Shell Programming and Scripting

Hello friends,I am new to Unix programming.
how do I achieve the following in Unix shell script (I am running ksh on AIX)
extract the number from name of file?

My file format is like "LongFileName-1234.020614-221030.txt"
now I want to extract value which is between (-) hyphen and (.) dot i.e. 1234,
once I get that value I need to assign it to variable and need to compare against oracle db.
i am looking for solution which works if value goes from 4 digit to 5 digit and onward. i.e. consider following file names

Code:

"LongFileName-123.020614-221030.txt"
"LongFileName-1234.020614-221030.txt"
"LongFileName-12345.020614-221030.txt"
"LongFileName-123456.020614-221030.txt"
 and I need to extract 123, 1234,12345,123456 etc

I appreciate your responses.

In ksh/bash:

Code:

while read f
do
  f=${f#*-}
  f=${f%%.*}
  echo $f
done <file_name_list
 
or exploiting the '-' an '.' more:
 
( IFS="$IFS-."
while read f1 f2 f3
do
  echo $f2
done <file_name_list )
 
but using 'sed' the stream editor, simplest of all:
 
sed 's/^[^-]*-\([^.]*\).*/\1/' file_name_list

The first uses the variable chewers: ${#NamePattern} ${#NamePattern} ${##NamePattern} ${%NamePattern} ${%%NamePattern} I use the mnemonic "Pound soft or hard on the nose, or take your small or big % off the end." The patterns are file flavored, with metacharacters * ? [] but not [^].

The second says the '-' and '.' is now a word separator like space (but just within the (subshell), we do not want it 'from now on and forever'), so you can read each line as 4 words into 3 variables.

Sed is instructed: substitute for "from beginning of line, anything not a '-' up to a '-', start a substring capture, anything not a dot, end substring, everything else", put back just the substring captured. The regex '.*' is greedy but the '^' BOL and '[^-]*' not a hyphen in any number is more restrained. Regex uses '.' where file patterns uses '?', '.*' where file uses '*', has the negative list and other more powerful metastrings. I expect I overlooked bash ways to use regex in shell.

Hi f150,
Welcome to forum, please use code tags while posting codes and commands in your post as per forum rules. Following may help you in same.

Code:

awk '{gsub(/\..*/,X,$0);gsub(/.*\-/,Y,$0);print $0}'  Input_file

Output will be as follows.

Code:

123
1234
12345
123456

Thanks,
R. Singh

Thank you Singh sir, I will do the same. will try the script and post the response.

---------- Post updated at 03:45 PM ---------- Previous update was at 03:40 PM ----------

Thnx a lot DGPickett ! I will try the script and post the result.
will it work even if the numbers between (-) and (.) does not follow any pattern? I mean if the numbers are like 231, 4823, 89103 will it still give same output ?

Yes, ther sed regex says anything but a dot '[^.]' in any quantity '*'. We could have been tighter and said just numbers '[0-9]'. The shell variable grinders are gentle up to the '-': '#*-' and agressive from end back to '.': '%%.*'. The two dots and two hyphens made the more complex and carefully limited expressions necessary.