sed help

[AstroLlama]

I've been playing around with sed, can somebody help figure this out?

I want to remove all text in an xml file between two <caption> tags. I've used the following command:

sed 's/<caption>[A-Za-z]*<\/caption>/<caption><\/caption>/g' file.xml >file2.xml

This works, but only if there is one word between the tags. I've tried modifying the command to remove multiple words in various ways but to no success. any sed gurus out there have advice?

[imbjr]

I've been playing around with sed, can somebody help figure this out?

I want to remove all text in an xml file between two <caption> tags. I've used the following command:

sed 's/<caption>[A-Za-z]*<\/caption>/<caption><\/caption>/g' file.xml >file2.xml

This works, but only if there is one word between the tags. I've tried modifying the command to remove multiple words in various ways but to no success. any sed gurus out there have advice?

I think you might need a space in that A-Za-z part of the clause:

sed 's/<caption>[A-Z a-z]*<\/caption>/<caption><\/caption>/g' file.xml >file2.xml

[DaithiF]

Hi,
another possibility would be:

sed 's/<caption>[^<]*<\/caption>/<caption><\/caption>/g'

which will handle cases where there are punctuation or other non-alphabetic characters in the caption tags

[mobilediesel]

I've been playing around with sed, can somebody help figure this out?

I want to remove all text in an xml file between two <caption> tags. I've used the following command:

sed 's/<caption>[A-Za-z]*<\/caption>/<caption><\/caption>/g' file.xml >file2.xml

This works, but only if there is one word between the tags. I've tried modifying the command to remove multiple words in various ways but to no success. any sed gurus out there have advice?

This works, too, and shortens the command a bit:

sed 's/\(<caption>\)[^<]*\(<\/caption>\)/\1\2/g' file.xml >file2.xml

Using [^<]* will match any character up to the </caption> tag. Putting <caption> and <\/caption> inside ( and ) means you only need \1\2 in the replacement end.

[mobilediesel]

Hi,
another possibility would be:

sed 's/<caption>[^<]*<\/caption>/<caption><\/caption>/g'

which will handle cases where there are punctuation or other non-alphabetic characters in the caption tags

Ah! I had the preview open while I went off and did something else and come back and someone posts what I had!

I should start refreshing the post again before hitting submit.

AND I should start editing my post so as not to post twice in a row...

[howlingmadhowie]

This works, too, and shortens the command a bit:

sed 's/\(<caption>\)[^<]*\(<\/caption>\)/\1\2/g' file.xml >file2.xml

Using [^<]* will match any character up to the </caption> tag. Putting <caption> and <\/caption> inside ( and ) means you only need \1\2 in the replacement end.

one thing worth noting is that the command separator for sed doesn't have to be a '/', so you could instead write it like this:

sed 's:\(<caption\)[^<]*\(</caption>\):\1\2:g' file.xml > file2.xml

which is a bit more legible

[mobilediesel]

one thing worth noting is that the command separator for sed doesn't have to be a '/', so you could instead write it like this:

sed 's:\(<caption\)[^<]*\(</caption>\):\1\2:g' file.xml > file2.xml

which is a bit more legible

I keep forgetting that for some reason. There are plenty of instances where using different separators make sed WAY easier to work with.

[ghostdog74]

I've been playing around with sed, can somebody help figure this out?

I want to remove all text in an xml file between two <caption> tags. I've used the following command:

sed 's/<caption>[A-Za-z]*<\/caption>/<caption><\/caption>/g' file.xml >file2.xml

This works, but only if there is one word between the tags. I've tried modifying the command to remove multiple words in various ways but to no success. any sed gurus out there have advice?

All the sed solutions provided thus far does not take care of multiline tags.
so why don't you ditch sed and use awk instead.

# more file
<tag1>
  i want text here
  <tag2>
    remove text here
  </tag2>
  you want text here too
</tag1>

$ awk -vRS='</tag2>' '{ gsub(/<tag2>.*/,"")}1' file
<tag1>
  i want text here


  you want text here too
</tag1>

Best of all, use a proper XML parser if you can.

[AstroLlama]

thanks for the quick replies, studying these examples has proven to be a great lesson in how to use sed.

DaithiF

sed 's/<caption>[^<]*<\/caption>/<caption><\/caption>/g'

which will handle cases where there are punctuation or other non-alphabetic characters in the caption tags

This works, too, and shortens the command a bit:

sed 's/\(<caption>\)[^<]*\(<\/caption>\)/\1\2/g' file.xml >file2.xml

Using [^<]* will match any character up to the </caption> tag. Putting <caption> and <\/caption> inside ( and ) means you only need \1\2 in the replacement end.

mobilediesel, though your command suggestion is slightly longer, the use of the parenthesis and numbers has other interesting applications that are sure to be helpful when doing more complicated sed work in the future.

thanks again!