pidgin2adium 3.0.0 → 3.0.1

data/lib/pidgin2adium/balance_tags.rb

@@ -0,0 +1,118 @@
+module Pidgin2Adium
+  # Balances tags of string using a modified stack. Returns a balanced
+  # string, but also affects the text passed into it!
+  # Use text = balance_tags(text).
+
+  # From Wordpress's formatting.php; rewritten in Ruby by Gabe
+  # Berke-Williams, 2009.
+  # Author:: Leonard Lin <leonard@acm.org>
+  # License:: GPL v2.0
+  # Copyright:: November 4, 2001
+  def Pidgin2Adium.balance_tags( text )
+    tagstack = []
+    stacksize = 0
+    tagqueue = ''
+    newtext = ''
+    single_tags = %w{br hr img input meta} # Known single-entity/self-closing tags
+    #nestable_tags = %w{blockquote div span} # Tags that can be immediately nested within themselves
+    nestable_tags = %w{blockquote div span font} # Tags that can be immediately nested within themselves
+    # 1: tagname, with possible leading "/"
+    # 2: attributes
+    tag_regex = /<(\/?\w*)\s*([^>]*)>/
+
+      # WP bug fix for comments - in case you REALLY meant to type '< !--'
+      text.gsub!('< !--', '<    !--')
+
+    # WP bug fix for LOVE <3 (and other situations with '<' before a number)
+    text.gsub!(/<([0-9]{1})/, '&lt;\1')
+
+    while ( pos = (text =~ tag_regex) )
+      newtext << tagqueue
+      tag = $1.downcase
+      attributes = $2
+      matchlen = $~[0].size
+
+      # clear the shifter
+      tagqueue = ''
+      # Pop or Push
+      if (tag[0,1] == "/") # End Tag
+        tag.slice!(0,1)
+        # if too many closing tags
+        if(stacksize <= 0)
+          tag = ''
+          #or close to be safe: tag = '/' << tag
+        elsif (tagstack[stacksize - 1] == tag) # found closing tag
+          # if stacktop value == tag close value then pop
+          tag = '</' << tag << '>' # Close Tag
+          # Pop
+          tagstack.pop
+          stacksize -= 1
+        else # closing tag not at top, search for it
+          (stacksize-1).downto(0) do |j|
+            if (tagstack[j] == tag)
+              # add tag to tagqueue
+              ss = stacksize - 1
+              ss.downto(j) do |k|
+                tagqueue << '</' << tagstack.pop << '>'
+                stacksize -= 1
+              end
+              break
+            end
+          end
+          tag = ''
+        end
+      else
+        # Begin Tag
+
+        # Tag Cleaning
+        if( (attributes[-1,1] == '/') || (tag == '') )
+          # If: self-closing or '', don't do anything.
+        elsif ( single_tags.include?(tag) )
+          # ElseIf: it's a known single-entity tag but it doesn't close itself, do so
+          attributes << '/'
+        else
+          # Push the tag onto the stack
+          # If the top of the stack is the same as the tag we want to push, close previous tag
+          if ((stacksize > 0) &&
+              ! nestable_tags.include?(tag) &&
+              (tagstack[stacksize - 1] == tag))
+            tagqueue = '</' << tagstack.pop << '>'
+            stacksize -= 1
+          end
+          tagstack.push(tag)
+          stacksize += 1
+        end
+
+        # Attributes
+        if(attributes != '')
+          attributes = ' ' << attributes
+        end
+        tag = '<' << tag << attributes << '>'
+        #If already queuing a close tag, then put this tag on, too
+        if (tagqueue)
+          tagqueue << tag
+          tag = ''
+        end
+      end
+      newtext << text[0,pos] << tag
+      text = text[pos+matchlen, text.length - (pos+matchlen)]
+    end
+
+    # Clear Tag Queue
+    newtext << tagqueue
+
+    # Add Remaining text
+    newtext << text
+
+    # Empty Stack
+    tagstack.reverse_each do |t|
+      newtext << '</' << t << '>' # Add remaining tags to close
+    end
+
+    # WP fix for the bug with HTML comments
+    newtext.gsub!("< !--", "<!--")
+    newtext.gsub!("<    !--", "< !--")
+
+    return newtext
+  end
+end

data/lib/pidgin2adium/log_converter.rb

@@ -1,72 +1,72 @@
 require 'pidgin2adium'
 
 module Pidgin2Adium
-    # An easy way to batch-process a directory. Used by the pidgin2adium
-    # command-line script.
-    class LogConverter
-	include Pidgin2Adium
-	# You can add options using the _opts_ hash, which can have the
-	# following keys, all of which are optional:
-	# * *overwrite*: If true, then overwrite even if log is found.
-	#   Defaults to false.
-	# * *output_dir*: The top-level dir to put the logs in.
-	#   Logs under output_dir are still each in their own folders, etc.
-	#   Defaults to Pidgin2Adium::ADIUM_LOG_DIR
-	def initialize(pidgin_log_dir, aliases, opts = {})
-	    # parse_and_generate will process it for us
-	    @opts = opts
+  # An easy way to batch-process a directory. Used by the pidgin2adium
+  # command-line script.
+  class LogConverter
+    include Pidgin2Adium
+    # You can add options using the _opts_ hash, which can have the
+    # following keys, all of which are optional:
+    # * *overwrite*: If true, then overwrite even if log is found.
+    #   Defaults to false.
+    # * *output_dir*: The top-level dir to put the logs in.
+    #   Logs under output_dir are still each in their own folders, etc.
+    #   Defaults to Pidgin2Adium::ADIUM_LOG_DIR
+    def initialize(pidgin_log_dir, aliases, opts = {})
+      # parse_and_generate will process it for us
+      @opts = opts
 
-	    @pidgin_log_dir = File.expand_path(pidgin_log_dir)
-	    @my_aliases = aliases
+      @pidgin_log_dir = File.expand_path(pidgin_log_dir)
+      @my_aliases = aliases
 
-	    unless File.directory?(@pidgin_log_dir)
-		puts "Source directory #{@pidgin_log_dir} does not exist or is not a directory."
-		raise Errno::ENOENT
-	    end
-	end
+      unless File.directory?(@pidgin_log_dir)
+        puts "Source directory #{@pidgin_log_dir} does not exist or is not a directory."
+        raise Errno::ENOENT
+      end
+    end
 
-	# Runs Pidgin2Adium::parse_and_generate on every log file in directory
-	# provided in new, then deletes Adium's search indexes to force
-	# it to rescan logs on startup.
-	def start
-	    log_msg "Begin converting."
-	    begin
-		files_path = get_all_chat_files(@pidgin_log_dir)
-	    rescue Errno::EACCES => bang
-		error("Sorry, permission denied for getting Pidgin chat files from #{@pidgin_log_dir}.")
-		error("Details: #{bang.message}")
-		raise Errno::EACCES
-	    end
+    # Runs Pidgin2Adium::parse_and_generate on every log file in directory
+    # provided in new, then deletes Adium's search indexes to force
+    # it to rescan logs on startup.
+    def start
+      log_msg "Begin converting."
+      begin
+        files_path = get_all_chat_files(@pidgin_log_dir)
+      rescue Errno::EACCES => bang
+        error("Sorry, permission denied for getting Pidgin chat files from #{@pidgin_log_dir}.")
+        error("Details: #{bang.message}")
+        raise Errno::EACCES
+      end
 
-	    total_files = files_path.size
-	    total_successes = 0
-	    log_msg("#{total_files} files to convert.")
-	    files_path.each_with_index do |fname, i|
-		log_msg(
-		    sprintf("[%d/%d] Converting %s...",
-			(i+1), total_files, fname)
-		)
-		result = parse_and_generate(fname, @my_aliases, @opts)
-		total_successes += 1 if result == true
-	    end
+      total_files = files_path.size
+      total_successes = 0
+      log_msg("#{total_files} files to convert.")
+      files_path.each_with_index do |fname, i|
+        log_msg(
+          sprintf("[%d/%d] Converting %s...",
+                  (i+1), total_files, fname)
+        )
+        result = parse_and_generate(fname, @my_aliases, @opts)
+        total_successes += 1 if result == true
+      end
 
-	    delete_search_indexes()
+      delete_search_indexes()
 
-	    log_msg "Finished converting! Converted #{total_successes} files of #{total_files} total."
-	    puts "Minor error messages:"
-	    puts @@oops_messages.join("\n")
-	    puts "Major error messages:"
-	    puts @@error_messages.join("\n")
-	end
+      log_msg "Finished converting! Converted #{total_successes} files of #{total_files} total."
+      puts "Minor error messages:"
+      puts @@oops_messages.join("\n")
+      puts "Major error messages:"
+      puts @@error_messages.join("\n")
+    end
 
-	###########
-	private
-	###########
+    ###########
+    private
+    ###########
 
-	def get_all_chat_files(dir)
-	    return [] if File.basename(dir) == ".system"
-	    # recurse into each subdir
-	    return (Dir.glob("#{@pidgin_log_dir}/**/*.{htm,html,txt}") - BAD_DIRS)
-	end
+    def get_all_chat_files(dir)
+      return [] if File.basename(dir) == ".system"
+      # recurse into each subdir
+      return (Dir.glob("#{@pidgin_log_dir}/**/*.{htm,html,txt}") - BAD_DIRS)
     end
+  end # END LogConverter class
 end

data/lib/pidgin2adium/log_file.rb

@@ -1,102 +1,102 @@
 require 'fileutils'
 
 module Pidgin2Adium
-    # A holding object for the result of LogParser.parse.  It makes the
-    # instance variable @chat_lines available, which is an array of Message
-    # subclass instances (XMLMessage, Event, etc.)
-    # Here is a list of the instance variables for each class in @chat_lines:
-    # 
-    # <b>All of these variables are read/write.</b>
-    # All::		 sender, time, buddy_alias
-    # XMLMessage::	 body
-    # AutoReplyMessage:: body
-    # Event::		 body, event_type
-    # StatusMessage::	 status
-    class LogFile
-	include Pidgin2Adium
-	def initialize(chat_lines, service, user_SN, partner_SN, adium_chat_time_start)
-	    @chat_lines = chat_lines
-	    @user_SN = user_SN
-	    @partner_SN = partner_SN
-	    @adium_chat_time_start = adium_chat_time_start
+  # A holding object for the result of LogParser.parse.  It makes the
+  # instance variable @chat_lines available, which is an array of Message
+  # subclass instances (XMLMessage, Event, etc.)
+  # Here is a list of the instance variables for each class in @chat_lines:
+  #
+  # <b>All of these variables are read/write.</b>
+  # All::		 sender, time, buddy_alias
+  # XMLMessage::	 body
+  # AutoReplyMessage:: body
+  # Event::		 body, event_type
+  # StatusMessage::	 status
+  class LogFile
+    include Pidgin2Adium
+    def initialize(chat_lines, service, user_SN, partner_SN, adium_chat_time_start)
+      @chat_lines = chat_lines
+      @user_SN = user_SN
+      @partner_SN = partner_SN
+      @adium_chat_time_start = adium_chat_time_start
 
-	    # @chat_str is generated when to_s is called
-	    @chat_str = nil
-	    
-	    # key is for Pidgin, value is for Adium
-	    # Just used for <service>.<screenname> in directory structure
-	    service_name_map = {'aim' => 'AIM',
-		'jabber' =>'Jabber',
-		'gtalk'=> 'GTalk',
-		'icq' => 'ICQ',
-		'qq' => 'QQ',
-		'msn' => 'MSN',
-		'yahoo' => 'Yahoo!'}
-	    
-	    @service = service_name_map[service.downcase]
-	end
-	
-	attr_reader :chat_lines, :service, :user_SN, :partner_SN, :adium_chat_time_start
+      # @chat_str is generated when to_s is called
+      @chat_str = nil
 
-	# Returns contents of log file
-	def to_s
-	    if @chat_str.nil?
-		# Faster than inject() or each()
-		@chat_str = @chat_lines.map{|l| l.to_s }.join
-	    end
-	    return @chat_str
-	end
-	
-	def each(&blk)
-	    @chat_lines.each{|l| yield l }
-	end
+      # key is for Pidgin, value is for Adium
+      # Just used for <service>.<screenname> in directory structure
+      service_name_map = {'aim' => 'AIM',
+        'jabber' =>'Jabber',
+        'gtalk'=> 'GTalk',
+        'icq' => 'ICQ',
+        'qq' => 'QQ',
+        'msn' => 'MSN',
+        'yahoo' => 'Yahoo!'}
 
-	# Set overwrite=true to create a logfile even if logfile already exists.
-	# Returns one of:
-	# * false (if an error occurred),
-	# * Pidgin2Adium::FILE_EXISTS if the file to be generated already exists and overwrite=false, or
-	# * the path to the new Adium log file.
-	def write_out(overwrite = false, output_dir_base = ADIUM_LOG_DIR)
-	    # output_dir_base + "/buddyname (2009-08-04T18.38.50-0700).chatlog"
-	    output_dir = File.join(output_dir_base, "#{@service}.#{@user_SN}", @partner_SN, "#{@partner_SN} (#{@adium_chat_time_start}).chatlog")
-	    # output_dir + "/buddyname (2009-08-04T18.38.50-0700).chatlog/buddyname (2009-08-04T18.38.50-0700).xml"
-	    output_path = output_dir + '/' + "#{@partner_SN} (#{@adium_chat_time_start}).xml"
-	    begin
-		FileUtils.mkdir_p(output_dir)
-	    rescue => bang
-		error "Could not create destination directory for log file. (Details: #{bang.class}: #{bang.message})"
-		return false
-	    end
-	    if overwrite
-		unless File.exist?(output_path) 
-		    # File doesn't exist, but maybe it does with a different
-		    # time zone. Check for a file that differs only in time
-		    # zone and, if found, change @output_path to target it.
-		    maybe_matches = Dir.glob(output_dir_base + '/' << File.basename(output_path).sub(/-\d{4}\)\.chatlog$/, '') << '/*')
-		    unless maybe_matches.empty?
-			output_path = maybe_matches[0]
-		    end
-		end
-	    else
-		if File.exist?(output_path)
-		    return FILE_EXISTS
-		end
-	    end
+      @service = service_name_map[service.downcase]
+    end
+
+    attr_reader :chat_lines, :service, :user_SN, :partner_SN, :adium_chat_time_start
+
+    # Returns contents of log file
+    def to_s
+      if @chat_str.nil?
+        # Faster than inject() or each()
+        @chat_str = @chat_lines.map{|l| l.to_s }.join
+      end
+      return @chat_str
+    end
+
+    def each(&blk)
+      @chat_lines.each{|l| yield l }
+    end
+
+    # Set overwrite=true to create a logfile even if logfile already exists.
+    # Returns one of:
+    # * false (if an error occurred),
+    # * Pidgin2Adium::FILE_EXISTS if the file to be generated already exists and overwrite=false, or
+    # * the path to the new Adium log file.
+    def write_out(overwrite = false, output_dir_base = ADIUM_LOG_DIR)
+      # output_dir_base + "/buddyname (2009-08-04T18.38.50-0700).chatlog"
+      output_dir = File.join(output_dir_base, "#{@service}.#{@user_SN}", @partner_SN, "#{@partner_SN} (#{@adium_chat_time_start}).chatlog")
+      # output_dir + "/buddyname (2009-08-04T18.38.50-0700).chatlog/buddyname (2009-08-04T18.38.50-0700).xml"
+      output_path = output_dir + '/' + "#{@partner_SN} (#{@adium_chat_time_start}).xml"
+      begin
+        FileUtils.mkdir_p(output_dir)
+      rescue => bang
+        error "Could not create destination directory for log file. (Details: #{bang.class}: #{bang.message})"
+        return false
+      end
+      if overwrite
+        unless File.exist?(output_path)
+          # File doesn't exist, but maybe it does with a different
+          # time zone. Check for a file that differs only in time
+          # zone and, if found, change @output_path to target it.
+          maybe_matches = Dir.glob(output_dir_base + '/' << File.basename(output_path).sub(/-\d{4}\)\.chatlog$/, '') << '/*')
+          unless maybe_matches.empty?
+            output_path = maybe_matches[0]
+          end
+        end
+      else
+        if File.exist?(output_path)
+          return FILE_EXISTS
+        end
+      end
 
-	    begin
-		outfile = File.new(output_path, 'w')
-	    rescue => bang
-		error "Could not open log file for writing. (Details: #{bang.class}: #{bang.message})"
-		return false
-	    end
+      begin
+        outfile = File.new(output_path, 'w')
+      rescue => bang
+        error "Could not open log file for writing. (Details: #{bang.class}: #{bang.message})"
+        return false
+      end
 
-	    # no \n before </chat> because @chat_str (from to_s) has it already
-	    outfile.printf('<?xml version="1.0" encoding="UTF-8" ?>'<<"\n"+
-			   '<chat xmlns="http://purl.org/net/ulf/ns/0.4-02" account="%s" service="%s">'<<"\n"<<'%s</chat>',
-			   @user_SN, @service, self.to_s)
-	    outfile.close
+      # no \n before </chat> because @chat_str (from to_s) has it already
+      outfile.printf('<?xml version="1.0" encoding="UTF-8" ?>'<<"\n"+
+                     '<chat xmlns="http://purl.org/net/ulf/ns/0.4-02" account="%s" service="%s">'<<"\n"<<'%s</chat>',
+                     @user_SN, @service, self.to_s)
+      outfile.close
 
-	    return output_path
-	end
+      return output_path
     end
+  end # END LogFile class
 end

data/lib/pidgin2adium/log_parser.rb

@@ -1,6 +1,6 @@
 # Contains the class BasicParser and its subclasses, HtmlLogParser and
 # TextFileParser, which parse the file passed into it and return a LogFile
-# object. 
+# object.
 #
 # Please use Pidgin2Adium.parse or Pidgin2Adium.parse_and_generate instead of
 # using these classes directly.
@@ -11,607 +11,608 @@ require 'balance_tags_c'
 require 'pidgin2adium/log_file'
 
 module Pidgin2Adium
-    # Empty class. Raise'd by LogParser if the first line of a log is not
-    # parseable.
-    class InvalidFirstLineError < StandardError; end
-
-    # BasicParser is a base class. Its subclasses are TextLogParser and
-    # HtmlLogParser.
-    #
-    # Please use Pidgin2Adium.parse or Pidgin2Adium.parse_and_generate instead of
-    # using this class directly.
-    class BasicParser
-	include Pidgin2Adium
-	def initialize(src_path, user_aliases)
-	    @src_path = src_path
-	    # Whitespace is removed for easy matching later on.
-	    @user_aliases = user_aliases.split(',').map!{|x| x.downcase.gsub(/\s+/,'') }.uniq
-	    # @user_alias is set each time get_sender_by_alias is called. It is a non-normalized
-	    # alias.
-	    # Set an initial value just in case the first message doesn't give
-	    # us an alias.
-	    @user_alias = user_aliases.split(',')[0]
-
-	    @tz_offset = get_time_zone_offset()
-	   
-	    file = File.new(@src_path, 'r')
-	    @first_line = file.readline
-	    @file_content = file.read
-	    file.close
-
-	    # Time regexes must be set before pre_parse().
-	    # "4/18/2007 11:02:00 AM" => %w{4, 18, 2007, 11, 02, 00, AM}
-	    # ONLY used (if at all) in first line of chat ("Conversation with...at...")
-	    @time_regex_first_line = %r{^(\d{1,2})/(\d{1,2})/(\d{4}) (\d{1,2}):(\d{2}):(\d{2}) ([AP]M)$}
-	    # "2007-04-17 12:33:13" => %w{2007, 04, 17, 12, 33, 13}
-	    @time_regex = /^(\d{4})-(\d{2})-(\d{2}) (\d{2}):(\d{2}):(\d{2})$/
-	    # sometimes a line in a chat doesn't have a full timestamp
-	    # "04:22:05 AM" => %w{04 22 05 AM}
-	    @minimal_time_regex = /^(\d{1,2}):(\d{2}):(\d{2})( [AP]M)?$/
-
-	    # Whether or not the first line is parseable.
-	    @first_line_is_valid = true
-	    begin
-		@service,
-		@user_SN,
-		@partner_SN,
-		# @basic_time_info is for files that only have the full
-		# timestamp at the top; we can use it to fill in the minimal
-		# per-line timestamps. It has only 3 elements (year, month,
-		# dayofmonth) because you should be able to fill everything
-		# else in. If you can't, something's wrong.
-		@basic_time_info,
-		# When the chat started, in Adium's format
-		@adium_chat_time_start = pre_parse()
-	    rescue InvalidFirstLineError
-		@first_line_is_valid = false
-		error("Failed to parse, invalid first line: #{@src_path}")
-		return # stop processing
-	    end
-	    
-	    # @status_map, @lib_purple_events, and @events are used in
-	    # create_status_or_event_msg
-	    @status_map = {
-		/(.+) logged in\.$/ => 'online',
-		/(.+) logged out\.$/ => 'offline',
-		/(.+) has signed on\.$/ => 'online',
-		/(.+) has signed off\.$/ => 'offline',
-		/(.+) has gone away\.$/ => 'away',
-		/(.+) is no longer away\.$/ => 'available',
-		/(.+) has become idle\.$/ => 'idle',
-		/(.+) is no longer idle\.$/ => 'available'
-	    }
-
-	    # lib_purple_events are all of event_type libPurple
-	    @lib_purple_events = [
-		# file transfer
-		/Starting transfer of .+ from (.+)/,
-		/^Offering to send .+ to (.+)$/,
-		/(.+) is offering to send file/,
-		/^Transfer of file .+ complete$/,
-		/Error reading|writing|accessing .+: .+/,
-		/You cancell?ed the transfer of/,
-		/File transfer cancelled/,
-		/(.+?) cancell?ed the transfer of/,
-		/(.+?) cancelled the file transfer/,
-		# Direct IM - actual (dis)connect events are their own types
-		/^Attempting to connect to (.+) at .+ for Direct IM\./,
-		/^Asking (.+) to connect to us at .+ for Direct IM\./,
-		/^Attempting to connect via proxy server\.$/,
-		/^Direct IM with (.+) failed/,
-		# encryption
-		/Received message encrypted with wrong key/,
-		/^Requesting key\.\.\.$/,
-		/^Outgoing message lost\.$/,
-		/^Conflicting Key Received!$/,
-		/^Error in decryption- asking for resend\.\.\.$/,
-		/^Making new key pair\.\.\.$/,
-		# sending errors
-		/^Last outgoing message not received properly- resetting$/,
-		/Resending\.\.\./,
-		# connection errors
-		/Lost connection with the remote user:.+/,
-		# chats
-		/^.+ entered the room\.$/,
-		/^.+ left the room\.$/
-	    ]
-
-	    # non-libpurple events
-	    # Each key maps to an event_type string. The keys will be matched against a line of chat
-	    # and the partner's alias will be in regex group 1, IF the alias is matched.
-	    @event_map = {
-		# .+ is not an alias, it's a proxy server so no grouping
-		/^Attempting to connect to .+\.$/ => 'direct-im-connect',
-		# NB: pidgin doesn't track when Direct IM is disconnected, AFAIK
-		/^Direct IM established$/ => 'directIMConnected',
-		/Unable to send message/ => 'chat-error',
-		/You missed .+ messages from (.+) because they were too large/ => 'chat-error',
-		/User information not available/ => 'chat-error'
-	    }
-
-	    @ignore_events = [
-		# Adium ignores SN/alias changes. 
-		/^.+? is now known as .+?\.<br\/?>$/
-	    ]
-	end
-
-	# This method returns a LogFile instance, or false if an error occurred.
-	def parse
-	    return false unless @first_line_is_valid
-	    @file_content = cleanup(@file_content).split("\n")
-
-	    @file_content.map! do |line|
-		# "next" returns nil which is removed by compact
-		next if line =~ /^\s+$/
-		if line =~ @line_regex
-		    create_msg($~.captures)
-		elsif line =~ @line_regex_status
-		    msg = create_status_or_event_msg($~.captures)
-		    # Error occurred while parsing
-		    return false if msg == false
-		else
-		    error "Could not parse line:"
-		    p line
-		    return false
-		end
-	    end
-	    @file_content.compact!
-	    return LogFile.new(@file_content, @service, @user_SN, @partner_SN, @adium_chat_time_start)
-	end
-	# Prevent parse from being called directly from BasicParser, since
-	# it uses subclassing magic.
-	protected :parse
-
-	#################
-	private
-	#################
-
-	def get_time_zone_offset()
-	    # We must have a tz_offset or else the Adium Chat Log viewer
-	    # doesn't read the date correctly and then:
-	    # 1) the log has an empty start date column in the viewer
-	    # 2) The timestamps are all the same for the whole log
-	    tz_match = /([-\+]\d+)[A-Z]{3}\.(?:txt|htm|html)/.match(@src_path)
-	    if tz_match and tz_match[1]
-		tz_offset = tz_match[1]
-	    else
-		# "-0500" (3d rather than 2d to allow for "+")
-		tz_offset = sprintf('%+03d00', Time.zone_offset(Time.now.zone) / 3600)
-	    end
-	    return tz_offset
-	end
-
-	#--
-	# Adium time format: YYYY-MM-DD\THH:MM:SS[+-]TZ_HRS like:
-	# 2008-10-05T22:26:20-0800
-	# HOWEVER:
-	# If it's the first line, then return it like this (note periods):
-	# 2008-10-05T22.26.20-0800
-	# because it will be used in the filename.
-	#++
-	# Converts a pidgin datestamp to an Adium one.
-	def create_adium_time(time, is_first_line = false)
-	    # parsed_date = [year, month, day, hour, min, sec]
-	    if time =~ @time_regex
-		year, month, day, hour, min, sec = $1.to_i,
-						   $2.to_i,
-						   $3.to_i,
-						   $4.to_i,
-						   $5.to_i,
-						   $6.to_i
-	    elsif is_first_line and time =~ @time_regex_first_line
-		hour = $4.to_i
-		if $7 == 'PM' and hour != 12
-		    hour += 12
-		end
-		year, month, day, min, sec = $3.to_i, # year
-					     $1.to_i, # month
-					     $2.to_i, # day
-						        # already did hour
-					     $5.to_i, # minutes
-					     $6.to_i  # seconds
-	    elsif time =~ @minimal_time_regex
-		# "04:22:05" => %w{04 22 05}
-		hour = $1.to_i
-		if $4 == 'PM' and hour != 12
-		    hour += 12
-		end
-		year, month, day = @basic_time_info
-		min = $2.to_i
-		sec = $3.to_i
-	    else
-		error("You have found an odd timestamp. Please report it to the developer.")
-		log_msg("The timestamp: #{time}")
-		log_msg("Continuing...")
-		year,month,day,hour,min,sec = ParseDate.parsedate(time)
-	    end
-	    if is_first_line
-		adium_time = Time.local(year,month,day,hour,min,sec).strftime("%Y-%m-%dT%H.%M.%S#{@tz_offset}")
-	    else
-		adium_time = Time.local(year,month,day,hour,min,sec).strftime("%Y-%m-%dT%H:%M:%S#{@tz_offset}")
-	    end
-	    return adium_time
-	end
-
-	# Extract required data from the file. Run by parse.
-	def pre_parse
-	    # Deal with first line.
-	    
-	    # the first line is special. It tells us (in order of regex groups):
-	    # 1) who we're talking to 
-	    # 2) what time/date
-	    # 3) what SN we used
-	    # 4) what protocol (AIM, icq, jabber...)
-	    first_line_match = /Conversation with (.+?) at (.+?) on (.+?) \((.+?)\)/.match(@first_line)
-	    if first_line_match.nil?
-		raise InvalidFirstLineError
-	    else
-		service = first_line_match[4]
-		# @user_SN is normalized to avoid "AIM.name" and "AIM.na me" folders
-		user_SN = first_line_match[3].downcase.tr(' ', '')
-		partner_SN = first_line_match[1]
-		pidgin_chat_time_start = first_line_match[2]
-		basic_time_info = case pidgin_chat_time_start
-				  when @time_regex: [$1.to_i, $2.to_i, $3.to_i]
-				  when @time_regex_first_line: [$3.to_i, $1.to_i, $2.to_i]
-				  end
-		adium_chat_time_start = create_adium_time(pidgin_chat_time_start, true)
-		return [service,
-			user_SN,
-			partner_SN,
-			basic_time_info,
-			adium_chat_time_start]
-	    end
-	end
-
-	def get_sender_by_alias(alias_name)
-	    no_action = alias_name.sub(/^\*{3}/, '')
-	    if @user_aliases.include? no_action.downcase.gsub(/\s+/, '')
-		# Set the current alias being used of the ones in @user_aliases
-		@user_alias = no_action
-		return @user_SN
-	    else
-		return @partner_SN
-	    end
-	end
-
-	#--
-	# create_msg takes an array of captures from matching against
-	# @line_regex and returns a Message object or one of its subclasses.
-	# It can be used for TextLogParser and HtmlLogParser because both of
-	# them return data in the same indexes in the matches array.
-	#++
-	def create_msg(matches)
-	    msg = nil
-	    # Either a regular message line or an auto-reply/away message.
-	    time = create_adium_time(matches[0])
-	    buddy_alias = matches[1]
-	    sender = get_sender_by_alias(buddy_alias)
-	    body = matches[3]
-	    if matches[2] # auto-reply
-		msg = AutoReplyMessage.new(sender, time, buddy_alias, body)
-	    else
-		# normal message
-		msg = XMLMessage.new(sender, time, buddy_alias, body)
-	    end
-	    return msg
-	end
-
-	#--
-	# create_status_or_event_msg takes an array of +MatchData+ captures from
-	# matching against @line_regex_status and returns an Event or Status.
-	# Returns nil if it's a message that should be ignored, or false if an
-	# error occurred.
-	#++
-	def create_status_or_event_msg(matches)
-	    # ["22:58:00", "BuddyName logged in."]
-	    # 0: time
-	    # 1: status message or event
-	    msg = nil
-	    time = create_adium_time(matches[0])
-	    str = matches[1]
-	    # Return nil, which will get compact'ed out
-	    return nil if @ignore_events.detect{|regex| str =~ regex }
-
-	    regex, status = @status_map.detect{|regex, status| str =~ regex}
-	    if regex and status
-		# Status message
-		buddy_alias = regex.match(str)[1]
-		sender = get_sender_by_alias(buddy_alias)
-		msg = StatusMessage.new(sender, time, buddy_alias, status)
-	    else
-		# Test for event
-		regex = @lib_purple_events.detect{|regex| str =~ regex }
-		event_type = 'libpurpleEvent' if regex
-		unless regex and event_type
-		    # not a libpurple event, try others
-		    if @event_map.detect{|regex,event_type| str =~ regex}
-			regex, event_type = $1, $2
-		    else
-			error(sprintf("Error parsing status or event message, no status or event found: %p", str))
-			return false
-		    end
-		end
-		if regex and event_type
-		    regex_matches = regex.match(str)
-		    # Event message
-		    if regex_matches.size == 1
-			# No alias - this means it's the user
-			buddy_alias = @user_alias
-			sender = @user_SN
-		    else
-			buddy_alias = regex_matches[1]
-			sender = get_sender_by_alias(buddy_alias)
-		    end
-		    msg = Event.new(sender, time, buddy_alias, str, event_type)
-		end
-	    end
-	    return msg
-	end
+  # Empty class. Raise'd by LogParser if the first line of a log is not
+  # parseable.
+  class InvalidFirstLineError < StandardError; end
+
+  # BasicParser is a base class. Its subclasses are TextLogParser and
+  # HtmlLogParser.
+  #
+  # Please use Pidgin2Adium.parse or Pidgin2Adium.parse_and_generate instead of
+  # using this class directly.
+  class BasicParser
+    include Pidgin2Adium
+    def initialize(src_path, user_aliases)
+      @src_path = src_path
+      # Whitespace is removed for easy matching later on.
+      @user_aliases = user_aliases.split(',').map!{|x| x.downcase.gsub(/\s+/,'') }.uniq
+      # @user_alias is set each time get_sender_by_alias is called. It is a non-normalized
+      # alias.
+      # Set an initial value just in case the first message doesn't give
+      # us an alias.
+      @user_alias = user_aliases.split(',')[0]
+
+      @tz_offset = get_time_zone_offset()
+
+      file = File.new(@src_path, 'r')
+      @first_line = file.readline
+      @file_content = file.read
+      file.close
+
+      # Time regexes must be set before pre_parse().
+      # "4/18/2007 11:02:00 AM" => %w{4, 18, 2007, 11, 02, 00, AM}
+      # ONLY used (if at all) in first line of chat ("Conversation with...at...")
+      @time_regex_first_line = %r{^(\d{1,2})/(\d{1,2})/(\d{4}) (\d{1,2}):(\d{2}):(\d{2}) ([AP]M)$}
+      # "2007-04-17 12:33:13" => %w{2007, 04, 17, 12, 33, 13}
+      @time_regex = /^(\d{4})-(\d{2})-(\d{2}) (\d{2}):(\d{2}):(\d{2})$/
+      # sometimes a line in a chat doesn't have a full timestamp
+      # "04:22:05 AM" => %w{04 22 05 AM}
+      @minimal_time_regex = /^(\d{1,2}):(\d{2}):(\d{2})( [AP]M)?$/
+
+      # Whether or not the first line is parseable.
+      @first_line_is_valid = true
+      begin
+        @service,
+        @user_SN,
+        @partner_SN,
+        # @basic_time_info is for files that only have the full
+        # timestamp at the top; we can use it to fill in the minimal
+        # per-line timestamps. It has only 3 elements (year, month,
+        # dayofmonth) because you should be able to fill everything
+        # else in. If you can't, something's wrong.
+        @basic_time_info,
+        # When the chat started, in Adium's format
+        @adium_chat_time_start = pre_parse()
+      rescue InvalidFirstLineError
+        @first_line_is_valid = false
+        error("Failed to parse, invalid first line: #{@src_path}")
+        return # stop processing
+      end
+
+      # @status_map, @lib_purple_events, and @events are used in
+      # create_status_or_event_msg
+      @status_map = {
+        /(.+) logged in\.$/ => 'online',
+        /(.+) logged out\.$/ => 'offline',
+        /(.+) has signed on\.$/ => 'online',
+        /(.+) has signed off\.$/ => 'offline',
+        /(.+) has gone away\.$/ => 'away',
+        /(.+) is no longer away\.$/ => 'available',
+        /(.+) has become idle\.$/ => 'idle',
+        /(.+) is no longer idle\.$/ => 'available'
+      }
+
+      # lib_purple_events are all of event_type libPurple
+      @lib_purple_events = [
+        # file transfer
+        /Starting transfer of .+ from (.+)/,
+        /^Offering to send .+ to (.+)$/,
+        /(.+) is offering to send file/,
+        /^Transfer of file .+ complete$/,
+        /Error reading|writing|accessing .+: .+/,
+        /You cancell?ed the transfer of/,
+        /File transfer cancelled/,
+        /(.+?) cancell?ed the transfer of/,
+        /(.+?) cancelled the file transfer/,
+        # Direct IM - actual (dis)connect events are their own types
+        /^Attempting to connect to (.+) at .+ for Direct IM\./,
+        /^Asking (.+) to connect to us at .+ for Direct IM\./,
+        /^Attempting to connect via proxy server\.$/,
+        /^Direct IM with (.+) failed/,
+        # encryption
+        /Received message encrypted with wrong key/,
+        /^Requesting key\.\.\.$/,
+        /^Outgoing message lost\.$/,
+        /^Conflicting Key Received!$/,
+        /^Error in decryption- asking for resend\.\.\.$/,
+        /^Making new key pair\.\.\.$/,
+        # sending errors
+        /^Last outgoing message not received properly- resetting$/,
+        /Resending\.\.\./,
+        # connection errors
+        /Lost connection with the remote user:.+/,
+        # chats
+        /^.+ entered the room\.$/,
+        /^.+ left the room\.$/
+      ]
+
+      # non-libpurple events
+      # Each key maps to an event_type string. The keys will be matched against a line of chat
+      # and the partner's alias will be in regex group 1, IF the alias is matched.
+      @event_map = {
+        # .+ is not an alias, it's a proxy server so no grouping
+        /^Attempting to connect to .+\.$/ => 'direct-im-connect',
+        # NB: pidgin doesn't track when Direct IM is disconnected, AFAIK
+        /^Direct IM established$/ => 'directIMConnected',
+        /Unable to send message/ => 'chat-error',
+        /You missed .+ messages from (.+) because they were too large/ => 'chat-error',
+        /User information not available/ => 'chat-error'
+      }
+
+      @ignore_events = [
+        # Adium ignores SN/alias changes.
+        /^.+? is now known as .+?\.<br\/?>$/
+      ]
     end
 
-    # Please use Pidgin2Adium.parse or Pidgin2Adium.parse_and_generate instead of
-    # using this class directly.
-    class TextLogParser < BasicParser
-	def initialize(src_path, user_aliases)
-	    super(src_path, user_aliases)
-	    @timestamp_rx = '\((\d{1,2}:\d{1,2}:\d{1,2})\)'
-	    
-	    # @line_regex matches a line in a TXT log file other than the first
-	    # @line_regex matchdata:
-	    # 0: timestamp
-	    # 1: screen name or alias, if alias set
-	    # 2: "<AUTO-REPLY>" or nil
-	    # 3: message body
-	    @line_regex = /#{@timestamp_rx} (.*?) ?(<AUTO-REPLY>)?: (.*)/o
-	    # @line_regex_status matches a status line
-	    # @line_regex_status matchdata:
-	    # 0: timestamp
-	    # 1: status message
-	    @line_regex_status = /#{@timestamp_rx} ([^:]+)/o
-	end
-
-	public :parse
-
-	#################
-	private
-	#################
-
-	def cleanup(text)
-	    text.tr!("\r", '')
-	    # Replace newlines with "<br/>" unless they end a chat line.
-	    text.gsub!(/\n(?!#{@timestamp_rx}|\Z)/, '<br/>')
-	    # Escape entities since this will be in XML
-	    text.gsub!('&', '&amp;') # escape '&' first
-	    text.gsub!('<', '&lt;')
-	    text.gsub!('>', '&gt;')
-	    text.gsub!('"', '&quot;')
-	    text.gsub!("'", '&apos;')
-	    return text
-	end
+    # This method returns a LogFile instance, or false if an error occurred.
+    def parse
+      return false unless @first_line_is_valid
+      @file_content = cleanup(@file_content).split("\n")
+
+      @file_content.map! do |line|
+        # "next" returns nil which is removed by compact
+        next if line =~ /^\s+$/
+        if line =~ @line_regex
+          create_msg($~.captures)
+        elsif line =~ @line_regex_status
+          msg = create_status_or_event_msg($~.captures)
+          # Error occurred while parsing
+          return false if msg == false
+        else
+          error "Could not parse line:"
+          p line
+          return false
+        end
+      end
+      @file_content.compact!
+      return LogFile.new(@file_content, @service, @user_SN, @partner_SN, @adium_chat_time_start)
+    end
+    # Prevent parse from being called directly from BasicParser, since
+    # it uses subclassing magic.
+    protected :parse
+
+    #################
+    private
+    #################
+
+    def get_time_zone_offset()
+      # We must have a tz_offset or else the Adium Chat Log viewer
+      # doesn't read the date correctly and then:
+      # 1) the log has an empty start date column in the viewer
+      # 2) The timestamps are all the same for the whole log
+      tz_match = /([-\+]\d+)[A-Z]{3}\.(?:txt|htm|html)/.match(@src_path)
+      if tz_match and tz_match[1]
+        tz_offset = tz_match[1]
+      else
+        # "-0500" (3d rather than 2d to allow for "+")
+        tz_offset = sprintf('%+03d00', Time.zone_offset(Time.now.zone) / 3600)
+      end
+      return tz_offset
+    end
+
+    #--
+    # Adium time format: YYYY-MM-DD\THH:MM:SS[+-]TZ_HRS like:
+    # 2008-10-05T22:26:20-0800
+    # HOWEVER:
+    # If it's the first line, then return it like this (note periods):
+    # 2008-10-05T22.26.20-0800
+    # because it will be used in the filename.
+    #++
+    # Converts a pidgin datestamp to an Adium one.
+    def create_adium_time(time, is_first_line = false)
+      # parsed_date = [year, month, day, hour, min, sec]
+      if time =~ @time_regex
+        year, month, day, hour, min, sec = $1.to_i,
+                                           $2.to_i,
+                                           $3.to_i,
+                                           $4.to_i,
+                                           $5.to_i,
+                                           $6.to_i
+      elsif is_first_line and time =~ @time_regex_first_line
+        hour = $4.to_i
+        if $7 == 'PM' and hour != 12
+          hour += 12
+        end
+        year, month, day, min, sec = $3.to_i, # year
+                                     $1.to_i, # month
+                                     $2.to_i, # day
+                                     # already did hour
+                                     $5.to_i, # minutes
+                                     $6.to_i  # seconds
+      elsif time =~ @minimal_time_regex
+        # "04:22:05" => %w{04 22 05}
+        hour = $1.to_i
+        if $4 == 'PM' and hour != 12
+          hour += 12
+        end
+        year, month, day = @basic_time_info
+        min = $2.to_i
+        sec = $3.to_i
+      else
+        error("You have found an odd timestamp. Please report it to the developer.")
+        log_msg("The timestamp: #{time}")
+        log_msg("Continuing...")
+        year,month,day,hour,min,sec = ParseDate.parsedate(time)
+      end
+      if is_first_line
+        adium_time = Time.local(year,month,day,hour,min,sec).strftime("%Y-%m-%dT%H.%M.%S#{@tz_offset}")
+      else
+        adium_time = Time.local(year,month,day,hour,min,sec).strftime("%Y-%m-%dT%H:%M:%S#{@tz_offset}")
+      end
+      return adium_time
+    end
+
+    # Extract required data from the file. Run by parse.
+    def pre_parse
+      # Deal with first line.
+
+      # the first line is special. It tells us (in order of regex groups):
+      # 1) who we're talking to
+      # 2) what time/date
+      # 3) what SN we used
+      # 4) what protocol (AIM, icq, jabber...)
+      first_line_match = /Conversation with (.+?) at (.+?) on (.+?) \((.+?)\)/.match(@first_line)
+      if first_line_match.nil?
+        raise InvalidFirstLineError
+      else
+        service = first_line_match[4]
+        # @user_SN is normalized to avoid "AIM.name" and "AIM.na me" folders
+        user_SN = first_line_match[3].downcase.tr(' ', '')
+        partner_SN = first_line_match[1]
+        pidgin_chat_time_start = first_line_match[2]
+        basic_time_info = case pidgin_chat_time_start
+                          when @time_regex then [$1.to_i, $2.to_i, $3.to_i]
+                          when @time_regex_first_line then [$3.to_i, $1.to_i, $2.to_i]
+                          end
+        adium_chat_time_start = create_adium_time(pidgin_chat_time_start, true)
+        return [service,
+          user_SN,
+          partner_SN,
+          basic_time_info,
+          adium_chat_time_start]
+      end
+    end
+
+    def get_sender_by_alias(alias_name)
+      no_action = alias_name.sub(/^\*{3}/, '')
+      if @user_aliases.include? no_action.downcase.gsub(/\s+/, '')
+        # Set the current alias being used of the ones in @user_aliases
+        @user_alias = no_action
+        return @user_SN
+      else
+        return @partner_SN
+      end
+    end
+
+    #--
+    # create_msg takes an array of captures from matching against
+    # @line_regex and returns a Message object or one of its subclasses.
+    # It can be used for TextLogParser and HtmlLogParser because both of
+    # them return data in the same indexes in the matches array.
+    #++
+    def create_msg(matches)
+      msg = nil
+      # Either a regular message line or an auto-reply/away message.
+      time = create_adium_time(matches[0])
+      buddy_alias = matches[1]
+      sender = get_sender_by_alias(buddy_alias)
+      body = matches[3]
+      if matches[2] # auto-reply
+        msg = AutoReplyMessage.new(sender, time, buddy_alias, body)
+      else
+        # normal message
+        msg = XMLMessage.new(sender, time, buddy_alias, body)
+      end
+      return msg
+    end
+
+    #--
+    # create_status_or_event_msg takes an array of +MatchData+ captures from
+    # matching against @line_regex_status and returns an Event or Status.
+    # Returns nil if it's a message that should be ignored, or false if an
+    # error occurred.
+    #++
+    def create_status_or_event_msg(matches)
+      # ["22:58:00", "BuddyName logged in."]
+      # 0: time
+      # 1: status message or event
+      msg = nil
+      time = create_adium_time(matches[0])
+      str = matches[1]
+      # Return nil, which will get compact'ed out
+      return nil if @ignore_events.detect{|regex| str =~ regex }
+
+      regex, status = @status_map.detect{|regex, status| str =~ regex}
+      if regex and status
+        # Status message
+        buddy_alias = regex.match(str)[1]
+        sender = get_sender_by_alias(buddy_alias)
+        msg = StatusMessage.new(sender, time, buddy_alias, status)
+      else
+        # Test for event
+        regex = @lib_purple_events.detect{|regex| str =~ regex }
+        event_type = 'libpurpleEvent' if regex
+        unless regex and event_type
+          # not a libpurple event, try others
+          if @event_map.detect{|regex,event_type| str =~ regex}
+            regex, event_type = $1, $2
+          else
+            error(sprintf("Error parsing status or event message, no status or event found: %p", str))
+            return false
+          end
+        end
+        if regex and event_type
+          regex_matches = regex.match(str)
+          # Event message
+          if regex_matches.size == 1
+            # No alias - this means it's the user
+            buddy_alias = @user_alias
+            sender = @user_SN
+          else
+            buddy_alias = regex_matches[1]
+            sender = get_sender_by_alias(buddy_alias)
+          end
+          msg = Event.new(sender, time, buddy_alias, str, event_type)
+        end
+      end
+      return msg
+    end
+  end # END BasicParser class
+
+  # Please use Pidgin2Adium.parse or Pidgin2Adium.parse_and_generate instead of
+  # using this class directly.
+  class TextLogParser < BasicParser
+    def initialize(src_path, user_aliases)
+      super(src_path, user_aliases)
+      @timestamp_rx = '\((\d{1,2}:\d{1,2}:\d{1,2})\)'
+
+      # @line_regex matches a line in a TXT log file other than the first
+      # @line_regex matchdata:
+      # 0: timestamp
+      # 1: screen name or alias, if alias set
+      # 2: "<AUTO-REPLY>" or nil
+      # 3: message body
+      @line_regex = /#{@timestamp_rx} (.*?) ?(<AUTO-REPLY>)?: (.*)/o
+
+      # @line_regex_status matches a status line
+      # @line_regex_status matchdata:
+      # 0: timestamp
+      # 1: status message
+      @line_regex_status = /#{@timestamp_rx} ([^:]+)/o
     end
 
-    # Please use Pidgin2Adium.parse or Pidgin2Adium.parse_and_generate instead
-    # of using this class directly.
-    class HtmlLogParser < BasicParser
-	def initialize(src_path, user_aliases) 
-	    super(src_path, user_aliases)
-	    @timestamp_rx = '\(((?:\d{4}-\d{2}-\d{2} )?\d{1,2}:\d{1,2}:\d{1,2}(?: [AP]M)?)\)'
-	    
-	    # @line_regex matches a line in an HTML log file other than the
-	    # first time matches on either "2008-11-17 14:12" or "14:12"
-	    # @line_regex match obj:
-	    # 0: timestamp, extended or not
-	    # 1: screen name or alias, if alias set
-	    # 2: "&lt;AUTO-REPLY&gt;" or nil
-	    # 3: message body
-	    # The ":" is optional to allow for strings like "(17:12:21) <b>***Gabe B-W</b> is confused<br/>"
-	    @line_regex = /#{@timestamp_rx} ?<b>(.+?) ?(&lt;AUTO-REPLY&gt;)?:?<\/b> ?(.+)<br ?\/>/o
-	    # @line_regex_status matches a status line
-	    # @line_regex_status match obj:
-	    # 0: timestamp
-	    # 1: status message
-	    @line_regex_status = /#{@timestamp_rx} ?<b> (.+)<\/b><br ?\/>/o
-	end
-	
-	public :parse
-
-	#################
-	private
-	#################
-
-	# Returns a cleaned string.
-	# Removes the following tags from _text_:
-	# * html
-	# * body
-	# * font
-	# * a with no innertext, e.g. <a href="blah"></a>
-	# And removes the following style declarations:
-	# * color: #000000 (just turns text black)
-	# * font-family
-	# * font-size
-	# * background
-	# * em (really it's changed to <span style="font-style: italic;">)
-	# Since each <span> has only one style declaration, spans with these
-	# declarations are removed (but the text inside them is preserved).
-	def cleanup(text)
-	    # Sometimes this is in there. I don't know why.
-	    text.gsub!(%r{&lt;/FONT HSPACE='\d'>}, '')
-	    # We can remove <font> safely since Pidgin and Adium both show bold
-	    # using <span style="font-weight: bold;"> except Pidgin uses single
-	    # quotes while Adium uses double quotes.
-	    text.gsub!(/<\/?(?:html|body|font)(?: .+?)?>/, '') # very important!
-
-	    text.tr!("\r", '')
-	    # Remove empty lines
-	    text.gsub!("\n\n", "\n")
-	    
-	    # Remove newlines that end the file, since they screw up the 
-	    # newline -> <br/> conversion
-	    text.gsub!(/\n\Z/, '')
-	    
-	    # Replace newlines with "<br/>" unless they end a chat line.
-	    # This must go after we remove <font> tags.
-	    text.gsub!(/\n(?!#{@timestamp_rx})/, '<br/>')
-	    
-	    # These empty links are sometimes appended to every line in a chat,
-	    # for some weird reason. Remove them.
-	    text.gsub!(%r{<a href=('").+?\1>\s*?</a>}, '')
-	
-	    # Replace single quotes inside tags with double quotes so we can
-	    # easily change single quotes to entities.
-	    # For spans, removes a space after the final declaration if it exists.
-	    text.gsub!(/<span style='([^']+?;) ?'>/, '<span style="\1">')
-	    text.gsub!(/([a-z]+=)'(.+?)'/, '\1"\2"')
+    public :parse
+
+    #################
+    private
+    #################
+
+    def cleanup(text)
+      text.tr!("\r", '')
+      # Replace newlines with "<br/>" unless they end a chat line.
+      text.gsub!(/\n(?!#{@timestamp_rx}|\Z)/, '<br/>')
+      # Escape entities since this will be in XML
+      text.gsub!('&', '&amp;') # escape '&' first
+      text.gsub!('<', '&lt;')
+      text.gsub!('>', '&gt;')
+      text.gsub!('"', '&quot;')
+      text.gsub!("'", '&apos;')
+      return text
+    end
+  end
+
+  # Please use Pidgin2Adium.parse or Pidgin2Adium.parse_and_generate instead
+  # of using this class directly.
+  class HtmlLogParser < BasicParser
+    def initialize(src_path, user_aliases)
+      super(src_path, user_aliases)
+      @timestamp_rx = '\(((?:\d{4}-\d{2}-\d{2} )?\d{1,2}:\d{1,2}:\d{1,2}(?: [AP]M)?)\)'
+
+      # @line_regex matches a line in an HTML log file other than the
+      # first time matches on either "2008-11-17 14:12" or "14:12"
+      # @line_regex match obj:
+      # 0: timestamp, extended or not
+      # 1: screen name or alias, if alias set
+      # 2: "&lt;AUTO-REPLY&gt;" or nil
+      # 3: message body
+      # The ":" is optional to allow for strings like "(17:12:21) <b>***Gabe B-W</b> is confused<br/>"
+      @line_regex = /#{@timestamp_rx} ?<b>(.+?) ?(&lt;AUTO-REPLY&gt;)?:?<\/b> ?(.+)<br ?\/>/o
+        # @line_regex_status matches a status line
+        # @line_regex_status match obj:
+        # 0: timestamp
+        # 1: status message
+        @line_regex_status = /#{@timestamp_rx} ?<b> (.+)<\/b><br ?\/>/o
+    end
+
+    public :parse
+
+    #################
+    private
+    #################
+
+    # Returns a cleaned string.
+    # Removes the following tags from _text_:
+    # * html
+    # * body
+    # * font
+    # * a with no innertext, e.g. <a href="blah"></a>
+    # And removes the following style declarations:
+    # * color: #000000 (just turns text black)
+    # * font-family
+    # * font-size
+    # * background
+    # * em (really it's changed to <span style="font-style: italic;">)
+    # Since each <span> has only one style declaration, spans with these
+    # declarations are removed (but the text inside them is preserved).
+    def cleanup(text)
+      # Sometimes this is in there. I don't know why.
+      text.gsub!(%r{&lt;/FONT HSPACE='\d'>}, '')
+      # We can remove <font> safely since Pidgin and Adium both show bold
+      # using <span style="font-weight: bold;"> except Pidgin uses single
+      # quotes while Adium uses double quotes.
+      text.gsub!(/<\/?(?:html|body|font)(?: .+?)?>/, '') # very important!
+
+      text.tr!("\r", '')
+      # Remove empty lines
+      text.gsub!("\n\n", "\n")
+
+      # Remove newlines that end the file, since they screw up the
+      # newline -> <br/> conversion
+      text.gsub!(/\n\Z/, '')
+
+      # Replace newlines with "<br/>" unless they end a chat line.
+      # This must go after we remove <font> tags.
+      text.gsub!(/\n(?!#{@timestamp_rx})/, '<br/>')
+
+      # These empty links are sometimes appended to every line in a chat,
+      # for some weird reason. Remove them.
+      text.gsub!(%r{<a href=('").+?\1>\s*?</a>}, '')
+
+      # Replace single quotes inside tags with double quotes so we can
+      # easily change single quotes to entities.
+      # For spans, removes a space after the final declaration if it exists.
+      text.gsub!(/<span style='([^']+?;) ?'>/, '<span style="\1">')
+      text.gsub!(/([a-z]+=)'(.+?)'/, '\1"\2"')
 =begin
-	    text.gsub!(/<a href='(.+?)'>/, '<a href="\1">')
-	    text.gsub!(/<img src='([^']+?)'/, '<img src="\1"')
-	    text.gsub!(/ alt='([^']+?)'/, ' alt="\1"')
+      text.gsub!(/<a href='(.+?)'>/, '<a href="\1">')
+      text.gsub!(/<img src='([^']+?)'/, '<img src="\1"')
+      text.gsub!(/ alt='([^']+?)'/, ' alt="\1"')
 =end
-	    text.gsub!("'", '&apos;')
-
-	    # This actually does match stuff, but doesn't group it correctly. :(
-	    # text.gsub!(%r{<span style="((?:.+?;)+)">(.*?)</span>}) do |s|
-	    text.gsub!(%r{<span style="(.+?)">(.*?)</span>}) do |s|
-		# Remove empty spans.
-		next if $2 == ''
-
-		# style = style declaration
-		# innertext = text inside <span>
-		style, innertext = $1, $2
-		# TODO: replace double quotes with "&quot;", but only outside tags; may still be tags inside spans
-		# innertext.gsub!("")
-		
-		styleparts = style.split(/; ?/)
-		styleparts.map! do |p|
-		    if p[0,5] == 'color'
-			if p.include?('color: #000000')
-			    next
-			elsif p =~ /(color: #[0-9a-fA-F]{6})(>.*)?/
-			    # Regarding the bit with the ">", sometimes this happens:
-			    # <span style="color: #000000>today;">today was busy</span>
-			    # Then p = "color: #000000>today"
-			    # Or it can end in ">;", with no text before the semicolon.
-			    # So keep the color but remove the ">" and anything following it.
-			    next($1)
-			end
-		    else
-			# don't remove font-weight
-			case p
-			when /^font-family/: next
-			when /^font-size/: next
-			when /^background/: next
-			end
-		    end
-		end.compact!
-		unless styleparts.empty?
-		    style = styleparts.join('; ')
-		    innertext = "<span style=\"#{style};\">#{innertext}</span>"
-		end
-		innertext
-	    end
-	    # Pidgin uses <em>, Adium uses <span>
-	    if text.gsub!('<em>', '<span style="font-style: italic;">')
-		text.gsub!('</em>', '</span>')
-	    end
-	    return text
-	end
+      text.gsub!("'", '&apos;')
+
+      # This actually does match stuff, but doesn't group it correctly. :(
+      # text.gsub!(%r{<span style="((?:.+?;)+)">(.*?)</span>}) do |s|
+      text.gsub!(%r{<span style="(.+?)">(.*?)</span>}) do |s|
+        # Remove empty spans.
+        next if $2 == ''
+
+        # style = style declaration
+        # innertext = text inside <span>
+        style, innertext = $1, $2
+        # TODO: replace double quotes with "&quot;", but only outside tags; may still be tags inside spans
+        # innertext.gsub!("")
+
+        styleparts = style.split(/; ?/)
+        styleparts.map! do |p|
+          if p[0,5] == 'color'
+            if p.include?('color: #000000')
+              next
+            elsif p =~ /(color: #[0-9a-fA-F]{6})(>.*)?/
+              # Regarding the bit with the ">", sometimes this happens:
+              # <span style="color: #000000>today;">today was busy</span>
+              # Then p = "color: #000000>today"
+              # Or it can end in ">;", with no text before the semicolon.
+              # So keep the color but remove the ">" and anything following it.
+              next($1)
+            end
+          else
+            # don't remove font-weight
+            case p
+            when /^font-family/ then next
+            when /^font-size/ then next
+            when /^background/ then next
+            end
+          end
+        end.compact!
+        unless styleparts.empty?
+          style = styleparts.join('; ')
+          innertext = "<span style=\"#{style};\">#{innertext}</span>"
+        end
+        innertext
+      end
+      # Pidgin uses <em>, Adium uses <span>
+      if text.gsub!('<em>', '<span style="font-style: italic;">')
+        text.gsub!('</em>', '</span>')
+      end
+      return text
+    end
+  end # END HtmlLogParser class
+
+  # A holding object for each line of the chat. It is subclassed as
+  # appropriate (eg AutoReplyMessage). Each subclass (but not Message
+  # itself) has its own to_s which prints out its information in a format
+  # appropriate for putting in an Adium log file.
+  # Subclasses: XMLMessage, AutoReplyMessage, StatusMessage, Event.
+  class Message
+    def initialize(sender, time, buddy_alias)
+      # The sender's screen name
+      @sender = sender
+      # The time the message was sent, in Adium format (e.g.
+      # "2008-10-05T22:26:20-0800")
+      @time = time
+      # The receiver's alias (NOT screen name)
+      @buddy_alias = buddy_alias
+    end
+    attr_accessor :sender, :time, :buddy_alias
+  end
+
+  # Basic message with body text (as opposed to pure status messages, which
+  # have no body).
+  class XMLMessage < Message
+    def initialize(sender, time, buddy_alias, body)
+      super(sender, time, buddy_alias)
+      @body = body
+      @styled_body = '<div><span style="font-family: Helvetica; font-size: 12pt;">%s</span></div>' % @body
+      normalize_body!()
+    end
+    attr_accessor :body
+
+    def to_s
+      return sprintf('<message sender="%s" time="%s" alias="%s">%s</message>' << "\n",
+                     @sender, @time, @buddy_alias, @styled_body)
+    end
+
+    #################
+    private
+    #################
+
+    # Balances mismatched tags, normalizes body style, and fixes actions
+    # so they are in Adium style (Pidgin uses "***Buddy waves at you", Adium uses
+    # "*Buddy waves at you*").
+    def normalize_body!
+      normalize_body_entities!()
+      # Fix mismatched tags. Yes, it's faster to do it per-message
+      # than all at once.
+      @body = Pidgin2Adium.balance_tags_c(@body)
+      if @buddy_alias[0,3] == '***'
+        # "***<alias>" is what pidgin sets as the alias for a /me action
+        @buddy_alias.slice!(0,3)
+        @body = '*' << @body << '*'
+      end
     end
 
-    # A holding object for each line of the chat. It is subclassed as
-    # appropriate (eg AutoReplyMessage). Each subclass (but not Message
-    # itself) has its own to_s which prints out its information in a format
-    # appropriate for putting in an Adium log file.
-    # Subclasses: XMLMessage, AutoReplyMessage, StatusMessage, Event.
-    class Message
-	def initialize(sender, time, buddy_alias)
-	    # The sender's screen name
-	    @sender = sender
-	    # The time the message was sent, in Adium format (e.g.
-	    # "2008-10-05T22:26:20-0800")
-	    @time = time
-	    # The receiver's alias (NOT screen name)
-	    @buddy_alias = buddy_alias
-	end
-	attr_accessor :sender, :time, :buddy_alias
+    # Escapes entities.
+    def normalize_body_entities!
+      # Convert '&' to '&amp;' only if it's not followed by an entity.
+      @body.gsub!(/&(?!lt|gt|amp|quot|apos)/, '&amp;')
     end
-   
-    # Basic message with body text (as opposed to pure status messages, which
-    # have no body).
-    class XMLMessage < Message
-	def initialize(sender, time, buddy_alias, body)
-	    super(sender, time, buddy_alias)
-	    @body = body
-	    @styled_body = '<div><span style="font-family: Helvetica; font-size: 12pt;">%s</span></div>' % @body
-	    normalize_body!()
-	end
-	attr_accessor :body
-
-	def to_s
-	    return sprintf('<message sender="%s" time="%s" alias="%s">%s</message>' << "\n",
-			   @sender, @time, @buddy_alias, @styled_body)
-	end
-
-	#################
-	private
-	#################
-
-	# Balances mismatched tags, normalizes body style, and fixes actions
-	# so they are in Adium style (Pidgin uses "***Buddy waves at you", Adium uses
-	# "*Buddy waves at you*").
-	def normalize_body!
-	    normalize_body_entities!()
-	    # Fix mismatched tags. Yes, it's faster to do it per-message
-	    # than all at once.
-	    @body = Pidgin2Adium.balance_tags_c(@body)
-	    if @buddy_alias[0,3] == '***'
-		# "***<alias>" is what pidgin sets as the alias for a /me action
-		@buddy_alias.slice!(0,3)
-		@body = '*' << @body << '*'
-	    end
-	end
-
-	# Escapes entities.
-	def normalize_body_entities!
-	    # Convert '&' to '&amp;' only if it's not followed by an entity.
-	    @body.gsub!(/&(?!lt|gt|amp|quot|apos)/, '&amp;')
-	end
+  end # END XMLMessage
+
+  # An auto reply message.
+  class AutoReplyMessage < XMLMessage
+    def to_s
+      return sprintf('<message sender="%s" time="%s" auto="true" alias="%s">%s</message>' << "\n",
+                     @sender, @time, @buddy_alias, @styled_body)
     end
+  end
 
-    # An auto reply message.
-    class AutoReplyMessage < XMLMessage
-	def to_s
-	    return sprintf('<message sender="%s" time="%s" auto="true" alias="%s">%s</message>' << "\n",
-			   @sender, @time, @buddy_alias, @styled_body)
-	end
+  # A message saying e.g. "Blahblah has gone away."
+  class StatusMessage < Message
+    def initialize(sender, time, buddy_alias, status)
+      super(sender, time, buddy_alias)
+      @status = status
     end
+    attr_accessor :status
 
-    # A message saying e.g. "Blahblah has gone away."
-    class StatusMessage < Message
-	def initialize(sender, time, buddy_alias, status)
-	    super(sender, time, buddy_alias) 
-	    @status = status
-	end
-	attr_accessor :status
-
-	def to_s
-	    return sprintf('<status type="%s" sender="%s" time="%s" alias="%s"/>' << "\n", @status, @sender, @time, @buddy_alias)
-	end
+    def to_s
+      return sprintf('<status type="%s" sender="%s" time="%s" alias="%s"/>' << "\n", @status, @sender, @time, @buddy_alias)
+    end
+  end
+
+  # Pidgin does not have Events, but Adium does. Pidgin mostly uses system
+  # messages to display what Adium calls events. These include sending a file,
+  # starting a Direct IM connection, or an error in chat.
+  class Event < XMLMessage
+    def initialize(sender, time, buddy_alias, body, event_type)
+      super(sender, time, buddy_alias, body)
+      @event_type = event_type
     end
-  
-    # Pidgin does not have Events, but Adium does. Pidgin mostly uses system
-    # messages to display what Adium calls events. These include sending a file,
-    # starting a Direct IM connection, or an error in chat.
-    class Event < XMLMessage
-	def initialize(sender, time, buddy_alias, body, event_type)
-	    super(sender, time, buddy_alias, body)
-	    @event_type = event_type
-	end
-	attr_accessor :event_type
-
-	def to_s
-	    return sprintf('<event type="%s" sender="%s" time="%s" alias="%s">%s</event>',
-			   @event_type, @sender, @time, @buddy_alias, @styled_body)
-	end
+    attr_accessor :event_type
+
+    def to_s
+      return sprintf('<event type="%s" sender="%s" time="%s" alias="%s">%s</event>',
+                     @event_type, @sender, @time, @buddy_alias, @styled_body)
     end
+  end
 end # end module