Update app.py

app.py

@@ -7,6 +7,7 @@ import asyncio
 import subprocess
 import tempfile
 import threading
+import time
 from datetime import datetime, timezone
 from pathlib import Path
 
@@ -59,6 +60,8 @@ LOG_HEADERS = [
     "parameters_requested",
     "model_raw_thinking",
     "extracted_json",
+    "vision_time_seconds",
+    "llm_time_seconds",
     "status",
     "error_detail",
 ]
@@ -124,8 +127,10 @@ def _download_or_create_workbook() -> openpyxl.Workbook:
             "F": 35,
             "G": 60,
             "H": 60,
-            "I": 12,
-            "J": 40,
+            "I": 18,
+            "J": 18,
+            "K": 12,
+            "L": 40,
         }
         for col, width in col_widths.items():
             ws.column_dimensions[col].width = width
@@ -159,6 +164,8 @@ def append_log(
     extracted_json: dict | None,
     status: str,
     error_detail: str = "",
+    vision_time: float | None = None,
+    llm_time: float | None = None,
 ):
     """Thread-safe: download → append row → upload."""
     if not HF_TOKEN:
@@ -173,6 +180,8 @@ def append_log(
         ", ".join(parameters),
         (raw_thinking or "")[:5000],
         json.dumps(extracted_json, ensure_ascii=False) if extracted_json else "",
+        round(vision_time, 2) if vision_time is not None else "",
+        round(llm_time, 2) if llm_time is not None else "",
         status,
         error_detail[:1000],
     ]
@@ -452,6 +461,8 @@ async def extract_document(
 
     raw_text  = ""
     extracted = {}
+    vision_time = None
+    llm_time = None
 
     try:
         processed_bytes, processed_name = prepare_file(file_bytes, content_type, filename)
@@ -464,7 +475,9 @@ async def extract_document(
         raise
 
     try:
+        t0 = time.time()
         raw_text = await asyncio.to_thread(extract_text_with_sdk, processed_bytes, processed_name)
+        vision_time = time.time() - t0
     except HTTPException as exc:
         asyncio.get_event_loop().run_in_executor(
             None, append_log,
@@ -477,12 +490,15 @@ async def extract_document(
     raw_text = clean_raw_text(raw_text)
 
     try:
+        t0 = time.time()
         extracted = await extract_json_via_chat(document_type, raw_text, param_list)
+        llm_time = time.time() - t0
     except HTTPException as exc:
         asyncio.get_event_loop().run_in_executor(
             None, append_log,
             filename, file_size, content_type, document_type, param_list,
-            raw_text, None, "error", str(exc.detail)
+            raw_text, None, "error", str(exc.detail),
+            vision_time
         )
         raise
 
@@ -490,7 +506,8 @@ async def extract_document(
     asyncio.get_event_loop().run_in_executor(
         None, append_log,
         filename, file_size, content_type, document_type, param_list,
-        raw_text, extracted, "success", ""
+        raw_text, extracted, "success", "",
+        vision_time, llm_time
     )
 
     return JSONResponse(content={
@@ -503,4 +520,4 @@ async def extract_document(
 
 @app.get("/health")
 def health():
-    return {"status": "ok"}
+    return {"status": "ok"}