Update app.py

app.py

@@ -47,8 +47,8 @@ Please analyze the freight rate table in the file I provide and convert it into
       "transit": "...",
       "transit_time": "...",
       "weight_breaks": {
-        "min": ...,
-        "-45kg": ...,
+        "m": ...,
+        "n": ...,
         "45kg": ...,
         "100kg": ...,
         "300kg": ...,
@@ -429,6 +429,8 @@ def run_process_internal_base_v2(file_bytes, filename, mime, question, model_cho
     user_prompt = (question or "").strip() or PROMPT_FREIGHT_JSON
     all_json_results, all_text_results = [], []
 
+    previous_header_json = None
+
     for i in range(0, len(pages), batch_size):
         batch = pages[i:i+batch_size]
         uploaded = []
@@ -438,16 +440,34 @@ def run_process_internal_base_v2(file_bytes, filename, mime, question, model_cho
                 up = genai.upload_file(path=tmp.name, mime_type="image/png")
                 up = _wait_file_active(up)
                 uploaded.append(up)
+    
+        # build dynamic prompt
+        if previous_header_json:
+            context_prompt = (
+                f"{user_prompt}\n"
+                "The previous page had this table structure:\n"
+                f"{json.dumps(previous_header_json, ensure_ascii=False, indent=2)}\n"
+                "If this page has no header, assume it continues with the same structure."
+            )
+        else:
+            context_prompt = user_prompt
+    
+        resp = model.generate_content([f"{context_prompt}\n(This is batch {i//batch_size+1})"] + uploaded)
+        text = _safe_text_from_gemini(resp)
+        json_text = _coerce_only_json(text)
+    
         try:
-            prompt = f"{user_prompt}\n(This is batch {i//batch_size+1})"
-            resp = model.generate_content([prompt] + uploaded)
-            text = _safe_text_from_gemini(resp)
-            json_text = _coerce_only_json(text)
-            try:
-                parsed = json.loads(json_text)
-                all_json_results.append(parsed)
-            except Exception:
-                all_text_results.append(text)
+            parsed = json.loads(json_text)
+            all_json_results.append(parsed)
+    
+            # ✅ update header context (for next page)
+            if i == 0:
+                # chỉ cần giữ phần "charges[0].weight_breaks" làm cấu trúc header
+                first_charge = (parsed.get("charges") or [{}])[0]
+                if "weight_breaks" in first_charge:
+                    previous_header_json = first_charge["weight_breaks"]
+        except Exception:
+            all_text_results.append(text)
         finally:
             for up in uploaded:
                 try: genai.delete_file(up.name)