Skip to content

feat: add v3 model#78

Merged
Muxv merged 51 commits intomainfrom
hsn/feat/add-v3-model
Mar 18, 2026
Merged

feat: add v3 model#78
Muxv merged 51 commits intomainfrom
hsn/feat/add-v3-model

Conversation

@HSn0918
Copy link
Collaborator

@HSn0918 HSn0918 commented Mar 16, 2026 •
edited
Loading

Summary

  • V3 模型支持：新增 v3-2026 解析模型选项，适用于 pdf2file 和 pic2file，支持 v3 JSON sidecar 输出及图片裁切辅助工具
  • 公式降级等级：新增 formula_level 枚举（0/1/2），用于控制 v2 导出时的公式降级行为
  • 安全修复：升级存在漏洞的依赖版本 — cryptography>=46.0.5、h2>=4.3.0、pypdf>=6.8.0、urllib3>=2.6.3、Pillow>=12.1.1（修复 CVE-2026-26007、CVE-2025-57804、CVE-2026-25990 等）

Menghuan1918 and others added 30 commits February 10, 2025 23:49
@Menghuan1918
[Doc2X] Add support to img APIs
8254b38
@Menghuan1918
[Doc2X] Finish the pic deal part
68dd68c
@Menghuan1918
[Doc2X] Fix many bugs in images process
790d8a4
@Menghuan1918
[Doc2X] Finished the pic ocr part
dee113c
@Menghuan1918
[Doc2X] Add support for merge_cross_page_forms
83d100c
@Menghuan1918
Update to 1.0.3
9c00c8e
@Menghuan1918
Add default limit
21a1e85
delete ocr
018e09f
@guixinW
还原
6c0112f
@guixinW
修改了V2中的parse_image_layout,原本该函数无法处理多个相同文件名的pic，因为zip必须不同名
2ab4794
@guixinW
保留了pdf2file中的ocr；另外增加了piclayout接口封装了ImageProcessor中的内容用于处理picture
e4974ed
@guixinW
增加了_get_lock，在ImageProcessor开启事件循环时才创建锁
8d5ed90
@guixinW
还原
8783f3c
@guixinW
利用添加的piclayout接口完成测试，修改了测试逻辑
e7df1ed
@guixinW
将img大小限制修改为<=7MB
8e0c0e3
@guixinW
修复了请求被限速后再次加入滑动窗口会添加新旧两个时间戳的bug；优化了滑动窗口在临界区sleep导致后续请求序列化的情况
ad6d838
@guixinW
修改highRPM测试次数为31用来测试接口30s/30req的限速是否正确
0a57955
@guixinW
修复了多线程判断文件是否存在造成的文件覆盖问题
0313f97
@guixinW
Merge branch 'dev' into main
2f03d38
@guixinW
Merge pull request #73 from guixinW/main
4856662 
贡献：删除了ocr在新版本中的接口，新增了piclayout；修改了一些piclayout中的限速bug
@guixinW
统一了pdf2file和pic2file的接口形式
7f46251
@guixinW
pdf2file增加了save_subdir选项
3167cdb
@guixinW
修复了doc2x_img.py的代码问题；增加了subdir选项的测试
2bab32b
@guixinW
为pdf2file添加了json字段，用于将pdf的信息直接保存为json文件
00f0eab
@guixinW
加入export_history关键字，记录下载状态，对于已经导出的文件无需再导出，未导出的类则轮询状态后导出
c7d004a
@Muxv
Merge pull request #74 from guixinW/main
19c1867 
为pdf2file增加了sub_dir关键字，用于将所有输出文件保存在以文件名命名的子文件夹下
@guixinW
修改了export_history存在的一些错误;弃用了asyncio.Lock(loop=self._loop)
34b7f34
@Muxv
Merge pull request #75 from guixinW/main
16fa079 
修改了export_history存在的一些错误;弃用了asyncio.Lock(loop=self._loop)
@Muxv
support save_subdir for json
89993bb
@Muxv
add file_type as suffix to md/tex zip save name
d0fffc2
Muxv and others added 17 commits June 13, 2025 21:00
@Muxv
simplify piclayout parameters
5b96ece
@Muxv
add image path to text output
a750abf
@Muxv
add zip path to text output
c0d8547
@Muxv
add zip output_format for piclayout
ef25c58
@Muxv
add subdir to piclayout and use tmp path to pytest
6578ed4
@Muxv
Merge pull request #76 from NoEdgeAI/feat/record
4a02b87 
1. piclayout 添加 output path， output_format， save_subdir 参数
2. 单元测试使用临时文件夹
3. 暂时屏蔽 record 功能
@HSn0918
feat: add v3 preupload model support for v2 client and CLI
bdff859
@HSn0918
feat(doc2x): add formula_level enum support for v2 export
08b4d4c
@HSn0918
update: add formula level
b9d39a5
@HSn0918
fix: update Base_URL to the correct v2 API endpoint
d5fb1e6
@HSn0918
feat(test): add formula level enum tests for v3 model support
bbb779a
@HSn0918
feat(test): enhance pic2file tests with v3 model integration and erro…
f296010 
…r handling
@HSn0918
feat(model): add V2 support and enhance normalization for V3 model
b3a4a98
@circlestarzero
feat: add v3 json sidecars and crop helpers
0905810
@circlestarzero
chore: ignore local history and tmp outputs
da07b3e
@circlestarzero
docs: add import examples for v3 crop helpers
2761602
@circlestarzero
refactor: move v3 crop helpers into package module
96b7bca
@github-actions
Copy link

github-actions bot commented Mar 16, 2026 •
edited
Loading

Dependency Review

The following issues were found:
  • ✅ 0 vulnerable package(s)
  • ✅ 0 package(s) with incompatible licenses
  • ✅ 0 package(s) with invalid SPDX license definitions
  • ⚠️ 6 package(s) with unknown licenses.
See the Details below.

License Issues

pyproject.toml

PackageVersionLicenseIssue Type
aiofiles>= 24.1.0NullUnknown License
cryptography>= 46.0.5NullUnknown License
h2>= 4.3.0NullUnknown License
pypdf>= 6.8.0NullUnknown License
pytest>= 8.3.5NullUnknown License
urllib3>= 2.6.3NullUnknown License

OpenSSF Scorecard

PackageVersionScoreDetails
pip/aiofiles >= 24.1.0 UnknownUnknown
pip/cryptography >= 46.0.5 UnknownUnknown
pip/h2 >= 4.3.0 UnknownUnknown
pip/pypdf >= 6.8.0 UnknownUnknown
pip/pytest >= 8.3.5 UnknownUnknown
pip/urllib3 >= 2.6.3 UnknownUnknown

Scanned Files

  • pyproject.toml

HSn0918 added 4 commits March 16, 2026 18:07
@HSn0918
fix(deps): bump vulnerable dependencies to patched versions
bb0dd88 
Address high/moderate severity CVEs reported in dependency review:
- cryptography: 46.0.4 → >=46.0.5 (CVE-2026-26007, SECT curves subgroup attack)
- h2: pin >=4.3.0 on Python>=3.9 (CVE-2025-57804, HTTP request smuggling)
- pypdf: pin >=6.8.0 on Python>=3.9 (multiple DoS/RAM CVEs); <6.0.0 on 3.8
- urllib3: pin >=2.6.3 on Python>=3.9 (decompression bomb CVEs); 1.x branch has no patch
- Pillow: pin >=12.1.1 on Python>=3.10 (CVE-2026-25990, PSD OOB write); use 10.x on <3.10

Note: h2, pypdf, urllib3 and Pillow vulnerabilities on Python 3.8 cannot be
fixed as upstream dropped Python 3.8 support before releasing patches.
@HSn0918
remove unfinished export_history from public pdf2file API
e1884a5
@HSn0918
remove: remove uv lock
4695875
@HSn0918
chore: drop Python 3.8 support, require >=3.9
e75e99c 
Remove vulnerable fallback constraints for urllib3 1.x and pypdf 5.x
that had no upstream security patches on Python 3.8.
@Muxv Muxv merged commit 3cf0b19 into main Mar 18, 2026
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

5 participants

@HSn0918 @Muxv @Menghuan1918 @guixinW @circlestarzero