fixes

app.py

@@ -1020,11 +1020,21 @@ def pattern_analyze_as_noun(word: str, hint_lemma: str = None) -> Dict[str, Any]
         analysis["gender"] = list(analysis["declension_by_gender"].keys())[0]
 
     return analysis
+
 def pattern_analyze_as_verb(word: str, hint_lemma: str = None) -> Dict[str, Any]:
     """Comprehensive verb conjugation analysis."""
     log(f"  Analyzing as verb (hint_lemma={hint_lemma})")
     verb_lemma = lemma(word)
     log(f"    lemma({word}) = {verb_lemma}")
+
+    # If the input word is already an infinitive (ends in 'en', 'n', 'ln'),
+    # and pattern.de gives a weird lemma, trust the input word.
+    # This fixes lemma('gießen') -> 'gaßen'
+    is_infinitive_form = word.endswith("en") or word.endswith("ln") or word.endswith("rn")
+    if is_infinitive_form and verb_lemma != word.lower():
+        log(f"    Pattern.de lemma '{verb_lemma}' is suspicious for infinitive '{word}'. Trusting input word.")
+        verb_lemma = word
+        
     if not verb_lemma or verb_lemma == word:
         if hint_lemma and hint_lemma != word:
             verb_lemma = hint_lemma
@@ -1111,6 +1121,7 @@ def pattern_analyze_as_verb(word: str, hint_lemma: str = None) -> Dict[str, Any]
             if form: analysis["conjugation"]["Konjunktiv II"][name] = form
         except: pass
     return analysis
+
 def pattern_analyze_as_adjective(word: str, hint_lemma: str = None) -> Dict[str, Any]:
     """Comprehensive adjective inflection analysis."""
     log(f"  Analyzing as adjective (hint_lemma={hint_lemma})")
@@ -1868,25 +1879,20 @@ 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 inflected forms.
+    NON-VARIANT, NON-AUXILIARY inflected forms.
     Returns a list of full entry reports.
     """
     log(f"Wiktionary: Querying for '{word}'...")
     found_entry_ids: Set[int] = set()
     
     # 1. Check if the word is a lemma (base form)
-    # e.g., input "Haus" finds "Haus (Substantiv)"
-    # e.g., input "gehe" finds "gehe (Konjugierte Form)"
     lemma_q = conn.execute(
         "SELECT id FROM entries WHERE word = ? AND lang = 'Deutsch'", (word,)
     ).fetchall()
     for row in lemma_q:
         found_entry_ids.add(row["id"])
         
-    # 2. Check if the word is a true inflected form, but NOT a "variant"
-    # e.g., input "gehe" finds "gehen (Verb)"
-    # e.g., input "Haus" finds "Hau (Substantiv)"
-    # This WILL NOT find "Häusle" from "Haus" anymore.
+    # 2. Check if the word is a true inflected form, but NOT a "variant" or "auxiliary"
     form_q = conn.execute(
         """
         SELECT DISTINCT e.id
@@ -1894,11 +1900,11 @@ def _wiktionary_find_all_entries(word: str, conn: sqlite3.Connection) -> List[Di
         JOIN entries e ON f.entry_id = e.id
         WHERE f.form_text = ? AND e.lang = 'Deutsch'
           AND f.id NOT IN (
-            -- Exclude all form_ids that are tagged as 'variant'
+            -- Exclude all form_ids that are tagged as 'variant' or 'auxiliary'
             SELECT ft.form_id
             FROM form_tags ft
             JOIN tags t ON ft.tag_id = t.id
-            WHERE t.tag = 'variant'
+            WHERE t.tag IN ('variant', 'auxiliary')
           )
         """, (word,)
     ).fetchall()
@@ -2116,18 +2122,21 @@ def _analyze_word_with_wiktionary(word: str, top_n: int) -> Dict[str, Any]:
             }
         }
         
-        # --- E. VALIDATION FILTER ---
+        # --- E. VALIDATION FILTER (REVISED LOGIC) ---
         is_valid = False
         is_inflected_entry = "Konjugierte Form" in pos_title or "Deklinierte Form" in pos_title
 
-        # Check 1: Is the input word the lemma OF A BASE FORM entry?
-        if not is_inflected_entry and lemma.lower() == word_lower:
+        # Check 1: Is the input word the lemma?
+        # This is true for base form entries (e.g., "Haus" -> "Haus (Substantiv)")
+        # AND for inflected form entries (e.g., "gießt" -> "gießt (Konjugierte Form)")
+        if lemma.lower() == word_lower:
             is_valid = True
-            log(f"[DEBUG] Wiktionary: KEEPING entry '{lemma}' ({pos_key}) because input word matches lemma of a base entry.")
+            log(f"[DEBUG] Wiktionary: KEEPING entry '{lemma}' ({pos_key}) because input word matches entry lemma.")
         
         # Check 2: Is the input word in the *bare* forms list?
-        # (This is the only check that should apply to inflected entries)
-        if not is_valid:
+        # (This applies to base entries where the input is an inflection, e.g., "gießt" -> "gehen (Verb)")
+        # We only run this if Check 1 failed AND this is not an inflected entry (which have no forms)
+        if not is_valid and not is_inflected_entry:
             for form_entry in inflections_wikt_block.get("forms_list", []):
                 form_text = form_entry.get("form_text", "")
                 bare_form = re.sub(r"\(.*\)", "", form_text).strip()