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 indexable_chunks.map { |c| c.lines },
275 ].flatten.compact.join " "
279 indexable_chunks.map { |c| c.lines }.flatten.compact.join " "
283 chunks.select { |c| c.is_a? Chunk::Text }
286 def indexable_subject
287 Message.normalize_subj(subj)
290 def quotable_body_lines
291 chunks.find_all { |c| c.quotable? }.map { |c| c.lines }.flatten
294 def quotable_header_lines
295 ["From: #{@from.full_address}"] +
296 (@to.empty? ? [] : ["To: " + @to.map { |p| p.full_address }.join(", ")]) +
297 (@cc.empty? ? [] : ["Cc: " + @cc.map { |p| p.full_address }.join(", ")]) +
298 (@bcc.empty? ? [] : ["Bcc: " + @bcc.map { |p| p.full_address }.join(", ")]) +
299 ["Date: #{@date.rfc822}",
303 def self.build_from_source source, source_info
304 m = Message.new :source => source, :source_info => source_info
311 ## here's where we handle decoding mime attachments. unfortunately
312 ## but unsurprisingly, the world of mime attachments is a bit of a
313 ## mess. as an empiricist, i'm basing the following behavior on
314 ## observed mail rather than on interpretations of rfcs, so probably
315 ## this will have to be tweaked.
317 ## the general behavior i want is: ignore content-disposition, at
318 ## least in so far as it suggests something being inline vs being an
319 ## attachment. (because really, that should be the recipient's
320 ## decision to make.) if a mime part is text/plain, OR if the user
321 ## decoding hook converts it, then decode it and display it
322 ## inline. for these decoded attachments, if it has associated
323 ## filename, then make it collapsable and individually saveable;
324 ## otherwise, treat it as regular body text.
326 ## everything else is just an attachment and is not displayed
329 ## so, in contrast to mutt, the user is not exposed to the workings
330 ## of the gruesome slaughterhouse and sausage factory that is a
331 ## mime-encoded message, but need only see the delicious end
334 def multipart_signed_to_chunks m
336 Redwood::log "warning: multipart/signed with #{m.body.size} parts (expecting 2)"
340 payload, signature = m.body
341 if signature.multipart?
342 Redwood::log "warning: multipart/signed with payload multipart #{payload.multipart?} and signature multipart #{signature.multipart?}"
346 ## this probably will never happen
347 if payload.header.content_type == "application/pgp-signature"
348 Redwood::log "warning: multipart/signed with payload content type #{payload.header.content_type}"
352 if signature.header.content_type != "application/pgp-signature"
353 ## unknown signature type; just ignore.
354 #Redwood::log "warning: multipart/signed with signature content type #{signature.header.content_type}"
358 [CryptoManager.verify(payload, signature), message_to_chunks(payload)].flatten.compact
361 def multipart_encrypted_to_chunks m
363 Redwood::log "warning: multipart/encrypted with #{m.body.size} parts (expecting 2)"
367 control, payload = m.body
368 if control.multipart?
369 Redwood::log "warning: multipart/encrypted with control multipart #{control.multipart?} and payload multipart #{payload.multipart?}"
373 if payload.header.content_type != "application/octet-stream"
374 Redwood::log "warning: multipart/encrypted with payload content type #{payload.header.content_type}"
378 if control.header.content_type != "application/pgp-encrypted"
379 Redwood::log "warning: multipart/encrypted with control content type #{signature.header.content_type}"
383 decryptedm, sig, notice = CryptoManager.decrypt payload
384 children = message_to_chunks(decryptedm, true) if decryptedm
385 [notice, sig, children].flatten.compact
388 ## takes a RMail::Message, breaks it into Chunk:: classes.
389 def message_to_chunks m, encrypted=false, sibling_types=[]
392 case m.header.content_type
393 when "multipart/signed"
394 multipart_signed_to_chunks m
395 when "multipart/encrypted"
396 multipart_encrypted_to_chunks m
400 sibling_types = m.body.map { |p| p.header.content_type }
401 chunks = m.body.map { |p| message_to_chunks p, encrypted, sibling_types }.flatten.compact
405 elsif m.header.content_type == "message/rfc822"
406 payload = RMail::Parser.read(m.body)
407 from = payload.header.from.first
408 from_person = from ? Person.from_address(from.format) : nil
409 [Chunk::EnclosedMessage.new(from_person, payload.to_s)] +
410 message_to_chunks(payload, encrypted)
413 ## first, paw through the headers looking for a filename
414 if m.header["Content-Disposition"] && m.header["Content-Disposition"] =~ /filename="?(.*?[^\\])("|;|$)/
416 elsif m.header["Content-Type"] && m.header["Content-Type"] =~ /name="?(.*?[^\\])("|;|$)/
419 ## haven't found one, but it's a non-text message. fake
422 ## TODO: make this less lame.
423 elsif m.header["Content-Type"] && m.header["Content-Type"] !~ /^text\/plain/
425 case m.header["Content-Type"]
426 when /text\/html/ then "html"
427 when /image\/(.*)/ then $1
430 ["sup-attachment-#{Time.now.to_i}-#{rand 10000}", extension].join(".")
433 ## if there's a filename, we'll treat it as an attachment.
435 # add this to the attachments list if its not a generated html
436 # attachment (should we allow images with generated names?).
437 # Lowercase the filename because searches are easier that way
438 @attachments.push filename.downcase unless filename =~ /^sup-attachment-/
439 add_label :attachment unless filename =~ /^sup-attachment-/
440 content_type = m.header.content_type || "application/unknown" # sometimes RubyMail gives us nil
441 [Chunk::Attachment.new(content_type, filename, m, sibling_types)]
443 ## otherwise, it's body text
445 ## if there's no charset, use the current encoding as the charset.
446 ## this ensures that the body is normalized to avoid non-displayable
448 body = Iconv.easy_decode($encoding, m.charset || $encoding, m.decode) if m.body
449 text_to_chunks((body || "").normalize_whitespace.split("\n"), encrypted)
454 ## parse the lines of text into chunk objects. the heuristics here
455 ## need tweaking in some nice manner. TODO: move these heuristics
456 ## into the classes themselves.
457 def text_to_chunks lines, encrypted
458 state = :text # one of :text, :quote, or :sig
462 lines.each_with_index do |line, i|
463 nextline = lines[(i + 1) ... lines.length].find { |l| l !~ /^\s*$/ } # skip blank lines
469 ## the following /:$/ followed by /\w/ is an attempt to detect the
470 ## start of a quote. this is split into two regexen because the
471 ## original regex /\w.*:$/ had very poor behavior on long lines
472 ## like ":a:a:a:a:a" that occurred in certain emails.
473 if line =~ QUOTE_PATTERN || (line =~ /:$/ && line =~ /\w/ && nextline =~ QUOTE_PATTERN)
475 elsif line =~ SIG_PATTERN && (lines.length - i) < MAX_SIG_DISTANCE
477 elsif line =~ BLOCK_QUOTE_PATTERN
478 newstate = :block_quote
482 chunks << Chunk::Text.new(chunk_lines) unless chunk_lines.empty?
492 if line =~ QUOTE_PATTERN || (line =~ /^\s*$/ && nextline =~ QUOTE_PATTERN)
494 elsif line =~ SIG_PATTERN && (lines.length - i) < MAX_SIG_DISTANCE
501 if chunk_lines.empty?
504 chunks << Chunk::Quote.new(chunk_lines)
510 when :block_quote, :sig
514 if !@have_snippet && state == :text && (@snippet.nil? || @snippet.length < SNIPPET_LEN) && line !~ /[=\*#_-]{3,}/ && line !~ /^\s*$/
516 @snippet += " " unless @snippet.empty?
517 @snippet += line.gsub(/^\s+/, "").gsub(/[\r\n]/, "").gsub(/\s+/, " ")
518 @snippet = @snippet[0 ... SNIPPET_LEN].chomp
519 @dirty = true unless encrypted && $config[:discard_snippets_from_encrypted_messages]
520 @snippet_contains_encrypted_content = true if encrypted
526 when :quote, :block_quote
527 chunks << Chunk::Quote.new(chunk_lines) unless chunk_lines.empty?
529 chunks << Chunk::Text.new(chunk_lines) unless chunk_lines.empty?
531 chunks << Chunk::Signature.new(chunk_lines) unless chunk_lines.empty?