This is a description of how to migrate an existing TWiki datastore to DokuWiki ready text files. I am in the process of migrating our 1000+ page TWiki site into DokuWiki and created this shell script to assist in the transition.

• An existing TWiki site
• Shell access on a Linux/UNIX box
• Perl

This script is able to translate the following items from TWiki Markup Language (TML) to DokuWiki:

• TWiki %META Information
• Attachments
• Monospaced/Bold Monospaced Text
• Headings
• Unordered Lists
• Ordered Lists
• Bold, Italic, Bold-Italic Text
• <b>, <i>, <u> to Doku syntax
• TML <verbatim>
• TML <nop>
• Table Headings

• I highly recommend running this script on a copy of your existing TWiki site…but that goes without saying.
• Using these two scripts in tandem will copy your existing Twiki one namespace at a time
  • Use the first script to copy the data and it will call the second script to do the transformations
• Save the code below into a file with a .sh extension e.g. twikiconverter.sh. The filename does not matter.
• Open a terminal window and browse to the directory into which you saved the twikiconverter.sh file.
• Change the permission of this file to allow execution

  • chmod u+x ./migratenamespace.sh
    chmod u+x ./twikiconverter.sh

• From the terminal, run the migratenamespace.sh script against a twiki namespace

  • ./migratenamespace.sh MyWikiNamespace

• Thats it! All of your TWiki documents are now DokuWiki documents imported into you new dokuwiki.

migratenamespace.sh

#!/bin/bash
#Twiki to DokuWiki File Converter
#Developed by Thomas Vachon <vachon-dokuwiki ATgamelogicDOTcom>
#Licensed under the GPL: (http://www.gnu.org/licenses/gpl.txt)
 
if [ -z "$1" ]; then
echo "Usage: $0 TwikiNamespace"
exit
fi
 
#Copy the files to your present directory (run this in the data directory of the doku desination)
rsync -arv /var/lib/twiki/data/$1 .
cd $1
rm *,v *.lock
 
#Makes all the filenames lowercase (as it is a dokuwiki requirement)
for f in *; do
g=`expr "xxx$f" : 'xxx\(.*\)' | tr '[A-Z]' '[a-z]'`
mv "$f" "$g"
done
 
cd ../
 
#make the namespace directory lowercase
namespace=`echo $1 | tr '[:upper:]' '[:lower:]'`
mv $1 $namespace
./twikiconverter $namespace/*.txt

twikiconverter

#!/bin/bash
#Twiki to DokuWiki Syntax Converter
#Developed by Ben Chapman <bkchapman atgmaildotcom>
#Supplemented by Thomas Vachon <vachon-dokuwiki ATgamelogicDOTcom>
#Licensed under the GPL: (http://www.gnu.org/licenses/gpl.txt)
 
 
if [ -z "$1" ]; then
echo "Usage: $0 twikifilename.txt or *.txt "
exit
fi
 
INPUT=$*
 
#Remove the ---+++++ !! (headed links)
perl -pi -e 's/\-\-\-\+\+\+\+\+ \!\!//g' $*
 
# Substituting newlines for carriage returns
perl -pi -e 's/\r/\n\n/g' $*
 
# Stripping Twiki %META information
perl -ni -e "print unless /^\s*%META:TOPIC*/;" $*
 
# Converting attachments
# Replace replaceme with the name of your namespace
perl -pi -e "s/\%META\:FILEATTACHMENT\{name\=\"/\{{replaceme\:/g" $*
perl -pi -e "s/\" attr\=\"\" comment\=\"/\|/g" $*
perl -pi -e 's/\" date\=\".+/}}/g' $*
 
# Converting monospaced text
perl -pi -e "s/\=/''/g" $*
 
# Converting bold, monospaced text
perl -pi -e "s/^\=\=/**''/g" $*
perl -pi -e "s/$\=\=/''**/g" $*
 
# Adding DokuWiki syntax to the headings
perl -pi -e 's/(---\+\s.*|---\+\w.*)/====== $1 ======/g' $*
perl -pi -e 's/(---\+\+\s.*|---\+\+\w.*)/===== $1 =====/g' $*
perl -pi -e 's/(---\+\+\+\s.*|---\+\+\+\w.*)/==== $1 ====/g' $*
perl -pi -e 's/(---\+\+\+\+\s.*|---\+\+\+\+\w.*)/=== $1 ===/g' $*
perl -pi -e 's/(---\+\+\+\+\+\s.*|---\+\+\+\+\+\w.*)/== $1 ==/g' $*
 
# Removing TML from the headings
perl -pi -e 's/(---\+\+\+\+\+\s|---\+\+\+\+\+)//g' $*
perl -pi -e 's/(---\+\+\+\+\s|---\+\+\+\+)//g' $*
perl -pi -e 's/(---\+\+\+\s|---\+\+\+)//g' $*
perl -pi -e 's/(---\+\+\s|---\+\+)//g' $*
perl -pi -e 's/(---\+\s|---\+)//g' $*
 
# Converting unordered lists
# I'm using %^ as a placeholder so the
# bold conversion below does not mess with
# our lists. Don't worry it gets fixed below.
perl -pi -e 's/\t\t\t\*/      \%\^/g' $*
perl -pi -e 's/\t\t\*/    \%\^/g' $*
perl -pi -e 's/\t\*/   \%\^/g' $*
perl -pi -e 's/            \*/        \%\^/g' $*
perl -pi -e 's/         \*/      \%\^/g' $*
perl -pi -e 's/      \*/    \%\^/g' $*
perl -pi -e 's/   \*/  \%\^/g' $*
 
# Commented out. See below
# Convert Ordered Lists v1
#perl -pi -e 's/\t\t\t\1/      \-/g' $*
#perl -pi -e 's/\t\t\1/    \-/g' $*
#perl -pi -e 's/\t\1/   \-/g' $*
#perl -pi -e 's/            \1/        \-/g' $*
#perl -pi -e 's/         \1/      \-/g' $*
#perl -pi -e 's/      \1/    \-/g' $*
#perl -pi -e 's/   \1/  \-/g' $*
 
# In Perl 5.1, the \1 in the capture group is supposed to reference a capturing group.  I assume the original author intended that to be a single "1" instead.  When the original set of commands are run against 5.1 an error is thrown and the text is not converted properly.  The modified version below throws no errors.
# Convert Ordered Lists v2
perl -pi -e 's/\t\t\t1/      \-/g' $*
perl -pi -e 's/\t\t1/    \-/g' $*
perl -pi -e 's/\t1/   \-/g' $*
perl -pi -e 's/            1/        \-/g' $*
perl -pi -e 's/         1/      \-/g' $*
perl -pi -e 's/      1/    \-/g' $*
perl -pi -e 's/   1/  \-/g' $*
 
# Converting bold-italic text
perl -pi -e 's/^\_{2}/\/\/\*\*/g' $*
perl -pi -e 's/$\_{2}/\*\*\/\//g' $*
 
# Converting italic text
perl -pi -e 's/(^\_{1})/\/\//g' $*
perl -pi -e 's/$\_{1}\s/\/\//g' $*
 
# Converting bold text
perl -pi -e 's/\*{1}| \*{1}/\*\*/g' $*
 
# Unordered lists are broken...fixing
perl -pi -e 's/\%\^/\*/g' $*
 
# Converting TML <verbatim> to Doku <file>
# You could use <code> but I like the highlighting
# of <file> better...your choice
perl -pi -e 's/<verbatim>/<file>/g' $*
perl -pi -e 's/<\/verbatim>/<\/file>/g' $*
# Removing <nop>...no more CamelCase
perl -pi -e 's/<nop>//g' $*
 
# Converting <b></b> to Doku bold
perl -pi -e 's/<b>/\*\*/g' $*
perl -pi -e 's/<\/b>/\*\*/g' $*
 
# Converting <i></i> to Doku italic
perl -pi -e 's/<i>/\/\//g' $*
perl -pi -e 's/<\/i>/\/\//g' $*
 
# Converting <u></u> to Doku underline
perl -pi -e 's/<u>/__/g' $*
perl -pi -e 's/<\/u>/__/g' $*
 
# Converting table headings
perl -pi -e 's/\| \*\*/\^  /g' $*
perl -pi -e 's/\*\*  \^/  \^/g' $*
perl -pi -e 's/\*\*  \|/  \^/g' $*
 
#Remove !!
perl -pi -e 's/\!\!//g' $*
 
#Remove %TOC%
perl -pi -e 's/\%TOC\%//g' $*
 
#Fix named links
perl -pi -e 's/\]\[/\|/g' $*
 
#Remove unneeded META info and HTML 4.0 Traditional br's
perl -pi -e 's/\%META.*//g' $*
perl -pi -e 's/\<br\>/\n/g' $*
 
#Name pre-formatted text as code
perl -pi -e 's/\<pre\>/\<code\>/g' $*
perl -pi -e 's/\<\/pre\>/\<\/code\>/g' $*
perl -pi -e 's/\<PRE\>/\<code\>/g' $*
perl -pi -e 's/\<\/PRE\>/\<\/code\>/g' $*
 
#Fix Twiki's abuse of spaces in links
#Yes, I know this is ugly, but it works, feel free to clean it up
perl -pi -e 's/\[\[(\w+)\s(\w+)/\[\[\1\2/g' $*
perl -pi -e 's/\[\[(\w+)\s(\w+)\s(\w+)/\[\[\1\2\3/g' $*
perl -pi -e 's/\[\[(\w+)\s(\w+)\s(\w+)\s(\w+)/\[\[\1\2\3\4/g' $*
perl -pi -e 's/\[\[(\w+)\s(\w+)\s(\w+)\s(\w+)\s(\w+)/\[\[\1\2\3\4\5/g' $*
perl -pi -e 's/\[\[(\w+)\s(\w+)\s(\w+)\s(\w+)\s(\w+)\s(\w+)/\[\[\1\2\3\4\5\6/g' $*
perl -pi -e 's/\[\[(\w+)\s(\w+)\s(\w+)\s(\w+)\s(\w+)\s(\w+)\s(\w+)/\[\[\1\2\3\4\5\6\7/g' $*
 
exit 0

• This script was written to convert TML from TWiki circa 2001. I don't think the syntax has changed that much, but if it does not work exactly as you would like take a look at the source and correct where needed.
• It just so happens that TML for tables and Doku syntax for tables is almost identical, so a quick modification to the header is all that was needed.
• If you don't need to convert a certain feature, simply comment out the code using a #. Look at the code for examples.
• This converted my 1000+ page TWiki site in about 30-45 seconds on a standard Dell P4. Once you start the script you won't get any feedback from your Terminal. You could open another window and check top - you should see the perl process running at about 40-60%.
• If you have questions my e-mail is in the source code.
• Good Luck!

— Ben Chapman

• This code has been updated (Nov 17, 2009) to support (at least in part) Twiki syntax as of 1.4.0. Although we had some legacy markup. This also now handles ordered lists and other annoyances we found in the use of the tool, such as Twiki allowing spaces in page links and dokuwiki not.
• This will now import entire namespaces, instead of just converting existing files.

— Thomas Vachon

• If you are running this script on Windows, you may encounter the “Can't do inplace edit without backup” error. You can fix this by replacing “perl -pi” by “perl -p -i'.bak'” and “perl -ni” by “perl -n -i'.bak'”. This will create backup files with “.bak” extension that you can remove afterwards,