Hey guys,

Searched here for this and found lots of answers, but none seems to work

The case: I have an ini file that lists a lot of settings like this:

and so on. In a bash script im trying to replace "SettingC=" if it looks exactly like that. So if anyone went and added "SettingC=true" it cant match.

So I have tried with sed and regex word boundaries, but it does not work due to some unknown reason.

This is what i try to run from inside a script:
However that does not work.

I have also tried with \<\>, which didn’t work either. Anyone that can enlighten me what im doing wrong here?

Cheers!

R

It's simple enough if you don't mind needlessly updating a file with the same setting it already had.To avoid that takes a bit more work.What that does is first test whether the line is already "SettingC=false" (explicitly anchored to the end of the line so that "false123" will not match) and branch to the end of the script if true, otherwise do the replacement.

[EDIT]And, a bit of testing reveals that sed is not as smart as I thought it was and will replace the file even if no changes were made, so there's no point in adding that test.[/EDIT]

In both cases, the regex ends with ".*" so that anything following the equals sign will be replaced.

There are lots of things that could be done to make this more robust with regard to whitespace, lines like "MySettingC=whatever", etc. While sed certainly could do all that (It is, after all, a Turing-complete language.), eventually you reach a point where it is just not a suitable language for the task at hand.

What about the following? It'll only match SettingC= followed by end of line.

1 members found this post helpful.

Thanks for the quick reply rknichols!

With the top example
sed -i "s/SettingC=.*/SettingC=false/" /path/to/ini_file
It replaces even if the string is "SettingC=" or "SettingC=true" and that seems to be ok and working but not what i was after.
The second example you had, did not actually replace at all. Please note that these strings are on single row and they start the row as my example above, so no need to take any of that in to consideration.

What im after is to ensure that the string is exactly "SettingC=" and if it, replace with "SettingC=false". I do fully agree with you on right tool for right job and this to me seemed like work for sed.

Thanks!

br,

R

Spot on and did the trick! Thanks for your help guys

Sorry, misunderstood what you wanted. This is easier -- just include the end-of-line mark in your regex.The "g" flag is pointless since there cannot be more than one match in the line. Note the difference in quoting. You need to get the "$" character past the shell. sed needs to see just the "$" and not "\$".

Again, you might want to ignore whitespace, especially at the end of the line:

1 members found this post helpful.

@rknichols: Thanks for explaining and the additional examples.

Thanks for very informative answers both!

Cheers,

R