[SOLVED] command to find/replace in a text file?

sneakyimp · 02-11-2011, 04:04 PM

I have an SQL dump, file.sql that has many references to a particular domain, d1.com. I would like to run a command that can replace every occurrence of d1.com with d2.com. I've tried looking into sed before but the man pages are quite daunting. Any help would be much appreciated.

stress_junkie · 02-11-2011, 04:09 PM

This may help.
http://tldp.org/LDP/abs/html/string-manipulation.html

Make a script to read each line, look for your target phrase, and if found then apply the change.

There are numerous examples in previous LQ threads. Use the LQ web based search to find them.

someshpr · 02-11-2011, 05:03 PM

Quote:

Originally Posted by sneakyimp

I have an SQL dump, file.sql that has many references to a particular domain, d1.com. I would like to run a command that can replace every occurrence of d1.com with d2.com. I've tried looking into sed before but the man pages are quite daunting. Any help would be much appreciated.

I haven't used sed much, but I believe something like this might work:

Code:

sed -i "s/d1.com/d2.com/g" file.sql

I am not sure, but you might have to use the escape character to get the period (.) properly parsed by the command.

As stress_junkie suggested there are numerous of sed related example in LQ. Also when a problem is solved, please mark it as solved.

HTH,

grail · 02-11-2011, 11:08 PM

As an add on to the last post, if you are not sure about the results you can append something like '.bak' after the '-i' option to create a backup.
Check the man page for this option.

micxz · 02-11-2011, 11:35 PM

Could also use perl:
perl -p -i.bak -e 's/oldstring/newstring/g' file

sneakyimp · 02-12-2011, 01:17 PM

Thank you for the good info all around.

stress_junkie, the bash stuff looks really helpful but I'm certainly not much of a shell script writer. Reading the file seems a bit confusing. There is a read example on that page but it's got some confusing comparison operators.

I'm kind of psyched to get more familiar with sed because it seems like a powerful tool. I have considerable familiarity with regular expressions in a PHP context (using preg_match) but don't really recognize the syntax of 's/oldstring/newstring/g'. I need to make this case-insensitive. Can someone explain the format of the expression? Is there somewhere to find documentation on this type of expression?

sneakyimp · 02-12-2011, 01:24 PM

I found this and it describes how to use the "y" command to change the case of strings but does not describe how to get a case-insensitive search and replace.

sneakyimp · 02-12-2011, 01:33 PM

FYI, trying the "I" flag and I'm getting error:

Code:

prompt$ sed -i .bak "s/d1.com/d2.com/gI" dump.sql
sed: 1: "s/d1.com/d2 ...": bad flag in substitute command: 'I'
prompt$

micxz · 02-12-2011, 02:59 PM

Here is somethin:
http://www.zytrax.com/tech/web/regex.htm

colucix · 02-12-2011, 03:40 PM

Quote:

Originally Posted by sneakyimp

FYI, trying the "I" flag and I'm getting error:

Code:

prompt$ sed -i .bak "s/d1.com/d2.com/gI" dump.sql
sed: 1: "s/d1.com/d2 ...": bad flag in substitute command: 'I'
prompt$

The 'I' flag is available in GNU sed only. Which version of sed are you running? And which OS? In alternative you can try the suggested perl code (see post #5 by micxz) with a slight modification:

Code:

perl -i.bak -pe 's/d1.com/d2.com/gi' dump.sql

micxz · 02-13-2011, 04:07 PM

here is a nice one liner or example page:
http://sed.sourceforge.net/sed1line.txt
look almost to the end of the file for what colucix means about the GNU versions.

sneakyimp · 02-14-2011, 01:58 PM

Quote:

Originally Posted by micxz

Here is somethin:
http://www.zytrax.com/tech/web/regex.htm

Thanks for that. I'm quite well acquainted with regular expressions generally speaking, but have a lot more experience working with them in PHP. What's puzzling about this SED regex syntax is that it appears to start with a command -- in this case "s" which stands for "substitute" -- and then has three slashes rather than being bracketed by just two slashes. I'm thinking I might need assistance with SED and or PERL regex command syntax rather than basic regex.

colucix: I don't *know* what version of sed I'm using. I've checked the man pages and there doesn't appear to be any way to determine what version I have. I get the feeling from the bits of reading I've done that there are many variations of sed. I am dealing with a variety of machines for this task: OSX 10.5.8, CentOS 5.5, Ubuntu, etc. It would be nice to come up with some kind of scheme that works on all of them with no surprises.

micxz: I read the stuff at the end but I still don't know how to get my sed version or how I might translate that version into a plan of action here.

I also tried perl with the I flag and it complained about a bad modifier.

sneakyimp · 02-14-2011, 02:15 PM

hm. I'm also having an issue where "somestring com" (note the space) is also being modified when I do a replace on "somestring.com" because the . is being treated as a wildcard. Grep doesn't seem to like it when I add a backslash to escape the period.

I am trying the following on CentOS 5.5 with some good results:

Code:

perl -p -i.bak -e 's/d1\.com/d2.com/gi' file.sql

This also appears to work on OSX -- I tried an uppercase "I" flag before and that caused the complaint.

THANKS for your help, gentlefolk! I'm cooking right along and the extra link resources should help me get better at this. Have a great week.

micxz · 02-14-2011, 09:44 PM

Glad it's working out. Try:

Code:

sed --version

sneakyimp · 02-14-2011, 11:22 PM

Code:

My_Mac:import sneakyimp$ sed --version
sed: illegal option -- -
usage: sed script [-Ealn] [-i extension] [file ...]
       sed [-Ealn] [-i extension] [-e script] ... [-f script_file] ... [file ...]