Hugging Face's logo

Datasets:
nyuuzyou
/
google-code-archive

Tasks:
Text Generation
Modalities:
Text
Formats:
parquet
Languages:
code
English
Size:
10M - 100M
Tags:
code
google-code
archive
Libraries:
Datasets
Dask
Polars
License:
Dataset card Data Studio Files
xet
Community
google-code-archive / README.md
nyuuzyou's picture
nyuuzyou
Upload folder using huggingface_hub
242084c verified
preview code
|
raw
history blame contribute delete
9.65 kB
metadata
annotations_creators:
  - machine-generated
language_creators:
  - found
language:
  - code
  - en
license: other
multilinguality:
  - multilingual
pretty_name: Google Code Archive Dataset
size_categories:
  - 10M<n<100M
source_datasets:
  - original
task_categories:
  - text-generation
tags:
  - code
  - google-code
  - archive
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

Google Code Archive Dataset

Dataset Description

This dataset was compiled from the Google Code Archive, a preserved snapshot of projects hosted on Google Code, Google's open-source project hosting service that operated from 2006 to 2016. Google Code was one of the major code hosting platforms of its era, hosting hundreds of thousands of open-source projects before its shutdown. The archive provides a unique historical record of open-source development during a formative period of modern software engineering.

Dataset Summary

Statistic Value
Total Files 65,825,565
Total Repositories 488,618
Total Size 47 GB (compressed Parquet)
Programming Languages 454
File Format Parquet with Zstd compression (71 files)

Key Features

  • Historical open-source corpus: Contains code from over 488K repositories hosted on Google Code during 2006-2016
  • Diverse language coverage: Spans 454 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
  • Quality filtered: Extensive filtering to remove vendor code, build artifacts, generated files, and low-quality content
  • Era-specific patterns: Captures coding conventions and library usage from the pre-modern era of software development

Languages

The dataset includes 454 programming languages. The top 30 languages by file count:

Rank Language File Count
1 Java 16,331,993
2 PHP 12,764,574
3 HTML 5,705,184
4 C++ 5,090,685
5 JavaScript 4,937,765
6 C 4,179,202
7 C# 3,872,245
8 Python 2,207,240
9 CSS 1,697,385
10 Objective-C 1,186,050
11 Shell 639,183
12 Java Server Pages 541,498
13 ActionScript 540,557
14 Makefile 481,563
15 ASP.NET 381,389
16 Smarty 339,555
17 Ruby 331,743
18 Go 316,427
19 Perl 307,960
20 Vim Script 216,236
21 Lua 215,226
22 HTML+PHP 150,781
23 HTML+Razor 149,131
24 MATLAB 145,686
25 Batchfile 138,523
26 Pascal 135,992
27 Visual Basic .NET 118,732
28 TeX 110,379
29 Less 98,221
30 Unix Assembly 94,758

Licenses

The dataset includes files from repositories with various licenses as specified in the Google Code Archive:

License File Count
Apache License 2.0 (asf20) 21,568,143
GNU GPL v3 (gpl3) 14,843,470
GNU GPL v2 (gpl2) 6,824,185
Other Open Source (oos) 5,433,436
MIT License (mit) 4,754,567
GNU LGPL (lgpl) 4,073,137
BSD License (bsd) 3,787,348
Artistic License (art) 1,910,047
Eclipse Public License (epl) 1,587,289
Mozilla Public License 1.1 (mpl11) 580,102
Multiple Licenses (multiple) 372,457
Google Summer of Code (gsoc) 63,292
Public Domain (publicdomain) 28,092

Dataset Structure

Data Fields

Field Type Description
code string Content of the source file (UTF-8 encoded)
repo_name string Name of the Google Code project
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 (Google Code license identifier)
size int64 Size of the source file in bytes

Data Format

  • Format: Apache Parquet with Zstd compression
  • File Structure: 71 files (google_code_0000.parquet to google_code_0070.parquet)

Data Splits

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

Example Data Point 

{
    'code': 'public class HundredIntegers {\n\tpublic static void main (String[] args) {\n\t\tfor (int i = 1; i<=100; i++) {\n\t\t\tSystem.out.println(i);\n\t\t}\n\t}\n}',
    'repo_name': '100integers',
    'path': 'HundredIntegers.java',
    'language': 'Java',
    'license': 'epl',
    'size': 147
}

Dataset Creation

Pipeline Overview

The dataset was created through a multi-stage pipeline:

  1. Project Discovery: Fetching project metadata from the Google Code Archive
  2. Source Filtering: Selecting projects that have source code available (hasSource: true)
  3. Archive Downloading: Downloading source archives from the Google Code Archive storage
  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 obtained directly from the Google Code Archive project metadata. The archive preserves the original license selection made by project owners when creating their repositories on Google Code.

File Filtering

Extensive filtering is applied to ensure data quality:

Size Limits

Limit Value
Max repository archive size 64 MB
Max single file size 2 MB
Max line length 1,000 characters

Excluded Directories

  • Configuration: .git/, .github/, .gitlab/, .vscode/, .idea/, .vs/, .settings/, .eclipse/, .project/, .metadata/
  • Vendor/Dependencies: node_modules/, bower_components/, jspm_packages/, vendor/, third_party/, 3rdparty/, external/, packages/, deps/, lib/vendor/, target/dependency/, Pods/
  • Build Output: build/, dist/, out/, bin/, target/, release/, debug/, .next/, .nuxt/, _site/, _build/, __pycache__/, .pytest_cache/, cmake-build-*, .gradle/, .maven/

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
  • Minified Files: Any file containing .min. in the name
  • Binary Files: .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
  • System Files: .DS_Store, thumbs.db

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, @generated, <auto-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(), IsTest(), and IsGenerated() functions
  • Documentation-only Repos: Repositories containing only documentation files (no actual code) are skipped

Source Data

All data originates from the Google Code Archive, which preserves projects hosted on Google Code before its shutdown in January 2016.

Considerations for Using the Data

Historical Context

This dataset represents code from 2006-2016 and may contain:

  • Outdated coding patterns and deprecated APIs
  • Legacy library dependencies that are no longer maintained
  • Security vulnerabilities that have since been discovered and patched
  • Code written for older language versions (Python 2, older Java versions, etc.)

Users should be aware that this code reflects historical practices and may not represent modern best practices.

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.