版本更新：

1、已实现多图识别并入库
2、增加图片上传时删除图片功能
3、改用模型glm-4.6v预计5月份到期
4、已对环境txt做更改

elastic/views.py

@@ -42,6 +42,29 @@ def _image_ref_to_url(request, image_ref: str) -> str:
         return ''
 
 
+def _parse_image_refs(image_ref):
+    if not image_ref:
+        return []
+    if isinstance(image_ref, (list, tuple)):
+        return [str(x) for x in image_ref if str(x).strip()]
+    if isinstance(image_ref, str):
+        s = image_ref.strip()
+        if not s:
+            return []
+        parsed = None
+        if s[:1] in ('[', '"'):
+            try:
+                parsed = json.loads(s)
+            except Exception:
+                parsed = None
+        if isinstance(parsed, list):
+            return [str(x) for x in parsed if str(x).strip()]
+        if isinstance(parsed, str):
+            s = parsed.strip()
+        return [s] if s else []
+    return []
+
+
 def _attach_image_urls(request, items):
     out = []
     for it in list(items or []):
@@ -49,7 +72,11 @@ def _attach_image_urls(request, items):
             d = dict(it or {})
         except Exception:
             continue
-        d['image_url'] = _image_ref_to_url(request, d.get('image', ''))
+        refs = _parse_image_refs(d.get('image', ''))
+        urls = [_image_ref_to_url(request, r) for r in refs if str(r).strip()]
+        urls = [u for u in urls if u]
+        d['image_urls'] = urls
+        d['image_url'] = urls[0] if urls else _image_ref_to_url(request, d.get('image', ''))
         out.append(d)
     return out
 
@@ -180,7 +207,11 @@ def update_data(request, doc_id):
         if "writer_id" in payload:
             updated["writer_id"] = payload["writer_id"]
         if "image" in payload:
-            updated["image"] = payload["image"]
+            img_val = payload["image"]
+            if isinstance(img_val, list):
+                updated["image"] = json_to_string(img_val)
+            else:
+                updated["image"] = img_val
         if "data" in payload:
             v = payload["data"]
             if isinstance(v, dict):
@@ -359,7 +390,7 @@ def string_to_json(s):
 
 # 移植自 a.py 的核心：调用大模型进行 OCR/信息抽取
 def ocr_and_extract_info(image_path: str):
-    from openai import OpenAI
+    # from openai import OpenAI
     def encode_image(path: str) -> str:
         with open(path, "rb") as f:
             return base64.b64encode(f.read()).decode("utf-8")
@@ -372,12 +403,42 @@ def ocr_and_extract_info(image_path: str):
     #     raise RuntimeError("缺少 AISTUDIO_API_KEY，请在环境变量或 settings 中配置")
 
 
-    api_key = getattr(settings, "AISTUDIO_API_KEY", "")
-    base_url = getattr(settings, "OPENAI_BASE_URL", "")
-    if not api_key or not base_url:
-        raise RuntimeError("缺少模型服务配置，请设置 AISTUDIO_API_KEY 与 OPENAI_BASE_URL")
-    client = OpenAI(api_key=api_key, base_url=base_url)
+    # api_key = getattr(settings, "AISTUDIO_API_KEY", "")
+    # base_url = getattr(settings, "OPENAI_BASE_URL", "")
+    # if not api_key or not base_url:
+    #     raise RuntimeError("缺少模型服务配置，请设置 AISTUDIO_API_KEY 与 OPENAI_BASE_URL")
+    # client = OpenAI(api_key=api_key, base_url=base_url)
+    # types = get_type_list()
+    # chat_completion = client.chat.completions.create(
+    #     messages=[
+    #         {"role": "system", "content": "你是一个能理解图片和文本的助手，请根据用户提供的信息进行回答。"},
+    #         {
+    #             "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}"}},
+    #             ],
+    #         },
+    #     ],
+    #     model="glm-5",
+    # )
+    # response_text = chat_completion.choices[0].message.content
 
+    from zai import ZhipuAiClient
+    import httpx
+    # api_key = (
+    #     getattr(settings, "ZHIPU_API_KEY", "")
+    #     or getattr(settings, "ZAI_API_KEY", "")
+    #     or getattr(settings, "AISTUDIO_API_KEY", "")
+    # )
+    # if not api_key:
+    #     raise RuntimeError("缺少模型服务配置，请设置 ZHIPU_API_KEY")
+    # base_url = (
+    #     getattr(settings, "ZHIPU_BASE_URL", "")
+    #     or getattr(settings, "ZAI_BASE_URL", "")
+    #     or "https://open.bigmodel.cn/api/paas/v4/"
+    # )
+    client = ZhipuAiClient(api_key="fb83a3f91e8c4e45af811236548765a2.cX4kUhigHm7VNowf")
     types = get_type_list()
     chat_completion = client.chat.completions.create(
         messages=[
@@ -390,9 +451,8 @@ def ocr_and_extract_info(image_path: str):
                 ],
             },
         ],
-        model=getattr(settings, "OPENAI_MODEL_NAME", "ernie-4.5-turbo-vl-32k"),
+        model="glm-4.6v",
     )
-
     response_text = chat_completion.choices[0].message.content
 
     def parse_response(text: str):
@@ -448,35 +508,66 @@ def upload(request):
         else:
             return JsonResponse({"status": "error", "message": "未登录"}, status=401)
 
-    file = request.FILES.get("file")
-    if not file:
+    files = request.FILES.getlist("file")
+    if not files:
+        one = request.FILES.get("file")
+        if one:
+            files = [one]
+    if not files:
         return JsonResponse({"status": "error", "message": "未选择文件"}, status=400)
 
     images_dir = os.path.join(settings.MEDIA_ROOT, "images")
     os.makedirs(images_dir, exist_ok=True)
-    filename = f"{uuid.uuid4()}_{file.name}"
-    abs_path = os.path.join(images_dir, filename)
-
-    with open(abs_path, "wb") as dst:
-        for chunk in file.chunks():
-            dst.write(chunk)
-
-    try:
-        data = ocr_and_extract_info(abs_path)
-        if not data:
-            return JsonResponse({"status": "error", "message": "无法识别图片内容"}, status=400)
-
+    rel_paths = []
+    image_urls = []
+    data_list = []
+    for file in files:
+        filename = f"{uuid.uuid4()}_{file.name}"
+        abs_path = os.path.join(images_dir, filename)
+        with open(abs_path, "wb") as dst:
+            for chunk in file.chunks():
+                dst.write(chunk)
+        try:
+            data = ocr_and_extract_info(abs_path)
+        except Exception as e:
+            return JsonResponse({"status": "error", "message": str(e)}, status=500)
+        if data:
+            data_list.append(data)
         rel_path = f"images/{filename}"
-        image_url = request.build_absolute_uri(settings.MEDIA_URL + rel_path)
-        return JsonResponse({
-            "status": "success",
-            "message": "识别成功，请确认数据后点击录入",
-            "data": data,
-            "image": rel_path,
-            "image_url": image_url,
-        })
-    except Exception as e:
-        return JsonResponse({"status": "error", "message": str(e)}, status=500)
+        rel_paths.append(rel_path)
+        image_urls.append(request.build_absolute_uri(settings.MEDIA_URL + rel_path))
+
+    if not data_list:
+        return JsonResponse({"status": "error", "message": "无法识别图片内容"}, status=400)
+
+    merged = {}
+    for item in data_list:
+        if not isinstance(item, dict):
+            continue
+        for k, v in item.items():
+            key = str(k).strip()
+            if not key:
+                continue
+            if key not in merged or merged.get(key) in (None, ''):
+                merged[key] = v
+                continue
+            if merged.get(key) == v:
+                continue
+            base = key
+            idx = 2
+            while f"{base}_{idx}" in merged:
+                idx += 1
+            merged[f"{base}_{idx}"] = v
+
+    return JsonResponse({
+        "status": "success",
+        "message": "识别成功，请确认数据后点击录入",
+        "data": merged,
+        "images": rel_paths,
+        "image_urls": image_urls,
+        "image": rel_paths[0] if rel_paths else "",
+        "image_url": image_urls[0] if image_urls else "",
+    })
 
 
 # 确认并入库
@@ -508,38 +599,44 @@ def confirm(request):
     ensure_type_in_list(edited.get("数据类型"))
     image_ref_to_store = ""
     temp_files_to_delete = []
-    if image_rel:
+    image_rels = _parse_image_refs(image_rel)
+    if image_rels:
         images_dir = os.path.join(settings.MEDIA_ROOT, "images")
         os.makedirs(images_dir, exist_ok=True)
-        src_abs = os.path.join(settings.MEDIA_ROOT, image_rel)
-        if not os.path.isfile(src_abs):
-            return JsonResponse({"status": "error", "message": "图片文件不存在"}, status=400)
-
-        webp_name = f"{uuid.uuid4().hex}.webp"
-        webp_abs = os.path.join(images_dir, webp_name)
-        try:
-            with Image.open(src_abs) as im:
-                if im.mode in ("RGBA", "LA", "P"):
-                    im = im.convert("RGBA")
-                else:
-                    im = im.convert("RGB")
-                im.save(webp_abs, format="WEBP", quality=80)
-        except Exception:
+        image_refs = []
+        for rel in image_rels:
+            src_abs = os.path.join(settings.MEDIA_ROOT, rel)
+            if not os.path.isfile(src_abs):
+                return JsonResponse({"status": "error", "message": "图片文件不存在"}, status=400)
+            webp_name = f"{uuid.uuid4().hex}.webp"
+            webp_abs = os.path.join(images_dir, webp_name)
             try:
-                if os.path.isfile(webp_abs):
-                    os.remove(webp_abs)
+                with Image.open(src_abs) as im:
+                    if im.mode in ("RGBA", "LA", "P"):
+                        im = im.convert("RGBA")
+                    else:
+                        im = im.convert("RGB")
+                    im.save(webp_abs, format="WEBP", quality=80)
             except Exception:
-                pass
-            return JsonResponse({"status": "error", "message": "图片转换WEBP失败"}, status=500)
+                try:
+                    if os.path.isfile(webp_abs):
+                        os.remove(webp_abs)
+                except Exception:
+                    pass
+                return JsonResponse({"status": "error", "message": "图片转换WEBP失败"}, status=500)
 
-        try:
-            object_name = f"images/{webp_name}"
-            from minio_storage.minio_connect import upload_file
-            upload_file(webp_abs, object_name, content_type="image/webp")
-            image_ref_to_store = f"minio:{object_name}"
-            temp_files_to_delete.extend([src_abs, webp_abs])
-        except Exception as e:
-            return JsonResponse({"status": "error", "message": f"上传到MinIO失败: {e}"}, status=500)
+            try:
+                object_name = f"images/{webp_name}"
+                from minio_storage.minio_connect import upload_file
+                upload_file(webp_abs, object_name, content_type="image/webp")
+                image_refs.append(f"minio:{object_name}")
+                temp_files_to_delete.extend([src_abs, webp_abs])
+            except Exception as e:
+                return JsonResponse({"status": "error", "message": f"上传到MinIO失败: {e}"}, status=500)
+        if len(image_refs) == 1:
+            image_ref_to_store = image_refs[0]
+        elif len(image_refs) > 1:
+            image_ref_to_store = json_to_string(image_refs)
 
     to_store = {
         "writer_id": str(request.session.get("user_id")),