xpash - Interactive XPath Shell

xpash.rb

#!/usr/bin/ruby
# -*- coding: utf-8 -*-
#
# xpash - Interactive XPath Shell

require 'readline'
require 'rexml/document'

include REXML

prompt = '> '

def help
  print <<-EOF
Usage: xpash [XMLFILE]

Commands:
- load FILE
  load XML file.
- parse STRING
  parse STRING as XML.
- QUERY, $VAR/QUERY
  query xpath and show result.
- VAR = QUERY
  and assign to VAR.
- ?, h, help
  show help.
- ^D, q, quit, exit
  quit shell.
  EOF
end

def dump_text(e)
  str = ''
  e.each_child do |c|
    case c.node_type
    when :text
      str += c.value
    when :element, :document
      str += dump_text(c)
    end
  end
  str
end

class XPashRunner

  def initialize(file = nil)
    @doc = nil
    @vars = {}
    load(file) if file
  end

  def load(file)
    $stderr.print "Loading [#{file}]..."
    begin
      @doc = Document.new(File.new file)
      $stderr.puts "done."
    rescue => e
      $stderr.puts e, "Failed."
    end
  end

  def parse(str)
    begin
      @doc = Document.new(str)
      $stderr.puts "Parsed: #{@doc}"
    rescue => e
      $stderr.puts e, "Failed."
    end
  end

  def run(line)
    if /\A([A-Za-z_]\w*)\s*=\s*(.*)\z/ =~ line
      var = $1
      expr = $2
    else
      var = nil
      expr = line
    end
    case expr
    when 'q', 'quit', 'exit'
      puts "Quit."
      exit
    when '?', 'h', 'help'
      help
    when /load\s+(.+)/
      load($1)
    when /parse\s+(.+)/
      parse($1)
    else
      if @doc.nil?
        $stderr.puts "Error: Any XML files are not loaded."
        break
      end
      result = query(expr)
      if var
        @vars[var] = result
        $stderr.puts "Assigned to '#{var}'"
      end
    end
  end

  def query(expr)
    if /\A\$([A-Za-z_]\w*)(?:\[(\d+)\])?(?:\/(.*))?\z/ =~ expr
      var = $1
      index = $2
      expr = $3
      if [email protected]_key?(var)
        $stderr.puts "Error: Missing variable '#{var}'"
        return
      end
      d = @vars[var]
      if index
        if d.instance_of? Array
          d = d[index.to_i - 1]
        else
          $stderr.puts "Error: '#{var}' is not array."
          return
        end
        $stderr.puts "Element from '#{var}[#{index}]'"
      else
        if !d.respond_to?(:elements)
          $stderr.puts "Error: '#{var}' is not element. (#{d.class})"
          return
        end
        $stderr.puts "Element from '#{var}'"
      end
    else
      d = @doc
    end
    text = false
    attr = nil
    case expr
    when /\A(?:(.*)\/)?text\(\)\z/
      expr = $1 || '.'
      text = true
      $stderr.puts "Query: #{expr}/text()"
    when /\A(?:(.*)\/)?@(\w+)\z/
      expr = $1 || '.'
      attr = $2
      $stderr.puts "Query: #{expr}/@#{attr}"
    else
      expr ||= '*'
      $stderr.puts "Query: #{expr}"
    end
    i = 0
    result = []
    begin
      d.elements.each(expr) do |e|
        case
        when text
          next unless e.has_text?
          value = dump_text(e)
        when attr
          next unless e.attributes.has_key?(attr)
          value = e.attributes[attr]
        else
          value = e
        end
        i += 1
        $stderr.print "#{i}: "
        puts value
        result.push(value)
      end
    rescue => e
      $stderr.puts "Error: #{e}"
      return nil
    end
    case result.length
    when 0
      $stderr.puts "Empty."
      nil
    when 1
      result[0]
    else
      result
    end
  end

end

$stdout.sync = true

runner = XPashRunner.new(*ARGV)

while line = Readline.readline(prompt, true)
  line.gsub!(/\A\s+|\s+\z/, '') # trim blank
  runner.run(line) unless line.empty?
end