UNIX for Dummies Questions & Answers

I've been playing around with sed for a few days now and find that regular expressions never work inside sed s/// on Mac OS 10.2.5. Has anyone else bumped up against this problem?

For example, the "sed" segment of the following always fails no matter how items are escaped:


cd /path/to/directory
egrep -a '<([\w]+)[^>]*>[^<]*</\1>' *.JPG | sed 's/<\([^>]+\)>/tag_attributes: \1/g'


Short of using Perl and laboriously opening each file walking through every line in every file and using Perl's s/// operator, is there another tool for doing this sort of stuff purely in unix?

Thanks

Ken

Could you explain further what you mean to accomplish with that line of code? Too confusing for me to decipher .. plus there's no -a option for the egrep command. And although egrep is used to search/display text from a file, it looks like you're using it on image files..

The actual line I'm using was more complex so I "simplified" it to the line in my original post:)

Actually what I'm doing is scanning jpeg files that have info entered in the Photoshop "file info" dialog box and extracting this info for inclusion in a database. The info I'm interested in is xml formatted with some Adobe namespace info.

This line works great for extracting the info:

egrep -a '<photoshop:([^>]+)>[^<]*</photoshop:\1>' "botanical-garden12-full.jpg"

yeilds lines like:

<photoshop:CaptionWriter>this is where the caption writer goes</photoshop:CaptionWriter>
<photoshop:Headline>this is where the headline goes</photoshop:Headline>
<photoshop:Instructions>this is where the instructions go</photoshop:Instructions>


If you then pipe it to sed to try and strip out the xml like so:

<egrep stuff from above...> | sed 's/<photoshop:\([^>]+\)>\([^<]*\)<\/photoshop:\1>/\1: \2>/g'

returns results as if the sed routine wasn't even there...

<photoshop:CaptionWriter>this is where the caption writer goes</photoshop:CaptionWriter>
<photoshop:Headline>this is where the headline goes</photoshop:Headline>
<photoshop:Instructions>this is where the instructions go</photoshop:Instructions>

I've tried various quotation, delimiter and escape styles on the above ("" as opposed to '', () as opposed to \(\) and s@@@ as opposed to s/// none of it produced the desired results.

I changed the search portion of the sed routine to a simple character class for testing and even that failed. The only thing that seems to work is a string literal.

Do you see any glaring errors in the above?

Thanks

Ken

I'm still trying to come up with a better solution, but in the meantime, awk would work like this:

<output from egrep...> | awk -F">" '{print $2}' | awk -F"<" '{print $1}'

Hmmm... My turn to be puzzled. No idea what your example is doing, but I'll go read up on awk...

Ken

What shell are you using?

tcsh for most stuff but the main script is running in sh (which I think is bash on OS X) not sure about that though.