mrjoshuak · March 22, 2025 19:18
diff --git a/get_contributors.sh b/get_contributors.sh
 #!/bin/bash

 # get_contributors.sh
 # A script to collect all contributors to a Git repository, quantify their contributions,
 # combine stats from the same person using different names/emails, and fetch GitHub profile information.
 #
 # Created by Joshua Kolden (https://github.com/mrjoshuak)
 #
 # Usage: ./get_contributors.sh [path_to_repo] [github_repo_owner/repo_name] [mapping_file]
 #
 # The optional mapping_file is a text file with lines in the format:
 # name_or_email_pattern|normalized_key
 #
 # Example mapping file:
 # # Lines starting with # are comments
 # John|johndoe
 # [email protected]|johndoe
 # Jane Smith|janesmith

 # Set the repository path and other parameters
 REPO_PATH=${1:-.}
 GITHUB_REPO=${2:-""}
 MAPPING_FILE=${3:-""}

 # Check if the path exists and is a git repository
 if [ ! -d "$REPO_PATH/.git" ]; then
 	echo "Error: $REPO_PATH is not a git repository"
 	exit 1
 fi

 # Change to the repository directory
 cd "$REPO_PATH" || exit 1

 echo "Analyzing contributors for repository: $(basename "$PWD")"
 echo "------------------------------------------------------"

 # Check for GitHub API credentials
 if [ -n "$GITHUB_TOKEN" ]; then
 	echo "GitHub API token found in environment"
 	GITHUB_AUTH="Authorization: token $GITHUB_TOKEN"
 elif [ -n "$GITHUB_API_TOKEN" ]; then
 	echo "GitHub API token found in environment"
 	GITHUB_AUTH="Authorization: token $GITHUB_API_TOKEN"
 else
 	echo "No GitHub API token found in environment. Will use unauthenticated API calls (rate limited)."
 	GITHUB_AUTH=""
 fi

 # If no GitHub repo was provided, try to guess it from the remote URL
 if [ -z "$GITHUB_REPO" ]; then
 	REMOTE_URL=$(git config --get remote.origin.url)
 	if [[ "$REMOTE_URL" == *"github.com"* ]]; then
 		# Extract owner/repo from the URL
 		GITHUB_REPO=$(echo "$REMOTE_URL" | sed -E 's|.*github\.com[:/]([^/]+/[^/]+)(\.git)?|\1|')
 		echo "Detected GitHub repository: $GITHUB_REPO"
 	else
 		echo "No GitHub repository specified and couldn't detect one from remotes."
 		echo "GitHub profile information will not be available."
 	fi
 fi

 # Get all contributors with their commit counts, lines added, and lines deleted
 echo "Collecting contribution data..."

 # Create temporary files
 TEMP_FILE=$(mktemp)
 NORM_FILE=$(mktemp)
 MANUAL_MAP=$(mktemp)
 GITHUB_INFO=$(mktemp)
 GITHUB_USERNAME_MAP=$(mktemp)

 # Create a default manual mapping file if no mapping file is provided
 if [ -z "$MAPPING_FILE" ] || [ ! -f "$MAPPING_FILE" ]; then
 	echo "# No custom mapping file provided. Using default empty mapping." >"$MANUAL_MAP"
 else
 	echo "Using mapping file: $MAPPING_FILE"
 	cp "$MAPPING_FILE" "$MANUAL_MAP"
 fi

 # Create a default GitHub username mapping file
 cat >"$GITHUB_USERNAME_MAP" <<EOF
 # Format: name_or_email_pattern|github_username
 # This file maps contributors to their GitHub usernames
 EOF

 # Get all authors and their emails
 git log --format='%an|%ae' | sort | uniq >/tmp/contributors.txt

 # Process each contributor and normalize names
 while IFS='|' read -r name email; do
 	# Check if this contributor matches any manual mapping
 	manual_key=""
 	while IFS='|' read -r pattern key; do
 		# Skip comments
 		if [[ "$pattern" == \#* ]]; then
 			continue
 		fi

 		# Check if pattern matches name or email
 		if [[ "$name" == *"$pattern"* || "$email" == *"$pattern"* ]]; then
 			manual_key="$key"
 			break
 		fi
 	done <"$MANUAL_MAP"

 	# If no manual mapping, create a normalized key from the name
 	if [[ -z "$manual_key" ]]; then
 		# Create a simple normalized name (lowercase, alphanumeric only)
 		manual_key=$(echo "$name" | tr '[:upper:]' '[:lower:]' | tr -cd 'a-z0-9')
 	fi

 	# Get commit count
 	commits=$(git log --author="$name" --pretty=oneline | wc -l | tr -d ' ')

 	# Get lines added and deleted
 	stats=$(git log --author="$name" --pretty=tformat: --numstat | awk '{ add += $1; del += $2 } END { printf "%d %d", add, del }')
 	added=$(echo "$stats" | cut -d' ' -f1)
 	deleted=$(echo "$stats" | cut -d' ' -f2)

 	# Calculate a "contribution score" (commits + added + deleted)
 	score=$((commits + added + deleted))

 	# Output to the temporary file with normalized name for later combining
 	echo "$manual_key|$score|$commits|$added|$deleted|$name|$email" >>"$NORM_FILE"

 done </tmp/contributors.txt

 # Combine entries with the same normalized key
 cat "$NORM_FILE" | sort -t'|' -k1,1 | awk -F'|' '
 BEGIN { OFS="|" }
 {
 	if ($1 == prev_key) {
 		# Add stats
 		total_score += $2
 		total_commits += $3
 		total_added += $4
 		total_deleted += $5

 		# Keep the longer name
 		if (length($6) > length(prev_name)) {
 			prev_name = $6
 			prev_email = $7
 		}
 	} else {
 		# Output previous combined entry
 		if (prev_key != "") {
 			print total_score, total_commits, total_added, total_deleted, prev_name, prev_email
 		}

 		# Start new entry
 		prev_key = $1
 		prev_name = $6
 		prev_email = $7
 		total_score = $2
 		total_commits = $3
 		total_added = $4
 		total_deleted = $5
 	}
 }
 END {
 	# Output the last entry
 	if (prev_key != "") {
 		print total_score, total_commits, total_added, total_deleted, prev_name, prev_email
 	}
 }' >"$TEMP_FILE"

 # Function to extract GitHub username from email
 extract_github_username() {
 	local email="$1"
 	local username=""

 	# Check if it's a GitHub noreply email
 	if [[ "$email" == *"@users.noreply.github.com"* ]]; then
 		# Extract the username or ID
 		username=$(echo "$email" | sed -E 's/^([0-9]+\+)?([^@]+)@users\.noreply\.github\.com$/\2/')
 	fi

 	echo "$username"
 }

 # Function to get GitHub info for a user
 get_github_info() {
 	local email="$1"
 	local name="$2"
 	local github_username=""
 	local full_name=""
 	local profile_url=""

 	# Check if this user has a manual GitHub username mapping
 	while IFS='|' read -r pattern username; do
 		# Skip comments
 		if [[ "$pattern" == \#* ]]; then
 			continue
 		fi

 		# Check if pattern matches name or email
 		if [[ "$name" == *"$pattern"* || "$email" == *"$pattern"* ]]; then
 			github_username="$username"
 			break
 		fi
 	done <"$GITHUB_USERNAME_MAP"

 	# If no manual mapping, try to extract from email
 	if [ -z "$github_username" ]; then
 		github_username=$(extract_github_username "$email")
 	fi

 	# If we have a GitHub repo and username or email, try to get more info
 	if [ -n "$GITHUB_REPO" ]; then
 		# Try to find the contributor in the GitHub repo contributors list
 		if [ -z "$github_username" ]; then
 			# Search by email
 			if [ -n "$GITHUB_AUTH" ]; then
 				# Use authenticated API call
 				response=$(curl -s -H "$GITHUB_AUTH" "https://api.github.com/search/users?q=$email+in:email")
 				if [ "$(echo "$response" | grep -c "total_count.*[1-9]")" -gt 0 ]; then
 					github_username=$(echo "$response" | grep -o '"login": *"[^"]*"' | head -1 | sed 's/"login": *"\([^"]*\)"/\1/')
 				fi
 			fi

 			# If still no username, try to search by name
 			if [ -z "$github_username" ]; then
 				if [ -n "$GITHUB_AUTH" ]; then
 					response=$(curl -s -H "$GITHUB_AUTH" "https://api.github.com/search/users?q=$name+in:name")
 					if [ "$(echo "$response" | grep -c "total_count.*[1-9]")" -gt 0 ]; then
 						github_username=$(echo "$response" | grep -o '"login": *"[^"]*"' | head -1 | sed 's/"login": *"\([^"]*\)"/\1/')
 					fi
 				fi
 			fi
 		fi

 		# If we found a GitHub username, get their profile info
 		if [ -n "$github_username" ]; then
 			profile_url="https://github.com/$github_username"

 			# Get full name from GitHub profile if we have auth
 			if [ -n "$GITHUB_AUTH" ]; then
 				user_info=$(curl -s -H "$GITHUB_AUTH" "https://api.github.com/users/$github_username")
 				full_name=$(echo "$user_info" | grep -o '"name": *"[^"]*"' | sed 's/"name": *"\([^"]*\)"/\1/')

 				# If no full name, use the username
 				if [ -z "$full_name" ] || [ "$full_name" = "null" ]; then
 					full_name="$github_username"
 				fi
 			else
 				full_name="$github_username"
 			fi
 		fi
 	fi

 	# If we couldn't get GitHub info, use the name from git
 	if [ -z "$full_name" ]; then
 		full_name="$name"
 	fi

 	# Return the info
 	echo "$full_name|$github_username|$profile_url"
 }

 # Get GitHub info for each contributor
 echo "Fetching GitHub profile information..."
 if [ -n "$GITHUB_REPO" ]; then
 	sort -t'|' -k1,1nr "$TEMP_FILE" | while IFS='|' read -r score commits added deleted name email; do
 		# Skip entries with zero score
 		if [[ $score -eq 0 ]]; then
 			continue
 		fi

 		# Get GitHub info
 		github_info=$(get_github_info "$email" "$name")
 		IFS='|' read -r full_name github_username profile_url <<<"$github_info"

 		# Output to GitHub info file
 		echo "$score|$commits|$added|$deleted|$name|$email|$full_name|$github_username|$profile_url" >>"$GITHUB_INFO"
 	done

 	# Display the results with GitHub info
 	echo "Name                            GitHub Username               Profile                         Commits    Added     Deleted   Score"
 	echo "--------------------------------------------------------------------------------------------------------------"
 	cat "$GITHUB_INFO" | while IFS='|' read -r score commits added deleted name email full_name github_username profile_url; do
 		# Use full name if available, otherwise use git name
 		display_name="$full_name"
 		if [ -z "$display_name" ]; then
 			display_name="$name"
 		fi

 		# Use username if available, otherwise blank
 		display_username="$github_username"
 		if [ -z "$display_username" ]; then
 			display_username="-"
 		fi

 		# Use profile URL if available, otherwise blank
 		display_url="$profile_url"
 		if [ -z "$display_url" ]; then
 			display_url="-"
 		fi

 		printf "%-30s %-30s %-30s %-10s %-10s %-10s %-10s\n" "$display_name" "$display_username" "$display_url" "$commits" "$added" "$deleted" "$score"
 	done
 else
 	# Display the results without GitHub info
 	echo "Name                            Email                           Commits    Added     Deleted   Score"
 	echo "--------------------------------------------------------------------------------------------"
 	sort -t'|' -k1,1nr "$TEMP_FILE" | while IFS='|' read -r score commits added deleted name email; do
 		# Skip entries with zero score
 		if [[ $score -eq 0 ]]; then
 			continue
 		fi
 		printf "%-30s %-30s %-10s %-10s %-10s %-10s\n" "$name" "$email" "$commits" "$added" "$deleted" "$score"
 	done
 fi

 # Clean up temporary files
 rm -f /tmp/contributors.txt "$TEMP_FILE" "$NORM_FILE" "$MANUAL_MAP" "$GITHUB_INFO" "$GITHUB_USERNAME_MAP"

 echo "------------------------------------------------------"
 echo "Note: 'Score' is a simple metric calculated as (commits + lines added + lines deleted)"
 echo "This is a basic quantification and doesn't account for the complexity or impact of contributions."
 echo "Contributors with multiple email addresses have been combined where possible."
	#!/bin/bash

	# get_contributors.sh
	# A script to collect all contributors to a Git repository, quantify their contributions,
	# combine stats from the same person using different names/emails, and fetch GitHub profile information.
	#
	# Created by Joshua Kolden (https://github.com/mrjoshuak)
	#
	# Usage: ./get_contributors.sh [path_to_repo] [github_repo_owner/repo_name] [mapping_file]
	#
	# The optional mapping_file is a text file with lines in the format:
	# name_or_email_pattern\|normalized_key
	#
	# Example mapping file:
	# # Lines starting with # are comments
	# John\|johndoe
	# [email protected]\|johndoe
	# Jane Smith\|janesmith

	# Set the repository path and other parameters
	REPO_PATH=${1:-.}
	GITHUB_REPO=${2:-""}
	MAPPING_FILE=${3:-""}

	# Check if the path exists and is a git repository
	if [ ! -d "$REPO_PATH/.git" ]; then
	echo "Error: $REPO_PATH is not a git repository"
	exit 1
	fi

	# Change to the repository directory
	cd "$REPO_PATH" \|\| exit 1

	echo "Analyzing contributors for repository: $(basename "$PWD")"
	echo "------------------------------------------------------"

	# Check for GitHub API credentials
	if [ -n "$GITHUB_TOKEN" ]; then
	echo "GitHub API token found in environment"
	GITHUB_AUTH="Authorization: token $GITHUB_TOKEN"
	elif [ -n "$GITHUB_API_TOKEN" ]; then
	echo "GitHub API token found in environment"
	GITHUB_AUTH="Authorization: token $GITHUB_API_TOKEN"
	else
	echo "No GitHub API token found in environment. Will use unauthenticated API calls (rate limited)."
	GITHUB_AUTH=""
	fi

	# If no GitHub repo was provided, try to guess it from the remote URL
	if [ -z "$GITHUB_REPO" ]; then
	REMOTE_URL=$(git config --get remote.origin.url)
	if [[ "$REMOTE_URL" == "github.com" ]]; then
	# Extract owner/repo from the URL
	GITHUB_REPO=$(echo "$REMOTE_URL" \| sed -E 's\|.*github\.com[:/]([^/]+/[^/]+)(\.git)?\|\1\|')
	echo "Detected GitHub repository: $GITHUB_REPO"
	else
	echo "No GitHub repository specified and couldn't detect one from remotes."
	echo "GitHub profile information will not be available."
	fi
	fi

	# Get all contributors with their commit counts, lines added, and lines deleted
	echo "Collecting contribution data..."

	# Create temporary files
	TEMP_FILE=$(mktemp)
	NORM_FILE=$(mktemp)
	MANUAL_MAP=$(mktemp)
	GITHUB_INFO=$(mktemp)
	GITHUB_USERNAME_MAP=$(mktemp)

	# Create a default manual mapping file if no mapping file is provided
	if [ -z "$MAPPING_FILE" ] \|\| [ ! -f "$MAPPING_FILE" ]; then
	echo "# No custom mapping file provided. Using default empty mapping." >"$MANUAL_MAP"
	else
	echo "Using mapping file: $MAPPING_FILE"
	cp "$MAPPING_FILE" "$MANUAL_MAP"
	fi

	# Create a default GitHub username mapping file
	cat >"$GITHUB_USERNAME_MAP" <<EOF
	# Format: name_or_email_pattern\|github_username
	# This file maps contributors to their GitHub usernames
	EOF

	# Get all authors and their emails
	git log --format='%an\|%ae' \| sort \| uniq >/tmp/contributors.txt

	# Process each contributor and normalize names
	while IFS='\|' read -r name email; do
	# Check if this contributor matches any manual mapping
	manual_key=""
	while IFS='\|' read -r pattern key; do
	# Skip comments
	if [[ "$pattern" == \#* ]]; then
	continue
	fi

	# Check if pattern matches name or email
	if [[ "$name" == "$pattern" \|\| "$email" == "$pattern" ]]; then
	manual_key="$key"
	break
	fi
	done <"$MANUAL_MAP"

	# If no manual mapping, create a normalized key from the name
	if [[ -z "$manual_key" ]]; then
	# Create a simple normalized name (lowercase, alphanumeric only)
	manual_key=$(echo "$name" \| tr '[:upper:]' '[:lower:]' \| tr -cd 'a-z0-9')
	fi

	# Get commit count
	commits=$(git log --author="$name" --pretty=oneline \| wc -l \| tr -d ' ')

	# Get lines added and deleted
	stats=$(git log --author="$name" --pretty=tformat: --numstat \| awk '{ add += $1; del += $2 } END { printf "%d %d", add, del }')
	added=$(echo "$stats" \| cut -d' ' -f1)
	deleted=$(echo "$stats" \| cut -d' ' -f2)

	# Calculate a "contribution score" (commits + added + deleted)
	score=$((commits + added + deleted))

	# Output to the temporary file with normalized name for later combining
	echo "$manual_key\|$score\|$commits\|$added\|$deleted\|$name\|$email" >>"$NORM_FILE"

	done </tmp/contributors.txt

	# Combine entries with the same normalized key
	cat "$NORM_FILE" \| sort -t'\|' -k1,1 \| awk -F'\|' '
	BEGIN { OFS="\|" }
	{
	if ($1 == prev_key) {
	# Add stats
	total_score += $2
	total_commits += $3
	total_added += $4
	total_deleted += $5

	# Keep the longer name
	if (length($6) > length(prev_name)) {
	prev_name = $6
	prev_email = $7
	}
	} else {
	# Output previous combined entry
	if (prev_key != "") {
	print total_score, total_commits, total_added, total_deleted, prev_name, prev_email
	}

	# Start new entry
	prev_key = $1
	prev_name = $6
	prev_email = $7
	total_score = $2
	total_commits = $3
	total_added = $4
	total_deleted = $5
	}
	}
	END {
	# Output the last entry
	if (prev_key != "") {
	print total_score, total_commits, total_added, total_deleted, prev_name, prev_email
	}
	}' >"$TEMP_FILE"

	# Function to extract GitHub username from email
	extract_github_username() {
	local email="$1"
	local username=""

	# Check if it's a GitHub noreply email
	if [[ "$email" == "@users.noreply.github.com" ]]; then
	# Extract the username or ID
	username=$(echo "$email" \| sed -E 's/^([0-9]+\+)?([^@]+)@users\.noreply\.github\.com$/\2/')
	fi

	echo "$username"
	}

	# Function to get GitHub info for a user
	get_github_info() {
	local email="$1"
	local name="$2"
	local github_username=""
	local full_name=""
	local profile_url=""

	# Check if this user has a manual GitHub username mapping
	while IFS='\|' read -r pattern username; do
	# Skip comments
	if [[ "$pattern" == \#* ]]; then
	continue
	fi

	# Check if pattern matches name or email
	if [[ "$name" == "$pattern" \|\| "$email" == "$pattern" ]]; then
	github_username="$username"
	break
	fi
	done <"$GITHUB_USERNAME_MAP"

	# If no manual mapping, try to extract from email
	if [ -z "$github_username" ]; then
	github_username=$(extract_github_username "$email")
	fi

	# If we have a GitHub repo and username or email, try to get more info
	if [ -n "$GITHUB_REPO" ]; then
	# Try to find the contributor in the GitHub repo contributors list
	if [ -z "$github_username" ]; then
	# Search by email
	if [ -n "$GITHUB_AUTH" ]; then
	# Use authenticated API call
	response=$(curl -s -H "$GITHUB_AUTH" "https://api.github.com/search/users?q=$email+in:email")
	if [ "$(echo "$response" \| grep -c "total_count.*[1-9]")" -gt 0 ]; then
	github_username=$(echo "$response" \| grep -o '"login": "[^"]"' \| head -1 \| sed 's/"login": "\([^"]\)"/\1/')
	fi
	fi

	# If still no username, try to search by name
	if [ -z "$github_username" ]; then
	if [ -n "$GITHUB_AUTH" ]; then
	response=$(curl -s -H "$GITHUB_AUTH" "https://api.github.com/search/users?q=$name+in:name")
	if [ "$(echo "$response" \| grep -c "total_count.*[1-9]")" -gt 0 ]; then
	github_username=$(echo "$response" \| grep -o '"login": "[^"]"' \| head -1 \| sed 's/"login": "\([^"]\)"/\1/')
	fi
	fi
	fi
	fi

	# If we found a GitHub username, get their profile info
	if [ -n "$github_username" ]; then
	profile_url="https://github.com/$github_username"

	# Get full name from GitHub profile if we have auth
	if [ -n "$GITHUB_AUTH" ]; then
	user_info=$(curl -s -H "$GITHUB_AUTH" "https://api.github.com/users/$github_username")
	full_name=$(echo "$user_info" \| grep -o '"name": "[^"]"' \| sed 's/"name": "\([^"]\)"/\1/')

	# If no full name, use the username
	if [ -z "$full_name" ] \|\| [ "$full_name" = "null" ]; then
	full_name="$github_username"
	fi
	else
	full_name="$github_username"
	fi
	fi
	fi

	# If we couldn't get GitHub info, use the name from git
	if [ -z "$full_name" ]; then
	full_name="$name"
	fi

	# Return the info
	echo "$full_name\|$github_username\|$profile_url"
	}

	# Get GitHub info for each contributor
	echo "Fetching GitHub profile information..."
	if [ -n "$GITHUB_REPO" ]; then
	sort -t'\|' -k1,1nr "$TEMP_FILE" \| while IFS='\|' read -r score commits added deleted name email; do
	# Skip entries with zero score
	if [[ $score -eq 0 ]]; then
	continue
	fi

	# Get GitHub info
	github_info=$(get_github_info "$email" "$name")
	IFS='\|' read -r full_name github_username profile_url <<<"$github_info"

	# Output to GitHub info file
	echo "$score\|$commits\|$added\|$deleted\|$name\|$email\|$full_name\|$github_username\|$profile_url" >>"$GITHUB_INFO"
	done

	# Display the results with GitHub info
	echo "Name GitHub Username Profile Commits Added Deleted Score"
	echo "--------------------------------------------------------------------------------------------------------------"
	cat "$GITHUB_INFO" \| while IFS='\|' read -r score commits added deleted name email full_name github_username profile_url; do
	# Use full name if available, otherwise use git name
	display_name="$full_name"
	if [ -z "$display_name" ]; then
	display_name="$name"
	fi

	# Use username if available, otherwise blank
	display_username="$github_username"
	if [ -z "$display_username" ]; then
	display_username="-"
	fi

	# Use profile URL if available, otherwise blank
	display_url="$profile_url"
	if [ -z "$display_url" ]; then
	display_url="-"
	fi

	printf "%-30s %-30s %-30s %-10s %-10s %-10s %-10s\n" "$display_name" "$display_username" "$display_url" "$commits" "$added" "$deleted" "$score"
	done
	else
	# Display the results without GitHub info
	echo "Name Email Commits Added Deleted Score"
	echo "--------------------------------------------------------------------------------------------"
	sort -t'\|' -k1,1nr "$TEMP_FILE" \| while IFS='\|' read -r score commits added deleted name email; do
	# Skip entries with zero score
	if [[ $score -eq 0 ]]; then
	continue
	fi
	printf "%-30s %-30s %-10s %-10s %-10s %-10s\n" "$name" "$email" "$commits" "$added" "$deleted" "$score"
	done
	fi

	# Clean up temporary files
	rm -f /tmp/contributors.txt "$TEMP_FILE" "$NORM_FILE" "$MANUAL_MAP" "$GITHUB_INFO" "$GITHUB_USERNAME_MAP"

	echo "------------------------------------------------------"
	echo "Note: 'Score' is a simple metric calculated as (commits + lines added + lines deleted)"
	echo "This is a basic quantification and doesn't account for the complexity or impact of contributions."
	echo "Contributors with multiple email addresses have been combined where possible."