[ILUG] Sed question

[Padraig Brady]

Padraig Brady wrote:
> Padraig Brady wrote:
> 
>> Rory Winston wrote:
>>
>>> Hi,
>>>
>>> I'm trying to use sed to do the following: search through a .jsp file 
>>> for
>>> any <img> references, and then generate a bare list of the image 
>>> filenames.
>>> So a .jsp page with 3 images inline would generate an output of:
>>>
>>> a.gif
>>> b.gif
>>> c.gif
>>>
>>> I'm trying to do it like the following (for this example, I'm 
>>> ignoring any
>>> complications due to case and/or whitespace):
>>>
>>> sed -n "/img src=\"/,/\">/p" foo.jsp
>>>
>>> But this doesnt just print out image filenames - it prints out entire 
>>> lines.
>>> Has anyone done anything like this already? If anyone has any grep-based
>>> solutions that would be great too. Correct me if I'm wrong, but is 
>>> sed (and
>>> Perl) able to handle certain types of multi-line matching that grep 
>>> cannot?
>>>
>>> Cheers!
>>> Rory
>>>
>>
>> How about:
>> sed -n 's/<.*img.*src="\([^"]*\)".*/\1/gp' foo.jsp
>> Padraig.
>>
>>
> 
> The script above doesn't deal correctly with multiple images
> on the same line, the following is better:
> 
> cat foo.jsp |
> sed -e 's/<[^>]*img/¬<img/g' |
> tr "¬" "\n" |
> sed -n 's/<.*img.*src="\([^"]*\)".*/\1/gp'
> 
Stephen's suggestion of not printing duplicate images is
obviously correct, so for completeness, and removing
the useless use of cat:

sed -e 's/<[^>]*img/¬<img/g' foo.jsp | #put each <img>...
tr "¬" "\n" |                          #on a new line.
sed -n 's/<.*img.*src="\([^"]*\)".*/\1/gp' |
sort -u

Padraig.