Shell Programming and Scripting

Help
I read a file that has 2 fields.
look for in a second file the first field and update it with the second field of first file.

file1
1131518fat11416.txt ../newaod/2001/04/2001-04-00129233-1.txt

file2
INSERT INTO tabric
VALUES
('00566605','00499181','ZPPLGU50S20F158O','1131518fat11416.txt','2000',1)

risultato
INSERT INTO tabric
VALUES
('00566605','00499181','ZPPLGU50S20F158O','../newaod/2001/04/2001-04-00129233-1.txt','2000',1)

I can not do it with awk.
Thanks to all

Assuming your files don't contain pipes ('|')

Code:

#!/bin/bash
while read A B
do
	sed "s|$A|$B|" file2
done < file1

Assuming $A doesn't contain "|", "\", ".", "*", "[", "]", "^", "$" and assuming $B doesn't contain "|", "\", "&". (If it's gnu sed, then there are a few other special characters to worry about.)

Regards,
Alister

---------- Post updated at 05:55 PM ---------- Previous update was at 04:31 PM ----------



The following attempts to backslash escape characters that are special in posix basic regular expressions when they appear in the first field of file1, and attempts to backslash escape characters that are special in sed replacement text when they appear in the second field of file1. I say "attempts" because I only tested it against a few contrived lines of text, and only using one sed implementation (not gnu, so any additional metacharacters in gnu sed will leak through unescaped and possibly break the solution), and because it was concocted for the fun of it.

The (hopefully correctly) sanitized input from file1 is used to build a stream of ed commands which modify file2:

Code:

{ sed 'h; s| .*||; s|[][^$/\.*]|\\&|g; s|.*|/&/|; x; s|.* ||; s|[\&/]|\\&|g; s|.*|s//&/|; H; g; s|\n||' file1; printf 'w\nq\n'; } | ed -s file2

For important tasks of this type, I would suggest using a tool that can match strings literally, like AWK's index/substr functions. Don't expand sh parameters into sed commands unless you are sure the incoming text is free of special characters.

Cheers,
Alister