TestingCodeExecution

app.py

@@ -9,6 +9,7 @@ from __future__ import annotations
 import json
 import os
 import tempfile
+import time
 from datetime import date
 from pathlib import Path
 from typing import Any, Optional
@@ -61,7 +62,6 @@ def get_whisper_model() -> Any:
 
     return _WHISPER_MODEL
 
-
 class ExpenseTextExtractor:
     """
     Главный экстрактор данных о расходах.
@@ -96,22 +96,37 @@ class ExpenseTextExtractor:
         Returns:
             Словарь со всеми извлечёнными данными
         """
+        timings = {}
+        
+        t0 = time.time()
         date_info = self.date_extractor.extract(text, reference_date=reference_date)
+        timings["date_extractor"] = round(time.time() - t0, 3)
+        
+        t0 = time.time()
         supplier_info = self.supplier_extractor.extract(
             text,
             date_phrase=date_info.get("matched_date_phrase"),
             debug=debug_supplier,
         )
+        timings["supplier_extractor"] = round(time.time() - t0, 3)
+        
+        t0 = time.time()
         user_info = self.user_extractor.extract(
             text,
             supplier_phrase=supplier_info.get("matched_supplier_phrase"),
             date_phrase=date_info.get("matched_date_phrase"),
         )
+        timings["user_extractor"] = round(time.time() - t0, 3)
+        
+        t0 = time.time()
         amount_info = self.amount_extractor.extract(
             text,
             matched_date_phrase=date_info["matched_date_phrase"],
             matched_supplier_phrase=supplier_info["matched_supplier_phrase"],
         )
+        timings["amount_extractor"] = round(time.time() - t0, 3)
+        
+        print(f"[TIMINGS] {timings}")
 
         result = {
             "text": text,
@@ -164,18 +179,21 @@ def polish_notes_text(text: str) -> str:
     return normalized
 
 
-def transcribe_audio_text(audio_path: str) -> str:
-    """Транскрибирует аудио в текст."""
+def transcribe_audio_text(audio_path: str) -> tuple[str, float]:
+    """Транскрибирует аудио в текст. Возвращает (текст, время в секундах)."""
     mock_text = os.getenv("EXPENSE_VOICE_MOCK_TEXT")
     if mock_text:
-        return mock_text.strip()
+        return mock_text.strip(), 0.0
 
     try:
+        t0 = time.time()
         whisper_model = get_whisper_model()
         segments, _ = whisper_model.transcribe(audio_path, language="ru", vad_filter=True)
         text = " ".join(segment.text.strip() for segment in segments if segment.text and segment.text.strip())
+        elapsed = round(time.time() - t0, 3)
+        print(f"[TIMINGS] whisper_transcribe: {elapsed}s")
         if text:
-            return text
+            return text, elapsed
     except Exception:
         pass
 
@@ -184,11 +202,13 @@ def transcribe_audio_text(audio_path: str) -> str:
 
 def process_voice_request(audio_path: str, mode: str, payload: dict[str, Any]) -> dict[str, Any]:
     """Обрабатывает голосовой запрос."""
+    total_start = time.time()
+    
     context = payload.get("context", {}) if isinstance(payload, dict) else {}
     supplier_names = extract_names(context.get("suppliers"))
     user_names = extract_names(context.get("users"))
 
-    transcript = transcribe_audio_text(audio_path)
+    transcript, whisper_time = transcribe_audio_text(audio_path)
 
     if mode == "notes":
         notes = polish_notes_text(transcript)
@@ -208,8 +228,15 @@ def process_voice_request(audio_path: str, mode: str, payload: dict[str, Any]) -
     if not user_names:
         raise RuntimeError("No users were provided by Laravel context.")
 
+    t0 = time.time()
     extractor = build_default_pipeline(suppliers=supplier_names, users=user_names)
+    pipeline_init_time = round(time.time() - t0, 3)
+    print(f"[TIMINGS] pipeline_init: {pipeline_init_time}s")
+    
     extracted = extractor.extract(transcript, reference_date=date.today().isoformat())
+    
+    total_time = round(time.time() - total_start, 3)
+    print(f"[TIMINGS] TOTAL: {total_time}s (whisper: {whisper_time}s)")
 
     return {
         "status": "ok",
@@ -264,8 +291,7 @@ def index():
         "message": "Voice processing API is running",
         "endpoints": {
             "POST /process-audio": "Process audio file",
-            "GET /health": "Health check",
-            "GET /date-test": "Test date parsing"
+            "GET /health": "Health check"
         }
     })
 
@@ -276,39 +302,6 @@ def health():
     return jsonify({"status": "ok"})
 
 
-@app.get("/date-test")
-def date_test():
-    """Тестирование парсера дат."""
-    test_phrases = [
-        "завтра",
-        "через 2 дня",
-        "на следующей неделе",
-        "15 января 2025",
-        "позавчера",
-        "в прошлый понедельник",
-        "оплата за март",
-        "5 марта",
-        "купил вчера",
-        "в конце месяца"
-    ]
-    
-    extractor = ExpenseDateExtractor()
-    results = []
-    for phrase in test_phrases:
-        result = extractor.extract(phrase)
-        results.append({
-            "phrase": phrase, 
-            "date": result.get("date_iso"),
-            "matched": result.get("matched_date_phrase")
-        })
-    
-    return jsonify({
-        "status": "ok",
-        "reference_date": date.today().isoformat(),
-        "results": results
-    })
-
-
 @app.post("/process-audio")
 def process_audio():
     """Обработка аудио файла."""