Subversion: Check out only those files affected during a specific commit

Question

In Subversion, is it possible to check out only those files affected during a specific commit, provided you know the specific revision number?

Answer 1

I think that a Subversion check out can only check out a specific folder and not individual files. That said, the command:

svn log -r *revision* -q -v

will list the changes associated with the specified revision so you could process the output of this command to help copy the desired files to somewhere after checking out a working folder for the full directory.

Answer 2

you can get a list of files modified since a specific revision like this:

svn diff -r [REVNUM]:HEAD --summarize > fileschanged.txt

then create a script to checkout each file in the fileschanged.txt

Answer 3

I think it's possible that you're using the term "check out" here in its CVS sense, that is, an "svn update". Is that right?

If so, then you can update to the particular revision that you want:

svn update -r12345

or even just those files to that revision:

svn update -r12345 file1 file2 file3

though you'll have to get the list of files from svn using some of the other suggestions.

Answer 4

I know it's a bit offtopic but I use TRAC to do this and it exports only the files modified between 2 revisions as a zip archive with the files in their original directory structure.

Answer 5

svn checkout only works at the directory level, not the file level. Of course, if all the changes in a specific changeset were from the same directory, then you can perform the svn checkout <url> <path> -r<revid>.

Answer 6

You can check the list of affected files using svn log -q -v -r123 url.

If you really only want to get the affected files you can use svn cat urlToFile -r123 > myfile for each file.

Answer 7

If you're using a recent version of svn you can checkout an empty folder, then update the specific files.

after getting your list of files to select,

svn co --depth=empty url dest
cd dest
svn up file1 file2 file3

Edit: pretty much the same solution as accepted answer here: Checkout one file from Subversion

Answer 8

Here is one way to do it.

svn co --depth empty svn://myrepository/mytrunk

cd mytrunk

svn diff -c[REVNUM] svn://myrepository/mytrunk | grep "Index:" | awk '{print $2}' | xargs -n 1 svn update --parents

The output from the awk one-liner is a list of files changed with relative paths. As previously mentioned, update works on individual files while checkout works on modules. The '--parents' option forces the creation of directories in your working copy as necessary.

Note that update also has a '--changelist' option (at least in v.1.7.1+), but it didn't seem to be doing what I wanted.

Answer 9

I needed a dump of all the files that had changed within a range of a dozen revisions. I did it like so (these commands may not be perfect, this is just a general explanation):

Create a diff that will show the files you need and write the list to a text file

svn diff --summarize -r219:232 > r219-232_Summary.txt

This will give you a text file with lines like

M      path/to/file.php

Massage the file format to replace the beginning of each line with an 'svn up' command, instead of 'A ' or 'M ' or whatever.

 sed -i 's/A      /svn up /g' ./r219-232_Summary.txt

... which will give you lines like

svn up path/to/file.php

Create a new directory and check out your project to it

svn co http://www.repo.net/whatever

Remove everything in the directory except the .svn file (I'm using a relatively recent svn client so if you're on an old svn client that has a .svn in every directory, not sure how that will work)

rm -rf ./*

Copy in your text file, make it executable, and run it as a script.

cp /path/to/wherever/r219-232_Summary.txt ./r219-232_Summary.sh
chmod 777 ./r219-232_Summary.sh
./r219-232_Summary.sh

If all goes as planned, then shell should parse each 'svn up' command in your text file and populate the directory with only the files that you're interested in.

Answer 10

Here is a simple bash script version. I used a "search pattern" to remove unnecessary lines. Revision, repository address and destination path are given as arguments but it could be hardcoded.

#!/bin/bash

REVISION=${1}
REPOSITORY="${2}"
DEST_PATH="${3}"
SEARCH_PATTERN="trunk"
FILES=$(svn log -q -v  $REPOSITORY -r $REVISION | grep $SEARCH_PATTERN)

mkdir -p $DEST_PATH
for FILE in $FILES ; do
    if [[ $FILE == *"$SEARCH_PATTERN"* ]]
    then
          FOLDER=$(dirname $FILE)
          mkdir -p $DEST_PATH/$FOLDER
      svn export -r $REVISION $REPOSITORY/$FILE $DEST_PATH/$FOLDER
    fi
done

#EOF

Answer 11

You can get just the list of files in a particular change using

svn log -v -q -r rev

You can then clean up the markers on the front of the line with a pipe like

sed -e "s/^ . / /" | egrep -v "Changed paths:|-----"

. In order to get just the files, you would probably use svn cat as suggested by Nikolai.

[edited to cleanup and simplify the sed pipeline.]

Answer 12

Sorry to resurrect a long-dead question, but it was never answered to my satisfaction.

You can build a simple tool to do this for you using SVNKit. I found this blog about it, and slightly modified the code for my needs. I'm using it to this day in my deployment process.

It does an export not a checkout, but I'm sure that would be a simple change.

Answer 13

svn mergeinfo --show-revs eligible http://base.com.rev1 http://base.com.rev2

This will give a lot of revisions

Now you would like to see the list of files in each revision. So take the list of revision and copy into note pad or editplus and modify so that they fit into one line

svn log -r r33521 -r33762 -3456 -r5623 -q -v >> missedFiles.txt

Note : I had 118 revisions and I was not able to get the list of files correctly. So i ran the above command by breaking into subsets of revision