I have a script i use to convert and move files regularly uploaded to an image gallery.
Now those that send over the files (a mac shop) have decided that they will be sending them in compressed .sea files... this has a great compression ratio and will save mucho bandwidth and time - unfortunately my script assumes that the files in question are in the root of the directory and also that they are all image files.

I need to extend the script so that it unstuffs the .sea file and performs the conversion and subsequent functions to all files in the root and it's recursive directories. The .sea file will be in the /home/image/convert directory but will no doubt uncompress into it's own. I.e. a file called test.sea will extract to /home/image/convert/test

Also - i need to extend the search and replace for any unconventional characters in the file names. Right now it replaces spaces with a _ , i need to also replace any periods with a _

Any help or pointers is greatly appreciated

Cheers,

S.



Here is the original script -



#!/bin/bash
#####################################################################
IMAGEDIR='/home/image/convert/'
DATE=`date +%m-%d`
JPG='/usr/local/apache/htdocs/imagegallery/admin/mainimages/tmp/'
TIF='/usr/local/apache/htdocs/imagegallery/admin/mainimages/'
ERRORS='/tmp/errors'
LIST='/tmp/list'
RESULTS='/tmp/results'
CONVERTED='/tmp/converted'
IMAGES="*"
#####################################################################
# Convert files and log actions
#####################################################################
#
# Create three files - error, list and results.
#
# list = file listing of all files converted that day (size and name)
# errors = errors generated by conversion
# converted = output of file conversion
# results = results or errors of the conversion process
#
#####################################################################
cd ${IMAGEDIR}
# Rename files with spaces - replace the spaces with a _
# NEED TO ADD replace statement for .
#
find ${IMAGEDIR} -name '* *' -type f | sort | while read FILE
do
NEWFILE=`dirname "${FILE}"`/`basename "${FILE}" | sed 's/ /_/g;'`
mv "${FILE}" "${NEWFILE}"
done
#
# Populate list of files > generated each time and cleaned out at the end of the daily script
ls -lh ${IMAGEDIR} | awk ' { print $9 " " $5 } '>>$LIST
#
# THERE HAS TO BE A BETTER WAY TO DO THIS other than a blanket do..
#
for files in ${IMAGES}
do
(
echo converting ${files} >>$CONVERTED
#
# echo output from conversion to converted log file/convert the files/move the files and set permissions
#
convert ${files} -resize 800x600 -colorspace rgb ${files}.jpg >>$CONVERTED 2>&1
#
# mv files to web space and set permissions
#
mv ${files} ${TIF}${files}_${DATE}.tif
mv ${files}.jpg ${JPG}${files}_${DATE}.jpg
)
done
cd ${TIF}
chown nobody:nobody *.tif
chmod 664 *.tif
cd ${JPG}
chown nobody:nobody *.jpg
chmod 664 *.jpg
#
# Generate Result section
#
#
cat $CONVERTED |egrep "converting|error" >>$RESULTS

find seems to be part of your solution. find <dir> -type f will list all the files in a directory recursively.

To replace periods I guess you can use sed s/'\.'/'_'/. Another way to do it - just exchange one character to another - is to use tr.

find <dir> -type f | tr '[:punct:][:space:]' '_'

will probably do approximately what you want :)

*testing*

Hm, it seems that the [:space:] set will also match newline, but that is not a big problem.

thanks for the reply - but i can't seem to get the syntax of the tr command correct...

[root@image test]# ls
file.1 file.2 file.3
[root@image test]# find /home/image/convert/test -type f | tr '[:punct:]' '_'
_home_image_convert_test_file_1
_home_image_convert_test_file_2
_home_image_convert_test_file_3
[root@image test]# ls
file.1 file.2 file.3
[root@image test]# find . -type f | tr '[:punct:]' '_'
__file_1
__file_2
__file_3
[root@image test]# ls
file.1 file.2 file.3

Oops... The underscores at the beginning come from the directory part. Maybe you should only apply the tr on the basename?

Ottherwise... Do it like this:

It'll be very ugly, but it works.

Great - that works (as you can see) ... but (dumb question alert)..

why is the change not applied?

[root@image convert]# ls
5459 54591 5459_6574_03 5459_6574_04 5459_6574_05 Test.sea stu.1.2 test
[root@image convert]# ls
5459 54591 5459_6574_03 5459_6574_04 5459_6574_05 Test.sea stu.1.2 test
[root@image convert]# find . -type f | tr '[:punct:]' '_' | sed s/'^_*'//
test_stu_2_3
5459_6574_03
54591
5459_6574_04
5459_6574_05
5459
Test_sea
stu_1_2
[root@image convert]# ls
5459 54591 5459_6574_03 5459_6574_04 5459_6574_05 Test.sea stu.1.2 test

i got it - Thanks Bebo

Yeah, you have to use mv or cp too, y'know

Ha, look at that, this silly remark is my hundredth post - wohoo!