README.md

metadata

annotations_creators:
  - machine-generated
language_creators:
  - found
language:
  - code
  - zh
  - en
license: other
multilinguality:
  - multilingual
pretty_name: GitCode Code Dataset
size_categories:
  - 10M<n<100M
source_datasets:
  - original
task_categories:
  - text-generation
tags:
  - code
  - chinese
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/*.parquet
    default: true
dataset_info:
  features:
    - name: code
      dtype: string
    - name: repo_name
      dtype: string
    - name: path
      dtype: string
    - name: language
      dtype: string
    - name: license
      dtype: string
    - name: size
      dtype: int64

GitCode Code Dataset

Dataset Description

This dataset was compiled from code repositories hosted on GitCode, a code hosting platform in China backed by CSDN (China Software Developer Network). GitCode serves as a domestic alternative to GitHub, widely used by Chinese developers, students, and enterprises for hosting open-source projects and educational resources, making this dataset particularly valuable for training code models with Chinese language understanding and Chinese coding conventions.

Dataset Summary

Statistic	Value
Total Files	48,142,567
Total Repositories	85,632
Total Size	40 GB (compressed Parquet)
Programming Languages	537
File Format	Parquet with Zstd compression (34 files)

Key Features

• Chinese code corpus: Contains code from over 85,000 repositories, many featuring Chinese comments, documentation, and variable names
• Diverse language coverage: Spans 537 programming languages identified by go-enry (based on GitHub Linguist rules)
• Rich metadata: Includes repository name, file path, detected language, license information, and file size
• Open-source and educational projects: Includes code from individual developers, students, and Chinese enterprises
• Quality filtered: Extensive filtering to remove vendor code, build artifacts, generated files, and low-quality content

Languages

The dataset includes 537 programming languages. The top 30 languages by file count:

Rank	Language	File Count
1	C++	9,513,619
2	C	8,220,317
3	Java	5,362,924
4	Python	3,428,302
5	TypeScript	3,166,959
6	JavaScript	2,540,280
7	HTML	1,578,824
8	Kotlin	1,413,651
9	C#	1,232,638
10	Go	1,159,708
11	Rust	812,959
12	Dart	767,731
13	TSX	749,355
14	PHP	663,953
15	Shell	629,436
16	Vue	563,754
17	Makefile	471,588
18	CMake	460,428
19	CSS	381,628
20	Ruby	350,213
21	Objective-C	347,251
22	LLVM	297,591
23	Unix Assembly	291,826
24	Swift	206,725
25	Objective-C++	160,526
26	Scala	157,367
27	QML	157,088
28	Lua	149,114
29	SCSS	141,661
30	GLSL	129,124

Licenses

The dataset includes files from repositories with various licenses. Repositories with restrictive licenses (CC-BY-ND variants, Commons Clause, SSPL) were excluded:

License	File Count
unknown	23,567,463
apache-2.0	8,722,445
mit	7,743,613
gpl-2.0	3,528,526
agpl-3.0	2,300,580
lgpl	1,013,654
bsd-3-clause	528,980
gpl-3.0	305,332
public-domain	163,493
bsd-2-clause	94,426
bsd	69,967
isc	36,117
unlicense	28,411
cc0-1.0	26,799
mpl-2.0	9,459
Other licenses	~5,000

Dataset Structure

Data Fields

Field	Type	Description
code	string	Content of the source file (UTF-8 encoded)
repo_name	string	Name of the GitCode repository (format: username/repo)
path	string	Path of the file within the repository (relative to repo root)
language	string	Programming language as identified by go-enry
license	string	License of the repository (SPDX identifier or "unknown")
size	int64	Size of the source file in bytes

Data Format

• Format: Apache Parquet with Zstd compression
• File Structure: 34 files (gitcode_0000.parquet to gitcode_0033.parquet)

Data Splits

All examples are in the train split. There is no validation or test split.

Example Data Point

{
    'code': '#!/bin/sh\n# defines variable\ngit=`ls -d */` && path=`pwd` && gitpath=$path/$git\n...',
    'repo_name': '00fly/git-auto-push',
    'path': 'all-in-one-cron-push.sh',
    'language': 'Shell',
    'license': 'apache-2.0',
    'size': 546
}

Dataset Creation

Pipeline Overview

The dataset was created through a multi-stage pipeline:

1. User Discovery: Collecting usernames via GitCode API
2. Repository Discovery: Fetching repository lists for each user via GitCode API
3. Repository Cloning: Using Git partial clone with blob size filtering
4. Content Extraction: Extracting and filtering source code files
5. Parquet Generation: Writing filtered records to Parquet shards with Zstd compression

Language Detection

Programming languages are detected using go-enry, a Go port of GitHub's Linguist library. Only files classified as Programming or Markup language types are included (Data and Prose types are excluded).

License Detection

Licenses are detected by:

1. Scanning for license files (LICENSE, LICENSE.txt, LICENSE.md, COPYING, etc.)
2. Matching license text against known patterns (MIT, Apache 2.0, GPL variants, BSD, Creative Commons, etc.)
3. Defaulting to "unknown" if no license can be detected

Blocked Licenses: The following restrictive licenses are excluded from the dataset:

• cc-by-nd, cc-by-nd-2.0, cc-by-nd-3.0, cc-by-nd-4.0 (Creative Commons No-Derivatives)
• commons-clause
• sspl, sspl-1.0 (Server Side Public License)

File Filtering

Extensive filtering is applied to ensure data quality:

Size Limits

Limit	Value
Max single file size	1 MB
Max line length	1,000 characters

Excluded Directories

• Configuration: .git/, .github/, .gitlab/, .vscode/, .idea/
• Vendor/Dependencies: node_modules/, vendor/, third_party/, bower_components/
• Build Output: build/, dist/, out/, bin/, target/, __pycache__/, .next/, .nuxt/, coverage/, .nyc_output/

Excluded Files

• Lock Files: package-lock.json, yarn.lock, pnpm-lock.yaml, Cargo.lock, go.sum, composer.lock, Gemfile.lock, poetry.lock, Pipfile.lock
• Minified Files: Any file containing .min. in the name
• Binary Files: .exe, .dll, .so, .dylib, .a, .o, .obj, .lib, .jar, .class, .pyc, .pyo, .wasm, .zip, .tar, .gz, .bz2, .7z, .rar, .png, .jpg, .jpeg, .gif, .bmp, .ico, .svg, .webp, .mp3, .mp4, .avi, .mov, .wav, .pdf, .doc, .docx, .xls, .xlsx, .ttf, .otf, .woff, .woff2, .eot

Content Filtering

• UTF-8 Validation: Files must be valid UTF-8 encoded text
• Binary Detection: Files detected as binary by go-enry are excluded
• Generated Files: Files with generation markers in the first 500 bytes are excluded:
  • generated by, do not edit, auto-generated, autogenerated, automatically generated, @generated, <auto-generated, this file is generated
• Empty Files: Files that are empty or contain only whitespace are excluded
• Long Lines: Files with any line exceeding 1,000 characters are excluded
• go-enry Filters: Additional filtering using go-enry's IsVendor(), IsImage(), IsDotFile(), and IsGenerated() functions
• Documentation-only Repos: Repositories containing only documentation files (no actual code) are skipped

Source Data

All data originates from public repositories hosted on GitCode.

Considerations for Using the Data

Personal and Sensitive Information

The dataset may contain:

• Email addresses in code comments or configuration files
• API keys or credentials that were accidentally committed
• Personal information in comments or documentation

Users should exercise caution and implement appropriate filtering when using this data.

Licensing Information

This dataset is a collection of source code from repositories with various licenses. Any use of all or part of the code gathered in this dataset must abide by the terms of the original licenses, including attribution clauses when relevant. The license field in each data point indicates the license of the source repository.