osm-labo/find_largest_commit.sh

#!/bin/bash

# Script to find which commit made the biggest change in repository size
# 
# This script analyzes git commit history to determine which commit caused
# the largest change in the repository's size. It checks out each commit,
# measures the repository size, and identifies the commit with the biggest
# size difference.
#
# Usage: ./find_largest_commit.sh [number_of_commits_to_check]
#
# Arguments:
#   number_of_commits_to_check: Optional. Number of recent commits to analyze.
#                              Defaults to 100 if not specified.
#
# Output:
#   - Detailed information about the commit with the largest size change
#   - A CSV file with data for all analyzed commits
#
# Requirements:
#   - git
#   - bc (for floating-point calculations)
#   - du (for measuring directory sizes)
#
# Author: Junie (JetBrains AI)
# Date: 2025-08-31

# Exit on error
set -e

# Trap for cleanup in case of unexpected exit
trap cleanup EXIT

cleanup() {
    # Make sure we return to the original branch
    if [ -n "$CURRENT_BRANCH" ]; then
        git checkout -q "$CURRENT_BRANCH" 2>/dev/null || true
        
        # Restore stashed changes if needed
        if [ "$STASH_NEEDED" = true ]; then
            echo "Restoring stashed changes..."
            git stash pop -q 2>/dev/null || true
        fi
    fi
}

# Default to checking the last 100 commits if not specified
NUM_COMMITS=${1:-100}

# Validate input
if ! [[ "$NUM_COMMITS" =~ ^[0-9]+$ ]]; then
    echo "Error: Number of commits must be a positive integer."
    echo "Usage: $0 [number_of_commits_to_check]"
    exit 1
fi

if [ "$NUM_COMMITS" -lt 1 ]; then
    echo "Error: Number of commits must be at least 1."
    echo "Usage: $0 [number_of_commits_to_check]"
    exit 1
fi

echo "Analyzing the last $NUM_COMMITS commits to find the largest size change..."
echo "This may take some time depending on repository size and history."
echo

# Get the list of commit hashes
COMMITS=$(git log --pretty=format:"%H" -n "$NUM_COMMITS")

# Initialize variables to track the largest change
LARGEST_CHANGE=0
LARGEST_COMMIT=""
LARGEST_SIZE_BEFORE=0
LARGEST_SIZE_AFTER=0

# Function to get repository size at a specific commit
get_repo_size() {
    local commit=$1
    # Checkout the commit
    git checkout -q "$commit"
    # Calculate size in bytes (excluding .git directory)
    local size=$(du -sb --exclude=.git . | cut -f1)
    echo "$size"
}

# Store current branch to return to it later
CURRENT_BRANCH=$(git rev-parse --abbrev-ref HEAD)

# Check if there are uncommitted changes
if [[ -n $(git status -s) ]]; then
    echo "Stashing uncommitted changes before proceeding..."
    STASH_NEEDED=true
    git stash push -m "Temporary stash by find_largest_commit.sh script"
else
    STASH_NEEDED=false
fi

# Temporary file to store results
TEMP_FILE=$(mktemp)

echo "Commit Hash,Author,Date,Size Before (bytes),Size After (bytes),Change (bytes),Change (%),Message" > "$TEMP_FILE"

# Counter for progress display
COUNTER=0
TOTAL_COMMITS=$(echo "$COMMITS" | wc -l)

# Process each commit
PREV_SIZE=""
for COMMIT in $COMMITS; do
    COUNTER=$((COUNTER + 1))
    echo -ne "Processing commit $COUNTER/$TOTAL_COMMITS...\r"
    
    # Get commit details
    AUTHOR=$(git show -s --format="%an" "$COMMIT")
    DATE=$(git show -s --format="%cd" --date=format:"%Y-%m-%d %H:%M:%S" "$COMMIT")
    MESSAGE=$(git show -s --format="%s" "$COMMIT" | sed 's/,/;/g') # Replace commas with semicolons
    
    # Get size after this commit
    SIZE_AFTER=$(get_repo_size "$COMMIT")
    
    # If this is the first commit we're checking, we don't have a previous size
    if [ -z "$PREV_SIZE" ]; then
        PREV_SIZE="$SIZE_AFTER"
        continue
    fi
    
    # Calculate size before (which is the size after the previous commit)
    SIZE_BEFORE="$PREV_SIZE"
    PREV_SIZE="$SIZE_AFTER"
    
    # Calculate change
    CHANGE=$((SIZE_AFTER - SIZE_BEFORE))
    ABS_CHANGE=${CHANGE#-} # Absolute value
    
    # Calculate percentage change
    if [ "$SIZE_BEFORE" -ne 0 ]; then
        PERCENT_CHANGE=$(echo "scale=2; 100 * $CHANGE / $SIZE_BEFORE" | bc)
    else
        PERCENT_CHANGE="N/A"
    fi
    
    # Record the data
    echo "$COMMIT,$AUTHOR,$DATE,$SIZE_BEFORE,$SIZE_AFTER,$CHANGE,$PERCENT_CHANGE%,$MESSAGE" >> "$TEMP_FILE"
    
    # Check if this is the largest change so far
    if [ "$ABS_CHANGE" -gt "$LARGEST_CHANGE" ]; then
        LARGEST_CHANGE="$ABS_CHANGE"
        LARGEST_COMMIT="$COMMIT"
        LARGEST_SIZE_BEFORE="$SIZE_BEFORE"
        LARGEST_SIZE_AFTER="$SIZE_AFTER"
    fi
done

# Return to the original branch
# (Cleanup function will handle restoring stashed changes)
git checkout -q "$CURRENT_BRANCH"

echo -e "\nAnalysis complete!"

# Function to format size in human-readable format
format_size() {
    local size=$1
    if [ "$size" -ge 1073741824 ]; then
        echo "$(echo "scale=2; $size / 1073741824" | bc) GB"
    elif [ "$size" -ge 1048576 ]; then
        echo "$(echo "scale=2; $size / 1048576" | bc) MB"
    elif [ "$size" -ge 1024 ]; then
        echo "$(echo "scale=2; $size / 1024" | bc) KB"
    else
        echo "$size bytes"
    fi
}

# Display the result
if [ -n "$LARGEST_COMMIT" ]; then
    echo
    echo "Commit with the largest size change:"
    echo "-----------------------------------"
    echo "Commit: $LARGEST_COMMIT"
    echo "Author: $(git show -s --format="%an" "$LARGEST_COMMIT")"
    echo "Date: $(git show -s --format="%cd" --date=format:"%Y-%m-%d %H:%M:%S" "$LARGEST_COMMIT")"
    echo "Message: $(git show -s --format="%s" "$LARGEST_COMMIT")"
    echo
    echo "Size before: $(format_size "$LARGEST_SIZE_BEFORE")"
    echo "Size after: $(format_size "$LARGEST_SIZE_AFTER")"
    
    CHANGE=$((LARGEST_SIZE_AFTER - LARGEST_SIZE_BEFORE))
    if [ "$CHANGE" -ge 0 ]; then
        echo "Change: +$(format_size "${CHANGE#-}") (increased)"
    else
        echo "Change: -$(format_size "${CHANGE#-}") (decreased)"
    fi
    
    if [ "$LARGEST_SIZE_BEFORE" -ne 0 ]; then
        PERCENT_CHANGE=$(echo "scale=2; 100 * $CHANGE / $LARGEST_SIZE_BEFORE" | bc)
        echo "Percentage change: $PERCENT_CHANGE%"
    fi
    
    echo
    echo "Files changed in this commit:"
    
    # Get the list of changed files
    CHANGED_FILES=$(git show --stat "$LARGEST_COMMIT" | grep '|' | sort -rn -k3)
    TOTAL_FILES=$(echo "$CHANGED_FILES" | wc -l)
    
    # If there are too many files, show only the top 10 with the most changes
    if [ "$TOTAL_FILES" -gt 10 ]; then
        echo "$CHANGED_FILES" | head -n 10
        echo "... and $(($TOTAL_FILES - 10)) more files (total: $TOTAL_FILES files changed)"
    else
        echo "$CHANGED_FILES"
    fi
else
    echo "No commits analyzed."
fi

echo
echo "Full results saved to: $TEMP_FILE"
echo "You can import this CSV file into a spreadsheet for further analysis."

# Make the script executable
chmod +x "$0"