htree 0.7.0

data.tar.gz.sig

@@ -0,0 +1,4 @@
+gS����B��ߦGgLAn���۱�� ���R�����nq�9繯�d�%���mZ��uo�ݷd�b��=j��4�6H����>�\��ٺ��9s���'�~�R�fh� 	����R%�\;�k
+��[����
+f��Z�<���lQ4XD�,�y�桾��ʛ���5��������g�Y8�^�<�_l�GknmL�-c�!�=�θ��4���:b4%u�L�N�8G�<0���c�
+������w��kXB�

data/Makefile

@@ -0,0 +1,20 @@
+RUBY=ruby
+
+all: README rdoc/index.html
+
+README: misc/README.erb
+	erb misc/README.erb > README
+
+check test:
+	$(RUBY) -I. test-all.rb
+
+install:
+	$(RUBY) install.rb
+
+.PHONY: check test all install
+
+RB = htree.rb htree/modules.rb $(wildcard htree/[a-l]*.rb) $(wildcard htree/[n-z]*.rb)
+rdoc/index.html: $(RB)
+	rm -rf doc
+	rdoc --op rdoc $(RB)
+

data/Manifest

@@ -0,0 +1,58 @@
+Makefile
+README
+Rakefile
+init.rb
+install.rb
+lib/htree.rb
+lib/htree/container.rb
+lib/htree/context.rb
+lib/htree/display.rb
+lib/htree/doc.rb
+lib/htree/elem.rb
+lib/htree/encoder.rb
+lib/htree/equality.rb
+lib/htree/extract_text.rb
+lib/htree/fstr.rb
+lib/htree/gencode.rb
+lib/htree/htmlinfo.rb
+lib/htree/inspect.rb
+lib/htree/leaf.rb
+lib/htree/loc.rb
+lib/htree/modules.rb
+lib/htree/name.rb
+lib/htree/output.rb
+lib/htree/parse.rb
+lib/htree/raw_string.rb
+lib/htree/regexp-util.rb
+lib/htree/rexml.rb
+lib/htree/scan.rb
+lib/htree/tag.rb
+lib/htree/template.rb
+lib/htree/text.rb
+lib/htree/traverse.rb
+test-all.rb
+test/assign.html
+test/template.html
+test/test-attr.rb
+test/test-charset.rb
+test/test-context.rb
+test/test-display_xml.rb
+test/test-elem-new.rb
+test/test-encoder.rb
+test/test-equality.rb
+test/test-extract_text.rb
+test/test-gencode.rb
+test/test-leaf.rb
+test/test-loc.rb
+test/test-namespace.rb
+test/test-output.rb
+test/test-parse.rb
+test/test-raw_string.rb
+test/test-rexml.rb
+test/test-scan.rb
+test/test-security.rb
+test/test-subnode.rb
+test/test-template.rb
+test/test-text.rb
+test/test-traverse.rb
+Manifest

data/README

@@ -0,0 +1,61 @@
+= htree - HTML/XML tree library
+
+htree provides a tree data structure which represent HTML and XML data.
+
+== Feature
+
+* Permissive unified HTML/XML parser
+* byte-to-byte round-tripping unparser
+* XML namespace support
+* Dedicated class for escaped string.  This ease sanitization.
+* HTML/XHTML/XML generator
+* template engine
+* recursive template expansion
+* converter to REXML document
+
+== Home Page
+
+http://www.a-k-r.org/htree/
+
+== Requirements
+
+* ruby : http://www.ruby-lang.org/
+
+== Download
+
+* latest release: http://www.a-k-r.org/htree/htree-0.7.tar.gz
+
+* development version in CVS repository:
+
+  http://cvs.m17n.org/viewcvs/ruby/htree/ or checkout with:
+
+    % cvs -d :pserver:anonymous@cvs.m17n.org:/cvs/ruby co htree
+
+== Install
+
+  % ruby install.rb
+
+== Reference Manual
+
+See rdoc/index.html or
+http://www.a-k-r.org/htree/rdoc/
+
+== Usage
+
+Following two-line script convert HTML to XHTML.
+
+  require 'htree'
+  HTree(STDIN).display_xml
+
+The conversion method to REXML is provided as to_rexml.
+
+  HTree(...).to_rexml
+
+== License
+
+Ruby's
+
+== Author
+Tanaka Akira <akr@fsij.org>
+
+[packaged as gem- andrew packer <agp@ppolitics.org>]

data/Rakefile

@@ -0,0 +1,37 @@
+require 'rubygems'
+require 'rake'
+require 'rake/testtask'
+require 'rake/rdoctask'
+require 'echoe'
+
+
+Echoe.new('htree', '0.7.0') do |p|
+  p.summary        = "HTML/XML tree library"
+  p.description    = "Htree provides a tree data structure which represent HTML and XML data"
+  p.url            = "http://www.a-k-r.org/htree/"
+  p.author         = "Tanaka Akira"
+  p.email          = "akr@fsij.org"
+  p.ignore_pattern = ["tmp/*", "script/*"]
+  p.development_dependencies = []
+end
+
+Dir["#{File.dirname(__FILE__)}/tasks/*.rake"].sort.each { |ext| load ext }
+
+desc 'Default: run unit tests.'
+task :default => :test
+
+desc 'Test'
+Rake::TestTask.new(:test) do |t|
+  t.libs << 'lib'
+  t.pattern = 'test/test-*.rb'
+  t.verbose = true
+end
+
+desc 'RDoc'
+Rake::RDocTask.new(:rdoc) do |rdoc|
+  rdoc.rdoc_dir = 'rdoc'
+  rdoc.title    = 'HTree'
+  rdoc.options << '-NS'
+  rdoc.rdoc_files.include('README')
+  rdoc.rdoc_files.include('lib/**/*.rb')
+end

data/htree.gemspec

@@ -0,0 +1,32 @@
+# -*- encoding: utf-8 -*-
+
+Gem::Specification.new do |s|
+  s.name = %q{htree}
+  s.version = "0.7.0"
+
+  s.required_rubygems_version = Gem::Requirement.new(">= 1.2") if s.respond_to? :required_rubygems_version=
+  s.authors = ["Tanaka Akira"]
+  s.date = %q{2010-01-27}
+  s.description = %q{Htree provides a tree data structure which represent HTML and XML data}
+  s.email = %q{akr@fsij.org}
+  s.extra_rdoc_files = ["README", "lib/htree.rb", "lib/htree/container.rb", "lib/htree/context.rb", "lib/htree/display.rb", "lib/htree/doc.rb", "lib/htree/elem.rb", "lib/htree/encoder.rb", "lib/htree/equality.rb", "lib/htree/extract_text.rb", "lib/htree/fstr.rb", "lib/htree/gencode.rb", "lib/htree/htmlinfo.rb", "lib/htree/inspect.rb", "lib/htree/leaf.rb", "lib/htree/loc.rb", "lib/htree/modules.rb", "lib/htree/name.rb", "lib/htree/output.rb", "lib/htree/parse.rb", "lib/htree/raw_string.rb", "lib/htree/regexp-util.rb", "lib/htree/rexml.rb", "lib/htree/scan.rb", "lib/htree/tag.rb", "lib/htree/template.rb", "lib/htree/text.rb", "lib/htree/traverse.rb"]
+  s.files = ["Makefile", "README", "Rakefile", "init.rb", "install.rb", "lib/htree.rb", "lib/htree/container.rb", "lib/htree/context.rb", "lib/htree/display.rb", "lib/htree/doc.rb", "lib/htree/elem.rb", "lib/htree/encoder.rb", "lib/htree/equality.rb", "lib/htree/extract_text.rb", "lib/htree/fstr.rb", "lib/htree/gencode.rb", "lib/htree/htmlinfo.rb", "lib/htree/inspect.rb", "lib/htree/leaf.rb", "lib/htree/loc.rb", "lib/htree/modules.rb", "lib/htree/name.rb", "lib/htree/output.rb", "lib/htree/parse.rb", "lib/htree/raw_string.rb", "lib/htree/regexp-util.rb", "lib/htree/rexml.rb", "lib/htree/scan.rb", "lib/htree/tag.rb", "lib/htree/template.rb", "lib/htree/text.rb", "lib/htree/traverse.rb", "test-all.rb", "test/assign.html", "test/template.html", "test/test-attr.rb", "test/test-charset.rb", "test/test-context.rb", "test/test-display_xml.rb", "test/test-elem-new.rb", "test/test-encoder.rb", "test/test-equality.rb", "test/test-extract_text.rb", "test/test-gencode.rb", "test/test-leaf.rb", "test/test-loc.rb", "test/test-namespace.rb", "test/test-output.rb", "test/test-parse.rb", "test/test-raw_string.rb", "test/test-rexml.rb", "test/test-scan.rb", "test/test-security.rb", "test/test-subnode.rb", "test/test-template.rb", "test/test-text.rb", "test/test-traverse.rb", "Manifest", "htree.gemspec"]
+  s.homepage = %q{http://www.a-k-r.org/htree/}
+  s.rdoc_options = ["--line-numbers", "--inline-source", "--title", "Htree", "--main", "README"]
+  s.require_paths = ["lib"]
+  s.rubyforge_project = %q{htree}
+  s.rubygems_version = %q{1.3.5}
+  s.summary = %q{HTML/XML tree library}
+  s.signing_key = '/Users/agp/.gem/gem-private_key.pem'
+  s.cert_chain  = ['gem-public_cert.pem']
+
+  if s.respond_to? :specification_version then
+    current_version = Gem::Specification::CURRENT_SPECIFICATION_VERSION
+    s.specification_version = 3
+
+    if Gem::Version.new(Gem::RubyGemsVersion) >= Gem::Version.new('1.2.0') then
+    else
+    end
+  else
+  end
+end

data/init.rb

@@ -0,0 +1 @@
+require 'htree'

data/install.rb

@@ -0,0 +1,112 @@
+#!/usr/bin/env ruby
+
+# usage: ruby install.rb [-n] [--destdir=DESTDIR]
+# options:
+#  -n : don't install
+#  --destdir=DESTDIR
+#
+# Author: Tanaka Akira <akr@m17n.org>
+
+require 'optparse'
+require 'fileutils'
+
+def target_directory
+  $:.each {|loc|
+    if %r{/site_ruby/[\d.]+\z} =~ loc
+      return loc
+    end
+  }
+  raise "could not find target install directory"
+end
+
+CVS_FILES = {}
+def cvs_files(dir)
+  return CVS_FILES[dir] if CVS_FILES.include? dir
+  if File.directory? "#{dir}/CVS"
+    result = {}
+    File.foreach("#{dir}/CVS/Entries") {|line|
+      case line
+      when %r{\A/([^/]+)/} then result[$1] = true
+      when %r{\AD/([^/]+)/} then result[$1] = true
+      end
+    }
+  else
+    result = nil
+  end
+  CVS_FILES[dir] = result
+  result
+end
+
+def each_target(&block)
+  target_set = {}
+  cvs = cvs_files('.')
+  Dir.glob("*.rb") {|filename|
+    next if /\Atest-/ =~ filename
+    next if /\Ainstall/ =~ filename
+    next if cvs && !cvs.include?(filename)
+    target_set[filename] = true
+    yield filename
+    each_require(filename, target_set, &block)
+  }
+end
+
+def each_require(file, target_set, &block)
+  File.foreach(file) {|line|
+    next if /\A\s*require\s+['"]([^'"]+)['"]/ !~ line
+    feature = $1
+    filename = "#{feature}.rb"
+    next if target_set.include? filename
+    next if !File.exist?(filename)
+    target_set[filename] = true
+    yield filename
+    each_require(filename, target_set, &block)
+  }
+end
+
+def collect_target
+  result = []
+  each_target {|filename| result << filename }
+  result.sort!
+  result
+end
+
+def install_file(src, dst)
+  ignore_exc(Errno::ENOENT) { return if FileUtils.compare_file src, dst }
+  # check shadow
+  ignore_exc(Errno::ENOENT) { File.unlink dst }
+  FileUtils.mkdir_p(File.dirname(dst), :mode=>0755)
+  FileUtils.cp(src, dst, :verbose => true)
+  File.chmod(0644, dst)
+end
+
+def ignore_exc(exc)
+  begin
+    yield
+  rescue exc
+  end
+end
+
+$opt_n = false
+$opt_destdir = ""
+ARGV.options {|q|
+  q.banner = 'ruby install.rb [opts]'
+  q.def_option('--help', 'show this message') {puts q; exit(0)}
+  q.def_option('-n', "don't install") { $opt_n = true }
+  q.def_option('--destdir=DESTDIR', "specify DESTDIR") {|destdir| $opt_destdir = destdir }
+  q.parse!
+}
+
+if $opt_n
+  dir = target_directory
+  collect_target.each {|filename|
+    puts "-> #{$opt_destdir}#{dir}/#{filename}"
+  }
+  exit
+else
+  File.umask 022
+  dir = target_directory
+  collect_target.each {|filename|
+    install_file filename, "#{$opt_destdir}#{dir}/#{filename}"
+  }
+end
+

data/lib/htree.rb

@@ -0,0 +1,97 @@
+#
+# = htree.rb
+#
+# HTML/XML document tree
+#
+# Author:: Tanaka Akira <akr@fsij.org>
+#
+# == Features
+#
+# - Permissive unified HTML/XML parser
+# - byte-to-byte round-tripping unparser
+# - XML namespace support
+# - Dedicated class for escaped string.  This ease sanitization.
+# - XHTML/XML generator
+# - template engine: link:files/htree/template_rb.html
+# - recursive template expansion
+# - REXML tree generator: link:files/htree/rexml_rb.html
+#
+# == Example
+#
+# The following one-liner prints parsed tree object.
+#
+#   % ruby -rhtree -e 'pp HTree(ARGF)' html-file
+#
+# The following two-line script convert HTML to XHTML.
+#
+#   require 'htree'
+#   HTree(STDIN).display_xml
+#
+# The conversion method to REXML is provided as to_rexml.
+#
+#     HTree(...).to_rexml
+#
+# == Module/Class Hierarchy
+#
+# * HTree
+#   * HTree::Name
+#   * HTree::Context
+#   * HTree::Location
+#   * HTree::Node
+#     * HTree::Doc
+#     * HTree::Elem
+#     * HTree::Text
+#     * HTree::XMLDecl
+#     * HTree::DocType
+#     * HTree::ProcIns
+#     * HTree::Comment
+#     * HTree::BogusETag
+# * HTree::Error
+#
+# == Method Summary
+#
+# HTree provides following methods.
+#
+# - Parsing Methods
+#   - HTree(<i>html_string</i>) -> HTree::Doc
+#   - HTree.parse(<i>input</i>) -> HTree::Doc
+#
+# - Generation Methods
+#   - HTree::Node#display_xml -> STDOUT
+#   - HTree::Node#display_xml(<i>out</i>) -> <i>out</i>
+#   - HTree::Node#display_xml(<i>out</i>, <i>encoding</i>) -> <i>out</i>
+#   - HTree::Text#to_s -> String
+#
+# - Template Methods
+#   - HTree.expand_template{<i>template_string</i>} -> STDOUT
+#   - HTree.expand_template(<i>out</i>){<i>template_string</i>} -> <i>out</i>
+#   - HTree.expand_template(<i>out</i>, <i>encoding</i>){<i>template_string</i>} -> <i>out</i>
+#   - HTree.compile_template(<i>template_string</i>) -> Module
+#   - HTree{<i>template_string</i>} -> HTree::Doc
+#
+# - Traverse Methods
+#   - HTree::Elem#attributes -> Hash[HTree::Name -> HTree::Text]
+#   - HTree::Elem::Location#attributes -> Hash[HTree::Name -> HTree::Location]
+#
+# - Predicate Methods
+#   - HTree::Traverse#doc? -> true or false
+#   - HTree::Traverse#elem? -> true or false
+#   - HTree::Traverse#text? -> true or false
+#   - HTree::Traverse#xmldecl? -> true or false
+#   - HTree::Traverse#doctype? -> true or false
+#   - HTree::Traverse#procins? -> true or false
+#   - HTree::Traverse#comment? -> true or false
+#   - HTree::Traverse#bogusetag? -> true or false
+#
+# - REXML Tree Generator
+#   - HTree::Node#to_rexml -> REXML::Child
+
+require 'htree/parse'
+require 'htree/extract_text'
+require 'htree/equality'
+require 'htree/inspect'
+require 'htree/display'
+require 'htree/loc'
+require 'htree/traverse'
+require 'htree/template'
+require 'htree/rexml'

data/lib/htree/container.rb

@@ -0,0 +1,8 @@
+require 'htree/modules'
+
+module HTree::Container
+  # +children+ returns children nodes as an array.
+  def children
+    @children.dup
+  end
+end

data/lib/htree/context.rb

@@ -0,0 +1,69 @@
+module HTree
+  class Context
+    # :stopdoc:
+    DefaultNamespaces = {'xml'=>'http://www.w3.org/XML/1998/namespace'}
+    DefaultNamespaces.default = ""
+    DefaultNamespaces.freeze
+    # :startdoc:
+
+    # The optional argument `namespaces' should be a hash or nil.
+    # HTree::DefaultNamespaces is used if nil is specified.
+    #
+    # If it is a hash, its key should be nil or a string.
+    # nil means default namespace.
+    # The string means some prefix which must not be empty.
+    #
+    # The hash value should be a string.
+    # The empty string "" means unbound namespace.
+    def initialize(namespaces=nil)
+      namespaces ||= DefaultNamespaces
+      namespaces.each_pair {|k, v|
+        check_namespace_prefix(k)
+        check_namespace_uri(v)
+      }
+      namespaces = namespaces.dup.freeze unless namespaces.frozen?
+      @namespaces = namespaces
+    end
+    attr_reader :namespaces
+
+    # return a namespace URI corresponding to _prefix_.
+    # It returns nil if _prefix_ is not defined.
+    def namespace_uri(prefix)
+      @namespaces[prefix]
+    end
+
+    # generate a new Context object which namespaces are substituted by
+    # a hash _declared_namespaces_.
+    def subst_namespaces(declared_namespaces)
+      namespaces = @namespaces.dup
+      declared_namespaces.each {|k, v|
+        check_namespace_prefix(k)
+        check_namespace_uri(v)
+        namespaces[k] = v
+      }
+      if namespaces == @namespaces
+        self
+      else
+        Context.new(namespaces)
+      end
+    end
+
+    private
+    def check_namespace_prefix(k)
+      unless (String === k && !k.empty?) || k == nil
+        raise ArgumentError, "invalid namespace prefix: #{k.inspect}"
+      end
+    end
+
+    def check_namespace_uri(v)
+      unless String === v
+        raise ArgumentError, "invalid namespace URI: #{v.inspect}"
+      end
+    end
+  end
+
+  # :stopdoc:
+  DefaultContext = Context.new
+  HTMLContext = DefaultContext.subst_namespaces(nil=>"http://www.w3.org/1999/xhtml")
+  # :startdoc:
+end