7 def initialize uri_or_fp, start_offset=nil, usual=true, archived=false, id=nil
12 @labels << :inbox unless archived?
16 raise ArgumentError, "not an mbox uri" unless uri_or_fp =~ %r!mbox://!
18 fn = uri_or_fp.sub(%r!^mbox://!, "")
19 ## heuristic: use the filename as a label, unless the file
20 ## has a path that probably represents an inbox.
21 @labels << File.basename(fn).intern unless File.dirname(fn) =~ /\b(var|usr|spool)\b/
31 def start_offset; 0; end
32 def end_offset; File.size @f; end
33 def total; end_offset; end
35 def load_header offset
41 self.broken_msg = "offset mismatch in mbox file offset #{offset.inspect}: #{l.inspect}. Run 'sup-import --rebuild #{to_s}' to correct this."
42 raise SourceError, self.broken_msg
44 header = MBox::read_header @f
49 def load_message offset
50 raise SourceError, self.broken_msg if broken?
54 RMail::Mailbox::MBoxReader.new(@f).each_message do |input|
55 return RMail::Parser.read(input)
57 rescue RMail::Parser::Error => e
58 raise SourceError, "error parsing message with rmail: #{e.message}"
64 raise SourceError, self.broken_msg if broken?
68 until @f.eof? || (l = @f.gets) =~ /^$/
75 def raw_full_message offset
76 raise SourceError, self.broken_msg if broken?
80 @f.gets # skip mbox header
81 until @f.eof? || (l = @f.gets) =~ BREAK_RE
89 raise SourceError, self.broken_msg if broken?
91 next_offset = cur_offset
96 ## cur_offset could be at one of two places here:
98 ## 1. before a \n and a mbox separator, if it was previously at
99 ## EOF and a new message was added; or,
100 ## 2. at the beginning of an mbox separator (in all other
103 l = @f.gets or raise "next while at EOF"
104 if l =~ /^\s*$/ # case 1
105 returned_offset = @f.tell
106 @f.gets # now we're at a BREAK_RE, so skip past it
108 returned_offset = cur_offset
109 ## we've already skipped past the BREAK_RE, to just go
112 while(line = @f.gets)
113 break if line =~ BREAK_RE
114 next_offset = @f.tell
118 self.cur_offset = next_offset
119 [returned_offset, @labels]
123 Redwood::register_yaml(Loader, %w(uri cur_offset usual archived id))