5 ## a Message is what's threaded.
7 ## it is also where the parsing for quotes and signatures is done, but
8 ## that should be moved out to a separate class at some point (because
9 ## i would like, for example, to be able to add in a ruby-talk
10 ## specific module that would detect and link to /ruby-talk:\d+/
11 ## sequences in the text of an email. (how sweet would that be?)
13 ## this class catches all source exceptions. if the underlying source
14 ## throws an error, it is caught and handled.
18 RE_PATTERN = /^((re|re[\[\(]\d[\]\)]):\s*)+/i
20 ## some utility methods
22 def normalize_subj s; s.gsub(RE_PATTERN, ""); end
23 def subj_is_reply? s; s =~ RE_PATTERN; end
24 def reify_subj s; subj_is_reply?(s) ? s : "Re: " + s; end
27 QUOTE_PATTERN = /^\s{0,4}[>|\}]/
28 BLOCK_QUOTE_PATTERN = /^-----\s*Original Message\s*----+$/
29 SIG_PATTERN = /(^-- ?$)|(^\s*----------+\s*$)|(^\s*_________+\s*$)|(^\s*--~--~-)|(^\s*--\+\+\*\*==)/
31 MAX_SIG_DISTANCE = 15 # lines from the end
33 DEFAULT_SENDER = "(missing sender)"
35 attr_reader :id, :date, :from, :subj, :refs, :replytos, :to, :source,
36 :cc, :bcc, :labels, :attachments, :list_address, :recipient_email, :replyto,
37 :source_info, :list_subscribe, :list_unsubscribe
39 bool_reader :dirty, :source_marked_read, :snippet_contains_encrypted_content
41 ## if you specify a :header, will use values from that. otherwise,
42 ## will try and load the header from the source.
44 @source = opts[:source] or raise ArgumentError, "source can't be nil"
45 @source_info = opts[:source_info] or raise ArgumentError, "source_info can't be nil"
46 @snippet = opts[:snippet]
47 @snippet_contains_encrypted_content = false
48 @have_snippet = !(opts[:snippet].nil? || opts[:snippet].empty?)
49 @labels = (opts[:labels] || []).to_set_of_symbols
55 ## we need to initialize this. see comments in parse_header as to
59 #parse_header(opts[:header] || @source.load_header(@source_info))
62 def parse_header header
63 ## forcibly decode these headers from and to the current encoding,
64 ## which serves to strip out characters that aren't displayable
65 ## (and which would otherwise be screwing up the display)
66 %w(from to subject cc bcc).each do |f|
67 header[f] = Iconv.easy_decode($encoding, $encoding, header[f]) if header[f]
70 @id = if header["message-id"]
71 mid = header["message-id"] =~ /<(.+?)>/ ? $1 : header["message-id"]
72 sanitize_message_id mid
74 id = "sup-faked-" + Digest::MD5.hexdigest(raw_header)
76 #Redwood::log "faking non-existent message-id for message from #{from}: #{id}"
80 @from = Person.from_address(if header["from"]
83 name = "Sup Auto-generated Fake Sender <sup@fake.sender.example.com>"
84 #Redwood::log "faking non-existent sender for message #@id: #{name}"
88 @date = case(date = header["date"])
94 rescue ArgumentError => e
95 #Redwood::log "faking mangled date header for #{@id} (orig #{header['date'].inspect} gave error: #{e.message})"
99 #Redwood::log "faking non-existent date header for #{@id}"
103 @subj = header.member?("subject") ? header["subject"].gsub(/\s+/, " ").gsub(/\s+$/, "") : DEFAULT_SUBJECT
104 @to = Person.from_address_list header["to"]
105 @cc = Person.from_address_list header["cc"]
106 @bcc = Person.from_address_list header["bcc"]
108 ## before loading our full header from the source, we can actually
109 ## have some extra refs set by the UI. (this happens when the user
110 ## joins threads manually). so we will merge the current refs values
112 refs = (header["references"] || "").scan(/<(.+?)>/).map { |x| sanitize_message_id x.first }
113 @refs = (@refs + refs).uniq
114 @replytos = (header["in-reply-to"] || "").scan(/<(.+?)>/).map { |x| sanitize_message_id x.first }
116 @replyto = Person.from_address header["reply-to"]
118 if header["list-post"]
119 @list_address = Person.from_address header["list-post"].gsub(/^<mailto:|>$/, "")
124 @recipient_email = header["envelope-to"] || header["x-original-to"] || header["delivered-to"]
125 @source_marked_read = header["status"] == "RO"
126 @list_subscribe = header["list-subscribe"]
127 @list_unsubscribe = header["list-unsubscribe"]
136 @dirty = true if @refs.delete ref
139 def snippet; @snippet || (chunks && @snippet); end
140 def is_list_message?; !@list_address.nil?; end
141 def is_draft?; @source.is_a? DraftLoader; end
143 raise "not a draft" unless is_draft?
144 @source.fn_for_offset @source_info
147 ## sanitize message ids by removing spaces and non-ascii characters.
148 ## also, truncate to 255 characters. all these steps are necessary
149 ## to make ferret happy. of course, we probably fuck up a couple
150 ## valid message ids as well. as long as we're consistent, this
151 ## should be fine, though.
153 ## also, mostly the message ids that are changed by this belong to
156 ## an alternative would be to SHA1 or MD5 all message ids on a regular basis.
158 def sanitize_message_id mid; mid.gsub(/(\s|[^\000-\177])+/, "")[0..254] end
162 index.sync_message self
167 def has_label? t; @labels.member? t; end
169 return if @labels.member? t
170 @labels = (@labels + [t]).to_set_of_symbols
174 return unless @labels.member? t
184 @labels = l.to_set_of_symbols
193 ## this is called when the message body needs to actually be loaded.
194 def load_from_source!
196 if @source.respond_to?(:has_errors?) && @source.has_errors?
197 [Chunk::Text.new(error_message(@source.error.message).split("\n"))]
200 ## we need to re-read the header because it contains information
201 ## that we don't store in the index. actually i think it's just
202 ## the mailing list address (if any), so this is kinda overkill.
203 ## i could just store that in the index, but i think there might
204 ## be other things like that in the future, and i'd rather not
206 ## actually, it's also the differentiation between to/cc/bcc,
207 ## so i will keep this.
208 parse_header @source.load_header(@source_info)
209 message_to_chunks @source.load_message(@source_info)
210 rescue SourceError, SocketError => e
211 Redwood::log "problem getting messages from #{@source}: #{e.message}"
212 ## we need force_to_top here otherwise this window will cover
213 ## up the error message one
215 Redwood::report_broken_sources :force_to_top => true
216 [Chunk::Text.new(error_message(e.message).split("\n"))]
221 def error_message msg
225 ***********************************************************************
226 An error occurred while loading this message. It is possible that
227 the source has changed, or (in the case of remote sources) is down.
228 You can check the log for errors, though hopefully an error window
229 should have popped up at some point.
231 The message location was:
232 #@source##@source_info
233 ***********************************************************************
235 The error message was:
240 ## wrap any source methods that might throw sourceerrors
241 def with_source_errors_handled
244 rescue SourceError => e
245 Redwood::log "problem getting messages from #{@source}: #{e.message}"
247 Redwood::report_broken_sources :force_to_top => true
248 error_message e.message
253 with_source_errors_handled { @source.raw_header @source_info }
257 with_source_errors_handled { @source.raw_message @source_info }
260 ## much faster than raw_message
261 def each_raw_message_line &b
262 with_source_errors_handled { @source.each_raw_message_line(@source_info, &b) }
265 ## returns all the content from a message that will be indexed
266 def indexable_content
269 from && from.indexable_content,
270 to.map { |p| p.indexable_content },
271 cc.map { |p| p.indexable_content },
272 bcc.map { |p| p.indexable_content },
273 chunks.select { |c| c.is_a? Chunk::Text }.map { |c| c.lines },
274 Message.normalize_subj(subj),
275 ].flatten.compact.join " "
278 def quotable_body_lines
279 chunks.find_all { |c| c.quotable? }.map { |c| c.lines }.flatten
282 def quotable_header_lines
283 ["From: #{@from.full_address}"] +
284 (@to.empty? ? [] : ["To: " + @to.map { |p| p.full_address }.join(", ")]) +
285 (@cc.empty? ? [] : ["Cc: " + @cc.map { |p| p.full_address }.join(", ")]) +
286 (@bcc.empty? ? [] : ["Bcc: " + @bcc.map { |p| p.full_address }.join(", ")]) +
287 ["Date: #{@date.rfc822}",
291 def self.build_from_source source, source_info
292 m = Message.new :source => source, :source_info => source_info
299 ## here's where we handle decoding mime attachments. unfortunately
300 ## but unsurprisingly, the world of mime attachments is a bit of a
301 ## mess. as an empiricist, i'm basing the following behavior on
302 ## observed mail rather than on interpretations of rfcs, so probably
303 ## this will have to be tweaked.
305 ## the general behavior i want is: ignore content-disposition, at
306 ## least in so far as it suggests something being inline vs being an
307 ## attachment. (because really, that should be the recipient's
308 ## decision to make.) if a mime part is text/plain, OR if the user
309 ## decoding hook converts it, then decode it and display it
310 ## inline. for these decoded attachments, if it has associated
311 ## filename, then make it collapsable and individually saveable;
312 ## otherwise, treat it as regular body text.
314 ## everything else is just an attachment and is not displayed
317 ## so, in contrast to mutt, the user is not exposed to the workings
318 ## of the gruesome slaughterhouse and sausage factory that is a
319 ## mime-encoded message, but need only see the delicious end
322 def multipart_signed_to_chunks m
324 Redwood::log "warning: multipart/signed with #{m.body.size} parts (expecting 2)"
328 payload, signature = m.body
329 if signature.multipart?
330 Redwood::log "warning: multipart/signed with payload multipart #{payload.multipart?} and signature multipart #{signature.multipart?}"
334 ## this probably will never happen
335 if payload.header.content_type == "application/pgp-signature"
336 Redwood::log "warning: multipart/signed with payload content type #{payload.header.content_type}"
340 if signature.header.content_type != "application/pgp-signature"
341 ## unknown signature type; just ignore.
342 #Redwood::log "warning: multipart/signed with signature content type #{signature.header.content_type}"
346 [CryptoManager.verify(payload, signature), message_to_chunks(payload)].flatten.compact
349 def multipart_encrypted_to_chunks m
351 Redwood::log "warning: multipart/encrypted with #{m.body.size} parts (expecting 2)"
355 control, payload = m.body
356 if control.multipart?
357 Redwood::log "warning: multipart/encrypted with control multipart #{control.multipart?} and payload multipart #{payload.multipart?}"
361 if payload.header.content_type != "application/octet-stream"
362 Redwood::log "warning: multipart/encrypted with payload content type #{payload.header.content_type}"
366 if control.header.content_type != "application/pgp-encrypted"
367 Redwood::log "warning: multipart/encrypted with control content type #{signature.header.content_type}"
371 decryptedm, sig, notice = CryptoManager.decrypt payload
372 children = message_to_chunks(decryptedm, true) if decryptedm
373 [notice, sig, children].flatten.compact
376 ## takes a RMail::Message, breaks it into Chunk:: classes.
377 def message_to_chunks m, encrypted=false, sibling_types=[]
380 case m.header.content_type
381 when "multipart/signed"
382 multipart_signed_to_chunks m
383 when "multipart/encrypted"
384 multipart_encrypted_to_chunks m
388 sibling_types = m.body.map { |p| p.header.content_type }
389 chunks = m.body.map { |p| message_to_chunks p, encrypted, sibling_types }.flatten.compact
393 elsif m.header.content_type == "message/rfc822"
394 payload = RMail::Parser.read(m.body)
395 from = payload.header.from.first
396 from_person = from ? Person.from_address(from.format) : nil
397 [Chunk::EnclosedMessage.new(from_person, payload.to_s)] +
398 message_to_chunks(payload, encrypted)
401 ## first, paw through the headers looking for a filename
402 if m.header["Content-Disposition"] && m.header["Content-Disposition"] =~ /filename="?(.*?[^\\])("|;|$)/
404 elsif m.header["Content-Type"] && m.header["Content-Type"] =~ /name="?(.*?[^\\])("|;|$)/
407 ## haven't found one, but it's a non-text message. fake
410 ## TODO: make this less lame.
411 elsif m.header["Content-Type"] && m.header["Content-Type"] !~ /^text\/plain/
413 case m.header["Content-Type"]
414 when /text\/html/ then "html"
415 when /image\/(.*)/ then $1
418 ["sup-attachment-#{Time.now.to_i}-#{rand 10000}", extension].join(".")
421 ## if there's a filename, we'll treat it as an attachment.
423 # add this to the attachments list if its not a generated html
424 # attachment (should we allow images with generated names?).
425 # Lowercase the filename because searches are easier that way
426 @attachments.push filename.downcase unless filename =~ /^sup-attachment-/
427 add_label :attachment unless filename =~ /^sup-attachment-/
428 content_type = m.header.content_type || "application/unknown" # sometimes RubyMail gives us nil
429 [Chunk::Attachment.new(content_type, filename, m, sibling_types)]
431 ## otherwise, it's body text
433 ## if there's no charset, use the current encoding as the charset.
434 ## this ensures that the body is normalized to avoid non-displayable
436 body = Iconv.easy_decode($encoding, m.charset || $encoding, m.decode) if m.body
437 text_to_chunks((body || "").normalize_whitespace.split("\n"), encrypted)
442 ## parse the lines of text into chunk objects. the heuristics here
443 ## need tweaking in some nice manner. TODO: move these heuristics
444 ## into the classes themselves.
445 def text_to_chunks lines, encrypted
446 state = :text # one of :text, :quote, or :sig
450 lines.each_with_index do |line, i|
451 nextline = lines[(i + 1) ... lines.length].find { |l| l !~ /^\s*$/ } # skip blank lines
457 ## the following /:$/ followed by /\w/ is an attempt to detect the
458 ## start of a quote. this is split into two regexen because the
459 ## original regex /\w.*:$/ had very poor behavior on long lines
460 ## like ":a:a:a:a:a" that occurred in certain emails.
461 if line =~ QUOTE_PATTERN || (line =~ /:$/ && line =~ /\w/ && nextline =~ QUOTE_PATTERN)
463 elsif line =~ SIG_PATTERN && (lines.length - i) < MAX_SIG_DISTANCE
465 elsif line =~ BLOCK_QUOTE_PATTERN
466 newstate = :block_quote
470 chunks << Chunk::Text.new(chunk_lines) unless chunk_lines.empty?
480 if line =~ QUOTE_PATTERN || (line =~ /^\s*$/ && nextline =~ QUOTE_PATTERN)
482 elsif line =~ SIG_PATTERN && (lines.length - i) < MAX_SIG_DISTANCE
489 if chunk_lines.empty?
492 chunks << Chunk::Quote.new(chunk_lines)
498 when :block_quote, :sig
502 if !@have_snippet && state == :text && (@snippet.nil? || @snippet.length < SNIPPET_LEN) && line !~ /[=\*#_-]{3,}/ && line !~ /^\s*$/
504 @snippet += " " unless @snippet.empty?
505 @snippet += line.gsub(/^\s+/, "").gsub(/[\r\n]/, "").gsub(/\s+/, " ")
506 @snippet = @snippet[0 ... SNIPPET_LEN].chomp
507 @dirty = true unless encrypted && $config[:discard_snippets_from_encrypted_messages]
508 @snippet_contains_encrypted_content = true if encrypted
514 when :quote, :block_quote
515 chunks << Chunk::Quote.new(chunk_lines) unless chunk_lines.empty?
517 chunks << Chunk::Text.new(chunk_lines) unless chunk_lines.empty?
519 chunks << Chunk::Signature.new(chunk_lines) unless chunk_lines.empty?