Class: MsgExtractor::Rtf::Decapsulator

Inherits:
Object
  • Object
show all
Defined in:
lib/msg_extractor/rtf/decapsulator.rb

Overview

Extracts HTML encapsulated in RTF ([MS-OXRTFEX]). Targets well-formed Outlook-generated RTF; returns nil on anything it cannot handle.

Constant Summary collapse

SKIP_DESTINATIONS = 
%w[
  fonttbl colortbl stylesheet info generator pntext listtable
  listoverridetable themedata colorschememapping datastore latentstyles
  xmlnstbl rsidtbl pgptbl background pict object header footer footnote
].freeze

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(rtf) ⇒ Decapsulator

Returns a new instance of Decapsulator.



20
21
22
23
24
25
26
27
28
29
 
# File 'lib/msg_extractor/rtf/decapsulator.rb', line 20

def initialize(rtf)
  @rtf = rtf.b
  @pos = 0
  @out = +"".encode(Encoding::UTF_8)   # accumulates decoded UTF-8 text
  @pending = +"".b                      # raw codepage bytes not yet decoded
  @codepage = 1252
  # Group-scoped state, saved on "{" and restored on "}".
  @state = { suppress: false, destination: :normal, uc: 1 }
  @stack = []
end

Class Method Details

.html_from(rtf) ⇒ Object 



12
13
14
15
16
17
18
 
# File 'lib/msg_extractor/rtf/decapsulator.rb', line 12

def self.html_from(rtf)
  return nil unless rtf
  return nil unless rtf.byteslice(0, 512).to_s.include?("\\fromhtml1")
  new(rtf).run
rescue StandardError
  nil # malformed RTF: html_body falls back to nil rather than raising
end

Instance Method Details

#run ⇒ Object 



31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
 
# File 'lib/msg_extractor/rtf/decapsulator.rb', line 31

def run
  while @pos < @rtf.bytesize
    byte = @rtf.getbyte(@pos)
    case byte
    when 0x7B # {
      @pos += 1
      @stack.push(@state.dup)
      handle_group_start
    when 0x7D # }
      @pos += 1
      @state = @stack.pop || @state
    when 0x5C # backslash
      handle_control
    when 0x0D, 0x0A # bare CR/LF are not document text in RTF
      @pos += 1
    else
      @pos += 1
      emit(byte.chr)
    end
  end
  finalize
end