fix(reader): prevent false phrasal verb detection for noun collocations and improve literary translation accuracy

2026-10-06 22:58:33 +03:00
parent 63e0c6fc07
commit e371cdd4e5
4 changed files with 224 additions and 33 deletions
@@ -1955,6 +1955,79 @@ def test_robust_sentence_split_and_ai_translation(monkeypatch):
    assert tr_data["target_word"] == "briefcase"


+def test_phrasal_verb_detection_no_false_positive_for_noun_phrase():
+    from app.phrasal import detect_phrasal_verb
+
+    # Seed temporary test dictionary with tawny owl
+    with get_db() as conn:
+        conn.execute(
+            "INSERT OR IGNORE INTO offline_dictionary (term, translation, phonetic, pos) VALUES (?, ?, '', ?)",
+            ("tawny owl", "неясыть, неясыть серая", "Noun"),
+        )
+        conn.execute(
+            "INSERT OR IGNORE INTO offline_dictionary (term, translation, phonetic, pos) VALUES (?, ?, '', ?)",
+            ("owl", "сыч, сова, филин", "Noun"),
+        )
+        conn.execute(
+            "INSERT OR IGNORE INTO offline_dictionary (term, translation, phonetic, pos) VALUES (?, ?, '', ?)",
+            ("tawny", "рыжевато-коричневый", "Adjective"),
+        )
+
+    sentence = "None of them noticed a large, tawny owl flutter past the window."
+    # tawny owl is a noun phrase (an animal species), NOT a phrasal verb!
+    assert detect_phrasal_verb(sentence, "owl") is None
+    assert detect_phrasal_verb(sentence, "tawny") is None
+    assert detect_phrasal_verb(sentence, "large") is None
+
+    # Genuine phrasal verbs must still be detected
+    s2 = "None of them noticed him take off his coat."
+    pv2 = detect_phrasal_verb(s2, "take")
+    assert pv2 is not None
+    assert pv2["term"] == "take off"
+
+    s3 = "He turned it off quietly."
+    pv3 = detect_phrasal_verb(s3, "turned")
+    assert pv3 is not None
+    assert pv3["term"] == "turn off"
+
+
+def test_sentence_dictionary_hints_and_fauna_precision():
+    from app.ai import get_sentence_dictionary_hints, build_translation_prompt
+
+    with get_db() as conn:
+        conn.execute(
+            "INSERT OR IGNORE INTO offline_dictionary (term, translation, phonetic, pos) VALUES (?, ?, '', ?)",
+            ("tawny owl", "неясыть, неясыть серая", "Noun"),
+        )
+        conn.execute(
+            "INSERT OR IGNORE INTO offline_dictionary (term, translation, phonetic, pos) VALUES (?, ?, '', ?)",
+            ("owl", "сыч, сова, филин", "Noun"),
+        )
+        conn.execute(
+            "INSERT OR IGNORE INTO offline_dictionary (term, translation, phonetic, pos) VALUES (?, ?, '', ?)",
+            ("large", "большой, крупный", "Adjective"),
+        )
+
+    sentence = "None of them noticed a large, tawny owl flutter past the window."
+    hints = get_sentence_dictionary_hints(sentence, "tawny")
+    assert "owl" in hints
+    assert "НЕ орёл" in hints
+    assert "large" in hints
+
+    prompt = build_translation_prompt(
+        sentence=sentence,
+        target_word="tawny",
+        context="",
+        dict_hint="рыжевато-коричневый",
+        sentence_hints=hints,
+    )
+    assert "Словарные подсказки по ключевым словам предложения" in prompt
+    assert "НЕ орёл" in prompt
+    assert "ПОЛНОТА И СВЯЗНОСТЬ" in prompt
+    assert "large" in prompt
+
+
+