Shell Programming and Scripting

maybe I'm missing something in the OP's requirement...

Both sed and grep are line-oriented utilities, so you'd need to orient the input into 'records' for it to be of any use. For example:

Code:

-> echo 1,2,3,4,5,6,10000,10987,16572 |tr ',' "\n"
1
2
3
4
5
6
10000
10987
16572

-> echo 1,2,3,4,5,6,10000,10987,16572 |tr ',' "\n" |egrep -v '10000|16572' # leaving 10987 in play for now... 
1
2
3
4
5
6
10987

Each line in the data is a record of comma-delimited fields. I don't see the problem with that.

You'd need to specify a full line match with -x or use ^...$ anchors to prevent partial matches from falsely eliminating keepers. It would also be a good idea to use -F, forcing fixed patterns, and -f, so that egrep reads the fixed strings from a file (that way, we can avoid a potentially lengthy command line): egrep -vxFf listfile

However, this approach irreversibly modifies the data, since after tr it's not possible to distinguish between newlines that were newlines and newlines that were commas, unless some restrictive assumptions can be made regarding the data (such as forbidding blank lines or null comma-delimited fields). This may not be a problem; I'm just mentioning it in case it is relevant to the original poster.

If an output format of one number per line is acceptable, the following might suffice:

Code:

tr , \\n < datafile | egrep -vxFf listfile

Regards,
Alister

---------- Post updated at 03:49 PM ---------- Previous update was at 03:42 PM ----------

If the data must be left intact, format unmodified except for the deleted numbers, the following script would do the job:

Code:

#!/bin/sh

list=$1
data=$2

{
    while read n; do
        printf 'g/%s/s/^%s,//\\\ns/,%s,/,/g\\\ns/,%s$//\n' $n $n $n $n
    done < "$list"
    printf 'wq\n'
} | ed -s "$data"

Important: I have not modified IFS in the read statement, nor set read to raw mode, nor quoted $n, nor bothered to escape the contents of $n before inserting it into the ed regular expressions because I am assuming that each line in the list file consists of nothing but numerical digits. In sed basic regular expression syntax, each line in the list file should match /^[0-9]\{1,\}$/. If the following doesn't yield the entire list file's contents, then don't use the above code:

Code:

sed -n '/^[0-9]\{1,\}$/p'

Test run (in what follows, the script above is rmnum.sh):

Code:

$ cat data
1,2,3,4,5,6,7,8,9,10
11,12,13,14,15,16,17,18,19,20
21,22,23,24,25,26,27,28,29,30
31,32,33,34,35,36,37,38,39,40
$ cat list  # odd numbers
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
$ ./rmnum.sh list data
$ cat data
2,4,6,8,10
12,14,16,18,20
22,24,26,28,30
32,34,36,38,40

If the list may contain floating point numbers, whose decimal point would be considered a wildcard, I would opt to use perl. I don't use it often, but, as I recall, it supports quoting variable text within regular expressions, so that metacharacters are treated literally.

Regards,
Alister