Get a list of repositories a user made commits to during a given date range. Make sure the script is executable, update GH_USERNAME and GH_TOKEN. Use `--help` for argument info.

gh_repo_search.sh

#!/bin/bash

# Get a list of repositories the defined user made commits to in the passed in date range
# Example: ./gh_repo_search.sh 2018-07-01 2018-08-01

# GitHub username and personal access token
GH_USERNAME="gh_username"
GH_TOKEN="***************************************"

# Take 2 passed in dates in yyyy-mm-dd format
START_DATE=""
END_DATE=""
DEBUG=false # Default is no debug output

# Repo Default Parameters
PAGE=1

# Array to store the repositories the user has made commits to in the date range
COMMITTED_REPOS=()

# Store list of repo users and names to ignore
IGNORE_PATTERNS=()

# Display help information
display_help() {
  echo "Usage: $0 [OPTIONS] START_DATE END_DATE"
  echo
  echo "This script retrieves a list of repositories the defined user made commits to during the passed-in date range."
  echo
  echo "Arguments:"
  echo "  START_DATE    The start date for the commit search (format: yyyy-mm-dd)"
  echo "  END_DATE      The end date for the commit search (format: yyyy-mm-dd)"
  echo
  echo "Options:"
  echo "  -d, --debug          Enable debug output"
  echo "  --ignore=PATTERNS    Comma-separated list of repository patterns to ignore, this accepts * as a wildcard and "
  echo "                       (e.g., --ignore=\"username/repo-*, anotheruser/repo\")"
  echo "  -h, --help           Display this help message and exit"
  echo
  echo "Examples:"
  echo "  $0 2021-12-01 2022-06-01 --ignore=\"username/repo-*, anotheruser/repo, user/*\""
  echo "  $0 2021-12-01 2022-06-01 -d"
  echo
  exit 0
}

# Make sure jq is installed
if ! command -v jq &>/dev/null; then
  echo "jq is not installed. Please install it to use this script." >&2
  exit 1
fi

# Parse command-line arguments
while [[ "$#" -gt 0 ]]; do
  case $1 in
  -d | --debug)
    DEBUG=true
    ;;
  --ignore=*)
    # Split the patterns by comma and trim any surrounding whitespace
    IFS=',' read -ra patterns <<<"${1#*=}"
    for pattern in "${patterns[@]}"; do
      pattern=$(echo "$pattern" | sed 's/^[[:space:]]*//;s/[[:space:]]*$//')
      IGNORE_PATTERNS+=("$pattern")
    done
    ;;
  --ignore)
    shift
    IFS=',' read -ra patterns <<<"$1"
    for pattern in "${patterns[@]}"; do
      pattern=$(echo "$pattern" | sed 's/^[[:space:]]*//;s/[[:space:]]*$//')
      IGNORE_PATTERNS+=("$pattern")
    done
    ;;
  -h | --help)
    display_help
    ;;
  *)
    if [ -z "$START_DATE" ]; then
      START_DATE="$1"
    elif [ -z "$END_DATE" ]; then
      END_DATE="$1"
    fi
    ;;
  esac
  shift
done

# Debugging statement to ensure the arguments are parsed correctly
if [ "$DEBUG" = true ]; then
  echo "Arguments parsed. Start Date: $START_DATE, End Date: $END_DATE, Ignore Patterns: ${IGNORE_PATTERNS[@]}" >&2
fi

# Make sure the date arguments are passed in
if [ -z "$START_DATE" ] || [ -z "$END_DATE" ]; then
  echo "Please provide a start date and end date in the format yyyy-mm-dd" >&2
  exit 1
fi

# Get list of GitHub repos belonging to the defined user
get_repos() {
  local page=$1
  local request_url="https://api.github.com/user/repos?per_page=100&page=${page}"

  # Debugging statement to ensure the URL is correct
  if [ "$DEBUG" = true ]; then
    echo "get_repos request_url: ${request_url}" >&2
  fi

  local response=$(curl -s -u "${GH_USERNAME}:${GH_TOKEN}" "${request_url}")

  # Check if response is valid JSON before proceeding
  if echo "$response" | jq . >/dev/null 2>&1; then
    # Parse the response with jq to return the list of repository full_names
    echo "$response" | jq -r '.[].full_name'
  else
    echo "Error: Unable to parse JSON response from GitHub API"
    echo "$response"
    exit 1
  fi
}

# Get the commits for a given repo using the passed in date range
get_commits() {
  local repo=$1
  local start_date=$2
  local end_date=$3
  local request_url="https://api.github.com/repos/${repo}/commits?author=${GH_USERNAME}&since=${start_date}T00:00:00Z&until=${end_date}T23:59:59Z"

  local response=$(curl -s -u "${GH_USERNAME}:${GH_TOKEN}" "${request_url}")

  # Handle potential errors or empty repositories
  if echo "$response" | jq . >/dev/null 2>&1; then
    if echo "$response" | jq 'has("message") and .message == "Git Repository is empty."' >/dev/null 2>&1; then
      if [ "$DEBUG" = true ]; then
        echo "Repository is empty: $repo" >&2
      fi
      echo "[]" # Return an empty array
    else
      echo "$response"
    fi
  else
    echo "Error: Unable to parse JSON response for commits"
    echo "$response"
    exit 1
  fi
}

# Check if a repository should be ignored
should_ignore_repo() {
  local repo_name=$1

  for pattern in "${IGNORE_PATTERNS[@]}"; do
    case "$repo_name" in
    $pattern)
      return 0 # Repo should be ignored
      ;;
    esac
  done
  return 1 # Repo should NOT be ignored
}

# Loop through each repo and get the commits for the passed in date range
while true; do
  # Get list of user's repos
  REPOS=$(get_repos $PAGE)

  # If no repos are returned, break the loop
  if [ -z "$REPOS" ]; then
    [ "$DEBUG" = true ] && echo "No repositories returned or reached end of list."
    break
  fi

  # Convert REPOS to an array to avoid subshell issues
  IFS=$'\n' read -rd '' -a repo_array <<<"$REPOS"

  # Loop through each repo in the array
  for REPO in "${repo_array[@]}"; do
    [ "$DEBUG" = true ] && echo "Processing repository: $REPO"

    # Skip the repo if it matches any ignore pattern
    if should_ignore_repo "$REPO"; then
      [ "$DEBUG" = true ] && echo "Ignoring repository: $REPO"
      continue
    fi

    # Get the commits made by the user to the repo during the passed in date range
    COMMITS=$(get_commits "$REPO" "$START_DATE" "$END_DATE")

    # Ensure COMMITS is valid JSON and not an error message
    if echo "$COMMITS" | jq . >/dev/null 2>&1; then
      commit_count=$(echo "$COMMITS" | jq length)
      [ "$DEBUG" = true ] && echo "Commit count for $REPO: $commit_count"

      if [ "$commit_count" -gt 0 ]; then
        [ "$DEBUG" = true ] && echo "Adding $REPO to COMMITTED_REPOS"
        # Add the repo full_name to the list of arrays
        #COMMITTED_REPOS+=("$REPO")
        COMMITTED_REPOS+=("$REPO | $commit_count")
      else
        [ "$DEBUG" = true ] && echo "No commits found for $REPO in the specified timeframe."
      fi
    else
      echo "Error: Unable to parse JSON response for commits"
      echo "$COMMITS"
      exit 1
    fi
  done

  # If fewer than 100 repos are returned, break out of the loop
  if [ "${#repo_array[@]}" -lt 100 ]; then
    break
  fi

  # Bump the page number to get the next group of 100 repos for the user
  PAGE=$((PAGE + 1))
done

# Echo the list of repositories the user has made commits to in the date range
[ "$DEBUG" = true ] && echo "Final COMMITTED_REPOS content: ${COMMITTED_REPOS[@]}"

if [ ${#COMMITTED_REPOS[@]} -gt 0 ]; then
  echo "Repositories ${GH_USERNAME} committed to:"
  for REPO in "${COMMITTED_REPOS[@]}"; do
    echo "${REPO}"
  done
else
  echo "No commits found in the specified timeframe."
fi