tidy-fork 1.1.3

data/CHANGES

@@ -0,0 +1,86 @@
+V 1.1.4 2009/5/13
+- Added a naive solution to having to do Tidy.path="/usr/lib/libtidy.so"
+- Automatically deduces the loadpath
+
+V 1.1.3 2008/10/20
+- Introduced Tidy.fresh_tidy_version = bool flag to enable/disable support
+  for fresh Tidylib versions. Disabled by default, turn it on if you
+  experience seg faults.
+- Segmentation fault when loading the TidyBuffer from the dynamic library.
+  Fix for 'tidybuf.rb:40: [BUG] Segmentation fault' error.
+  [Patch by Ben Walding (bwalding)]
+  
+- The line separator is hard-encoded instead of using the $/ magic variable, 
+  so it breaks on unix platforms.
+  [Patch by Damien Merenne <dam@cosinux.org>]
+
+V 1.1.2 2005/03/14
+
+- Tidybuf.to_s bugfix: added nil check
+  @struct.bp ? @struct.bp.to_s(@struct.size) : ''
+
+V 1.1.1 2005/03/07
+
+- Bugfix
+
+  Tidybuf.to_s assumes that TidyBuffer.bp is null-terminated. This is
+  only true if the length of the data is evenly divisable by 256, but
+  if Tidy output is e.g. 512 bytes long the buffer is not null-terminated
+  and the code reads out of bounds. Result: garbage chars in output.
+  
+  Fix: Code changed to read only TidyBuffer.size bytes from the buffer. 
+  In TidyBuffer#to_s: @struct.bp.to_s => @struct.bp.to_s(@struct.size)
+  '.'*345 = 512 char output (for testing)
+
+V 1.1.0 2005/02/21
+
+  - $TIDYLIB global eliminated
+    Path to the library is now specified as follows:
+    Tidy.path = '/path/to/tidylib.so'
+
+  - minor code/docs/gemspec cleanup
+
+V 1.0.1
+
+  .sub calls changed to .gsub
+
+V 1.0.0
+
+  Oct/18/2004 - Declared stable, now packaged as a RubyGem
+
+VB.2.5
+
+  - Auto-detection and ENV variables removed, adds too much complexity
+    Module location is now specified by defining $TIDYLIB before require 'tidy'
+    $TIDYLIB is the system path to the library (ex: /usr/lib/tidylib.so)
+    Raises an error if $TIDYLIB is not defined
+
+V.B2.4
+
+  Now uses ENV['TIDY_LIB'] instead of $TIDY_LIB
+
+V B2.3
+
+  - Added library auto-detection
+    a) use $TIDY_LIB if defined
+    b) Search $: and ENV['PATH'] paths for (tidy|tidylib|libtidy|htmltidy).(dll|so)
+    c) Raise an error if not found, otherwise loads
+
+V.B2.2
+
+  - Tidy.to_b added
+
+V.B2.1
+
+  - load_config method added
+
+V.B2
+
+  - load(path) method added, library must be loaded explicity
+    this avoids hard coding the library name
+    require 'tidy'
+    Tidy.load('path/to/tidylib.so')
+    
+V.B1
+
+  First release

data/MANIFEST

@@ -0,0 +1,12 @@
+lib/tidy.rb
+lib/tidy/tidybuf.rb
+lib/tidy/tidyerr.rb
+lib/tidy/tidylib.rb
+lib/tidy/tidyobj.rb
+lib/tidy/tidyopt.rb
+test/usage.rb
+CHANGES
+MANIFEST
+README.txt.en
+tidy.gemspec
+VERSION

data/README.txt.en

@@ -0,0 +1,36 @@
+Tidy README
+============
+
+  Ruby interface to HTML Tidy Library Project (http://tidy.sf.net)
+
+Requirements
+------------
+
+  * Recent version of Ruby
+  * RubyGems 1.8+ (http://rubygems.rubyforge.org)
+  * HTML Tidy Library (compiled)
+
+Install
+-------
+
+  - Download library from http://tidy.sf.net (pre-compiled versions available).
+  
+  - Install files using one of the following:
+
+    $ gem install tidy-x-x-x.gem
+    $ ruby install.rb
+
+  - Open test/usage.rb, change Tidy.path to point to your compiled Tidy library, run.
+
+Usage
+-----
+
+  See API docs
+
+License
+-------
+
+  Distributes under the same terms as Ruby
+  http://www.ruby-lang.org/en/LICENSE.txt
+
+Kevin Howe <kh at newclear.ca>

data/VERSION

@@ -0,0 +1 @@
+1.1.3

data/install.rb

@@ -0,0 +1,11 @@
+require 'rbconfig'
+require 'ftools'
+
+# Install lib
+dst_dir = Config::CONFIG['sitelibdir']
+Dir.chdir('lib') { 
+  Dir['**/*'].reject { |f| f =~ /\.(cvs|gem|svn)($|\/)/i or not File.file?(f) }.each { |file|
+    File.mkpath File.join(dst_dir, File.dirname(file)), true
+    File.install file, File.join(dst_dir, file), 0644, true
+  }
+}

data/lib/tidy/tidybuf.rb

@@ -0,0 +1,50 @@
+# Buffer structure.
+#
+class Tidybuf
+
+  extend DL::Importable
+  
+  # Access TidyBuffer instance.
+  #
+  attr_reader(:struct)
+
+  def initialize
+    @struct = self.class.struct(construct_tidy_buffer).malloc
+    Tidylib.buf_init(@struct)
+  end
+    
+  # Free current contents and zero out.
+  #
+  def free
+    Tidylib.buf_free(@struct)
+  end
+
+  # Convert to array.
+  #
+  def to_a
+    to_s.split($/)
+  end
+
+  # Convert to string.
+  #
+  def to_s
+    @struct.bp ? @struct.bp.to_s(@struct.size) : ''
+  end
+
+protected
+
+  def construct_tidy_buffer
+    struct_rows = [
+      "TidyAllocator* allocator",
+      "byte* bp",
+      "uint size",
+      "uint allocated",
+      "uint next" 
+    ]
+    if Tidy.fresh_tidy_version
+      struct_rows.unshift "int* allocator"
+    end
+    struct_rows
+  end
+  
+end

data/lib/tidy/tidyerr.rb

@@ -0,0 +1,30 @@
+# Parameterized error message.
+#
+class Tidyerr < String
+  
+  # Error parameter.
+  #
+  attr_reader :severity, :line, :column, :message
+
+  # Create new instance.
+  #
+  def initialize(error)
+    super(error.to_s)
+    parameterize
+  end
+  
+  # Parse error message into parameters (where applicable).
+  #
+  def parameterize
+    if to_str[0,4] == 'line'
+      tokens    = to_str.split(' ', 7)
+      @severity = tokens[5][0,1] # W or E
+      @line     = tokens[1].to_i
+      @column   = tokens[3].to_i
+      @message  = tokens[6]
+    end
+  end
+
+  protected :parameterize
+
+end

data/lib/tidy/tidylib.rb

@@ -0,0 +1,117 @@
+# Ruby wrapper for HTML Tidy Library Project (http://tidy.sf.net).
+#
+module Tidylib
+
+  extend DL::Importable
+
+  module_function
+  
+  # Load the library.
+  #
+  def load(path)
+    begin
+      dlload(path)
+    rescue
+      raise LoadError, "Unable to load #{path}"
+    end
+    extern "void *tidyCreate()"
+    extern "void tidyBufFree(void*)"
+    extern "void tidyBufInit(void*)"
+    extern "int tidyCleanAndRepair(void*)"
+    extern "int tidyLoadConfig(void*, char*)"
+    extern "int tidyOptGetIdForName(char*)"
+    extern "char tidyOptGetValue(void*, unsigned int)"
+    extern "int tidyOptParseValue(void*, char*, char*)"
+    extern "int tidyParseString(void*, char*)"
+    extern "void tidyRelease(void*)"
+    extern "char* tidyReleaseDate()"
+    extern "int tidyRunDiagnostics(void*)"
+    extern "int tidySaveBuffer(void*, void*)"
+    extern "int tidySetErrorBuffer(void*, void*)"
+  end
+  
+  # tidyBufInit
+  def buf_init(buf)
+    tidyBufInit(buf)
+  end
+
+  # tidyBufFree
+  #
+  def buf_free(buf)
+    tidyBufFree(buf)
+  end
+  
+  # tidyCreate
+  #
+  def create
+    tidyCreate
+  end
+
+  # tidyCleanAndRepair
+  #
+  def clean_and_repair(doc)
+    tidyCleanAndRepair(doc)
+  end
+  
+  # tidyLoadConfig
+  #
+  def load_config(doc, file)
+    tidyLoadConfig(doc, file.to_s)
+  end
+
+  # tidyOptParseValue
+  #
+  def opt_parse_value(doc, name, value)
+    tidyOptParseValue(doc, translate_name(name), value.to_s)
+  end
+
+  # tidyOptGetValue (returns true/false instead of 1/0)
+  #
+  def opt_get_value(doc, name)
+    value = tidyOptGetValue(doc, tidyOptGetIdForName(translate_name(name)))
+    Tidy.to_b(value)
+  end
+
+  # tidyParseString
+  #
+  def parse_string(doc, str)
+    tidyParseString(doc, str.to_s)
+  end
+  
+  # tidyRelease
+  #
+  def release(doc)
+    tidyRelease(doc)
+  end
+  
+  # tidyReleaseDate
+  #
+  def release_date
+    tidyReleaseDate
+  end
+  
+  # tidyRunDiagnostics
+  #
+  def run_diagnostics(doc)
+    tidyRunDiagnostics(doc)
+  end
+  
+  # tidySaveBuffer
+  #
+  def save_buffer(doc, buf)
+    tidySaveBuffer(doc, buf)
+  end
+  
+  # tidySetErrorBuffer
+  #
+  def set_error_buffer(doc, buf)
+    tidySetErrorBuffer(doc, buf)
+  end
+
+  # Convert to string, replace underscores with dashes (:output_xml => 'output-xml').
+  #
+  def translate_name(name)
+    name.to_s.gsub('_', '-')
+  end
+
+end

data/lib/tidy/tidyobj.rb

@@ -0,0 +1,103 @@
+# Ruby interface to Tidylib.
+#
+class Tidyobj
+
+  # Diagnostics Buffer (Array of String).
+  #
+  attr_reader(:diagnostics)
+
+  # Access the tidy instance.
+  #
+  attr_reader(:doc)
+
+  # Error Buffer (Array of Tidyerr).
+  #
+  attr_reader(:errors)
+
+  # Options interface (Tidyopt).
+  #
+  attr_reader(:options)
+  
+  # Construct a new instance.
+  # Receives a hash of options to be set.
+  #
+  def initialize(options=nil)
+    @diagnostics = Array.new
+    @doc = Tidylib.create
+    @errors = Array.new
+    @errbuf = Tidybuf.new
+    @outbuf = Tidybuf.new
+    @options = Tidyopt.new(@doc)
+    rc = Tidylib.set_error_buffer(@doc, @errbuf.struct)
+    verify_severe(rc)
+    unless options.nil?
+      options.each { |name, value| Tidylib.opt_parse_value(@doc, name, value) }
+    end
+  end
+
+  # Clean and Repair.
+  #
+  def clean(str)
+    verify_doc
+    rc = -1
+
+    # Clean and repair the string.
+    #
+    rc = Tidylib.parse_string(@doc, str)                                               # Parse the input
+    rc = Tidylib.clean_and_repair(@doc) if rc >= 0                                     # Tidy it up!
+    rc = (Tidylib.opt_parse_value(@doc, :force_output, true) == 1 ? rc : -1) if rc > 1 # If error, force output
+    rc = Tidylib.save_buffer(@doc, @outbuf.struct) if rc >= 0                          # Pretty Print
+    verify_severe(rc)
+
+    # Save and clear output/errors.
+    #
+    output = @outbuf.to_s
+    @errors = @errbuf.to_a.collect { |e| Tidyerr.new(e) }
+    @outbuf.free
+    @errbuf.free
+    
+    # Save diagnostics.
+    #
+    rc = Tidylib.run_diagnostics(@doc)
+    verify_severe(rc)
+    @diagnostics = @errbuf.to_a
+    @errbuf.free
+
+    output
+  end
+  
+  # Load a tidy config file.
+  #
+  def load_config(file)
+    verify_doc
+    rc = Tidylib.load_config(@doc, file)
+    case rc
+      when -1 then raise LoadError, "#{file} does not exist"
+      when  1 then raise LoadError, "errors parsing #{file}"
+    end
+    rc
+  end
+
+  # Clear the tidy instance.
+  #
+  def release
+    verify_doc
+    Tidylib.release(@doc)
+    @doc = nil
+  end
+  
+  # Raise an error if the tidy document is invalid.
+  #
+  def verify_doc
+    raise TypeError, 'Invalid Tidy document' unless @doc.class == DL::PtrData
+  end
+
+  # Raise severe error based on tidy status value.
+  #
+  def verify_severe(rc)
+    raise "A severe error (#{rc}) occurred.\n" if rc < 0
+  end
+
+  protected :verify_doc, :verify_severe
+
+end

data/lib/tidy/tidyopt.rb

@@ -0,0 +1,31 @@
+# Ruby interface to Tidylib options.
+#
+class Tidyopt
+
+  # Construct a new instance.
+  #
+  def initialize(doc)
+    @doc = doc
+  end
+    
+  # Reader for options (Hash syntax).
+  #
+  def [](name)
+    Tidylib.opt_get_value(@doc, name)
+  end
+    
+  # Writer for options (Hash syntax).
+  #
+  def []=(name, value)
+    Tidylib.opt_parse_value(@doc, name, value)
+  end
+    
+  # Reader/Writer for options (Object syntax).
+  #
+  def method_missing(name, value=:none, *args)
+    name = name.to_s.gsub('=', '')
+    return self[name] if value == :none
+    self[name] = value
+  end
+
+end

data/lib/tidy.rb

@@ -0,0 +1,83 @@
+# Ruby interface to HTML Tidy Library Project (http://tidy.sf.net).
+#
+# =Usage
+#
+#   require 'tidy'
+#   Tidy.path = '/usr/lib/tidylib.so'
+#   html = '<html><title>title</title>Body</html>'
+#   xml = Tidy.open(:show_warnings=>true) do |tidy|
+#     tidy.options.output_xml = true
+#     puts tidy.options.show_warnings
+#     xml = tidy.clean(html)
+#     puts tidy.errors
+#     puts tidy.diagnostics
+#     xml
+#   end
+#   puts xml
+#
+# Author::  Kevin Howe
+# License:: Distributes under the same terms as Ruby
+#
+module Tidy
+
+  require 'dl/import'
+  require 'dl/struct'
+  require 'tidy/tidybuf'
+  require 'tidy/tidyerr'
+  require 'tidy/tidylib'
+  require 'tidy/tidyobj'
+  require 'tidy/tidyopt'
+
+  module_function
+
+  # Return a Tidyobj instance.
+  #
+  def new(options=nil)
+    Tidyobj.new(options)
+  end
+  
+  # Path to Tidylib.
+  #
+  def path() @path end
+  
+  # Set the path to Tidylib (automatically loads the library).
+  #
+  def path=(path)
+    Tidylib.load(path)
+    @path = path
+  end
+  
+  def fresh_tidy_version=(bool)
+    @fresh_tidy_version = bool
+  end
+  
+  def fresh_tidy_version
+    !!@fresh_tidy_version
+  end
+  
+  # With no block, open is a synonym for Tidy.new.
+  # If a block is present, it is passed aTidy as a parameter.
+  # aTidyObj.release is ensured at end of the block.
+  #
+  def open(options=nil)
+    raise "Tidy.path was not specified." unless @path
+    tidy = Tidy.new(options)
+    if block_given?
+      begin
+        yield tidy
+      ensure
+        tidy.release
+      end
+    else
+      tidy
+    end
+  end
+  
+  # Convert to boolean.
+  # 0, false and nil return false, anything else true.
+  #
+  def to_b(value)
+    [0,false,nil].include?(value) ? false : true
+  end
+
+end

data/test/usage.rb

@@ -0,0 +1,13 @@
+$LOAD_PATH.unshift('../lib')
+require 'tidy'
+Tidy.path = '/usr/lib/tidylib.so'
+html = '<html><title>title</title>Body</html>'
+xml = Tidy.open(:show_warnings=>true) do |tidy|
+  tidy.options.output_xml = true
+  puts tidy.options.show_warnings
+  xml = tidy.clean(html)
+  puts tidy.errors
+  puts tidy.diagnostics
+  xml
+end
+puts xml

data/tidy-fork.gemspec

@@ -0,0 +1,33 @@
+MANIFEST = %w(
+CHANGES
+install.rb
+lib
+lib/tidy
+lib/tidy/tidybuf.rb
+lib/tidy/tidyerr.rb
+lib/tidy/tidylib.rb
+lib/tidy/tidyobj.rb
+lib/tidy/tidyopt.rb
+lib/tidy.rb
+MANIFEST
+README.txt.en
+test
+test/usage.rb
+tidy-fork.gemspec
+VERSION
+)
+
+spec = Gem::Specification.new do |s|
+  s.name = 'tidy-fork'
+  s.version = "1.1.3"
+  s.authors = ['Kevin Howe', 'Dima Sabanin']
+  s.email = 'kh@newclear.ca'
+  s.homepage = 'http://github.com/railsmonk/tidy-fork'
+  s.platform = Gem::Platform::RUBY
+  s.summary = 'Forked Ruby interface to HTML Tidy Library Project with important bugfixes'
+  s.files = MANIFEST
+  s.require_path = 'lib'
+  s.rdoc_options << '--all' << '--inline-source' << '--main' << 'lib/tidy.rb'
+  s.has_rdoc = true
+#  s.rubyforge_project = 'tidy'
+end

metadata

@@ -0,0 +1,71 @@
+--- !ruby/object:Gem::Specification 
+name: tidy-fork
+version: !ruby/object:Gem::Version 
+  version: 1.1.3
+platform: ruby
+authors: 
+- Kevin Howe
+- Dima Sabanin
+autorequire: 
+bindir: bin
+cert_chain: []
+
+date: 2009-11-18 00:00:00 +07:00
+default_executable: 
+dependencies: []
+
+description: 
+email: kh@newclear.ca
+executables: []
+
+extensions: []
+
+extra_rdoc_files: []
+
+files: 
+- CHANGES
+- install.rb
+- lib/tidy/tidybuf.rb
+- lib/tidy/tidyerr.rb
+- lib/tidy/tidylib.rb
+- lib/tidy/tidyobj.rb
+- lib/tidy/tidyopt.rb
+- lib/tidy.rb
+- MANIFEST
+- README.txt.en
+- test/usage.rb
+- tidy-fork.gemspec
+- VERSION
+has_rdoc: true
+homepage: http://github.com/railsmonk/tidy-fork
+licenses: []
+
+post_install_message: 
+rdoc_options: 
+- --all
+- --inline-source
+- --main
+- lib/tidy.rb
+require_paths: 
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement 
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      version: "0"
+  version: 
+required_rubygems_version: !ruby/object:Gem::Requirement 
+  requirements: 
+  - - ">="
+    - !ruby/object:Gem::Version 
+      version: "0"
+  version: 
+requirements: []
+
+rubyforge_project: 
+rubygems_version: 1.3.5
+signing_key: 
+specification_version: 3
+summary: Forked Ruby interface to HTML Tidy Library Project with important bugfixes
+test_files: []
+