shell script to count number of transcription and text files and documents in pgp-text git repo over time

count_pgptxtfiles.sh

#!/bin/bash

# based on https://blog.benoitblanchon.fr/git-file-count-vs-time/

OUTPUT=stats.csv

# create output file with a CSV header
# echo "date;transcription_count;transcribed_documents;translation_count;translated_documents" > $OUTPUT
echo "date,transcriptions,transcribed_documents,translations,translated_documents" > $OUTPUT

# function that counts files matching the specified regex
count() {
    git ls-tree -r --name-only $COMMIT | grep -e $1 | wc -l | sed 's/ //g'
}

# adapt file count to count unique pgpids
count_documents() {
    git ls-tree -r --name-only $COMMIT  | grep -e $1 | sed -re "s/^.*(PGPID[0-9]+)_.*$/\1/" | uniq | wc -l | sed 's/ //g'
}

   
# for each commit in log
git log --pretty="%H %cd" --date=short | while read COMMIT DATE
do
    # skip commits made on the same day
    [ "$PREV_DATE" == "$DATE" ] && continue
    PREV_DATE="$DATE"

    # count files
    XSCRIPT_FILES=$(count ".*/*_transcription.txt$")
    TRANSLATION_FILES=$(count ".*/*_translation.txt$")    
    # count documents
    XSCRIPT_DOCS=$(count_documents ".*/*_transcription.txt$")
    TRANSLATION_DOCS=$(count_documents ".*/*_translation.txt$")    

    # print to console
    echo $DATE
    echo " $XSCRIPT_FILES	transcription.txt files"
    echo " $XSCRIPT_DOCS    transcription.txt PGPIDs"    
    echo " $TRANSLATION_FILES	translation.txt files"    
    echo " $TRANSLATION_DOCS	translation.txt PGPIDs"    

    # append to CSV file
    # echo "$DATE;$XSCRIPT_FILES;$TRANSLATION_FILES" >> $OUTPUT     
    echo "$DATE,$XSCRIPT_FILES,$XSCRIPT_DOCS,$TRANSLATION_FILES,$TRANSLATION_DOCS" >> $OUTPUT 
done