ansaurus

Question

how to print a section of file between two regular expressions only if a line within the section contains a certain string within it

Answer 1

+2 A:

awk -vRS="</event>" '/<event>/ && /1279939300.862594_PFM_1_1912320699|1281686522.353435_PFM_1_988171542/{print}' file

ghostdog74 2010-08-27 13:32:36

This worked brilliant thanks. I am now checking times against the sed one above to see which one is better timewise (as my files are huge). Here are the times for this: real 3m4.890suser 3m2.273ssys 0m2.568s

amadain 2010-08-27 14:53:22

Answer 2

A:

You should be able to embed the unique ids directly into the regular expression, using the | character to allow either uniqueid. I did a quick test and the following regular expression seems to find the correct entries:

 <event.*?uniqueid=("1279939300\.862594_PFM_1_1912320699"|"1281686522\.353435_PFM_1_988171542").*?</event>

Ryan Brunner 2010-08-27 13:33:28

the uniqueId is not necessarily on the same line as the <event> tag

amadain 2010-08-27 13:50:23

I have to admit that I'm not familiar with sed, but is there not a toggle to enable multi-line regexes?

Ryan Brunner 2010-08-27 14:44:03

yea. I'm trying Dennis's multi-line sed from above. The awk one worked in around 3 mins

amadain 2010-08-27 14:59:02

Answer 3

+1 A:

Give this a try:

sed -n ':a; /<event>/,/<\/event>/ N; /<event>/,/<\/event>/!b; /<\/event>/ {/uniqueId="1279939300.862594_PFM_1_1912320699"\|uniqueId="1281686522.353435_PFM_1_988171542"/p;d}; ba'

Dennis Williamson 2010-08-27 14:30:50

after 15mins of waiting this still hasn't returned an answer. Its probably more suited to smaller files. Thanks though for the multi line sed. This is at a level that I have not used yet so I will study it and learn from it

amadain 2010-08-27 15:02:38

@amadain: I had a couple of errors. See the edited version.

Dennis Williamson 2010-08-27 16:54:21

dennis a means append a line and is used after a pattern match (from what I could find). Could you explain what it does here as a sole command :a ?

amadain 2010-08-29 13:47:26

@amadain: `:` is the label command, `a` is the name of the label, `b` is the branch command. So when `sed` reaches `ba` it branches to label "a" which is `:a` creating a loop. The `b` without a label branches to the end (past the last command) thus exiting the loop. The `d` (delete) command also branches to the end exiting the loop.

Dennis Williamson 2010-08-29 14:44:42

that's wonderful sedding. Thank you

amadain 2010-08-29 14:51:28

for one of the first times the sed answer is considerably more complicated than the awk answer.

amadain 2010-08-29 14:53:42

ansaurus

tags:

views:

answers:

how to print a section of file between two regular expressions only if a line within the section contains a certain string within it

related questions