Quick and dirty bash script for running a multi-find-and-replace on a csv (using another csv containing mapping definitions)

csv_multi_find_replace.sh

#!/bin/bash

# Takes a mapping csv with the first column containing the values to replace-with and the 2nd to nth columns containing the patterns
# to replace-by. The target document is matched by cell against each pattern, and a cell that matches is replaced in the output (mfr_output.csv)
# by the replace value corresponding to the first pattern it matches on.

# Known issues:
#   Seems to be bad at matching certain accented chars, encoding issues? Maybe deburr before matching if you have that problem

mapping_file=$1

target_file=$2

mapping_file_length=$( cat $mapping_file | wc -l )

# likely need to change the file separator (-F) option to suit your needs
gawk -v map_length="${mapping_file_length}" -F "," '
  BEGIN {
    target_file_string = ""
  }

  NR <= map_length {
    pattern = "^"$2"$"
    for (i = 3; i <= NF; i++){
      if ($i != ""){
        pattern = pattern"|^"$i"$"
      }
    }
    patern_map[NR] = pattern
    replace_map[NR] = $1
  }
 
  NR > map_length {
    mapped_record = ""
    for (i = 1; i <= NF; i++){
      sub(RS, $(i), "") 
      mapped_cell = $(i)
      for (j = 1; j <= map_length; j++){
        if ( mapped_cell ~ patern_map[j] ){
          mapped_cell = replace_map[j]
          j = map_length + 1
        } 
      }
      if (i <NF) {
        mapped_record = mapped_record""mapped_cell""FS""
      } else {
        mapped_record = mapped_record""mapped_cell""RS""
      }
    } 
    target_file_string = target_file_string""mapped_record
  }
 
  END {
    print target_file_string
  }
' $mapping_file $target_file > mfr_output.csv