- Update pack strings for endian fixes.

- Add ChangeLog.


git-svn-id: https://ruby-msg.googlecode.com/svn/trunk@114 c30d66de-b626-0410-988f-81f6512a6d81

Author: aquasync

ChangeLog

@@ -0,0 +1,74 @@
+== 1.3.2 / ...
+
+- Initial simple msg test case.
+- Update TODO, stripping out all the redundant ole stuff.
+- Fix property set guids to use the new Ole::Types::Clsid type.
+- Add block form Msg.open
+- Fix file requires for running tests individually.
+- Update RangesIO subclasses for changes in ruby-ole.
+- Start to flesh out move to mapi to clean up the way pst hijacks the msg
+  classes currently.
+- Pretty big refactoring, adding initial outlook 2003 pst support.
+- Merge initial pst reading code (converted from libpst).
+- Add a ChangeLog :).
+
+== 1.3.1 / 2007-08-21
+
+- Add fix for issue #2, and #4.
+- Move ole code to ruby-ole project, and depend on it.
+
+== 1.2.17 / 2007-05-13
+
+(this was last release before splitting out ruby-ole. subsequent bug fix
+point releases 1-3 were made directly on the gem, not reflected in the
+repository, though the fixes were also forward-ported.)
+
+- Update Ole::Storage backend, finalising api for split to separate 
+  library.
+
+== 1.2.16 / 2007-04-28
+
+- Some minor fixes to msg parser.
+- Extending RTF and body conversion support.
+- Initial look at possible wmf conversion for embedded images.
+- Add initial cli converter tool
+- Add rdoc to ole/storage, and msg/properties
+- Add streaming IO support to Ole::Storage, and use it in Msg::Properties
+- Updates to test cases
+- Add README, and update TODO
+- Convert rtf support tools in c to small ruby class.
+- Merge preliminary write support for Ole::Storage, as well as preliminary
+  filesystem api.
+
+== 1.2.13 / 2007-01-22
+
+- Nested msg support
+
+== 1.2.10 / 2007-01-21
+
+- Add initial vcard support.
+- Implement a named properties map, for vcard conversion.
+- Add orderedhash to Mime for keeping header order
+- Fix line endings in lib/mime
+- First released version
+
+== <= 1.2.9 / 2007-01-11..2007-01-19
+
+(haven't bothered to note exact versions and dates - nothing here was released.
+can look at history of lib/msg.rb to see exact VERSION at each commit.)
+
+- Merged most of the named property work.
+- Added some test files.
+- Update svn:ignore, to exclude test messages and ole files which I can't
+  release. Need to get some clean files for use in test cases.
+  Also excluding source to the mapitags files for the moment.
+  A lot of it is not redistributable
+- Added a converter to extract embedded html in rtf. Downloaded somewhere,
+  source unknown.
+- Minor fix to ole/storage.rb, after new OleDir#type behaviour
+- Imported support.rb, replacing previously required std.rb
+- Added initial support for parsing times in Msg::Properties.
+- Imported some rtf decompression code and minor updates.
+- Cleaned up the ole class a bit
+- Fixed OleDir#data method using sb_blocks map (see POLE).
+

lib/msg/properties.rb

@@ -239,16 +239,16 @@ def self.parse_nameid obj
 			# not sure about any of this stuff really.
 			# should flip a few bits in the real msg, to get a better understanding of how this works.
 			props = props_obj.read.scan(/.{8}/m).map do |str|
-				flags, offset = str[4..-1].unpack 'S2'
+				flags, offset = str[4..-1].unpack 'v2'
 				# the property will be serialised as this pseudo property, mapping it to this named property
 				pseudo_prop = 0x8000 + offset
 				named = flags & 1 == 1
 				prop = if named
-					str_off = *str.unpack('L')
-					len = *names_data[str_off, 4].unpack('L')
+					str_off = *str.unpack('V')
+					len = *names_data[str_off, 4].unpack('V')
 					Ole::Types::FROM_UTF16.iconv names_data[str_off + 4, len]
 				else
-					a, b = str.unpack('S2')
+					a, b = str.unpack('v2')
 					Log.debug "b not 0" if b != 0
 					a
 				end
@@ -275,7 +275,7 @@ def parse_substg key, encoding, offset, obj
 					# equal to the lengths of all the sub parts. gives an implied array size i suppose.
 					# maybe you can initialize the array at this time. the sizes are the same as all the
 					# ole object sizes anyway, its to pre-allocate i suppose.
-					#p obj.data.unpack('L*')
+					#p obj.data.unpack('V*')
 					# ignore this one
 					return
 				else
@@ -305,7 +305,7 @@ def parse_properties obj
 				Log.warn "padding was not as expected #{pad} (#{data.length}) -> #{data[0...pad].inspect}"
 			end
 			data[pad..-1].scan(/.{16}/m).each do |data|
-				property, encoding = ('%08x' % data.unpack('L')).scan /.{4}/
+				property, encoding = ('%08x' % data.unpack('V')).scan /.{4}/
 				key = property.hex
 				# doesn't make any sense to me. probably because its a serialization of some internal
 				# outlook structure...
@@ -316,12 +316,12 @@ def parse_properties obj
 					# multivalue versions ignored also
 				when '0003' # long
 					# don't know what all the other data is for
-					add_property key, *data[8, 4].unpack('L')
+					add_property key, *data[8, 4].unpack('V')
 				when '000b' # boolean
 					# again, heaps more data than needed. and its not always 0 or 1.
 					# they are in fact quite big numbers. this is wrong.
 #					p [property, data[4..-1].unpack('H*')[0]]
-					add_property key, data[8, 4].unpack('L')[0] != 0
+					add_property key, data[8, 4].unpack('V')[0] != 0
 				when '0040' # systime
 					# seems to work:
 					add_property key, Ole::Types.load_time(data[8..-1])
@@ -459,7 +459,10 @@ def body_html
 			@body_html = (self[:body_html].read rescue nil)
 			@body_html = (Msg::RTF.rtf2html body_rtf rescue nil) if !@body_html or @body_html.strip.empty?
 			# last resort
-			@body_html = (::RTF::Converter.rtf2text body_rtf, :html rescue nil) if !@body_html or @body_html.strip.empty?
+			if !@body_html or @body_html.strip.empty?
+				Log.warn 'creating html body from rtf'
+				@body_html = (::RTF::Converter.rtf2text body_rtf, :html rescue nil)
+			end
 			@body_html
 		end
 

lib/msg/rtf.rb

@@ -1,7 +1,7 @@
 require 'stringio'
 require 'strscan'
 
-require 'rtf.rb'
+require 'rtf'
 
 class Msg
 	#
@@ -35,7 +35,7 @@ def rtfdecompr data
 			rtf = ''
 
 			# get header fields (as defined in RTFLIB.H)
-			compr_size, uncompr_size, magic, crc32 = io.read(16).unpack 'L*'
+			compr_size, uncompr_size, magic, crc32 = io.read(16).unpack 'V*'
 			#warn "compressed-RTF data size mismatch" unless io.size == data.compr_size + 4
 
 			# process the data
@@ -46,7 +46,6 @@ def rtfdecompr data
 				flag_count = -1
 				flags = nil
 				while rtf.length < uncompr_size and !io.eof?
-					#p [rtf.length, uncompr_size]
 					# each flag byte flags 8 literals/references, 1 per bit
 					flags = ((flag_count += 1) % 8 == 0) ? io.getc : flags >> 1
 					if 1 == (flags & 1) # each flag bit is 1 for reference, 0 for literal
@@ -55,18 +54,24 @@ def rtfdecompr data
 						rp = (rp << 4) | (l >> 4) # the offset relative to block start
 						l = (l & 0xf) + 2 # the number of bytes to copy
 						l.times do
-							rtf << (buf[wp] = buf[rp])
+							rtf << buf[wp] = buf[rp]
 							wp = (wp + 1) % 4096
 							rp = (rp + 1) % 4096
 						end
 					else
-						rtf << (buf[wp] = io.getc)
+						rtf << buf[wp] = io.getc
 						wp = (wp + 1) % 4096
 					end
 				end
 			else # unknown magic number
 				raise "Unknown compression type (magic number 0x%08x)" % magic
 			end
+			
+			# not sure if its due to a bug in the above code. doesn't seem to be
+			# in my tests, but sometimes there's a trailing null. we chomp it here,
+			# which actually makes the resultant rtf smaller than its advertised
+			# size (+uncompr_size+).
+			rtf.chomp! 0.chr
 			rtf
 		end
 

lib/pst.rb

@@ -829,6 +829,7 @@ def initialize pst, list
 				# NOTE we take the last value seen value if there are duplicates. this "fixes"
 				# test4-o1997.pst for the time being.
 				warn "there are duplicate id2 records with id #{id2.id2}" if @id_from_id2[id2.id2]
+				next if @id_from_id2[id2.id2]
 				@id_from_id2[id2.id2] = id2.id
 			end
 		end