The following is untested, but in principle it should work (I will test it if I have more time).
You could convert the djvu file for example to jpg's like this:
#!/bin/sh
# djvu -> jpgs converter
i=1
# number of pages (392)
while [ $i -ne 392 ]
do
ddjvu -page=$i -format=pnm 1.djvu $i.pnm
pnmtojpeg $i.pnm > $i.jpg
rm -f $i.pnm
echo "page $i done"
i=`expr $i + 1`
done
(from http://caree.livejournal.com/74639.html)
Then you could use scantailor to split the pages and produce a new output (consisting of tif files).
Apply in a third step djvubind to that folder and you get your desired djvu file.
Here is my complete working script. It needs djvulibre-bin library for ddjvu (decode to tiff format), cjb2 (because i don't know how to directly convert tiff to djvu), djvm (for insertions/deletions of pages in djvu file) tools and imagemagick library for convert (for splitting page and converting to mystical pbm file format) tool. All libraries available through apt-get.
Scheme:
file.djvu --./ddjvu--> bifold tiff --./convert--> single page tiff --./convert--> pbm --./djvm--> out.djvu
Example: djvusplit 3 10 file.djvu which means split pages from 3 to 10.
Complete code:
#!/bin/bash
echo Usage: djvusplit PAGEFROM PAGETO FILE.DJVU
#make temp folder
mkdir ./tmp
cp "$3" ./tmp/
cd tmp
for i in $(seq -w $2 -1 $1) # Descending
do
ddjvu -format=tiff -page=$i "$3" t$i.tiff
convert -crop 2x1@ t$i.tiff t$i-%d.tiff
convert t$i-0.tiff t$i-0.pbm
convert t$i-1.tiff t$i-1.pbm
cjb2 t$i-0.pbm t$i-0.djvu
cjb2 t$i-1.pbm t$i-1.djvu
djvm -i t$i-0.djvu t$i-1.djvu 2
djvm -d $3 $i
djvm -i $3 t$i-0.djvu $i
rm ./t*
done
# total clean
mv $3 ../out.djvu
cd ..
rm -r ./tmp
Unfortunately...
I use this script with some other to maintaining my electronic library. They located here on GitHub.
There aren't so many tools that can operate directly on DjVu files, compared with other more common formats such as PDF or JPEG. With image manipulation programs, there's the added hurdle that most of these operate on a single image at a time, but the DjVu file contains multiple pages.
One possibility is to go via pdf. With ddjvu from DjVuLibre, a PDF un2up filter, and pdf2djvu:
ddjvu -format=pdf 2up.djvu 2up.pdf
un2up <2up.pdf | pdf2djvu /dev/stdin >1up.djvu
You might be able to cobble up an un2up for djvu inspired by my pdf version using python-djvulibre. I haven't checked how hard that API is to get into.
You probably want this method.
In bash, you would do something like this:
input_file=inputfile.djvu
output_dir=outdir
mkdir "$output_dir"
page_count=$( djvused "$input_file" -e n )
for page_num in $( seq 1 "$page_count" ); do
printf 'select %d; save-page outdir/page-%04d.djvu\n' "$page_num" "$page_num"
done
| djvused "$input_file"
That would generate a djvused script like this:
select 1; save-page-with outdir/page-0001.djvu
select 2; save-page-with outdir/page-0002.djvu
select 3; save-page-with outdir/page-0003.djvu
select 4; save-page-with outdir/page-0004.djvu
select 5; save-page-with outdir/page-0005.djvu
select 6; save-page-with outdir/page-0006.djvu
which would produce a directory a directory of files, like this:
$ ls -l outdir/
total 400
-rw-r--r-- 1 dwon dwon 59145 Apr 26 17:30 page-0001.djvu
-rw-r--r-- 1 dwon dwon 69848 Apr 26 17:30 page-0002.djvu
-rw-r--r-- 1 dwon dwon 60037 Apr 26 17:30 page-0003.djvu
-rw-r--r-- 1 dwon dwon 68312 Apr 26 17:30 page-0004.djvu
-rw-r--r-- 1 dwon dwon 71849 Apr 26 17:30 page-0005.djvu
-rw-r--r-- 1 dwon dwon 68368 Apr 26 17:30 page-0006.djvu
The DjVu format allows documents to be stored as multiple files, with references between them. This is what is known as "indirect" format (as opposed to the usual "bundled" format, which stores everything together). It's possible to convert a document between bundled and indirect format.
The benefit of this method is that any complex document structure (e.g. shared dictionaries for JB2 compression, page repetition, metadata covering the whole file, etc) will be preserved. This makes it handy for adding information & metadata (e.g. OCR text) since it preserves the overall content structure.
The drawbacks are that you don't get any control over filenames, and some software might not be able to handle it. (e.g. Okular 21.12.3 chokes on this sample file when converted to indirect format, even though DjView4 works fine.)
The conversion can be done using the djvmcvt command, or by using the save-indirect subcommand of djvused.
Using djvmcvt(1):
mkdir output_directory
djvmcvt -i doc_in.djvu output_directory index_filename.djvu
Using djvused(1), subcommand save-indirect:
mkdir output_directory
djvused doc_in.djvu -e "save-indirect output_directory/index_filename.djvu"
These commands produce output like this:
$ mkdir outdir
$ djvmcvt -i sample2.djvu outdir index.djvu
$ ls -l outdir/
total 56
-rw-r--r-- 1 dwon dwon 22173 Apr 26 18:32 dict0084.iff
-rw-r--r-- 1 dwon dwon 71 Apr 26 18:32 index.djvu
-rw-r--r-- 1 dwon dwon 22653 Apr 26 18:32 p0001.djvu
-rw-r--r-- 1 dwon dwon 3688 Apr 26 18:32 p0002.djvu
$ djvudump outdir/index.djvu
FORM:DJVM [59]
DIRM [47] Document directory (indirect, 3 files 2 pages)
dict0084.iff -> dict0084.iff
p0001.djvu -> p0001.djvu
p0002.djvu -> p0002.djvu
$ djvudump outdir/dict0084.iff
FORM:DJVI [22161]
Djbz [22149] JB2 shared dictionary
$ djvudump outdir/p0001.djvu
FORM:DJVU [22641]
INFO [10] DjVu 3300x2550, v21, 300 dpi, gamma=2.2
INCL [12] Indirection chunk --> {dict0084.iff}
Sjbz [3054] JB2 bilevel data
FGbz [22] JB2 colors data, v0, 2 colors
BG44 [3774] IW4 data #1, 72 slices, v1.2 (color), 1100x850
BG44 [3400] IW4 data #2, 11 slices
BG44 [5270] IW4 data #3, 10 slices
BG44 [6635] IW4 data #4, 6 slices
TXTz [387] Hidden text (text, etc.)
$ djvudump outdir/p0002.djvu
FORM:DJVU [3676]
INFO [10] DjVu 3300x2550, v21, 300 dpi, gamma=2.2
INCL [12] Indirection chunk --> {dict0084.iff}
Sjbz [2792] JB2 bilevel data
FGbz [404] JB2 colors data, v0, 43 colors
BG44 [87] IW4 data #1, 97 slices, v1.2 (b&w), 275x213
TXTz [318] Hidden text (text, etc.)
Taking bot2417's script as base, here my own
#!/bin/bash
echo "################################################"
echo Usage: djvusplit2 LASTPAGE FILE.DJVU
echo "################################################"
if [ ! -f $2 ];
then
echo "file $2 not exists!\n"
exit
fi
start=1
mkdir ./tmp
for i in $(seq $start +1 $1)
do
j=$(($i*2-1))
k=$(($i*2))
# extract pages to tiff format
ddjvu -format=pbm -page=$i $2 ./tmp/$i.tiff
# split pages
convert -crop 2x1@ ./tmp/$i.tiff ./tmp/$i-%d.tiff
#delete extracted tiff
#rm ./tmp/$i.tiff
# convert tiff to djvu pages
cjb2 ./tmp/$i-0.tiff ./tmp/$j.djvu
cjb2 ./tmp/$i-1.tiff ./tmp/$k.djvu
#delete splitted tiffs
#rm ./tmp/$i-0.tiff
#rm ./tmp/$i-1.tiff
# create djvu file and add pages
if [ $i -eq 1 ];
then
djvm -c "(new) $2" ./tmp/$i.djvu
echo "create new $2 OK"
djvm -i "(new) $2" ./tmp/$(($i+1)).djvu
echo "insert page $(($i+1)) OK"
else
djvm -i "(new) $2" ./tmp/$j.djvu
echo "insert page $j OK"
djvm -i "(new) $2" ./tmp/$k.djvu
echo "insert page $k OK"
fi
#delete djvu pages
#rm ./tmp/$j.djvu
#rm ./tmp/$k.djvu
done
echo "\nfile (new) $2 created!!!"
echo "\n"
# cleanup temp dir
rm -r ./tmp
http://en.wikisource.org/wiki/Help:DjVu_files#Splitting_DjVu_files
Hope you find your answer here.
