From: Robert Klemme <shortcutter@...>
Date: 2011-02-03T01:52:41+09:00
Subject: Re: Editing a text file and then replacing existing file

On Wed, Feb 2, 2011 at 5:31 PM, Jeremy Bopp <jeremy@bopp.net> wrote:
> On 2/2/2011 9:53 AM, Bob Hatch wrote:
>> I have a script that I use to find text in a file and then delete the
>> line(s) the text is on. That part works. What I have been unsuccessfully
>> trying to do is take the new text and overwrite the existing file that I
>> did the search and delete on. Here is my existing code that finds the
>> lines and deletes them:
>>
>> path='c:\ruby192\my_projects\IIS_Logs\ex11012607.log'
>> search_text = %r{/memberinfo/downline/tree/can.asp}
>> search_text2 = %r{/memberinfo/downline/tree/disto.asp}
>> search_text3 = %r{/styles/default.css}
>> search_text4 = %r{/scripts/generalscript.js}
>> search_text5 = %r{/memberinfo/downline/tree/upl.asp}
>> search_text6 = %r{/scripts/wz_jsgraphics.js}
>> search_text7 = %r{/memberinfo/downline/tree/disfo2.asp}
>> search_text8 = %r{/styles/png.htc}
>> File.open(path, 'a+') do |file|
>>   file.each_line do |line|
>>       puts(line) unless line =~ search_text ||
>>   puts(line) unless line =~ search_text2 ||
>>       puts(line) unless line =~ search_text3 ||
>>       puts(line) unless line =~ search_text4 ||
>>       puts(line) unless line =~ search_text5 ||
>>       puts(line) unless line =~ search_text6 ||
>>       puts(line) unless line =~ search_text7 ||
>>       puts(line) unless line =~ search_text8
>>    end
>> file.write(line)
>> file.close
>> end
>
> First of all, let's simplify this code by using a single regexp that
> will match all lines of interest.  This uses the x option for the regexp
> construction so that the format is a little more legible:
>
> search_text = %r{
>  /memberinfo/downline/tree/can.asp    |
>  /memberinfo/downline/tree/disto.asp  |
>  /styles/default.css                  |
>  /scripts/generalscript.js            |
>  /memberinfo/downline/tree/upl.asp    |
>  /scripts/wz_jsgraphics.js            |
>  /memberinfo/downline/tree/disfo2.asp |
>  /styles/png.htc
> }x
>
> Beware that the x option will cause any unescaped whitespace in your
> regexp to be ignored, so you need to take care if you need to match a
> file path that has spaces in it.
>
> With this regexp you would be able to use a single puts(line) call.
> BTW, you should really try to avoid duplicated code like you have with
> your puts lines in general.  Alarm bells should go off in your head if
> you see that sort of thing.  More importantly though, you need to take a
> closer look at what you're doing there with the || operator.  That's not
> exactly doing what you seem to intend. ;-)
>
> Now we get to the meat of your problem.  You can't write to the file
> from which you're reading without manually maintaining at least 2
> pointers into the file, one for your current read position and another
> for your write position.  It's *much* easier actually create a new file
> to hold the filtered data and then use that file to replace the original:
>
> File.open(source_path) do |infile|
>  File.open(temp_path, "w") do |outfile|
>    infile.each_line do |line|
>      outfile.write(line) unless line =~ search_text
>    end
>  end
> end
> File.rename(temp_path, source_path)

Just an additional perspective: for things like this it's probably
easier to use GNU fgrep:

$ fgrep -v -f search_texts input > output
$ mv input input~ && mv output input

Kind regards

robert

-- 
remember.guy do |as, often| as.you_can - without end
http://blog.rubybestpractices.com/