#!/bin/bash

# Function to get the list of extensions from the arguments passed to the script
get_extensions() {
  # Check for input:
  # If the user provides any arguments, store those as the list of file extensions to look for.
  # If no arguments are given, leave extensions array empty to let the regex pattern search for everything.
  if [ $# -gt 0 ]; then
    EXTENSIONS=()
    for var in $@
    do
      # Strip the . from the given extension if user gives the extension with a period (like .md)
      var=${var/"."/}
      EXTENSIONS+=("${var}")
    done
  else
    EXTENSIONS=()
  fi
}

# Function to get the current date in YYYY-MM-DD format
get_current_date() {
  date +%F
}

# Function to ignore files based on .gitignore
is_ignored() {
  local filepath="$1"
  git check-ignore -q "$filepath"  # Returns 0 if ignored, 1 otherwise
  return $?
}

# Function to process a single file
process_file() {
  local filepath="$1"
  local header="# $filepath"
  {
    echo "$header"
    echo ""  # Print a blank line
    cat "$filepath"
    echo ""  # Print a blank line after file content
  } >> "$output_file"
}

# Function to traverse directories recursively
traverse_directory() {
  local dir="$1"

  # The regex pattern starts with '.*', which searches for any matching characters
  # '.' is a wildcard, while '*' means match as many occurrences of the preceding char
  # '\.' escapes the '.' special char to search for the actual char
  # () is a capturing group. The | character is an OR operator
  # '$' indicates the preceding string should be followed by the end of line
  if [ ${#EXTENSIONS[@]} -gt 0 ]; then
    EXTENSIONS_PATTERN=$(printf '\\|%s' "${EXTENSIONS[@]}")
    pattern=".*\.\($EXTENSIONS_PATTERN\)$"
  else
    pattern=".*$"
  fi

  # Loop through each item in the directory
  for entry in "$dir"/*; do
    # Check if entry exists (handles empty directories)
    if [[ -e "$entry" ]]; then
      if [[ -d "$entry" ]]; then
        # Check if the directory is .git
        if [[ "$(basename "$entry")" != ".git" ]]; then
          # Recursively traverse the directory
          traverse_directory "$entry"
        fi
      elif [[ -f "$entry" ]]; then
        # Ignore .gitignore files and files in .gitignore
        if [[ "$(basename "$entry")" != ".gitignore" ]] && 
          ! is_ignored "$entry" && 
          [[ "$entry" =~ $pattern ]]; then # Check against regex pattern
          process_file "$entry"
        fi
      fi
    fi
  done
}


# Main function
main() {
  get_extensions

  output_file="output-$(get_current_date).txt"

  # Clear previous output file (if exists)
  > "$output_file"

  # Start traversing from the current directory
  traverse_directory "."

  echo "Consolidation complete! Output saved in: $output_file"
}

# Start the process.
echo "Consolidating..."

# Execute main function
main

# Complete the process:
# Inform the user that the operation is complete and let them know where the consolidated output has been saved.
echo "Consolidation complete! See your output file at: "