CODEWITHSUNDEEP
svnmergejira
Nikolay Kuznetsov
Nikolay Kuznetsov

Reputation: 9579

SVN automatic merge from trunk to branch by specifying JIRA ticket id

I use JIRA as bug-tracking system. If there is a newly revealed bug then it is commited to the trunk first. Commits are done under a specific ticket. For example, ticket #37 is fixed by commits: 65, 68 and 74.

Sometimes I would need to backport a ticket to a branch. So I have to download a repository for a branch, i.e. #3 and apply following command 

svn merge -c65,68,74  http://.../svn/trunk

then to commit and so on for every ticket.

What I would like to do is specify a list of tickets from JIRA and branch id. So they would be merged automatically from trunk to a specific branch.

What are the options to accomplish that?

Upvotes: 1

Views: 907

Answers (1)

Dan
Dan

Reputation: 4502

If you're referencing your Jira tickets in a consistent way from the commit messages, using the Jira ABC-123 ticket number format you can get a list of revisions by searching the SVN log.

svn log --search=FOO-57 http://.../svn/trunk

This will give you the log messages and revision numbers for each relevant commit. Assuming a unix-like environment, you can use some pipes and shell manipulation to convert the svn log output to a list of revision numbers like so:

svn log --search=FOO-57 http://.../svn/trunk | grep ^r | cut '-d|' -f1 | sed s,r,,

A little bit more shell manipulation can give you that same revision list in a comma-separated format:

REVISIONS=($(svn log --search=FOO-57 http://.../svn/trunk | grep ^r | cut '-d|' -f1 | sed s,r,,))
echo "${REVISIONS[*]}" | sed 's/ /,/g'

So now you can merge just by using the output of that second command as your -c argument to svn merge

REVISIONS=($(svn log --search=FOO-57 http://.../svn/trunk | grep ^r | cut '-d|' -f1 | sed s,r,,))
svn merge -c$(echo "${REVISIONS[*]}" | sed 's/ /,/g') http://.../svn/trunk

You can convert the above into a re-usable shell script by replacing the search term and the URL with "$1" and "$2". Or, if you want to apply this to multiple Jira tickets at once, you can wrap the entire thing in a loop like so:

#!/bin/bash
TRUNK_URL="http://.../svn/trunk"
REVISIONS=($((for search_term in "$@"; do
    svn log --search="$search_term" "$TRUNK_URL" | grep ^r | cut '-d|' -f1 | sed s,r,,;
done) | sort -u))
svn merge -c$(echo "${REVISIONS[*]}" | sed 's/ /,/g') "$TRUNK_URL"

If you'd prefer to avoid shell scripting to get the revision list, you can achieve a similar results using any language with libsvn bindings. Here's an example using Python and pysvn:

#!/usr/bin/env python
import pysvn
logs = pysvn.Client().log('http://.../svn/trunk')
to_merge = { x['revision'].number for x in logs if 'FOO-57' in x['message'] }

Upvotes: 3

Related Questions