barg-parser/barg

#!/usr/bin/env bash

# Bash Argument Parser (BARG)
# Requires Bash 4.3 or higher for associative arrays

# Check Bash version
if [ "${BASH_VERSINFO:-0}" -lt 4 ]; then
  echo "Error: This script requires Bash 4.0 or higher (current: $BASH_VERSION)" >&2
  exit 1
fi

# Main parsing function - all logic contained here
parse_arguments() {
  local -a spec_input=("$@")
  local input="${spec_input[-1]}"
  unset 'spec_input[-1]'

  # Local storage (no globals!)
  local -A schema
  local -A aliases
  local -a tokens

  # Tokenize input into tagged tokens (format: tag::value)
  tokenize() {
    local input_str="$1"
    local -a args
    tokens=()

    # Split input into array
    read -ra args <<<"$input_str"

    # Root command
    local root="${args[0]}"
    tokens+=("root::${root}")

    local i=1
    while [[ $i -lt ${#args[@]} ]]; do
      local token="${args[$i]}"

      # Rest arguments (everything after --)
      if [[ "$token" == "--" ]]; then
        ((i++))
        local rest_args=""
        while [[ $i -lt ${#args[@]} ]]; do
          if [[ -n "$rest_args" ]]; then
            rest_args="${rest_args} ${args[$i]}"
          else
            rest_args="${args[$i]}"
          fi
          ((i++))
        done
        if [[ -n "$rest_args" ]]; then
          tokens+=("rest::${rest_args}")
        fi
        break
      fi

      # Long option: --key or --key=value
      if [[ "$token" =~ ^--(.+)$ ]]; then
        local full="${BASH_REMATCH[1]}"
        if [[ "$full" =~ ^([^=]+)=(.*)$ ]]; then
          tokens+=("narg::${BASH_REMATCH[1]}")
          tokens+=("value::${BASH_REMATCH[2]}")
        else
          tokens+=("narg::${full}")
        fi
        ((i++))
        continue
      fi

      # Short option: -k or -abc or -k=value
      if [[ "$token" =~ ^-(.+)$ ]]; then
        local full="${BASH_REMATCH[1]}"
        if [[ "$full" =~ ^([^=]+)=(.*)$ ]]; then
          local flags="${BASH_REMATCH[1]}"
          local value="${BASH_REMATCH[2]}"
          # Split flags into individual characters
          for ((j = 0; j < ${#flags}; j++)); do
            tokens+=("narg::${flags:$j:1}")
          done
          tokens+=("value::${value}")
        else
          # Split all flags
          for ((j = 0; j < ${#full}; j++)); do
            tokens+=("narg::${full:$j:1}")
          done
        fi
        ((i++))
        continue
      fi

      # Unknown: could be command, positional, or value
      tokens+=("unk::${token}")
      ((i++))
    done
  }

  # Generate schema from spec
  generate_schema() {
    local -a spec_array=("$@")
    local entry_id=100
    local level=-1
    local -a cmd_stack=()
    local -a pos_stack=()

    for entry in "${spec_array[@]}"; do
      # Split entry by semicolons
      IFS=';' read -ra elements <<<"$entry"
      local entry_type="${elements[0]}"

      case "$entry_type" in
      command)
        local id="$entry_id"
        local aliases_str="${elements[1]}"
        local help_text="${elements[2]}"

        # Schema entries
        schema["${id}.entryType"]="command"
        schema["${id}.name"]="${aliases_str%%,*}" # First alias
        schema["${id}.help"]="$help_text"
        schema["${id}.args"]=""

        # Register aliases
        if [[ ${#cmd_stack[@]} -eq 0 ]]; then
          aliases["cmd::root"]="$id"
        else
          IFS=',' read -ra alias_list <<<"$aliases_str"
          for alias in "${alias_list[@]}"; do
            local parent="${cmd_stack[-1]}"
            aliases["${parent}.cmd::${alias}"]="$id"
          done
        fi

        # Push to stacks
        ((level++))
        cmd_stack+=("$id")
        pos_stack+=(0)
        ((entry_id++))
        ;;

      end)
        ((level--))
        unset 'cmd_stack[-1]'
        unset 'pos_stack[-1]'
        ;;

      argument)
        local id="$entry_id"
        local aliases_str="${elements[1]}"
        local arg_name="${aliases_str%%,*}"

        # Default attributes
        schema["${id}.entryType"]="argument"
        schema["${id}.name"]="$arg_name"
        schema["${id}.dest"]="$arg_name"
        schema["${id}.required"]="false"
        schema["${id}.type"]="positional"

        # Parse additional attributes
        for ((i = 2; i < ${#elements[@]}; i++)); do
          local element="${elements[$i]}"
          if [[ "$element" =~ ^([^:]+):(.*)$ ]]; then
            local attr_key="${BASH_REMATCH[1]}"
            local attr_value="${BASH_REMATCH[2]}"
            if [[ "$attr_key" == "default" ]]; then
              schema["${id}.value"]="$attr_value"
            else
              schema["${id}.${attr_key}"]="$attr_value"
            fi
          elif [[ "$element" =~ ^([^:]+)$ ]]; then
            local attr_key="${BASH_REMATCH[1]}"
            schema["${id}.${attr_key}"]="true"
          fi
        done

        # Add to parent command's args list
        local parent="${cmd_stack[-1]}"
        local cmd_args="${schema["${parent}.args"]}"
        if [[ -n "$cmd_args" ]]; then
          schema["${parent}.args"]="${cmd_args},${id}"
        else
          schema["${parent}.args"]="$id"
        fi

        # Register aliases based on type
        local arg_type="${schema["${id}.type"]}"
        case "$arg_type" in
        positional)
          local pos="${pos_stack[-1]}"
          aliases["${parent}.pos::${pos}"]="$id"
          pos_stack[-1]=$((pos + 1))
          ;;
        rest)
          aliases["${parent}.rest"]="$id"
          ;;
        option | flag)
          IFS=',' read -ra alias_list <<<"$aliases_str"
          for alias in "${alias_list[@]}"; do
            aliases["${parent}.narg::${alias}"]="$id"
          done
          ;;
        esac

        ((entry_id++))
        ;;

      *)
        echo "Error: Invalid entry type: $entry_type" >&2
        return 1
        ;;
      esac
    done
  }

  # Parse tokens based on schema
  process_tokens() {
    local -a cmd_stack=()
    local -a pos_stack=()
    local token_idx=0

    # Process each token
    while [[ $token_idx -lt ${#tokens[@]} ]]; do
      local tagged_token="${tokens[$token_idx]}"
      local token_tag="${tagged_token%%::*}"
      local token="${tagged_token#*::}"

      local found=false
      local level

      # Special handling for root - must be processed first
      if [[ "$token_tag" == "root" ]]; then
        local root_id="${aliases["cmd::root"]}"
        if [[ -n "$root_id" ]]; then
          schema["${root_id}.name"]="$token"
          cmd_stack+=("$root_id")
          pos_stack+=(0)
          found=true
        fi
        ((token_idx++))
        continue
      fi

      # Bubble up: try matching from current level to root
      for ((level = ${#cmd_stack[@]} - 1; level >= 0; level--)); do
        local entry_id=""
        local current_cmd="${cmd_stack[$level]}"

        case "$token_tag" in
        rest)
          entry_id="${aliases["${current_cmd}.rest"]}"
          ;;
        narg)
          entry_id="${aliases["${current_cmd}.narg::${token}"]}"
          ;;
        unk)
          # Try as command
          entry_id="${aliases["${current_cmd}.cmd::${token}"]}"

          # Try as positional (only on last level)
          if [[ -z "$entry_id" && $level -eq $((${#cmd_stack[@]} - 1)) ]]; then
            local pos="${pos_stack[$level]}"
            entry_id="${aliases["${current_cmd}.pos::${pos}"]}"
            if [[ -n "$entry_id" ]]; then
              pos_stack[$level]=$((pos + 1))
            fi
          fi
          ;;
        esac

        # No match at this level, try parent
        if [[ -z "$entry_id" ]]; then
          continue
        fi

        # Found match - process it
        local entry_type="${schema["${entry_id}.entryType"]}"

        case "$entry_type" in
        command)
          cmd_stack+=("$entry_id")
          pos_stack+=(0)
          ;;

        argument)
          local arg_type="${schema["${entry_id}.type"]}"
          local value=""

          case "$arg_type" in
          rest | positional)
            value="$token"
            ;;
          flag)
            value="true"
            ;;
          option)
            # Next token should be the value
            local next_idx=$((token_idx + 1))

            if [[ $next_idx -ge ${#tokens[@]} ]]; then
              echo "Error: No value provided for option: $token" >&2
              return 1
            fi

            local next_tagged="${tokens[$next_idx]}"
            local next_tag="${next_tagged%%::*}"
            local next_token="${next_tagged#*::}"

            if [[ "$next_tag" != "value" && "$next_tag" != "unk" ]]; then
              echo "Error: Expected value for option, got: $next_token ($next_tag)" >&2
              return 1
            fi

            value="$next_token"
            # Skip next token
            ((token_idx++))
            ;;
          esac

          schema["${entry_id}.value"]="$value"
          ;;
        esac

        found=true
        break
      done

      if [[ "$found" == "false" ]]; then
        echo "Warning: Invalid argument: \"$token\" ($token_tag). Skipping..." >&2
      fi

      ((token_idx++))
    done

    # Extract and output results
    local -a commands=()
    local -A values

    for cmd_id in "${cmd_stack[@]}"; do
      local cmd_name="${schema["${cmd_id}.name"]}"
      local cmd_args="${schema["${cmd_id}.args"]}"
      commands+=("$cmd_name")

      if [[ -z "$cmd_args" ]]; then
        continue
      fi

      IFS=',' read -ra arg_ids <<<"$cmd_args"
      for arg_id in "${arg_ids[@]}"; do
        local name="${schema["${arg_id}.name"]}"
        local dest="${schema["${arg_id}.dest"]}"
        local value="${schema["${arg_id}.value"]}"
        local required="${schema["${arg_id}.required"]}"

        # Check required
        if [[ -z "$value" && "$required" == "true" ]]; then
          echo "Error: Argument \"$name\" is required in command \"$cmd_name\"" >&2
          return 1
        fi

        # Handle duplicate destinations
        if [[ -n "${values[$dest]}" ]]; then
          values["${cmd_name}_${dest}"]="$value"
        else
          values["$dest"]="$value"
        fi
      done
    done

    # Output results
    echo "Commands: ${commands[*]}"
    echo "Values:"
    for key in "${!values[@]}"; do
      echo "  ${key}=${values[$key]}"
    done
  }

  # Execute the pipeline
  tokenize "$input"
  generate_schema "${spec_input[@]}" || return 1
  process_tokens
}

# Wrapper to handle errors gracefully
parse_input_wrap() {
  parse_arguments "$@" 2>&1
}

# Example spec
SPEC=(
  "command;barg;Barg - Bash Argument Parser"
  "argument;global;type:flag"
  "command;build;Build a dev container"
  "argument;from;type:option;dest:fromName;help:Name of the base container"
  "argument;name;type:positional;required;help:Name of the container"
  "argument;nametwo;help:Name of the container"
  "argument;image,i;type:option;required;dest:imageName;help:Base image"
  "argument;include,I;type:option;repeatable;dest:includePaths;help:Include paths"
  "argument;verbose,v;type:flag;repeatable;default:0;help:Increase verbosity (-v, -vv, ...)"
  "argument;quiet,q;type:flag;default:false;help:Suppress all output"
  "argument;cmd;type:rest;help:Command and args to run inside the container"
  "argument;name;dest:nameOpt;type:option;help:Building container name"
  "command;container;Container building"
  "argument;dev;type:flag;help:Is dev or not"
  "argument;name;required;help:Building container name"
  "argument;cmd;type:rest;dest:cmdRest;help:Command and args to run inside the container"
  "end"
  "end"
  "command;stop;Stop a dev container"
  "argument;name;type:positional;required;help:Name of the container"
  "argument;kill,k;type:flag;default:false;help:Force kill the container"
  "end"
  "end"
)

# Test inputs
INPUTS=(
  "dev build -i myimage buildtwo --from base-dev --image node:18 --include src --include test -v -v -q mycontainer -- npm run start"
  "dev build buildname container -i myimage --dev --name webapp externalContainer"
  "dev --global build --image debian mybox container --dev --name webapp externalContainer -- echo hi"
  "dev stop mycontainer --kill"
  "dev stop mycontainer --kill -- not specified one"
  "dev build --from alpine --image node:20 mybox -- echo hi"
  "dev build --aaaaa --from alpine --image node:20 --bbbbb orb mybox ccccc -- echo hi"
)

# Run tests
for input in "${INPUTS[@]}"; do
  echo ""
  echo "------------------------"
  echo "> $input"
  parse_input_wrap "${SPEC[@]}" "$input"
done