Git History to CSV - Enhanced Branch Detection

git_history_to_csv.sh

#!/bin/bash

#####################################################################
# Git History to CSV - Enhanced Branch Detection
#####################################################################
#
# DESCRIPTION:
#   This script generates a CSV file containing the history of git commits
#   with comprehensive branch information. It tracks which commits belong to
#   which branches by using multiple detection methods, ensuring accurate
#   branch attribution even for commits without direct references.
#
# FEATURES:
#   - Enhanced branch detection (uses both direct refs and branch ancestry)
#   - Supports all branches or specific branch filtering
#   - Date range filtering capability
#   - Multiple sorting options
#   - Detailed branch statistics
#   - Debug mode for troubleshooting
#   - Handles complex commit messages with commas and quotes
#
# USAGE:
#   ./git_history_to_csv.sh [options]
#
# OPTIONS:
#   -o FILE       Specify output file (default: git_history.csv)
#   -a            Include all branches (default)
#   -c            Current branch only
#   -b BRANCHES   Specific branches only (comma-separated list)
#   -s DATE       Include commits since DATE (format: YYYY-MM-DD)
#   -u DATE       Include commits until DATE (format: YYYY-MM-DD)
#   -d ORDER      Sort order: 'default', 'author-date', or 'committer-date'
#   -v            Enable debug output
#   -h            Display help message
#
# EXAMPLES:
#   # Put thiss script in your root project directory
#   # Generate CSV with all branches (default)
#   ./git_history_to_csv.sh
#
#   # Specify output file
#   ./git_history_to_csv.sh -o project_history.csv
#
#   # Only include commits from specific branches
#   ./git_history_to_csv.sh -b main,develop
#
#   # Include commits from the last 30 days
#   ./git_history_to_csv.sh -s "$(date -d '30 days ago' +%Y-%m-%d)"
#
#   # Debug mode with date range and specific output file
#   ./git_history_to_csv.sh -v -s 2023-01-01 -u 2023-12-31 -o year_2023.csv
#
#####################################################################

# Check if git is installed
if ! command -v git &> /dev/null; then
    echo "Error: git is not installed or not in the PATH" >&2
    exit 1
fi

# Check if we're in a git repository
if ! git rev-parse --is-inside-work-tree &> /dev/null; then
    echo "Error: Not a git repository" >&2
    exit 1
fi

# Default output file name
OUTPUT_FILE="git_history.csv"

# Default values for options
ALL_BRANCHES=true
SPECIFIC_BRANCHES=""
SINCE_DATE=""
UNTIL_DATE=""
SORT_ORDER="default"  # Can be 'default', 'author-date', 'committer-date'
DEBUG=false

# Function to display usage information
show_usage() {
  echo "Usage: $0 [options]"
  echo "Options:"
  echo "  -o FILE       Specify output file (default: git_history.csv)"
  echo "  -a            Include all branches (default)"
  echo "  -c            Current branch only"
  echo "  -b BRANCHES   Specific branches only (comma-separated list)"
  echo "  -s DATE       Include commits since DATE (format: YYYY-MM-DD)"
  echo "  -u DATE       Include commits until DATE (format: YYYY-MM-DD)"
  echo "  -d ORDER      Sort order: 'default', 'author-date', or 'committer-date'"
  echo "  -v            Enable debug output"
  echo "  -h            Display this help message"
  exit 1
}

# Parse command line arguments
while getopts ":o:acb:s:u:d:vh" opt; do
  case $opt in
    o)
      OUTPUT_FILE="$OPTARG"
      ;;
    a)
      ALL_BRANCHES=true
      SPECIFIC_BRANCHES=""
      ;;
    c)
      ALL_BRANCHES=false
      SPECIFIC_BRANCHES=""
      ;;
    b)
      ALL_BRANCHES=false
      SPECIFIC_BRANCHES="$OPTARG"
      ;;
    s)
      SINCE_DATE="$OPTARG"
      ;;
    u)
      UNTIL_DATE="$OPTARG"
      ;;
    d)
      SORT_ORDER="$OPTARG"
      ;;
    v)
      DEBUG=true
      ;;
    h)
      show_usage
      ;;
    \?)
      echo "Invalid option: -$OPTARG" >&2
      show_usage
      ;;
    :)
      echo "Option -$OPTARG requires an argument." >&2
      show_usage
      ;;
  esac
done

# Debug function
debug() {
  if [ "$DEBUG" = true ]; then
    echo "[DEBUG] $*" >&2
  fi
}

# Function to get branch information from a commit hash
get_branches_for_commit() {
  local commit_hash="$1"
  local branches=""
  local local_branches=""
  local remote_branches=""

  debug "Getting branches for commit $commit_hash"
  
  # Get all branches containing this commit
  while IFS= read -r branch; do
    # Skip empty lines
    if [ -z "$branch" ]; then
      continue
    fi

    # Trim leading whitespace
    branch="${branch## }"
    
    # Skip the current/HEAD marker (* prefix)
    branch="${branch#* }"

    debug "  Found branch: $branch"
    
    # Separate local and remote branches
    if [[ "$branch" == *"/"* ]] && [[ "$branch" != "HEAD" ]]; then
      remote_branches="$remote_branches,$branch"
    else
      local_branches="$local_branches,$branch"
    fi
  done < <(git branch --all --contains "$commit_hash" 2>/dev/null)

  # Remove leading comma
  local_branches="${local_branches#,}"
  remote_branches="${remote_branches#,}"

  debug "  Local branches: $local_branches"
  debug "  Remote branches: $remote_branches"

  # Prefer local branches over remote branches
  if [ -n "$local_branches" ]; then
    branches="$local_branches"
  elif [ -n "$remote_branches" ]; then
    branches="$remote_branches"
  fi

  echo "$branches"
}

# Function to extract branch from reference
extract_branch_from_refs() {
  local refs="$1"
  local branch="(none)"
  
  debug "Extracting branch from refs: $refs"
  
  # Extract branch name from refs using string operations instead of regex
  if [ -n "$refs" ]; then
    # Try to find "HEAD -> branch" pattern
    if [[ "$refs" == *"HEAD -> "* ]]; then
      # Extract the branch name after "HEAD -> "
      head_branch="${refs#*HEAD -> }"
      # Take everything up to the next comma or end of string
      branch="${head_branch%%,*}"
      branch="${branch%% *}"  # Remove any trailing space
      debug "  Found HEAD -> branch: $branch"
    # Try to find "tag: tagname" pattern
    elif [[ "$refs" == *"tag: "* ]]; then
      # Extract the tag name after "tag: "
      tag_name="${refs#*tag: }"
      # Take everything up to the next comma or end of string
      tag_name="${tag_name%%,*}"
      tag_name="${tag_name%% *}"  # Remove any trailing space
      branch="tag: $tag_name"
      debug "  Found tag: $branch"
    # Check for individual branch names
    elif [[ "$refs" == *","* ]]; then
      # Split by comma and look for local branches first
      local found_local=false
      IFS=',' read -ra REFS <<< "$refs"
      for ref in "${REFS[@]}"; do
        ref="${ref## }"  # Remove leading spaces
        # Skip remotes if we find a local branch
        if [[ "$ref" != *"/"* ]] && [[ "$ref" != "HEAD" ]] && [[ "$ref" != *"tag:"* ]]; then
          branch="$ref"
          found_local=true
          debug "  Found local branch in list: $branch"
          break
        fi
      done
      
      # If no local branch found, use the first remote branch
      if [ "$found_local" = false ]; then
        for ref in "${REFS[@]}"; do
          ref="${ref## }"  # Remove leading spaces
          if [[ "$ref" != "HEAD" ]] && [[ "$ref" != *"tag:"* ]]; then
            branch="$ref"
            debug "  Using remote branch: $branch"
            break
          fi
        done
      fi
    # Take the single ref as fallback
    else
      branch="${refs## }"  # Remove leading space
      branch="${branch%% *}"   # Remove anything after a space
      debug "  Using single ref: $branch"
    fi
  else
    debug "  No refs found in git log output"
  fi
  
  echo "$branch"
}

# Construct the git command with the specified options
GIT_CMD="git log"

# Handle branch selection
if [ "$ALL_BRANCHES" = true ]; then
  GIT_CMD="$GIT_CMD --all"
elif [ -n "$SPECIFIC_BRANCHES" ]; then
  # Convert comma-separated list to space-separated for git
  BRANCH_LIST=$(echo "$SPECIFIC_BRANCHES" | sed 's/,/ /g')
  GIT_CMD="$GIT_CMD $BRANCH_LIST"
fi

# Add date range filters if specified
if [ -n "$SINCE_DATE" ]; then
  GIT_CMD="$GIT_CMD --since=$SINCE_DATE"
fi

if [ -n "$UNTIL_DATE" ]; then
  GIT_CMD="$GIT_CMD --until=$UNTIL_DATE"
fi

# Add sorting options
case "$SORT_ORDER" in
  "author-date")
    GIT_CMD="$GIT_CMD --author-date-order"
    ;;
  "committer-date")
    GIT_CMD="$GIT_CMD --date-order"
    ;;
  *)
    # Default sort order, no additional flag needed
    ;;
esac

# Add headers to the CSV file
echo "Date,Author,Branch,Message,CommitHash" > "$OUTPUT_FILE"

# Display the options being used
echo "Generating CSV with the following options:"
echo "Output file: $OUTPUT_FILE"
if [ "$ALL_BRANCHES" = true ]; then
  echo "Branches: All branches"
elif [ -n "$SPECIFIC_BRANCHES" ]; then
  echo "Branches: $SPECIFIC_BRANCHES"
else
  echo "Branches: Current branch only"
fi
[ -n "$SINCE_DATE" ] && echo "Since date: $SINCE_DATE"
[ -n "$UNTIL_DATE" ] && echo "Until date: $UNTIL_DATE"
echo "Sort order: $SORT_ORDER"
if [ "$DEBUG" = true ]; then
  echo "Debug mode: Enabled"
fi
echo "Processing..."

# Get git log and format it as CSV, now including the commit hash (%H)
$GIT_CMD --pretty=format:'%ad,%an,%D,\"%s\",%H' --date=short | 
while IFS= read -r line; do
  # Parse the line properly, handling quoted message field
  date=$(echo "$line" | cut -d, -f1)
  author=$(echo "$line" | cut -d, -f2)
  refs=$(echo "$line" | cut -d, -f3)
  
  # Extract the commit hash (last field)
  commit_hash=$(echo "$line" | awk -F, '{print $NF}')
  
  # Get the message (everything between the 4th field and the last field)
  message_with_commas=$(echo "$line" | awk -F, '{
    # Remove the first three fields and the last field
    for(i=4; i<NF; i++) {
      printf "%s", $i
      if(i<NF-1) printf ","
    }
    print ""
  }')
  
  debug "Processing commit $commit_hash from $date by $author"
  
  # Extract branch from refs first
  branch=$(extract_branch_from_refs "$refs")
  
  # If no branch found from refs (%D) or branch is "(none)", try fallback method
  if [ "$branch" = "(none)" ] || [ -z "$branch" ]; then
    debug "No branch info from refs, trying git branch --contains fallback"
    branches_from_git=$(get_branches_for_commit "$commit_hash")
    
    if [ -n "$branches_from_git" ]; then
      branch="$branches_from_git"
      debug "Found branches using fallback: $branch"
    else
      debug "Fallback also returned no branches"
    fi
  fi
  
  # Print the processed line, escaping any commas in the message field
  # Replace commas inside message with escaped commas
  message_escaped="${message_with_commas//,/\\,}"
  
  debug "Final branch assignment: $branch"
  echo "$date,$author,$branch,$message_escaped,$commit_hash" 
done >> "$OUTPUT_FILE"

# Check if the command was successful
if [ $? -eq 0 ]; then
    echo "Git history successfully exported to $OUTPUT_FILE"
else
    echo "Error: Failed to export git history" >&2
    exit 1
fi

# Count the number of commits exported
COMMIT_COUNT=$(wc -l < "$OUTPUT_FILE")
COMMIT_COUNT=$((COMMIT_COUNT - 1))  # Subtract 1 for the header line
echo "Total commits exported: $COMMIT_COUNT"

# List branch statistics
if [ "$ALL_BRANCHES" = true ] || [ -n "$SPECIFIC_BRANCHES" ]; then
  echo ""
  echo "Commit count by branch:"
  tail -n +2 "$OUTPUT_FILE" | cut -d, -f3 | sort | uniq -c | sort -nr | 
  while read -r count branch; do
    branch="${branch#\"}"  # Remove leading quote if present
    branch="${branch%\"}"  # Remove trailing quote if present
    printf "  %-20s %s\n" "$branch:" "$count commits"
  done
fi

exit 0