nyuuzyou/ms-codeplex-archive
Microsoft CodePlex Archive Dataset Dataset Description Source code from the Microsoft CodePlex Archive on the Internet Archive. CodePlex was Microsoft's open-source project hosting service from 2006 to 2017, popular for .NET and Windows projects. Dataset Summary Statistic Value Total Files 5,043,730 Total Repositories 38,087 Total Size 3.6 GB (compressed Parquet) Programming Languages 91 File Format Parquet with Zstd compression… See the full description on the dataset page: https://huggingface.co/datasets/nyuuzyou/ms-codeplex-archive.
Microsoft CodePlex Archive Dataset
Dataset Description
Source code from the Microsoft CodePlex Archive on the Internet Archive. CodePlex was Microsoft's open-source project hosting service from 2006 to 2017, popular for .NET and Windows projects.
Dataset Summary
Features
- 38K repositories from the 2006-2017 period
- 91 programming languages, dominated by C# and .NET
- Metadata: repository name, file path, language, license, file size
- Filtered to remove vendor code, build artifacts, and generated files
Languages
Top 30 of 91 languages by file count:
Licenses
Dataset Structure
Data Fields
Data Format
- Format: Apache Parquet with Zstd compression
- File Structure: 10 files (
codeplex_0000.parquettocodeplex_0009.parquet)
Data Splits
Train split only.
Example Data Point
{
"code": "using System;\nusing System.Collections.Generic;\nusing System.ComponentModel;\nusing System.Data;\nusing System.Drawing;\nusing System.Text;\nusing System.Windows.Forms;\n\nnamespace KeygenApp\n{\n public partial class KeygenForm : Form\n {\n public KeygenForm()\n {\n InitializeComponent();\n }\n }\n}",
"repo_name": "2atgroup",
"path": "TiffBrowserTestCSharp/Keygen/KeygenForm.cs",
"language": "C#",
"license": "Microsoft Reciprocal License (Ms-RL)",
"size": 1259
}Dataset Creation
Pipeline
- Load project metadata from
zips.csv(108,508 repositories) - Extract 122 tar archives from the Internet Archive
tars/directory - Extract source code from
sourceCode/sourceCode.zipwithin each project - Parse
license/license.jsonfor license metadata - Filter non-code files
- Write to Parquet with Zstd compression
Language Detection
By file extension.
License Detection
Parsed from license/license.json in each project archive.
File Filtering
Size Limits
Excluded Directories
- Config:
.git/,.github/,.gitlab/,.vscode/,.idea/,.vs/,.settings/,.eclipse/,.project/,.metadata/ - Vendor:
node_modules/,bower_components/,jspm_packages/,vendor/,third_party/,3rdparty/,external/,packages/,deps/,lib/vendor/,target/dependency/,Pods/ - Build:
build/,dist/,out/,bin/,target/,release/,debug/,.next/,.nuxt/,_site/,_build/,__pycache__/,.pytest_cache/,cmake-build-*,.gradle/,.maven/,obj/ - CodePlex metadata:
discussions/,issues/,releases/,wiki/,wikiRender/ - Tests:
test/,tests/,spec/,specs/,__tests__/
Excluded Files
- Lock files:
package-lock.json,yarn.lock,pnpm-lock.yaml,Gemfile.lock,Cargo.lock,poetry.lock,Pipfile.lock,composer.lock,go.sum,mix.lock,packages.lock.json - Minified: files containing
.min. - Hidden: files starting with
.(except.htaccess) - Binary:
.exe,.dll,.so,.dylib,.a,.lib,.o,.obj,.jar,.war,.ear,.class,.pyc,.pyo,.wasm,.bin,.dat,.pdf,.doc,.docx,.xls,.xlsx,.ppt,.pptx,.zip,.tar,.gz,.bz2,.7z,.rar,.jpg,.jpeg,.png,.gif,.bmp,.ico,.svg,.mp3,.mp4,.avi,.mov,.wav,.flac,.ttf,.otf,.woff,.woff2,.eot,.pdb,.nupkg,.snk - System:
.DS_Store,thumbs.db
Content Filtering
- UTF-8 Validation: must be valid UTF-8
- Binary Detection: no null bytes, <30% non-printable characters in first 8KB
- Generated Files: excluded if first 500 bytes contain
generated by,do not edit,auto-generated,@generated, etc. - Empty Files: excluded
- Long Lines: excluded if any line >1,000 characters in first 10 lines
Source Data
Microsoft CodePlex Archive on Internet Archive, tars/ directory (122 tar archives). 108,508 repositories accessible out of 108,516 listed.
Archive Structure
Each CodePlex project archive contains:
sourceCode/sourceCode.zip- source code (extracted)license/license.json- license metadata (parsed)discussions/,issues/,releases/,wiki/- excluded
Considerations
Sensitive Information
May contain email addresses, accidentally committed credentials, or personal information in comments. Filter accordingly.
Licensing
Use must comply with the original repository licenses. The license field indicates each file's source license.
