|
|
@@ -0,0 +1,21 @@
|
|
|
+data_source:
|
|
|
+ name: "山西云集科技有限公司"
|
|
|
+ description: "山西云集科技有限公司 bank_statement_wired_unet"
|
|
|
+ json_format: "mineru"
|
|
|
+ base_dir: "/Users/zhch158/workspace/data/流水分析"
|
|
|
+ json_dir: "{{name}}/bank_statement_wired_unet"
|
|
|
+ image_dir: "{{name}}/bank_statement_wired_unet/{{name}}"
|
|
|
+ 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"
|
|
|
+
|
|
|
+ # 批量处理配置
|
|
|
+ batch:
|
|
|
+ enabled: true
|
|
|
+ template_name: null # 可选:指定默认模板
|
|
|
+ auto_apply: false # 是否自动应用模板
|
|
|
+ parallel_workers: 4 # 并行处理线程数
|