require 'rmail'
+require 'uri'
module Redwood
module MBox
class Loader < Source
- attr_reader_cloned :labels
-
- def initialize uri_or_fp, start_offset=nil, usual=true, archived=false, id=nil
- super
+ yaml_properties :uri, :cur_offset, :usual, :archived, :id, :labels
+ ## uri_or_fp is horrific. need to refactor.
+ def initialize uri_or_fp, start_offset=nil, usual=true, archived=false, id=nil, labels=[]
@mutex = Mutex.new
- @labels = [:unread]
- @labels << :inbox unless archived?
+ @labels = ((labels || []) - LabelManager::RESERVED_LABELS).uniq.freeze
case uri_or_fp
when String
- raise ArgumentError, "not an mbox uri" unless uri_or_fp =~ %r!mbox://!
-
- fn = uri_or_fp.sub(%r!^mbox://!, "")
- ## heuristic: use the filename as a label, unless the file
- ## has a path that probably represents an inbox.
- @labels << File.basename(fn).intern unless File.dirname(fn) =~ /\b(var|usr|spool)\b/
- @f = File.open fn
+ uri = URI(Source.expand_filesystem_uri(uri_or_fp))
+ raise ArgumentError, "not an mbox uri" unless uri.scheme == "mbox"
+ raise ArgumentError, "mbox uri ('#{uri}') cannot have a host: #{uri.host}" if uri.host
+ @f = File.open uri.path
+ @path = uri.path
else
@f = uri_or_fp
+ @path = uri_or_fp.path
+ end
+
+ super uri_or_fp, start_offset, usual, archived, id
+ end
+
+ def file_path; @path end
+ 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
+
+ def self.suggest_labels_for path
+ ## heuristic: use the filename as a label, unless the file
+ ## has a path that probably represents an inbox.
+ if File.dirname(path) =~ /\b(var|usr|spool)\b/
+ []
+ else
+ [File.basename(path).intern]
end
end
+ def check
+ if (cur_offset ||= start_offset) > end_offset
+ raise OutOfSyncSourceError, "mbox file is smaller than last recorded message offset. Messages have probably been deleted by another client."
+ end
+ end
+
def start_offset; 0; end
def end_offset; File.size @f; end
- def total; end_offset; end
def load_header offset
header = nil
@f.seek offset
l = @f.gets
unless l =~ BREAK_RE
- self.broken_msg = "offset mismatch in mbox file offset #{offset.inspect}: #{l.inspect}. Run 'sup-import --rebuild #{to_s}' to correct this."
- raise SourceError, self.broken_msg
+ raise OutOfSyncSourceError, "mismatch in mbox file offset #{offset.inspect}: #{l.inspect}."
end
header = MBox::read_header @f
end
end
def load_message offset
- raise SourceError, self.broken_msg if broken?
@mutex.synchronize do
@f.seek offset
begin
return RMail::Parser.read(input)
end
rescue RMail::Parser::Error => e
- raise SourceError, "error parsing message with rmail: #{e.message}"
+ raise FatalSourceError, "error parsing mbox file: #{e.message}"
end
end
end
def raw_header offset
- raise SourceError, self.broken_msg if broken?
ret = ""
@mutex.synchronize do
@f.seek offset
ret
end
- def raw_full_message offset
- raise SourceError, self.broken_msg if broken?
+ def raw_message offset
ret = ""
+ each_raw_message_line(offset) { |l| ret += l }
+ ret
+ end
+
+ ## apparently it's a million times faster to call this directly if
+ ## we're just moving messages around on disk, than reading things
+ ## into memory with raw_message.
+ ##
+ ## i hoped never to have to move shit around on disk but
+ ## sup-sync-back has to do it.
+ def each_raw_message_line offset
@mutex.synchronize do
@f.seek offset
- @f.gets # skip mbox header
+ yield @f.gets
until @f.eof? || (l = @f.gets) =~ BREAK_RE
- ret += l
+ yield l
end
end
- ret
end
def next
- raise SourceError, self.broken_msg if broken?
returned_offset = nil
next_offset = cur_offset
- @mutex.synchronize do
- @f.seek cur_offset
-
- ## cur_offset could be at one of two places here:
-
- ## 1. before a \n and a mbox separator, if it was previously at
- ## EOF and a new message was added; or,
- ## 2. at the beginning of an mbox separator (in all other
- ## cases).
-
- l = @f.gets or raise "next while at EOF"
- if l =~ /^\s*$/ # case 1
- returned_offset = @f.tell
- @f.gets # now we're at a BREAK_RE, so skip past it
- else # case 2
- returned_offset = cur_offset
- ## we've already skipped past the BREAK_RE, to just go
- end
+ begin
+ @mutex.synchronize do
+ @f.seek cur_offset
+
+ ## cur_offset could be at one of two places here:
- while(line = @f.gets)
- break if line =~ BREAK_RE
- next_offset = @f.tell
+ ## 1. before a \n and a mbox separator, if it was previously at
+ ## EOF and a new message was added; or,
+ ## 2. at the beginning of an mbox separator (in all other
+ ## cases).
+
+ l = @f.gets or raise "next while at EOF"
+ if l =~ /^\s*$/ # case 1
+ returned_offset = @f.tell
+ @f.gets # now we're at a BREAK_RE, so skip past it
+ else # case 2
+ returned_offset = cur_offset
+ ## we've already skipped past the BREAK_RE, so just go
+ end
+
+ while(line = @f.gets)
+ break if line =~ BREAK_RE
+ next_offset = @f.tell
+ end
end
+ rescue SystemCallError, IOError => e
+ raise FatalSourceError, "Error reading #{@f.path}: #{e.message}"
end
self.cur_offset = next_offset
- [returned_offset, labels]
+ [returned_offset, (@labels + [:unread]).uniq]
end
end
-Redwood::register_yaml(Loader, %w(uri cur_offset usual archived id))
-
end
end
projects / sup / blobdiff