From 56166abff6f65e6e8ad7751b7edd0eda3253e142 Mon Sep 17 00:00:00 2001
From: Aw <62973442+Awei-996@users.noreply.github.com>
Date: Mon, 11 May 2026 21:54:25 +0800
Subject: [PATCH] =?UTF-8?q?fix(parser):=20=E4=BF=AE=E5=A4=8D=20Docling=20?=
 =?UTF-8?q?=E8=A7=A3=E6=9E=90=E6=96=87=E6=A1=A3=E6=97=B6=E5=9B=BE=E7=89=87?=
 =?UTF-8?q?=E9=94=99=E4=BD=8D=E9=97=AE=E9=A2=98?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

遍历所有图片生成等长替换列表，非 data URI 图片填空字符串，
确保占位符替换严格 1:1，避免混合图片类型时顺序错乱。
---
 .../package/yuxi/plugins/parser/unified.py    | 35 ++++++++-----------
 1 file changed, 14 insertions(+), 21 deletions(-)

diff --git a/backend/package/yuxi/plugins/parser/unified.py b/backend/package/yuxi/plugins/parser/unified.py
index 7e263256..a7b82ad6 100644
--- a/backend/package/yuxi/plugins/parser/unified.py
+++ b/backend/package/yuxi/plugins/parser/unified.py
@@ -127,31 +127,24 @@ def _convert_with_docling(file_path: Path, params: dict | None = None) -> str:
     doc = result.document
 
     if hasattr(doc, "pictures") and doc.pictures:
-        image_refs: list[tuple[str, bytes]] = []
-
+        replacements: list[str] = []
         for pic in doc.pictures:
-            if hasattr(pic, "image") and hasattr(pic.image, "uri"):
-                uri = str(pic.image.uri)
-                if uri.startswith("data:"):
+            uri = str(pic.image.uri) if hasattr(pic, "image") and hasattr(pic.image, "uri") else ""
+            if uri.startswith("data:"):
+                try:
                     image_data, mime_type = _parse_data_uri(uri)
-                    timestamp = int(time.time() * 1000000)
-                    filename = f"image_{timestamp}.{mime_type.split('/')[-1]}"
-                    image_refs.append((filename, image_data))
-
-        image_urls: list[str] = []
-        for filename, image_data in image_refs:
-            try:
-                url = _upload_image_to_minio(image_data, filename, image_bucket, image_prefix)
-                image_urls.append(f"![{filename}]({url})")
-            except Exception as e:  # noqa: BLE001
-                logger.error(f"上传图片失败 {filename}: {e}")
-                image_urls.append(f"[图片: {filename}]")
+                    filename = f"image_{int(time.time() * 1000000)}.{mime_type.split('/')[-1]}"
+                    url = _upload_image_to_minio(image_data, filename, image_bucket, image_prefix)
+                    replacements.append(f"![{filename}]({url})")
+                except Exception as e:  # noqa: BLE001
+                    logger.error(f"上传图片失败 {filename}: {e}")
+                    replacements.append("")
+            else:
+                replacements.append("")
 
         markdown = doc.export_to_markdown()
-
-        for url in image_urls:
-            markdown = re.sub(r"<!--\s*image\s*-->", url, markdown, count=1)
-
+        for replacement in replacements:
+            markdown = re.sub(r"<!--\s*image\s*-->", replacement, markdown, count=1)
         return markdown
 
     return doc.export_to_markdown()