部分修改[0.2.8.14][ci]

elastic/views.py

@@ -8,6 +8,7 @@ import base64
 import json
 import csv
 import io
+import mimetypes
 from datetime import datetime, timezone, timedelta
 import tempfile
 import concurrent.futures
@@ -40,6 +41,8 @@ except ImportError as e:
         HAS_PDF_SUPPORT = False
         PDF_ERROR = str(e)
 
+MAX_SINGLE_UPLOAD_COUNT = int(getattr(settings, "MAX_SINGLE_UPLOAD_COUNT", 3))
+
 
 def _filter_results_for_user(request, results):
     session_user_id = request.session.get("user_id")
@@ -614,6 +617,7 @@ def ocr_and_extract_info(image_path: str):
             return base64.b64encode(f.read()).decode("utf-8")
 
     base64_image = encode_image(image_path)
+    mime_type = mimetypes.guess_type(image_path)[0] or "image/jpeg"
 
     # api_key = getattr(settings, "AISTUDIO_API_KEY", "188f57db3766e02ed2c7e18373996d84f4112272")
     # base_url = getattr(settings, "OPENAI_BASE_URL", "https://aistudio.baidu.com/llm/lmapi/v3")
@@ -665,7 +669,7 @@ def ocr_and_extract_info(image_path: str):
                 "role": "user",
                 "content": [
                     {"type": "text", "text": f"请识别这张图片中的信息，将你认为重要的数据转换为不包含嵌套的json，不要显示其它信息以便于解析，直接输出json结果即可。使用“数据类型”字段表示这个东西的大致类型，除此之外你可以自行决定使用哪些json字段。“数据类型”的内容有严格规定，请查看{json.dumps(types, ensure_ascii=False)}中是否包含你所需要的类型，确定不包含后你才可以填入你觉得合适的大致分类。"},
-                    {"type": "image_url", "image_url": {"url": f"data:image/png;base64,{base64_image}"}},
+                    {"type": "image_url", "image_url": {"url": f"data:{mime_type};base64,{base64_image}"}},
                 ],
             },
         ],
@@ -715,6 +719,7 @@ def upload_page(request):
     context = {
         "user_id": user_id_qs or session_user_id,
         "username": me.get("username"),
+        "max_single_upload_count": MAX_SINGLE_UPLOAD_COUNT,
     }
     return render(request, "elastic/upload.html", context)
 
@@ -738,6 +743,14 @@ def upload(request):
                 files = [one]
         if not files:
             return JsonResponse({"status": "error", "message": "未选择文件"}, status=400)
+        if len(files) > MAX_SINGLE_UPLOAD_COUNT:
+            return JsonResponse(
+                {
+                    "status": "error",
+                    "message": f"单次最多上传 {MAX_SINGLE_UPLOAD_COUNT} 个文件，请分批上传",
+                },
+                status=400,
+            )
 
         images_dir = os.path.join(settings.MEDIA_ROOT, "images")
         os.makedirs(images_dir, exist_ok=True)
@@ -784,17 +797,20 @@ def upload(request):
                 abs_p, fname = img_info
                 try:
                     data = ocr_and_extract_info(abs_p)
-                    return data
-                except Exception:
-                    return None
+                    return data, None
+                except Exception as e:
+                    return None, f"{fname}: {str(e)}"
 
             group_data_list = []
+            group_errors = []
             with concurrent.futures.ThreadPoolExecutor(max_workers=min(len(group_images), 8)) as executor:
                 futures = [executor.submit(run_ocr, img_info) for img_info in group_images]
                 for future in concurrent.futures.as_completed(futures):
-                    res = future.result()
+                    res, err = future.result()
                     if res:
                         group_data_list.append(res)
+                    elif err:
+                        group_errors.append(err)
 
             merged_group_data = {}
             for item in group_data_list:
@@ -814,7 +830,12 @@ def upload(request):
                         merged_group_data[f"{base}_{idx}"] = v
 
             if not merged_group_data:
-                merged_group_data = {"文件名": f.name, "提示": "未识别到具体内容"}
+                merged_group_data = {
+                    "文件名": f.name,
+                    "提示": "未识别到具体内容" if not group_errors else "识别失败",
+                }
+                if group_errors:
+                    merged_group_data["错误信息"] = "；".join(group_errors[:3])
 
             rel_paths = [f"images/{img[1]}" for img in group_images]
             image_urls = [request.build_absolute_uri(settings.MEDIA_URL + rp) for rp in rel_paths]