8 yaml_properties :uri, :cur_offset, :usual, :archived, :id, :labels
11 ## uri_or_fp is horrific. need to refactor.
12 def initialize uri_or_fp, start_offset=nil, usual=true, archived=false, id=nil, labels=[]
14 @labels = ((labels || []) - LabelManager::RESERVED_LABELS).uniq.freeze
18 uri = URI(Source.expand_filesystem_uri(uri_or_fp))
19 raise ArgumentError, "not an mbox uri" unless uri.scheme == "mbox"
20 raise ArgumentError, "mbox URI ('#{uri}') cannot have a host: #{uri.host}" if uri.host
21 raise ArgumentError, "mbox URI must have a path component" unless uri.path
22 @f = File.open uri.path
26 @path = uri_or_fp.path
29 super uri_or_fp, start_offset, usual, archived, id
32 def file_path; @path end
33 def is_source_for? uri; super || (self.uri.is_a?(String) && (URI(Source.expand_filesystem_uri(uri)) == URI(Source.expand_filesystem_uri(self.uri)))) end
35 def self.suggest_labels_for path
36 ## heuristic: use the filename as a label, unless the file
37 ## has a path that probably represents an inbox.
38 if File.dirname(path) =~ /\b(var|usr|spool)\b/
41 [File.basename(path).downcase.intern]
46 if (cur_offset ||= start_offset) > end_offset
47 raise OutOfSyncSourceError, "mbox file is smaller than last recorded message offset. Messages have probably been deleted by another client."
51 def start_offset; 0; end
52 def end_offset; File.size @f; end
54 def load_header offset
60 raise OutOfSyncSourceError, "mismatch in mbox file offset #{offset.inspect}: #{l.inspect}."
62 header = MBox::read_header @f
67 def load_message offset
71 RMail::Mailbox::MBoxReader.new(@f).each_message do |input|
72 m = RMail::Parser.read(input)
73 if m.body && m.body.is_a?(String)
74 m.body.gsub!(/^>From /, "From ")
78 rescue RMail::Parser::Error => e
79 raise FatalSourceError, "error parsing mbox file: #{e.message}"
84 ## scan forward until we're at the valid start of a message
89 until @f.eof? || (l = @f.gets) =~ BREAK_RE
92 self.cur_offset += string.length
100 until @f.eof? || (l = @f.gets) =~ /^\r*$/
107 def raw_message offset
109 each_raw_message_line(offset) { |l| ret += l }
113 ## apparently it's a million times faster to call this directly if
114 ## we're just moving messages around on disk, than reading things
115 ## into memory with raw_message.
117 ## i hoped never to have to move shit around on disk but
118 ## sup-sync-back has to do it.
119 def each_raw_message_line offset
120 @mutex.synchronize do
123 until @f.eof? || (l = @f.gets) =~ BREAK_RE
130 returned_offset = nil
131 next_offset = cur_offset
134 @mutex.synchronize do
137 ## cur_offset could be at one of two places here:
139 ## 1. before a \n and a mbox separator, if it was previously at
140 ## EOF and a new message was added; or,
141 ## 2. at the beginning of an mbox separator (in all other
144 l = @f.gets or raise "next while at EOF"
145 if l =~ /^\s*$/ # case 1
146 returned_offset = @f.tell
147 @f.gets # now we're at a BREAK_RE, so skip past it
149 returned_offset = cur_offset
150 ## we've already skipped past the BREAK_RE, so just go
153 while(line = @f.gets)
154 break if line =~ BREAK_RE
155 next_offset = @f.tell
158 rescue SystemCallError, IOError => e
159 raise FatalSourceError, "Error reading #{@f.path}: #{e.message}"
162 self.cur_offset = next_offset
163 [returned_offset, (self.labels + [:unread]).uniq]