podgraph 0.1.5 → 1.0.1

data/lib/podgraph/posterous.rb

@@ -1,163 +0,0 @@
-require 'mail'
-require 'rexml/document'
-require 'yaml'
-require 'optparse'
-
-require_relative 'trestle'
-include Podgraph
-
-# :include: ../../README.rdoc
-module Podgraph
-  # Reads XHTML file, analyses it, finds images, checks if they can be inlined,
-  # generates multipart/relative or multipart/mixed MIME mail.
-  class Posterous
-
-    # some options for mail generator; change with care
-    attr_accessor :o
-
-    # a Trestle object
-    attr_accessor :trestle
-
-    # Analyses _filename_. It must be a XHTML file.
-    # _to_, _from_ are email.
-    # _mode_ is 1 of 'related' or 'mixed' string.
-    def initialize(trestle, filename, to, from, mode)
-      @o = Hash.new()
-      @o[:user_agent] = Podgraph::Meta::NAME + ?/ + Podgraph::Meta::VERSION
-      @o[:subject] = ''
-      @o[:body] = []
-      @o[:attachment] = []
-      @o[:a_marks] = {}
-      @o[:mode] = mode
-      @o[:to] = to
-      @o[:from] = from
-
-      @trestle = trestle
-
-      fp = (filename == STDIN ? STDIN : File.new(filename))
-      begin
-        make(fp)
-      rescue
-        raise $!
-      ensure
-        fp.close unless fp == STDIN
-      end
-    end
-
-    def make(fp)
-      xml = REXML::Document.new(fp)
-      begin
-        @o[:subject].replace(REXML::XPath.first(xml, "/html/body/div/h1").text.gsub(/\s+/, " "))
-        raise if @o[:subject] =~ /^\s*$/
-      rescue
-        raise 'cannot extract the subject from <h1>'
-      end
-
-      img_collect = ->(i, a) {
-        if i.name == 'img'
-          if (src = i.attributes['src']) =~ /^\s*$/
-            raise '<img> tag with missing or empty src attribute'
-          elsif src =~ /\s*(https?|s?ftp):\/\//
-            # we are ignoring URL's
-            return
-          else
-            a << src
-            if @o[:mode] == 'related'
-              # replace src attribute with a random chars--later
-              # we'll replace such marks with corrent content-id
-              random = Mail.random_tag()
-              i.attributes['src'] = random
-              @o[:a_marks][src] = random   # save an act of the replacement
-              
-              @o.rehash()                  # is this really necessary?
-            end
-          end
-        end
-      }
-      
-      f = 1
-      xml.elements.each('/html/body/div/*') { |i|
-        if f == 1
-          f = 0                     # skip first <h1>
-          next
-        end
-
-        @trestle.veputs(2, "node: #{i.name}")
-        img_collect.call(i, @o[:attachment])
-        i.each_recursive { |j|
-          @trestle.veputs(2, "node recursive: #{j.name}")
-          img_collect.call(j, @o[:attachment])
-        }
-        
-        @o[:body] << i
-      }
-
-      raise "body is empty or filled with nonsence" if @o[:body].size == 0
-    end
-    private :make
-
-    # Returns ready for delivery Mail object.
-    def generate()
-      m = Mail.new()
-      m.from(@o[:from])
-      m.to(@o[:to])
-      m.transport_encoding = Mail::Encodings.get_encoding('8bit')
-#      m.content_transfer_encoding('8bit')
-      m.subject(@o[:subject])
-      m.headers({'User-Agent' => @o[:user_agent]})
-
-      @trestle.veputs(2, "Body lines=#{@o[:body].size}, bytes=#{@o[:body].to_s.bytesize}")
-      if @o[:attachment].size == 0
-        m.content_disposition('inline')
-        m.content_type('text/html; charset="UTF-8"')
-        m.body(@o[:body])
-      else
-        if @o[:mode] == 'related'
-          m.content_type('Multipart/Related')
-        end
-        m.html_part = Mail::Part.new {
-          content_type('text/html; charset=UTF-8')
-        }
-        m.html_part.body = @o[:body]
-        m.html_part.content_disposition('inline') if @o[:mode] == 'mixed'
-        
-        begin
-          @o[:attachment].each { |i| m.add_file(i) }
-        rescue
-          raise("cannot attach: #{$!}")
-        end
-
-        if @o[:mode] == 'related'
-          if (fqdn = Socket.gethostname() ) == ''
-            raise 'hostname is not set!'
-          end
-          cid = {}
-          m.parts[1..-1].each { |i|
-            i.content_disposition('inline')
-            cid[i.filename] = i.content_id("<#{Mail.random_tag}@#{fqdn}.NO_mail>")
-          }
-
-          @o[:a_marks].each { |k, v|
-            if cid.key?(k)
-              @trestle.veputs(2, "mark #{k} = #{v}; -> to #{cid[k]}")
-              # replace marks with corresponding content-id
-              m.html_part.body.raw_source.sub!(v, "cid:#{cid[k][1..-1]}")
-            else
-              raise("orphan key in cid: #{k}")
-            end
-          }
-        end
-      end # a.size
-      
-      return m
-    end
-  
-    # Print Mail object to stdout.
-    # _e_ is an optional encoding.
-    def dump(e = '')
-      puts (e == '' ? generate().to_s : generate().to_s.encode(e))
-    end
-    
-  end # Posterous
-  
-end # Podgraph

data/lib/podgraph/trestle.rb

@@ -1,216 +0,0 @@
-# :erb:
-require 'yaml'
-require 'shellwords.rb'
-require 'optparse'
-require 'pp'
-require 'open4'
-
-require_relative 'meta'
-
-# :include: ../../README.rdoc
-module Podgraph
-
-  class Trestle
-
-    # Execute _cmd_ and return a list [exit_status, stderr,
-    # stdout]. Very handy.
-    def self.cmd_run(cmd)
-      so = sr = ''
-      status = Open4::popen4(cmd) { |pid, stdin, stdout, stderr|
-        so = stdout.read
-        sr = stderr.read
-      }
-      [status.exitstatus, sr, so]
-    end
-
-    # Return a directory with program libraries.
-    def self.gem_libdir
-      t = ["#{File.dirname(File.expand_path($0))}/../lib/#{Podgraph::Meta::NAME}",
-           "#{Gem.dir}/gems/#{Podgraph::Meta::NAME}-#{Podgraph::Meta::VERSION}/lib/#{Podgraph::Meta::NAME}",
-           "lib/#{Podgraph::Meta::NAME}"]
-      t.each {|i| return i if File.readable?(i) }
-      fail "all paths are invalid: #{t}"
-    end
-
-    # Analogue to shell command +which+.
-    def self.in_path?(file)
-      return true if file =~ %r%\A/% and File.exist? file
-
-      ENV['PATH'].split(File::PATH_SEPARATOR).any? do |path|
-        File.exist? File.join(path, file)
-      end
-    end
-
-    # Print an error message _t_ and exit if _ec_ > 0.
-    def self.errx(ec, t)
-      STDERR.puts File.basename($0) + ' error: ' + t.to_s
-      exit ec if ec > 0
-    end
-
-    # Print a warning.
-    def self.warnx(t)
-      STDERR.puts File.basename($0) + ' warning: ' + t.to_s
-    end
-
-    # #veputs uses this to decide to put a newline or not to put.
-    NNL_MARK = '__NNL__'
-
-    # Use this in your CL options to check if modifying some variable is
-    # not an idempotent act.
-    attr_reader :cl_opt_protect
-
-    # [conf] Typically must be a reference to some global variable.
-    def initialize(conf)
-      @conf = conf
-      @conf[:verbose] = 0
-      @conf[:banner] = "Usage: #{File.basename($0)} [options]"
-      @conf[:config] =  Meta::NAME + '.yaml'
-      @conf[:config_dirs] =  [ENV['HOME']+'/.'+Meta::NAME,
-                              File.absolute_path("#{File.dirname(File.expand_path($0))}/../etc"),
-                              '/usr/etc', '/usr/local/etc', '/etc',
-                              "#{Gem.dir}/gems/#{Meta::NAME}-#{Meta::VERSION}/etc"
-                             ]
-      @conf[:config_env] = [Meta::NAME.upcase + '_CONF']
-
-      @cl_parsing_times = 0 # not used
-      @cl_opt_protect = false
-    end
-
-    # [level] A verbose level.
-    # [t]     A string to print.
-    #
-    # Don't print _t_ with a newline if it contains NNL_MARK at the end.
-    def veputs(level, t)
-      t = t.dup
-      nnl = nil
-      if t.match(/#{NNL_MARK}$/)
-        t.sub!(/#{$&}/, '')
-        nnl = 1
-      end
-
-      if @conf[:verbose] >= level
-        nnl ? print(t) : puts(t)
-        STDOUT.flush
-      end
-    end
-    
-    # Run all configuration parsing in a batch.
-    #
-    # [rvars] A list of variable names which must be in the
-    #         configuration file.
-    #
-    # If no block is given, only standard CL options will be analysed.
-    def config_parse(rvars, &block)
-      cb = ->(b, src) {
-        if b
-          block.call src
-        else
-          # very basic default options
-          cl_parse(src, nil, true)
-        end
-      }
-      
-      # 1. parse env
-      @conf[:config_env].each {|i|
-#        puts '0 run:'
-        cb.call(block_given?, ENV[i].shellsplit) if ENV.key?(i)
-      }
-
-      # 2. parse CL in case of '--config' option
-#      puts "\n1 run"
-      @cl_opt_protect = true
-      cb.call(block_given?, ARGV.dup)
-      @cl_opt_protect = false
-
-      # 3. load the configuration file & do the final CL parsing
-      begin
-#        puts "\n2 run"
-        r = config_flat_load(rvars)
-      rescue
-        Trestle.errx(1, "cannot load config: #{$!}")
-      end
-      veputs(1, "Loaded config: #{r}")
-      cb.call(block_given?, ARGV)
-    end
-
-    # Load a config file immediately if it contains '/' in its name,
-    # otherwise search through several dirs for it.
-    #
-    # [rvars] a list of requied variables in the config
-    #
-    # Return a loaded filename or nil on error.
-    def config_flat_load(rvars)
-      p = ->(f) {
-        if File.readable?(f)
-          begin
-            myconf = YAML.load_file(f)
-          rescue
-            abort("cannot parse #{f}: #{$!}")
-          end
-          rvars.each { |i|
-            fail "missing or nil '#{i}' in #{f}" if ! myconf.key?(i.to_sym) || ! myconf[i.to_sym]
-          }
-          @conf.merge!(myconf)
-          return @conf[:config]
-        end
-        return nil
-      }
-
-      if @conf[:config].index('/')
-        return p.call(@config[:config])
-      else
-        @conf[:config_dirs].each {|dir|
-          return dir+'/'+@conf[:config] if p.call(dir + '/' + @conf[:config])
-        }
-      end
-
-      return nil
-    end
-
-
-    # Parses CL-like options.
-    #
-    # [src] An array of options (usually +ARGV+).
-    #
-    # If _o_ is non nil function parses _src_ immediately, otherwise it
-    # only creates +OptionParser+ object and return it (if _simple_ is
-    # false).
-    def cl_parse(src, o = nil, simple = false)
-      if ! o then
-#        puts "NEW o (#{cl_opt_protect})" + src.to_s 
-        o = OptionParser.new
-        o.banner = @conf[:banner]
-        o.on('-v', 'Be more verbose.') { |i|
-#          puts "cl_parsing_times "+cl_parsing_times.to_s
-          @conf[:verbose] += 1 unless cl_opt_protect
-        }
-        o.on('-V', 'Show version & exit.') { |i|
-          puts Meta::VERSION
-          exit 0
-        }
-        o.on('--config NAME', "Set a config name (default is #{@conf[:config]})") {|i|
-          @conf[:config] = i
-        }
-        o.on('--config-dirs', 'Show possible config locations') {
-          @conf[:config_dirs].each { |j|
-            f = j + '/' + @conf[:config]
-            puts (File.readable?(f) ? '* ' : '  ') +  f
-          }
-          exit 0
-        }
-
-        return o if ! simple
-      end
-
-      begin
-        o.parse!(src)
-        @cl_parsing_times += 1
-      rescue
-        Trestle.errx(1, $!.to_s)
-      end
-    end
-    
-  end # trestle
-end
-
-# Don't remove this: falsework/0.2.2/naive/2010-12-26T04:50:00+02:00

data/test/config.yaml

@@ -1,3 +0,0 @@
-:to: alex@goliard
-# :to: post@podgraph-test.posterous.com
-:from: alexander.gromnitsky@gmail.com

data/test/garbage_01.html

@@ -1 +0,0 @@
-fukuyama

data/test/garbage_02.html

@@ -1 +0,0 @@
-<html><lolipop>1</lolipop>

data/test/garbage_03.html

@@ -1,8 +0,0 @@
-<html>
-<body>
-<div>
-<h1>Really, dude</h1>
-This doesn't work.
-</div>
-</body>
-</html>

data/test/garbage_04.html

@@ -1,9 +0,0 @@
-<html>
-<body>
-<div>
-<h1>zzz</h1>
-<p>Missing <img src='yobo'/> inline image.</p>
-(hidden text)
-</div>
-</body>
-</html>

data/test/helper.rb

@@ -1,3 +0,0 @@
-require_relative 'helper_trestle'
-
-require 'digest/md5'

data/test/helper_trestle.rb

@@ -1,37 +0,0 @@
-# :erb:
-# Various staff for minitest. Include this file into your 'helper.rb'.
-
-require 'fileutils'
-include FileUtils
-
-require_relative '../lib/podgraph/trestle'
-include Podgraph
-
-# don't run tests automatically if they were invoked as 'gem check -t ...'
-if $0 =~ /gem/
-  require 'minitest/unit'
-else
-  require 'minitest/autorun'
-end
-
-# Return the right directory for (probably executable) _c_.
-def cmd(c)
-  case File.basename(Dir.pwd)
-  when Meta::NAME.downcase
-    # test probably is executed from the Rakefile
-    Dir.chdir('test')
-  when 'test'
-    # we are in the test directory, there is nothing special to do
-  else
-    # tests were invoked by 'gem check -t podgraph'
-    begin
-      Dir.chdir(Trestle.gem_libdir + '/../../test')
-    rescue
-      raise "running tests from '#{Dir.pwd}' isn't supported: #{$!}"
-    end
-  end
-
-  '../bin/' + c
-end
-
-# Don't remove this: falsework/0.2.2/naive/2010-12-26T04:50:00+02:00