[Cygwin, sed] Using filenames as both files and search strings within sed

lingh · 10-18-2012, 02:57 AM

I have a collection of documents within a folder, and I want to go through each of the files and then do the following:

1. Use the file as an input to sed.
2. Use the filename as a parameter in the sed command, with the extension stripped off.

So, lets assume my files are:

file1.txt
file2.txt
file3.sh

I can go through the files easily with the line code:

[pre]for file in `dir`; do echo $file;done[/pre]

I can extend that and strip the extension as well:

[pre]for file in `dir`; do echo $file to ${file%%.*}; done[/pre]

But how can I add these together? Let's say my requirement is as simple as this for sed:

Within the file file3.sh is the name of the document, followed by a version number. e.g. file3.sh contains somewhere within it the string "file3 Vxxx" where xxx is the version number. For each file I want to print this string and this string only from the file. There is text before and after this string.

I need to do this same task for every file in the current directory, but I cannot seem to get the sed line to work, nor can I seem to get sed to pick up the substituted filename (without the extension)

Can anyone offer some insight into this?

ZackFair · 10-18-2012, 03:20 AM

Not sure what you want to do with sed.
Also not sure I got your question right but here is my suggestion:

Code:

for filename in `ls`; do 
  name=`echo $filename|awk -F'.' {'print $1'}`
  cat $filename | grep "$name V"
done

If this is a standalone string (no extra text at beginning and at the end of the string) - this should do the trick.

ntubski · 10-18-2012, 03:21 AM

Quote:

Originally Posted by lingh

I can go through the files easily with the line code:

[pre]for file in `dir`; do echo $file;done[/pre]

Even easier is

Code:

for file in *; do echo "$file"; done

If you remember to quote $file, this will handle spaces in filenames as well.

Quote:

But how can I add these together? Let's say my requirement is as simple as this for sed:

Within the file file3.sh is the name of the document, followed by a version number. e.g. file3.sh contains somewhere within it the string "file3 Vxxx" where xxx is the version number. For each file I want to print this string and this string only from the file. There is text before and after this string.

You haven't written what you tried with sed, but I think grep -o will be the most convenient for this:

Code:

for file in *; do
    grep -o "${file%%.*} V..." "$file"
done

Although that won't work so well if your filenames have special regex characters in them ("[", "*", "^", or "$").

Quote:

From man grep(1):
-o, --only-matching

Print only the matched (non-empty) parts of a matching line, with each such part on a separate output line.

lingh · 10-18-2012, 04:32 AM

Thanks very much, ntubski, your information really helped. I ended up doing this:

Quote:

for file in *; do grep -o --text "${file%%.*} V[0-9]\{3\}" "$file";done

Because I wanted to make sure the version number was really numeric and I had to force grep to treat the file as text, because it was coming up with "file3.sh matches" and not giving me the output - presumably because it thought the file was binary. But it works perfectly. Thanks for that. I think I need to check out grep in more detail.

David the H. · 10-20-2012, 10:30 AM

Please use ***[code][/code]*** tags around your code and data, to preserve the original formatting and to improve readability. Do not use quote tags, bolding, colors, "start/end" lines, or other creative techniques.

To expand slightly on ntubski's post above, never use commands like ls or cat with for loops. See here for why:

Don't Read Lines With For

When operating on files, simple globbing is usually all you need. And when reading the contents of files or commands, use a while+read loop.

David the H. · 10-20-2012, 10:38 AM

( Deleted accidental double post )