic_lemma = pattern_block.get("infinitive")
+        elif pos_key == "noun":
+            semantic_lemma = pattern_block.get("base_form")
+        elif pos_key == "adjective":
+            semantic_lemma = pattern_block.get("predicative")
+    
+    # Final fallback
     if not semantic_lemma:
-         semantic_lemma = wikt_report.get("lemma", "") 
+         semantic_lemma = wikt_report.get("word", "") # Use the original word as last resort
     
     log(f"[DEBUG] Wiktionary Semantics: Building block for lemma='{semantic_lemma}', pos='{pos_key}'")
     # --- END OF FIX ---
@@ -2078,7 +2126,7 @@ def _wiktionary_format_semantics_block(
     wiktionary_senses = []
     for sense in wikt_report.get("senses", []):
         wiktionary_senses.append({
-            "definition": sense.get("gloss_text"),
+            "definition": sense.get("glosses"), # <-- Corrected from gloss_text
             "source": "wiktionary"
         })
 
@@ -2121,7 +2169,6 @@ def _wiktionary_format_semantics_block(
         "wiktionary_translations": wikt_report.get("translations", []),
         "wiktionary_derived_terms": wikt_report.get("derived_terms", []),
         "wiktionary_related_terms": wikt_report.get("related_terms", [])
-        
     }
 
 def _analyze_word_with_wiktionary(word: str, top_n: int) -> Dict[str, Any]: