Skip to content

feat(dataset): 知识库级 pdfParseConfig 文档解析开关(#68624) - #7721

Merged
c121914yu merged 3 commits into
labring:mainfrom
ohowandanliao:feat/68624/backend-p2t-switches
Sep 14, 2026
Merged

c121914yu merged 3 commits into
labring:mainfrom
ohowandanliao:feat/68624/backend-p2t-switches

Conversation

@ohowandanliao

Copy link
Copy Markdown
Contributor

需求背景

68624 黄逸轩
知识库级文档解析开关:知识库可配置四个解析开关(保留页眉页脚 /
保留附录 / 图片识别 / 图表转表格),在外部解析服务(p2t)解析 PDF
时逐项生效。

主要改动

  • 新增 pdfParseConfig 数据集级配置:zod + mongo 双
    schema,create / createWithFiles / update 三个入口支持,update
    为整体替换语义,未设置的字段由读取层按固定默认值补全
  • 解析链路透传:外部解析请求以独立表单字段下发四个开关(值统一转
    字符串,服务端自行 str→bool),与内网 develop-1.5.0
    的下发方式对齐
  • S3 rawText 缓存 key 收敛:解析配置纳入缓存 key
    指纹,避免同文件不同开关共享缓存;未传配置保持旧 key
    格式,兼容存量缓存
  • sangfor 解析
    provider(feat: add third-party sangfor parsing #7715)同步透传四开关,保证两条解析路径行为一致
  • 单测:S3 key 稳定性 / 区分度 / 旧 key 兼容、配置默认值补全

验证情况

  • 测试环境全量部署实测:四开关解析行为符合预期(页眉页脚 /
    附录即时生效,图表转表格依赖 MinerU VLM
    输出,结果以折叠表格呈现);sangfor 链路 E2E 通过

备注

  • sangfor provider 命中时按上游设计绕过集合「增强解析」开关(全站
    pdf 接管),此为上游既有语义,本 MR 仅保证该路径下开关不丢失

@cla-assistant

cla-assistant Bot commented Sep 11, 2026

Copy link
Copy Markdown

CLA assistant check
All committers have signed the CLA.

@github-actions

github-actions Bot commented Sep 11, 2026

Copy link
Copy Markdown

Build and publish successful - Preview fastgpt Image:

ghcr.io/labring/fastgpt-pr:fastgpt_037de007a1e74bc11409994e4cf42990aba781ae

🕒 Time: 2026-09-14 19:40:23 (UTC+8)

@ohowandanliao
ohowandanliao force-pushed the feat/68624/backend-p2t-switches branch from ac52dd8 to d102ee4 Compare September 12, 2026 04:00
ctlaltlaltc
ctlaltlaltc previously approved these changes Sep 12, 2026
@ohowandanliao
ohowandanliao force-pushed the feat/68624/backend-p2t-switches branch 2 times, most recently from 72ec228 to 8771da9 Compare September 13, 2026 15:49
@ohowandanliao
ohowandanliao force-pushed the feat/68624/backend-p2t-switches branch from 8771da9 to bfbb346 Compare September 14, 2026 01:51
Comment thread packages/global/core/dataset/type.ts Outdated
Comment thread packages/service/core/dataset/read.ts Outdated
@ohowandanliao
ohowandanliao force-pushed the feat/68624/backend-p2t-switches branch from 9752e18 to 26c4ad5 Compare September 14, 2026 09:13
c121914yu
c121914yu previously approved these changes Sep 14, 2026
@github-actions

github-actions Bot commented Sep 14, 2026

Copy link
Copy Markdown

Docs preview deployed

🔗 👀 Click here to visit preview

ghcr.io/labring/fastgpt-docs-pr:037de007a1e74bc11409994e4cf42990aba781ae

🕒 Time: 2026-09-14 19:36:53 (UTC+8)

上游同步 0d3630a..12cf1fc。按 review 意见将定制解析配置与上游原有链路
区分开, 命名与熊涛 labring#7715/labring#7706 的 sangfor 体系对齐:
- 数据集字段 pdfParseConfig 改名 sangforFileParseConfig, schema/类型/函数同步
- 补全(getDatasetIultmzhFileParseConfig)与表单协议(appendIultmzhFileParseFields)
  收拢 thirdProvider/sangfor/parseConfig.ts, 通用解析链仅保留单行委托
- 移除无引用 DefaultPdfParseConfig, 测试迁移 thirdProvider 测试文件

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
@c121914yu
c121914yu merged commit 3002087 into labring:main Sep 14, 2026
12 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants