Run a command or a script recursively

Luvz2Fly · 12-04-2008, 10:15 AM

Good day everyone...
I'm trying to figure this out on my own but I surrender! I have a command line program that converts HTML files to PDF but I need it to start in the main directory and traverse through each directory hitting every HTML file (basically do a batch HTML to PDF conversion)

The basic command I need to run is:

htmldoc -f output.pdf --webpage *.html

How do I execute this command starting at top directory and going to each subdirectory? If it helps there is only ONE level of subdirectory:

>START
|__Dir1
|file1.html
|file2.html
|__Dir2
|file1.html
|__Dir3
|file1.html
|file2.html
|file3.html

etc...

Thanks!

TB0ne · 12-04-2008, 10:47 AM

Quote:

Originally Posted by Luvz2Fly

Good day everyone...
I'm trying to figure this out on my own but I surrender! I have a command line program that converts HTML files to PDF but I need it to start in the main directory and traverse through each directory hitting every HTML file (basically do a batch HTML to PDF conversion)

The basic command I need to run is:

htmldoc -f output.pdf --webpage *.html

How do I execute this command starting at top directory and going to each subdirectory? If it helps there is only ONE level of subdirectory:

You don't say what you've already tried, or if you need a script file to do this, but if all the files are in one directory/sub-directory location, you could do:

htmldoc -f output.pdf --webpage `find ./ -name *.html`

Note those are backticks, not single quotes, around the find command.

rayfordj · 12-04-2008, 10:49 AM

if it'll take args from stdin you could try something like this

Code:

find . -type f -iname \*.html -print | htmldoc -f output.pdf --webpage

i92guboj · 12-04-2008, 11:02 AM

There are two basic approaches that you could use.

One is the real recursive one. The script would need to process the files into a dir one by one. The basic scheme would be:

Code:

if [ "$1" == "" ]; then
  dir=$PWD
else
  dir="$1"
fi

for file in "$dir"
do
  if [ -d file ]; then
    $0 "$dir" # launches the script again on the next dir
  else if [ ! "${file/.html/}" == "${file}" ]; then
    # if the filename without ".html" is not the same than the filename
    # then the file is an html file, there are many other ways to check this
    # Here you must insert the commands to convert $file to pdf
  else
    : # do nothing since it's not a dir, and it's not html either.
  fi
done

That's just the basic idea and might not be perfect but it should put you on the right track if you really want to do this recursively.

The other way (usually simpler and probably much less intensive in ram usage) is to just use find, with -exec. Or better, use it to feed a while loop which brings you more possibilities. For example:

Code:

cd /whatever/dir/; find . -name \*.html | while read file; do html2pdf "$file" "${file/.html/.pdf}"; done

I have no idea if html2pdf is a real tool, and if it is I have no idea of the syntax. The above snippets are meant only as a rough example on how to implement this on a bourne compatible shell.

zer0x333 · 12-04-2008, 11:11 AM

This will do the files one by one, keeping the PDF in the same location as the source HTML:

Code:

#!/bin/bash
IFS="

"
for htmlfile in $(find . -type f -name '*.html')
do
        pdffile=$(echo $htmlfile | sed s/\.html$/\.pdf/g)
        htmldoc -f $pdffile --webpage $htmlfile
        echo Converted $htmlfile to $pdffile
done