replacing text with an http link with sed?

crypted · 12-10-2010, 10:36 AM

Been at this all morning and fail on every attempt.

I have a file called sitemapindex.xml

Its contents just are names of the other sitemaps. Google gets pissy because it isn't in a http:// format.

So, I need "<loc>sitemap-c1mediawiki-............</loc>" to be replaced with "<loc>http://domain.tld/sitemap-c1mediawiki-............</loc>".

Here's the expression I've been trying:

find /path/to/site/sitemap-index-c1mediawiki* -type f | xargs sed -i 's/[<loc>sitemap]/[<loc>http://domain.tld/]/'

It fails.

I've tried with and without brackets.

Thoughts?

crts · 12-10-2010, 11:44 AM

Try this:

Code:

sed -r 's|(<loc>)(sitemap)|\1http://domain.tld/\2|'

If you are satisfied with the result you can add the -i option.
BTW, you do not need to pipe to xargs. You can use find's -exec option instead:

Code:

find /path/to/dir/ -type f -exec sed -ri '...' '{}' +

thesnow · 12-10-2010, 11:45 AM

Code:

echo "<loc>sitemap-c1mediawiki-" | sed 's/<loc>/<loc>http:\/\/domain.tld\//'
<loc>http://domain.tld/sitemap-c1mediawiki-

crypted · 12-10-2010, 11:51 AM

Code:

sed: -e expression #1, char 48: invalid reference \2 on `s' command's RHS

It looked perfect with the -r option output. Removed the -r and put in -i, and it won't work.

crts · 12-10-2010, 12:03 PM

Quote:

Originally Posted by crypted

Code:

sed: -e expression #1, char 48: invalid reference \2 on `s' command's RHS

It looked perfect with the -r option output. Removed the -r and put in -i, and it won't work.

The '-r' option tells sed to use extended regular expressions. Do not remove it. I said add the '-i' option; not replace '-r' for '-i'.

Code:

sed -ri ...

Also have a look at my suggestion for the alternative 'find'.

Hope this helps.