gitqlite

gitqlite is a tool for running SQL queries on git repositories. It implements SQLite virtual tables and uses go-git. It's meant for ad-hoc querying of git repositories on disk through a common interface (SQL), as an alternative to patching together various shell commands.

Installation

go install -v -tags=sqlite_vtable github.com/augmentable-dev/gitqlite

Will use the go tool chain to install a binary to $GOBIN.

GOBIN=$(pwd) go install -v -tags=sqlite_vtable github.com/augmentable-dev/gitqlite

Will produce a binary in your current directory.

TODO: more installation instructions

Usage

gitqlite -h

Will output the most up to date usage instructions for your version of the CLI. Typically the first argument is a SQL query string:

gitqlite "SELECT * FROM commits"

Your current working directory will be used as the path to the git repository to query by default. Use the --repo flag to specify an alternate path, or even a remote repository reference (http(s) or ssh). gitqlite will clone the remote repository to a temporary directory before executing a query.

You can also pass a query in via stdin:

cat query.sql | gitqlite

By default, output will be an ASCII table. Use --format json or --format csv for alternatives. See -h for all the options.

Tables

commits

Similar to git log, the commits table includes all commits in the history of the currently checked out commit.

Column	Type
id	TEXT
message	TEXT
summary	TEXT
author_name	TEXT
author_email	TEXT
author_when	DATETIME
committer_name	TEXT
committer_email	TEXT
committer_when	DATETIME
parent_id	TEXT
parent_count	INT
tree_id	TEXT
additions	INT
deletions	INT

files

The files table iterates over ALL the files in a commit history, by default from what's checked out in the repository. The full table is every file in every tree of a commit history. Use the commit_id column to filter for files that belong to the work tree of a specific commit.

Column	Type
commit_id	TEXT
tree_id	TEXT
name	TEXT
mode	TEXT
type	TEXT
contents	TEXT

refs

Column	Type
name	TEXT
type	TEXT
hash	TEXT

Example Queries

This will return all commits in the history of the currently checked out branch/commit of the repo.

SELECT * FROM commits

Return the (de-duplicated) email addresses of commit authors:

SELECT DISTINCT author_email FROM commits

Return the commit counts of every author (by email):

SELECT author_email, count(*) FROM commits GROUP BY author_email ORDER BY count(*) DESC

Same as above, but excluding merge commits:

SELECT author_email, count(*) FROM commits WHERE parent_count < 2 GROUP BY author_email ORDER BY count(*) DESC

This is an expensive query. It will iterate over every file in every tree of every commit in the current history:

SELECT * FROM files

Outputs the set of files in the tree of a certain commit:

SELECT * FROM files WHERE commit_id='some_commit_id'

Same as above if you just have the commit short id:

SELECT * FROM files WHERE commit_id LIKE 'shortened_commit_id%'

Returns author emails with lines added/removed, ordered by total number of commits in the history:

SELECT count(*) AS commits, SUM(additions) AS additions, SUM(deletions) AS  deletions, author_email FROM commits GROUP BY author_email ORDER BY commits

Returns commit counts by author, broken out by day of the week:

SELECT
    count(*) AS commits,
    count(case when strftime('%w',author_when)='0' then 1 end) as sunday,
    count(case when strftime('%w',author_when)='1' then 1 end) as monday,
    count(case when strftime('%w',author_when)='2' then 1 end) as tuesday,
    count(case when strftime('%w',author_when)='3' then 1 end) as wednesday,
    count(case when strftime('%w',author_when)='4' then 1 end) as thursday,
    count(case when strftime('%w',author_when)='5' then 1 end) as friday,
    count(case when strftime('%w',author_when)='6' then 1 end) as saturday,
    author_email
FROM commits GROUP BY author_email ORDER BY commits