Mine Software Repositories with Repoblick.
Just run repoblick plot bb. It gets the changesets of the first
twenty repositories at Bitbucket and generates a plot of when contributors
joined the projects. Check your ~/.repoblick/output/bitbucket directory for the
image file.
Repoblick does a number of things.
- List projects in a repository (currently only Bitbucket)
- Download projects, either cloning the whole repository, or only a part of it upto a certain changeset
- Import log data into an sqlite database for later processing
- Convert a raw, almost normalized database to a more verbose statistical database for easier querying
- Plot statistical database using a built-in query or plain sql
Since each step requires the previous ones, performing any of them will
perform the previous ones, too. Unless it was done earlier, or you
explicitly tell repoblick not to (with the --only-this-step option).
Running repoblick/cmdline.py -h prints
usage: repoblick/cmdline.py [-h] {plot,statdb,list,splot,mirror,import} ...
Explore version controlled repositories
positional arguments:
{plot,statdb,list,splot,mirror,import}
list List repositories and store their names and attributes
mirror Make a local mirror of repositories for later
processing
import Process logs of repositories and put them into the
database
statdb Convert logdata database to a statistical database
plot Make a plot of log data
splot Make a plot of a single project using SVNPlot (not
working yet)
optional arguments:
-h, --help show this help message and exit
Running repoblick/cmdline.py plot -h prints
usage: repoblick/cmdline.py plot [-h] [-d DIR] [-f FROM_FILE] [-o] [-c COMMITS]
[-s START_INDEX] [-i INDEX_COUNT] [-q QUERY]
host [projects [projects ...]]
positional arguments:
host Either a known host (see cmdline.py plot show hosts),
or a local path
projects A specific project at the host. If not given, all
projects at host are used
optional arguments:
-h, --help show this help message and exit
-d DIR, --dir DIR Directory to store databases and repository mirrors
-f FROM_FILE, --from-file FROM_FILE
File to read for projects
-o, --only-this-step Do not perform operations of previous steps even if
needed
-c COMMITS, --commits COMMITS
Maximum number of commits to mirror
-s START_INDEX, --start-index START_INDEX
Start listing at this index
-i INDEX_COUNT, --index-count INDEX_COUNT
-q QUERY, --query QUERY
On systems with a debian-style package management (Debian, Ubuntu, etc), run
apt-get install python python-argparse python-pysqlite2 python-matplotlib mercurial.
If you have python>=2.7, you won't need argparse, it's part of the standard library.
If all you want is crunch data without plotting, you won't need matplotlib.