A simple script to count net changes in the git repository for the defined file extensions (useful for daily goals when writing thesis/articles)

git_time_word_count.sh

#!/bin/bash
# Git Word Count Analyzer
# Copyright (c) 2024 Daniel Borek
# MIT License
# Inspired by https://www.gijsvandam.nl/post/measuring-your-writing-progress-with-a-git-word-count/

# Configuration
# FILE_EXTENSIONS: Array of file extensions to analyze (e.g., "typ" "md")
# Separate multiple extensions with spaces
FILE_EXTENSIONS=("typ" "md" )

# DAYS_TO_SEARCH: Number of days to analyze, starting from today
# The script will stop at the repository's creation date if reached
# Default is set to 1 week (7 days)
DAYS_TO_SEARCH=7

# Function to calculate word changes for a single commit
calculate_word_changes() {
    local sha=$1
    local parent_sha=$(git rev-parse $sha^@ 2>/dev/null || git hash-object -t tree /dev/null)
    local file_patterns=()
    for ext in "${FILE_EXTENSIONS[@]}"; do
        file_patterns+=("*.$ext")
    done

    if [ "$parent_sha" = "$sha" ]; then
        # This is the first commit
        local added=$(git show --format= --name-only $sha | grep -E "\.($(IFS=\|; echo "${FILE_EXTENSIONS[*]}"))\$" | xargs cat | wc -w | xargs)
        local deleted=0
        local duplicated=0
    else
        local added=$(git diff --word-diff=porcelain $parent_sha..$sha -- "${file_patterns[@]}" | grep -e"^+[^+]" | wc -w | xargs)
        local deleted=$(git diff --word-diff=porcelain $parent_sha..$sha -- "${file_patterns[@]}" | grep -e"^-[^-]" | wc -w | xargs)
        local duplicated=$(git diff $parent_sha..$sha -- "${file_patterns[@]}" | grep -e"^+[^+]" -e"^-[^-]" | sed -e's/.//' | sort | uniq -d | wc -w | xargs)
    fi
    echo "$added $deleted $duplicated"
}

# Function to format numbers with leading spaces
format_number() {
    printf "%6d" $1
}

# Get the date of the earliest commit
earliest_date=$(git log --reverse --format=%ad --date=short | head -1)

# Print header
printf "%-10s %8s %8s %11s %11s\n" "Date" "Added" "Deleted" "Duplicated" "Net Change"
printf "%-10s %8s %8s %11s %11s\n" "----------" "--------" "--------" "-----------" "-----------"

for i in $(seq 0 $((DAYS_TO_SEARCH - 1))); do
    # Calculate the date we're checking
    check_date=$(date -v-${i}d +%Y-%m-%d)

    # If we've gone past the earliest commit date, break the loop
    if [[ "$check_date" < "$earliest_date" ]]; then
        break
    fi

    j=$((i+1))
    next_date=$(date -v-${j}d +%Y-%m-%d)

    # Get the list of commits for this day
    commits=$(git rev-list --since="$next_date" --until="$check_date" main)

    total_added=0
    total_deleted=0
    total_duplicated=0

    if [ -n "$commits" ]; then
        for sha in $commits; do
            if git diff-tree --no-commit-id --name-only -r $sha | grep -qE "\.($(IFS=\|; echo "${FILE_EXTENSIONS[*]}"))\$"; then
                read added deleted duplicated <<< $(calculate_word_changes $sha)
                total_added=$((total_added + added))
                total_deleted=$((total_deleted + deleted))
                total_duplicated=$((total_duplicated + duplicated))
            fi
        done
    fi

    net_change=$((total_added - total_deleted))

    # Output the results for this day
    printf "%-10s %8s %8s %11s " "$check_date" "$(format_number $total_added)" "$(format_number $total_deleted)" "$(format_number $total_duplicated)"
    if [ $net_change -gt 0 ]; then
        printf "%11s\n" "+$(format_number $net_change)"
    elif [ $net_change -lt 0 ]; then
        printf "%11s\n" "-$(format_number ${net_change#-})"
    else
        printf "%11s\n" "$(format_number 0)"
    fi
done