feat(unitproc/bkt-replace_contractions):Add Bash function

[BK-2020-03.git] / unitproc / bkt-replace_contractions

#!/bin/bash

# Function to replace contractions
replace_contractions() {
    # Desc: Replace ' with ʼ in contractions
    # Note: In contractions of UTF-8 text file, replaces U+0027
    #   APOSTROPHE with U+02BC MODIFIER LETTER APOSTROPHE
    # Input: stdin  
    #        arg1   file path
    # Output: stdout
    # Version: 0.0.1
    # Depends: GNU sed 4.8

    # Check input
    if [[ "$#" -gt 1 ]]; then
        echo "FATAL:Incorrect argument count:$#" 1>&2;
        return 1;
    fi;

    if [[ -f "$1" ]]; then
        # Use specified file
        input="$1";
    else
        # Use standard input
        input="-";
    fi;

    # Perform substitutions
    sed -E \
        -e "s/(you're|You're|YOU'RE)/youʼre/gI" \
        -e "s/(i'm|I'm|I'M)/Iʼm/gI" \
        -e "s/(you've|You've|YOU'VE)/youʼve/gI" \
        -e "s/(they're|They're|THEY'RE)/theyʼre/gI" \
        -e "s/(we're|We're|WE'RE)/weʼre/gI" \
        -e "s/(they've|They've|THEY'VE)/theyʼve/gI" \
        -e "s/(we've|We've|WE'VE)/weʼve/gI" \
        -e "s/(i've|I've|I'VE)/Iʼve/gI" \
        -e "s/(that's|That's|THAT'S)/thatʼs/gI" \
        -e "s/(what's|What's|WHAT'S)/whatʼs/gI" \
        -e "s/(here's|Here's|HERE'S)/hereʼs/gI" \
        -e "s/(there's|There's|THERE'S)/thereʼs/gI" \
        -e "s/(where's|Where's|WHERE'S)/whereʼs/gI" \
        -e "s/(who's|Who's|WHO'S)/whoʼs/gI" \
        -e "s/(how's|How's|HOW'S)/howʼs/gI" \
        -e "s/(doesn't|Doesn't|DOESN'T)/doesnʼt/gI" \
        -e "s/(don't|Don't|DON'T)/donʼt/gI" \
        -e "s/(i'll|I'll|I'LL)/Iʼll/gI" \
        -e "s/(we'll|We'll|WE'LL)/weʼll/gI" \
        -e "s/(they'll|They'll|THEY'LL)/theyʼll/gI" \
        "$input";
}; # replace ' with ʼ in contractions