unitproc/bkt-replace_contractions

   1 #!/bin/bash
   2
   3 # Function to replace contractions
   4 replace_contractions() {
   5     # Desc: Replace ' with ʼ in contractions
   6     # Note: In contractions of UTF-8 text file, replaces U+0027
   7     #   APOSTROPHE with U+02BC MODIFIER LETTER APOSTROPHE
   8     # Input: stdin
   9     #        arg1   file path
  10     # Output: stdout
  11     # Version: 0.0.1
  12     # Depends: GNU sed 4.8
  13
  14     # Check input
  15     if [[ "$#" -gt 1 ]]; then
  16         echo "FATAL:Incorrect argument count:$#" 1>&2;
  17         return 1;
  18     fi;
  19
  20     if [[ -f "$1" ]]; then
  21         # Use specified file
  22         input="$1";
  23     else
  24         # Use standard input
  25         input="-";
  26     fi;
  27
  28     # Perform substitutions
  29     sed -E \
  30         -e "s/(you're|You're|YOU'RE)/youʼre/gI" \
  31         -e "s/(i'm|I'm|I'M)/Iʼm/gI" \
  32         -e "s/(you've|You've|YOU'VE)/youʼve/gI" \
  33         -e "s/(they're|They're|THEY'RE)/theyʼre/gI" \
  34         -e "s/(we're|We're|WE'RE)/weʼre/gI" \
  35         -e "s/(they've|They've|THEY'VE)/theyʼve/gI" \
  36         -e "s/(we've|We've|WE'VE)/weʼve/gI" \
  37         -e "s/(i've|I've|I'VE)/Iʼve/gI" \
  38         -e "s/(that's|That's|THAT'S)/thatʼs/gI" \
  39         -e "s/(what's|What's|WHAT'S)/whatʼs/gI" \
  40         -e "s/(here's|Here's|HERE'S)/hereʼs/gI" \
  41         -e "s/(there's|There's|THERE'S)/thereʼs/gI" \
  42         -e "s/(where's|Where's|WHERE'S)/whereʼs/gI" \
  43         -e "s/(who's|Who's|WHO'S)/whoʼs/gI" \
  44         -e "s/(how's|How's|HOW'S)/howʼs/gI" \
  45         -e "s/(doesn't|Doesn't|DOESN'T)/doesnʼt/gI" \
  46         -e "s/(don't|Don't|DON'T)/donʼt/gI" \
  47         -e "s/(i'll|I'll|I'LL)/Iʼll/gI" \
  48         -e "s/(we'll|We'll|WE'LL)/weʼll/gI" \
  49         -e "s/(they'll|They'll|THEY'LL)/theyʼll/gI" \
  50         "$input";
  51 }; # replace ' with ʼ in contractions