Convert encoding through subdirectories?

dl7und · 04-11-2006, 12:11 PM

I am trying to convert the encoding of a bunch of php files scattered over several subdirectories. (a Postnuke language pack) For the recoding part I planned to use htmlrecode, which I had been using before on files in one directory.

Since I had problems getting along with the piping (htmlrecode reads and writes to standard in/output), the author was so kind to help me with a small script for all files in one directory:

Code:

#!/bin/bash
mkdir o
for s in *.html; do
	htmlrecode -IBIG5 -OUTF-8 < "$s" > o/"$s"
done

This will put the recoded files into a new subdirectory. This time however I had to modify files in many subdirectories, and I must modify the original file, not just copy a recoded version somewhere. So I tried this, inspired by another problem's solution:

Code:

find /directory_name -name *.php | htmlrecode -OUTF8

All php files were listed, but none was recoded. I admit I am pretty weak at pipes (never smoked), so some people will probably laugh seeing my obvious mistake. My personal guess is that htmlrecode does not get anything piped in (and out), but I cannot find a solution for this problem and would be grateful for a helpful answer...

jlinkels · 04-12-2006, 09:16 PM

There is a similar thread about recursing into subdirectories here.

If you want to do something with find, do it like this:

find /directory_name -name *.php -exec yourcommand {} \;

The curly braces are substituted with the file name found.

jlinkels

dl7und · 04-16-2006, 01:06 AM

Thanks for the response! Unfortunately, the command got stuck again. But the curly brackets remined me of something - and now I finally did it.

The problem was not so much the cruise through subdirectories as more the correct piping. Htmlrecode (and probably iconv too) does not accept filenames as regular arguments, they have to be piped. Plus, input and output file may not be the same. So in the end, I solved this in two steps:

Code:

for f in `find /start_directory -name *.php`; do htmlrecode -OUTF8 < "$f" > "${f/php/phpu}"; done
for f in `find /start_directory -name *.phpu`; do mv "$f" "${f/phpu/php}"; done

The first step would convert 8859-1 PHP files to utf-8 and save the recoded ones with the extension "phpu". The second step then searches for those files and renames them back to "php", overwriting the previous ones.

I guess this can even be done in one line/step, but for the moment I'm happy it worked.