0

I have this simple script, which does nothing more than:

  • check if an email matches a specific pattern
  • in that case, add a tag to a taglist
  • before quitting, print that taglist
set -e

lista_tag=()
in_file="/tmp/grepmail-classify.txt"

# save stdin to file, to use it multiple times
cp /dev/stdin $in_file


# CLASSIFY

res=$(grepmail -B "some regex pattern" < $in_file)
if [ ! -z "$res" ]
then
    lista_tag+=("PUSH")
fi

res=$(grepmail -B "some other regex pattern" < $in_file)
if [ ! -z "$res" ]
then
    lista_tag+=("MERGIFY")
fi

# ⁝ Many many more similar patterns

# output them comma separated
echo ${lista_tag[*]}

As you can see there is a case for refactoring and abstraction. res and if .. fi parts are repeated. But I am not sure how to safely pass commands around. What I think I would like to do is to invoke a function like this (or similar):

classify '"grepmail -B "somepattern"' 'MYTAG'

But it is tricky! I have read the FAQ but I am not sure it covers my case.

So here is the question: what is the correct way to pass commands around (if there is any)? How would the res= part of such function look like?

Improve this question

2 Answers 2

Reset to default
2 
classify '"grepmail -B "somepattern"' 'MYTAG'

This is hard to get to work, for exactly the reasons mentioned in BashFAQ 050.

But we can make it work if we put the "tag" argument first, since that allows us to use the rest for the command:

#!/bin/bash
lista_tag=()
classify() {
    local tag="$1"
    shift
    res=$( "$@" < "$in_file")
    if [ -n "$res" ]; then
        lista_tag+=("$tag")
    fi
}
classify PUSH    grepmail -B "some regex pattern"
classify MERGIFY grepmail -B "some other regex pattern"

The key here being that we don't stick the arguments of the command into one string, but keep them separate. "$@" is magic: it expands to all the positional parameters separately. After shifting out the tag, the rest are your command.

You can't stick the redirection there in the same way, though, as that would require running the command through eval and quoting it appropriately. Which you'd also need to do carefully for any user-provided input, since otherwise you'd have a high chance of leaving a command execution vulnerability there.

Anyway, since the grepmail -B part seems constant, just pass the tag and the pattern:

#!/bin/bash
lista_tag=()
in_file=foo.txt
classify() {
    local pattern="$1"
    local tag="$2"
    if [[ -n "$(grepmail -B "$pattern" < "$in_file")" ]]; then
        lista_tag+=("$tag")
    fi
}
classify "some regex pattern" PUSH
classify "some other regex pattern" MERGIFY
Improve this answer
1

This is a good opportunity to use an associative array:

#!/bin/bash
lista_tag=()

declare -A patterns=(
    [PUSH]='some regex pattern'
    [MERGIFY]='some other pattern'
    # ... other [tag]=pattern pairs ...
)

# capture stdin
in_file=$(mktemp)
cat > $in_file

# CLASSIFY

# iterate over the array indices
for tag in "${!patterns[@]}"; do
    if [[ -n "$( grepmail -B "${patterns[$tag]}" < "$in_file" )" ]]; then
        lista_tag+=("$tag")
    fi
done

# print comma-separated list
( IFS=","; echo "${lista_tag[*]}" )

If grepmail can have a non-zero exit status if there's no match (like grep -q) then it's simpler:

for tag in "${!patterns[@]}"; do
    grepmail -q -B "${patterns[$tag]}" < "$in_file" && lista_tag+=("$tag")
done
Improve this answer
1
  • Need more details. Please edit your question Commented Jan 31, 2023 at 21:54

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.