SVN command to delete all locally missing files

Question

In SVN is there a command I can use to delete all locally missing files in a directory?

Or failing that, some way of listing only those files that are missing (or, in the more general case, have status A, D, ?, etc.)

I maintain 1 working folder, and 1 checkout folder. And move only files from working folder to the check out folder. If a folder must be added/removed, I do it manually and then move files into that folder (for example new packages, old packages). Most svn issues happen when you move folders on your working copy. Even when coping files from your working copy to checkout copy, you should only move files, not folders. — Siddharth, Mar 09 '12 at 05:37

score 186 · Answer 1 · edited Dec 24 '16 at 08:38

186

If you're using Mac (Darwin) or Linux you can pipe the outputs of the following commands to svn rm for all missing files. You can set the current working directory to the appropriate directory or subdirectory before running these - dependent on whether you want to run this your entire working copy, or only a subset.

Run an svn status
Search for lines that begin with "!" (missing)
Print the "--force" (svn argument) and the second column (the file name) of the output from #2
Run svn rm using the output of #3 as arguments

So the full command is:

svn st | grep ^! | awk '{print " --force "$2}' | xargs svn rm

References:

edited Dec 24 '16 at 08:38

Peter Mortensen

30,738
21
105
131

answered Dec 12 '12 at 19:07

Peter Ajtai

56,972
13
121
140

4

This doesn't work if there are spaces in the file name. I don't know enough awk to fix it, though. – bugloaf Oct 08 '13 at 15:34
@bugloaf - thanks for the heads up - think you'd have to incorporate some sort of loop to get all columns 2+ - something like http://stackoverflow.com/a/7918051/186636 - not sure exactly - I'll try to play around with it at some point – Peter Ajtai Oct 08 '13 at 18:33
4

Use `awk '{print " --force "$2"@"}'` if your filename contains a `@` (if you're an iOS dev for example) – Guillaume Algis Nov 21 '13 at 16:35
1

And handle a filename with whitespace(s), `awk '{$1=""; print " --force \""substr($0,2)"@\"" }'` should do the trick (and begins to look ugly) – Guillaume Algis Nov 21 '13 at 16:42
Here's another option for handling files with whitespace (the \x22 prints a straight double-quote): delete - `svn status | grep ^! | awk -F ' ' '{print " \x22"$2"\x22"}' | xargs svn delete`, add - `svn status | grep ^? | awk -F ' ' '{print " \x22"$2"\x22"}' | xargs svn add`. Note that's seven spaces in the `-F` argument. – Nick Williams Nov 25 '13 at 22:02
14

The full command taking care of the space is "svn st | grep ^! | awk '{$1=""; print " --force \""substr($0,2)"@\"" }' | xargs svn rm" – ABCD Jan 03 '14 at 01:27
4

Why don't you put the `--force` into the xargs part? – jwg Feb 21 '14 at 17:25
Under windows and cygwin use: svn st | grep ^! | awk '{gsub(/\\/, "/");print " --force \""$2"\""}' | xargs svn rm (gsub to replace / with \; filename with in " to work with whitespaces) – leo Apr 16 '15 at 08:02
This is pretty great! thanks everyone. It can be VERY easily modified to work for newly added files by changing the `^!` to `^?` – Tony Adams Aug 19 '15 at 19:42
1

This filename hack is so terrible, stop using this old 1970th unix hacks, it does not work with all possible file names that require more complex escapes. – Lothar May 22 '18 at 13:54
1

I stopped using SVN. – Peter Ajtai May 25 '18 at 00:00
@Lothar, right... `svn status --xml` would be a better start for the pipeline. – David L. Mar 10 '21 at 10:15

score 109 · Accepted Answer · edited Dec 24 '16 at 08:30

109

If you are using TortoiseSVN, just do a Check for Modifications, sort by the Status column, select all the entries marked missing, right-click to open the context menu, and select Delete. Finally, commit to publish the changes to the repository.

If you are on Windows, but prefer the command-line and enjoy dabbling in PowerShell, this one-liner will do the trick:

svn status | ? { $_ -match '^!\s+(.*)' } | % { svn rm $Matches[1] }

That is, filter the output to only those lines showing missing files (denoted by an exclamation at the start of the line), capture the associated file name, and perform an svn rm on that file name.

(Blog post Remove all “missing” files from a SVN working copy does something similar for Unix/Linux.)

edited Dec 24 '16 at 08:30

Peter Mortensen

30,738
21
105
131

answered Mar 09 '12 at 04:24

Michael Sorens

35,361
26
116
172

is there a way in eclipse using subclipse plugin? it is so bad to update a then find them in project and delete them again. – Stan May 29 '14 at 06:41
FYI, if a subversion password is needed here is the syntax: svn status --username user_name_here --password password_here | ? { $_ -match '^!\s+(.*)' } | % { svn rm $Matches[1] --username user_name_here--password password_here } – Jeff Sep 26 '16 at 14:03
should you first commit after this line before executing `svn up` ? because the files are restored from the svn server – Pierre Oct 23 '18 at 09:38
Will this action delete the files from the repository too? – Sebastián Ulloa Feb 05 '19 at 18:08

score 65 · Answer 3 · edited Dec 24 '16 at 08:44

65

svn st | grep ! | cut -d! -f2| sed 's/^ *//' | sed 's/^/"/g' | sed 's/$/"/g' | xargs svn rm

svn status
Filter only on missing files
Cut out exclamation point
Filter out trailing whitespaces
Add leading quote
Add trailing quote
svn remove each file

edited Dec 24 '16 at 08:44

Peter Mortensen

30,738
21
105
131

answered Aug 25 '14 at 19:06

ivrin

1,707
1
14
11

This solution works if there are spaces in the files/directories. – yuric Jan 21 '16 at 18:43
How does one add verbosity to this? xargs -verbose svn rm ? – TechFanDan May 09 '17 at 14:52
1

Just a hint for users of scripts etc., this command fails if no files to delete are found: svn: E205001: Try 'svn help delete' for more information svn: E205001: Not enough arguments provided – Thorsten Schöning Jul 09 '17 at 10:42
2

@ThorstenSchöning's problem can be fixed by adding `-r` to the `xargs` options (do not run, if nor arguments are given). Additionally, the numer of arguments should be limited and removals should be batched because the invocation fails with very large number of files to remove, e.g. `-n 500`. And last but not least, the quoting performed is suboptimal for files containing shell specials like `$` - better use `\n` as delimiter and leave quoting to `xargs`: `svn status | grep "!" | cut -d! -f2 | sed 's/^ *//' | xargs -n 500 -d "\n" -r svn rm` – haui Jun 09 '20 at 06:35

score 32 · Answer 4 · edited May 26 '13 at 07:28

32

I just found this, which does the trick, Remove all “missing” files from a SVN working copy:

svn rm $( svn status | sed -e '/^!/!d' -e 's/^!//' )

edited May 26 '13 at 07:28

brasofilo

25,496
15
91
179

answered Mar 07 '12 at 11:16

wheresrhys

22,558
19
94
162

Worked for me on OS X. Thanks :) – Gustavo de Souza Feb 20 '19 at 13:35

score 22 · Answer 5 · edited Dec 24 '16 at 08:43

22

Thanks to Paul Martin for the Windows version.

Here is a slight modification to the script so Windows files with spaces are taken into account as well. Also the missing.list file will be removed at the end.

I saved the following in svndel.bat in my SVN bin directory (set in my %%PATH environment) so it can be called from any folder at the command prompt.

### svndel.bat
svn status | findstr /R "^!" > missing.list
for /F "tokens=* delims=! " %%A in (missing.list) do (svn delete "%%A")
del missing.list 2>NUL

edited Dec 24 '16 at 08:43

Peter Mortensen

30,738
21
105
131

answered May 29 '14 at 23:15

user3689460

229
2
2

nice to see some innovative batch code still appearing :) – user230910 Sep 25 '17 at 09:04
Wow, great answer. I was about to go with powershell, but this was so much easier. – Jim Andrakakis Jan 26 '18 at 15:21
I added another line after `del missing.list 2>NUL`, because I added all new files directly after first removing the deleted files so it added `missing.list` to svn: `svn delete "missing.list"` worked for me – Pierre Oct 24 '18 at 06:33
I had modified this script to attempt to run from an arbitrary location then encountered this issue https://stackoverflow.com/questions/1137219/aptana-1-5-svn-error-resolving-case-of/1137297#1137297 so it appears you MUST use the Current Working Directory to perform this action. – aolszowka Dec 02 '19 at 17:10

score 8 · Answer 6 · edited Dec 24 '16 at 08:41

8

I like the PowerShell option... But here's another option if you're using Windows batch scripts:

svn status | findstr /R "^!" > missing.list
for /F "tokens=2 delims= " %%A in (missing.list) do (svn delete %%A)

edited Dec 24 '16 at 08:41

Peter Mortensen

30,738
21
105
131

answered Feb 07 '13 at 16:52

Paul Martin

316
3
5

score 8 · Answer 7 · answered Sep 22 '17 at 07:13

8

It is actually possible to completely remove the missing.list from user3689460 and Paul Martin

for /F "tokens=* delims=! " %%A in ('svn status ^| findstr /R "^!"') do (svn delete "%%A")

answered Sep 22 '17 at 07:13

Amfasis

3,932
2
18
27

for reference - this is DOS / cmd.exe in windows land – user230910 Oct 10 '17 at 07:01

score 6 · Answer 8 · answered Mar 11 '16 at 10:20

6

An alternative that works on Linux (bash) for to-be-removed files not containg spaces in path:

svn delete `svn status | grep ! | awk '{print $2}'`

answered Mar 11 '16 at 10:20

gilyen

81
1
4

score 5 · Answer 9 · answered Jan 27 '13 at 08:28

This shell script, recursively examines (svn status) directories in your project, removing missing files (as the question demands) and adding new files to the repository. It is some sort of "store into the repository the current snapshot of the project".

if [ $# != 1 ]
then
    echo  "usage: doSVNsnapshot.sh DIR"
    exit 0
fi

ROOT=$1

for i in `find ${ROOT} -type d \! -path "*.svn*" `
do

    echo
    echo "--------------------------"
    ( cd $i ; 
    echo $i
    echo "--------------------------"


    svn status | awk '  
            /^[!]/ { system("svn rm " $2) }
            /^[?]/ { system("svn add " $2) }
        '
    )
    echo

done

score 5 · Answer 10 · edited Dec 24 '16 at 08:46

5

A slight modification of the command line, which works on Mac OS (hopefully even on Linux) and copes with the files the command "svm sr" reports, like "!M" (missing and modified).

It copes with spaces in the files.

It is based on a modification of a previous answer:

svn st | grep ! | sed 's/!M/!/' | cut -d! -f2| sed 's/^ *//' | sed 's/^/"/g' | sed 's/$/"/g' | xargs svn --force rm

edited Dec 24 '16 at 08:46

Peter Mortensen

30,738
21
105
131

answered Apr 22 '16 at 14:30

Stefano Bossi

51
1
4

score 3 · Answer 11 · answered Oct 31 '14 at 14:27

When dealing with a lot of files, it can happen that the argument input to xargs is getting too long. I went for a more naive implementation which works in that case, too.

This is for Linux:

#! /bin/bash
# 1. get all statii in the working copy
# 2. filter out only missing files
# 3. cut off the status indicator (!) and only return filepaths
MISSING_PATHS=$(svn status $1 | grep -E '^!' | awk '{print $2}')
# iterate over filepaths
for MISSING_PATH in $MISSING_PATHS; do
    echo $MISSING_PATH
    svn rm --force "$MISSING_PATH"
done

score 3 · Answer 12 · answered Jan 08 '19 at 06:21

3

Improved Version

So the full command is:

svn st | grep ^! | sed 's/![[:space:]]*//' |tr '\n' '\0' | xargs -0 svn --force rm

answered Jan 08 '19 at 06:21

Zeta

65
4

The part "sed 's/![[:space:]]*//'" should read "sed 's/^![[:space:]]*//'" (notice the extra caret "^") so that it only matches "!" + spaces at the beginning of each line and not in the middle. – stux Oct 09 '21 at 16:49

SVN command to delete all locally missing files

12 Answers12

Linked