GIT Repository History scanner

repository-history.sh

#!/bin/bash

# Analyzes git commit history to find ticket-related commits
# Usage: ./repository-history.sh [-h COMMIT_HASH] [-t TICKET] [-a AUTHOR] [-d DAYS]
# - Without arguments: Shows ticket commits from current branch (last 30 days)
# - -h COMMIT_HASH: Shows ticket commits up to that hash in current branch (within specified days)
# - -t TICKET: Filters results to show only commits for specific ticket (e.g., OV-12345)
# - -a AUTHOR: Filters results by author name/email (case-insensitive partial match)
# - -d DAYS: Number of days to look back (default: 30)
# - All flags can be combined: -h HASH -t TICKET -a AUTHOR -d DAYS
# Note: COMMIT_HASH must exist in current branch
# Tickets follow the configured pattern (configurable in CONFIGURATION SECTION)

set -e

# =============================================================================
# CONFIGURATION SECTION
# =============================================================================
# Customize these variables for different projects or ticket systems

# Ticket pattern configuration
TICKET_PREFIX="AB"                    # Ticket prefix (e.g., AB, JIRA, PROJ)
TICKET_PATTERN="$TICKET_PREFIX-[0-9]+"   # Full regex pattern for tickets
TICKET_EXAMPLE="$TICKET_PREFIX-12345"    # Example ticket for help messages

# Default values
DEFAULT_DAYS=30                       # Default number of days to look back

# Display configuration
DATE_FORMAT="%Y-%m-%d %H:%M:%S"       # Format for commit dates
SEPARATOR_LINE="================================================================="

# =============================================================================

# Function to display ticket commits from git log output
display_ticket_commits() {
    local ticket_filter="$1"
    local author_filter="$2"
    
    local filter_pipeline="grep -E '$TICKET_PATTERN'"
    
    # Add ticket filter if specified
    if [ -n "$ticket_filter" ]; then
        filter_pipeline="$filter_pipeline | grep -i '$ticket_filter'"
    fi
    
    # Add author filter if specified
    if [ -n "$author_filter" ]; then
        filter_pipeline="$filter_pipeline | grep -i '$author_filter'"
    fi
    
    eval "$filter_pipeline" | \
    while IFS='|' read -r hash author date message; do
        # Extract ticket numbers from the commit message
        tickets=$(echo "$message" | grep -oE "$TICKET_PATTERN" | sort -u | tr '\n' ', ' | sed 's/,$//')
        
        # Clean commit message by removing ticket prefix if it starts the message
        # Handles patterns like: "OV-1234 message", "OV-1234 - message", "OV-1234: message", etc.
        clean_message="$message"
        if [[ "$message" =~ ^$TICKET_PREFIX-[0-9]+[[:space:]]*[-:]*[[:space:]]* ]]; then
            clean_message=$(echo "$message" | sed -E "s/^$TICKET_PREFIX-[0-9]+[[:space:]]*[-:]*[[:space:]]*//" | sed 's/^[[:space:]]*//')
        fi
        
        echo "Commit: $hash"
        echo "Author: $author"
        echo "Date:   $date"
        echo "Ticket(s): $tickets"
        echo "Message: $clean_message"
        echo "---"
    done
}

# Function to show summary statistics
show_summary() {
    local context_desc="$1"
    local ticket_filter="$2"
    local author_filter="$3"
    shift 3
    local git_log_args=("$@")
    
    echo
    echo "Analysis complete!"
    echo "Summary:"
    echo "========"
    
    total_commits=$(git log "${git_log_args[@]}" --oneline | wc -l | tr -d ' ')
    
    # Build filter pipeline for counting
    local filter_pipeline="git log \"${git_log_args[@]}\" --pretty=format:\"%h|%an|%ad|%s\" --date=format:\"$DATE_FORMAT\" | grep -E '$TICKET_PATTERN'"
    
    if [ -n "$ticket_filter" ]; then
        filter_pipeline="$filter_pipeline | grep -i '$ticket_filter'"
    fi
    
    if [ -n "$author_filter" ]; then
        filter_pipeline="$filter_pipeline | grep -i '$author_filter'"
    fi
    
    ticket_commits=$(eval "$filter_pipeline" | wc -l | tr -d ' ')
    
    echo "Total commits $context_desc: $total_commits"
    
    # Build description of filters
    local filter_desc=""
    if [ -n "$ticket_filter" ] && [ -n "$author_filter" ]; then
        filter_desc="with ticket $ticket_filter and author matching '$author_filter'"
    elif [ -n "$ticket_filter" ]; then
        filter_desc="with ticket $ticket_filter"
    elif [ -n "$author_filter" ]; then
        filter_desc="by author matching '$author_filter'"
    else
        filter_desc="with ticket references"
    fi
    
    echo "Commits $filter_desc: $ticket_commits"
    
    # List unique tickets found
    echo
    echo "Unique tickets found:"
    eval "$filter_pipeline" | while IFS='|' read -r hash author date message; do
        echo "$message" | grep -oE "$TICKET_PATTERN"
    done | sort -u | while read ticket; do
        echo "  - $ticket"
    done
}

# Function to analyze commits with given git log arguments
analyze_commits() {
    local context_desc="$1"
    local ticket_filter="$2"
    local author_filter="$3"
    shift 3
    local git_log_args=("$@")
    
    git log "${git_log_args[@]}" --pretty=format:"%h|%an|%ad|%s" --date=format:"$DATE_FORMAT" | display_ticket_commits "$ticket_filter" "$author_filter"
    show_summary "$context_desc" "$ticket_filter" "$author_filter" "${git_log_args[@]}"
}

# Parse arguments using flags
COMMIT_HASH=""
TICKET_FILTER=""
AUTHOR_FILTER=""
DAYS="$DEFAULT_DAYS"

while [[ $# -gt 0 ]]; do
    case $1 in
        -h|--hash)
            COMMIT_HASH="$2"
            shift 2
            ;;
        -t|--ticket)
            TICKET_FILTER="$2"
            shift 2
            ;;
        -a|--author)
            AUTHOR_FILTER="$2"
            shift 2
            ;;
        -d|--days)
            DAYS="$2"
            # Validate that DAYS is a positive integer
            if ! [[ "$DAYS" =~ ^[0-9]+$ ]] || [ "$DAYS" -eq 0 ]; then
                echo "Error: Days must be a positive integer"
                exit 1
            fi
            shift 2
            ;;
        --help)
            echo "Usage: $0 [-h COMMIT_HASH] [-t TICKET] [-a AUTHOR] [-d DAYS]"
            echo "  -h, --hash COMMIT_HASH    Analyze commits up to specified hash"
            echo "  -t, --ticket TICKET       Filter by specific ticket (e.g., $TICKET_EXAMPLE)"
            echo "  -a, --author AUTHOR       Filter by author name/email (case-insensitive)"
            echo "  -d, --days DAYS           Number of days to look back (default: $DEFAULT_DAYS)"
            echo "  --help                    Show this help message"
            echo ""
            echo "Examples:"
            echo "  $0                        # Show all tickets from current branch (last $DEFAULT_DAYS days)"
            echo "  $0 -d 7                   # Show tickets from last 7 days"
            echo "  $0 -h 8b8e791b -d 14      # Show tickets up to commit within last 14 days"
            echo "  $0 -t $TICKET_EXAMPLE            # Show only commits for ticket $TICKET_EXAMPLE"
            echo "  $0 -a john -d 60          # Show commits by 'john' from last 60 days"
            echo "  $0 -h 8b8e791b -t $TICKET_EXAMPLE -a john -d 7 # Combine all filters"
            exit 0
            ;;
        *)
            echo "Unknown option: $1"
            echo "Use --help for usage information"
            exit 1
            ;;
    esac
done

# Handle commit hash validation and analysis
if [ -n "$COMMIT_HASH" ]; then
    CURRENT_BRANCH=$(git rev-parse --abbrev-ref HEAD)
    
    # Validate if the commit hash exists in current branch
    if ! git merge-base --is-ancestor "$COMMIT_HASH" HEAD 2>/dev/null && [ "$COMMIT_HASH" != "$(git rev-parse HEAD 2>/dev/null)" ]; then
        # Commit hash is not in current branch or doesn't exist - return empty
        exit 0
    fi
    
    echo "Analyzing commits reachable from: $COMMIT_HASH (last $DAYS days only)"
    if [ -n "$TICKET_FILTER" ]; then
        echo "Filtering by ticket: $TICKET_FILTER"
    fi
    if [ -n "$AUTHOR_FILTER" ]; then
        echo "Filtering by author: $AUTHOR_FILTER"
    fi
    echo "Time range: last $DAYS days"
    echo "Looking for commits containing ticket references ($TICKET_PREFIX-XXXXX)"
    echo "$SEPARATOR_LINE"
    echo
    
    analyze_commits "reachable from $COMMIT_HASH (last $DAYS days)" "$TICKET_FILTER" "$AUTHOR_FILTER" "$COMMIT_HASH" --since="$DAYS days ago"
    
else
    # Default behavior - last N days from current branch
    CURRENT_BRANCH=$(git rev-parse --abbrev-ref HEAD)
    echo "Analyzing commits from branch: $CURRENT_BRANCH"
    if [ -n "$TICKET_FILTER" ]; then
        echo "Filtering by ticket: $TICKET_FILTER"
    fi
    if [ -n "$AUTHOR_FILTER" ]; then
        echo "Filtering by author: $AUTHOR_FILTER"
    fi
    echo "Time range: last $DAYS days"
    echo "Looking for commits from the last $DAYS days containing ticket references ($TICKET_PREFIX-XXXXX)"
    echo "$SEPARATOR_LINE"
    echo
    
    analyze_commits "in last $DAYS days" "$TICKET_FILTER" "$AUTHOR_FILTER" --since="$DAYS days ago"
fi