Parcourir la source

refactor: Add descriptions and json_format parameters to data sources in table_line_generator.yaml for improved clarity and consistency

zhch158_admin il y a 10 heures
Parent
commit
a78355fe6b
1 fichiers modifiés avec 19 ajouts et 2 suppressions
  1. 19 2
      table_line_generator/table_line_generator.yaml

+ 19 - 2
table_line_generator/table_line_generator.yaml

@@ -29,7 +29,8 @@ table_editor:
 
   data_sources:
     - name: "B用户_扫描流水"
-      tool: "mineru"  # 格式同 MinerU
+      description: "{{name}} MinerU + PaddleOCR 坐标"
+      json_format: "mineru"  # 格式同 MinerU
       base_dir: "/Users/zhch158/workspace/data/流水分析"
       json_dir: "{{name}}/mineru_vllm_results_cell_bbox"
       image_dir: "{{name}}/mineru_vllm_results/{{name}}"
@@ -43,7 +44,8 @@ table_editor:
         image_suffix: ".png"
 
     - name: "康强_北京农村商业银行"
-      tool: "ppstructure"
+      description: "{{name}} PPStructure 坐标"
+      json_format: "ppstructure"
       base_dir: "/Users/zhch158/workspace/data/流水分析"
       json_dir: "{{name}}/ppstructurev3_client_results"
       image_dir: "{{name}}/ppstructurev3_client_results/{{name}}"
@@ -52,6 +54,21 @@ table_editor:
       image_pattern: "{{ name }}_page_(?P<page>\\d{3})\\.png"
       sort_key: "page"
       output:
+        directory: "{{base_dir}}/{{ name }}.clustered_wiredtable"
+        structure_suffix: "_structure.json"
+        image_suffix: ".png"
+
+    - name: "康强_北京农村商业银行"
+      description: "{{name}} MinerU + PaddleOCR 坐标"
+      json_format: "mineru"  # 格式同 MinerU
+      base_dir: "/Users/zhch158/workspace/data/流水分析"
+      json_dir: "{{name}}/mineru_vllm_results_cell_bbox"
+      image_dir: "{{name}}/mineru_vllm_results/{{name}}"
+      # {{ name }}_page_(?P<page>\d{3})\.json 来匹配 JSON 文件名。模板变量 {{ name }} 会在运行时被替换成实际前缀;随后 _page_ 是字面固定部分。命名捕获组 (?P<page>\d{3}) 强制页面编号恰好为三位数字,并将其存入 page 组  
+      json_pattern: "{{ name }}_page_(?P<page>\\d{3})\\.json"
+      image_pattern: "{{ name }}_page_(?P<page>\\d{3})\\.png"
+      sort_key: "page"
+      output:
         directory: "{{base_dir}}/{{ name }}.wiredtable"
         structure_suffix: "_structure.json"
         image_suffix: ".png"