Execute independent greps in a single shell script

Question

I have several grep filters that I usually use to parse specific information.

1st grep: grep "pattern1\|pattern2\|pattern3\|" file.txt

2nd grep: grep "patternA\|patternB\|patternC\|" file.txt

etc.

I apply each grep usually to the same file.txt to get an independent output.

I would like to know how I can group this bunch of greps into a single bash script in order to get independent outputs based on each type of grep.

For example: input file.txt looks as follows:

This line1 is the first line in here1
This line2 is the second line in here2
This line3 is the third line in here3
This line4 is the fourth line in here4

I usually apply separate greps here to get specific patterns.

grep -h -r --color=always "line1\|here1" file.txt >>pattern1.txt

or

grep -h -r --color=always "line2\|here2" file.txt >>pattern2.txt

This will highlight only the information required and will give me separate pattern*.txt files to work on. The objective here is to run all these greps in a single time to evaluate the same file and print in shell as follows:

• Pattern1
• Pattern2
• Pattern3

etc.

Each grep should evaluate the complete file independently.

• Eric Renouf: This question would benefit a lot from you editting it to include sample input and desired output. Also, check out unix.stackexchange.com/help/formatting
• terdon: Yes, please edit and clarify. Give us a sample input and the specific grep commands you would run and show us the output you want to see. I'm afraid I don't understand what you are asking right now.
• raver83: I have input a sample, hope now is more clear.
• Jeff Schaller: How are you getting from your grep criteria to the name "Pattern1" as the output?

Answer 1

If I understand the question correctly, it’s about how to do the same command (grep), with the same options and the same input, but multiple times, with different regex arguments and different outputs. And I guess you want to avoid unnecessary repetition/duplication.

It seems like you want an array of regular expressions (search strings):

declare -A regex
regex[1]="line1\|here1"
regex[2]="line2\|here2"
regex[3]="line3\|here3"
regex[4]="line4\|here4"
for i in "${!regex[@]}"
do
 grep -h -r --color=always "${regex["$i"]}" file.txt >> "pattern$i.txt"
done

The first line (declare -A regex) declares an associative array called regex; that means it creates the array as a placeholder, but doesn't enter any information (elements) into it. The next four lines populate the array with four elements, which are the regular expressions, indexed by the numbers 1, 2, 3, and 4. (I’m using those indices because that’s what you seem to want, but you can use any distinct strings as indices: for example, uno, dos, tres and cuatro, or ant, bat, cat and dog.^†) The for statement, for i in "${!regex[@]}", causes the variable i to iterate through the index values 1, 2, 3, and 4. (If I had left out the ! and said for i in "${regex[@]}", it would have iterated through the element values, line1\|here1, line2\|here2, line3\|here3 and line4\|here4.) When $i is 1, ${regex["$i"]} reduces to ${regex[1]}, which expands to line1\|here1. So, the loop iterates (executes) four times, executing the four grep commands that you want.

If you want to run the grep processes in parallel, just do:

for i in "${!regex[@]}"
do
 grep -h -r --color=always "${regex["$i"]}" file.txt>> "pattern$i.txt" &
done
wait

______________
^† If the indices are numerically distinct non-negative integers, you can leave out the declare statement.

• raver83: Thanks!, it is working as expected, however, if I use strings as indices for the regex, it doesn't work, for example: regex[test_one]="line1\|here1", is there something I'm missing here?. When using strings the for loop exits after evaluating the first grep.
• G-Man Says 'Reinstate Monica': Sorry; it's my fault — I left out a line. I've fixed the answer.
• raver83: Thanks G-man, I believe it should work but doing some research, declare -A is available from bash v4.0, unfortunately I'm a mac OS X user and bash is available up to v3.2.57, I'm thinking to use python instead to use dictionaries as associative arrays, currently this is what I get as an output: declare: usage: declare [-afFirtx] [-p] [name[=value] ...]
• G-Man Says 'Reinstate Monica': Well, as I said, if you use numbers as indices, you don’t need to declare the array at all.