Skip to content

Latest commit

 

History

History
182 lines (136 loc) · 5.98 KB

ReadMe.md

File metadata and controls

182 lines (136 loc) · 5.98 KB
Raw

MCMD | Multi-programming-language Commit Message Dataset

Download_URL

This dataset has two version: Raw and Filtered

All information can be found in Raw Data.

Raw Data

tar -zxvf raw_data.tar.gz
|-- raw_data
    |-- cpp
    |   |-- ...
    |   |   |-- ...
    |   |-- Tencent
    |   |   |-- MMKV.pickle
    |   |   |-- mars.pickle
    |   |   |-- ncnn.pickle
    |   |   |-- rapidjson.pickle
    |   |-- ...
    |   |   |-- ...
    |   |-- zhongyang219
    |       |-- TrafficMonitor.pickle
    |-- csharp
    |   |-- ...
    |   |   |-- ...
    |   |-- xupefei
    |       |-- Locale-Emulator.pickle
    |-- java
    |   |-- elastic
    |   |   |-- elasticsearch.pickle
    |   |-- ...
    |   |   |-- ...
    |   |-- zxing
    |       |-- zxing.pickle
    |-- javascript
    |   |-- ...
    |   |   |-- ...
    |   |-- vuejs
    |       |-- vue-cli.pickle
    |       |-- vue.pickle
    |       |-- vuex.pickle
    |-- python
        |-- ...
        |   |-- ...
        |-- yunjey
        |   |-- pytorch-tutorial.pickle
        |-- zulip
            |-- zulip.pickle

Under the folder raw_data, there are 5 folders named with Programming Language including Java(java), C#(csharp), C++(cpp), Python(python), and JavaScript(javascript).

Under each Programming Language folder, there are many folders named with Owner Name such as elastic.

Under each Owner Name folder, there are many .pickle files named with Repo Name such as elasticsearch.pickle.

(RepoFullName = Owner Name + / + Repo Name .

For example, RepoFullName:elastic/elasticsearch means its Owner Name is elastic and its Repo Name is elasticsearch.)

.pickle files contain <Diff,Message, SHA,Timestamp> for each commit which created before 2021.

For example, if you want to see commits of the repository (RepoFullName:elastic/elasticsearch), you can use the code below,

import pickle

repo_raw_data = pickle.load(open("raw_data/java/elastic/elasticsearch.pickle","rb"))

where java is the Programming Language, elastic/elasticsearch is the RepoFullName.

The variable repo_raw_data stores all of the commits in elastic/elasticsearch before 2021.

If you want to see one of the commits, you can use the code below,

repo_raw_data[618]

where 618 is the index of the commit.

You can get its Diff,Message, SHA,Timestamp by

repo_raw_data[618]['diff']
repo_raw_data[618]['msg']
repo_raw_data[618]['sha']
repo_raw_data[618]['date']

Using SHA and RepoFullName, you can find the original webpage by https://github.com/`RepoFullName`/commit/`sha` as https://github.com/elastic/elasticsearch/commit/63f7fc7cb843799042e5bdb66e28eb6be0de2d7a.

The Diff is equal to the content in https://github.com/`RepoFullName`/commit/`sha` .diff as https://github.com/elastic/elasticsearch/commit/63f7fc7cb843799042e5bdb66e28eb6be0de2d7a.diff.

The Msg is equal to the sentences in https://github.com/elastic/elasticsearch/commit/63f7fc7cb843799042e5bdb66e28eb6be0de2d7a.

The Timestamp is in ISO 8601 format: YYYY-MM-DDTHH:MM:SSZ.

Example

Code and the results can be seen here.

Filtered Data

tar -zxvf filtered_data.tar.gz
|-- filtered_data
    |-- cpp
    |   |-- sort_random_train80_valid10_test10
    |   |   |-- test.diff.txt
    |   |   |-- test.msg.txt
    |   |   |-- test.repo.txt
    |   |   |-- test.sha.txt
    |   |   |-- test.time.txt
    |   |   |-- train.diff.txt
    |   |   |-- train.msg.txt
    |   |   |-- train.repo.txt
    |   |   |-- train.sha.txt
    |   |   |-- train.time.txt
    |   |   |-- valid.diff.txt
    |   |   |-- valid.msg.txt
    |   |   |-- valid.repo.txt
    |   |   |-- valid.sha.txt
    |   |   |-- valid.time.txt
    |   |-- sort_time_train80_valid10_test10
    |       |-- ...
    |-- csharp
    |   |-- sort_random_train80_valid10_test10
    |   |   |-- ...
    |   |-- sort_time_train80_valid10_test10
    |       |-- ...
    |-- java
    |   |-- sort_random_train80_valid10_test10
    |   |   |-- ...
    |   |-- sort_time_train80_valid10_test10
    |       |-- ...
    |-- javascript
    |   |-- sort_random_train80_valid10_test10
    |   |   |-- ...
    |   |-- sort_time_train80_valid10_test10
    |       |-- ...
    |-- python
        |-- sort_random_train80_valid10_test10
        |   |-- ...
        |-- sort_time_train80_valid10_test10
            |-- ...

Under the folder filtered_data, there are 5 folders named with Programming Language including Java(java), C#(csharp), C++(cpp), Python(python), and JavaScript(javascript).

Under each Programming Language folder, there are filtered data after different splitting strategies such as sort_random_train80_valid10_test10.

Under sort_random_train80_valid10_test10 folder, there are 3 classes of files including train, valid, and test.

For each class, there are .txt files including diff.txt,msg.txt,repo.txt,sha.txt,time.txt.

For example, if you want to read a commit in filtered_data/java/sort_random_train80_valid10_test10/train.msg.txt

train_msg = open("filtered_data/java/sort_random_train80_valid10_test10/train.msg.txt").read().split("\n")
train_msg[18]

More code and the results can be seen here.