7 attr_reader_cloned :labels
9 def initialize uri_or_fp, start_offset=nil, usual=true, archived=false, id=nil
14 @labels << :inbox unless archived?
18 raise ArgumentError, "not an mbox uri" unless uri_or_fp =~ %r!mbox://!
20 fn = uri_or_fp.sub(%r!^mbox://!, "")
21 ## heuristic: use the filename as a label, unless the file
22 ## has a path that probably represents an inbox.
23 @labels << File.basename(fn).intern unless File.dirname(fn) =~ /\b(var|usr|spool)\b/
33 def start_offset; 0; end
34 def end_offset; File.size @f; end
35 def total; end_offset; end
37 def load_header offset
43 self.broken_msg = "offset mismatch in mbox file offset #{offset.inspect}: #{l.inspect}. Run 'sup-import --rebuild #{to_s}' to correct this."
44 raise SourceError, self.broken_msg
46 header = MBox::read_header @f
51 def load_message offset
52 raise SourceError, self.broken_msg if broken?
56 RMail::Mailbox::MBoxReader.new(@f).each_message do |input|
57 return RMail::Parser.read(input)
59 rescue RMail::Parser::Error => e
60 raise SourceError, "error parsing message with rmail: #{e.message}"
66 raise SourceError, self.broken_msg if broken?
70 until @f.eof? || (l = @f.gets) =~ /^$/
77 def raw_full_message offset
78 raise SourceError, self.broken_msg if broken?
82 @f.gets # skip mbox header
83 until @f.eof? || (l = @f.gets) =~ BREAK_RE
91 raise SourceError, self.broken_msg if broken?
93 next_offset = cur_offset
98 ## cur_offset could be at one of two places here:
100 ## 1. before a \n and a mbox separator, if it was previously at
101 ## EOF and a new message was added; or,
102 ## 2. at the beginning of an mbox separator (in all other
105 l = @f.gets or raise "next while at EOF"
106 if l =~ /^\s*$/ # case 1
107 returned_offset = @f.tell
108 @f.gets # now we're at a BREAK_RE, so skip past it
110 returned_offset = cur_offset
111 ## we've already skipped past the BREAK_RE, to just go
114 while(line = @f.gets)
115 break if line =~ BREAK_RE
116 next_offset = @f.tell
120 self.cur_offset = next_offset
121 [returned_offset, labels]
125 Redwood::register_yaml(Loader, %w(uri cur_offset usual archived id))