1 ;;; nndoc.el --- single file access for Gnus
3 ;; Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
4 ;; 2004, 2005, 2006, 2007, 2008, 2009, 2010 Free Software Foundation, Inc.
6 ;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
7 ;; Masanobu UMEDA <umerin@flab.flab.fujitsu.junet>
10 ;; This file is part of GNU Emacs.
12 ;; GNU Emacs is free software: you can redistribute it and/or modify
13 ;; it under the terms of the GNU General Public License as published by
14 ;; the Free Software Foundation, either version 3 of the License, or
15 ;; (at your option) any later version.
17 ;; GNU Emacs is distributed in the hope that it will be useful,
18 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
19 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20 ;; GNU General Public License for more details.
22 ;; You should have received a copy of the GNU General Public License
23 ;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>.
27 ;; For Outlook mail boxes format, see http://mbx2mbox.sourceforge.net/
37 (eval-when-compile (require 'cl))
41 (defvoo nndoc-article-type 'guess
43 One of `mbox', `babyl', `digest', `news', `rnews', `mmdf', `forward',
44 `rfc934', `rfc822-forward', `mime-parts', `standard-digest',
45 `slack-digest', `clari-briefs', `nsmail', `outlook', `oe-dbx',
46 `mailman', `exim-bounce', or `guess'.")
48 (defvoo nndoc-post-type 'mail
49 "*Whether the nndoc group is `mail' or `post'.")
51 (defvoo nndoc-open-document-hook 'nnheader-ms-strip-cr
52 "Hook run after opening a document.
53 The default function removes all trailing carriage returns
56 (defvar nndoc-type-alist
58 (article-begin . "^\^A\^A\^A\^A\n")
59 (body-end . "^\^A\^A\^A\^A\n"))
62 (head-begin . "^ ?\n")
66 (subtype digest guess))
68 (generate-head-function . nndoc-generate-mime-parts-head)
69 (article-transform-function . nndoc-transform-mime-parts))
71 (article-begin . "^From - "))
73 (article-begin . "^Path:"))
75 (article-begin . "^#! *rnews +\\([0-9]+\\) *\n")
76 (body-end-function . nndoc-rnews-body-end))
78 (article-begin-function . nndoc-mbox-article-begin)
79 (body-end-function . nndoc-mbox-body-end))
81 (article-begin . "\^_\^L *\n")
83 (body-begin-function . nndoc-babyl-body-begin)
84 (head-begin-function . nndoc-babyl-head-begin))
86 (article-begin . "^------ This is a copy of the message, including all the headers. ------\n\n")
87 (body-end-function . nndoc-exim-bounce-body-end-function))
89 (article-begin . "^--.*\n+")
91 (prepare-body-function . nndoc-unquote-dashes))
93 (article-begin . "^--__--__--\n\nMessage:")
94 (body-end . "^--__--__--$")
95 (prepare-body-function . nndoc-unquote-dashes))
97 (article-begin . "^ \\*")
98 (body-end . "^\t------*[ \t]^*\n^ \\*")
101 (generate-head-function . nndoc-generate-clari-briefs-head)
102 (article-transform-function . nndoc-transform-clari-briefs))
105 (first-article . ,(concat "^" (make-string 70 ?-) "\n *\n+"))
106 (article-begin . ,(concat "^\n" (make-string 30 ?-) "\n *\n+"))
107 (prepare-body-function . nndoc-unquote-dashes)
108 (body-end-function . nndoc-digest-body-end)
110 (body-begin . "^ *\n")
111 (file-end . "^End of .*digest.*[0-9].*\n\\*\\*\\|^End of.*Digest *$")
112 (subtype digest guess))
114 (article-begin . "^------------------------------*[\n \t]+")
116 (body-end-function . nndoc-digest-body-end)
117 (body-begin . "^ ?$")
118 (file-end . "^End of")
119 (prepare-body-function . nndoc-unquote-dashes)
120 (subtype digest guess))
122 (pre-dissection-function . nndoc-decode-content-transfer-encoding)
123 (article-begin . "^== [0-9]+ of [0-9]+ ==$")
124 (head-begin . "^Date:")
126 (body-end-function . nndoc-digest-body-end)
128 (file-end . "^==============================================================================$")
129 (prepare-body-function . nndoc-unquote-dashes)
130 (subtype digest guess))
132 (article-begin . "^\\\\\\\\\n")
133 (head-begin . "^\\(Paper.*:\\|arXiv:\\)")
134 (head-end . "\\(^\\\\\\\\.*\n\\|-----------------\\)")
136 (body-end . "\\(-------------------------------------------------\\|%-%-%-%-%-%-%-%-%-%-%-%-%-%-\\|%%--%%--%%--%%--%%--%%--%%--%%--\\|%%%---%%%---%%%---%%%---\\)")
137 (file-end . "\\(^Title: Recent Seminal\\|%%%---%%%---%%%---%%%---\\)")
138 (generate-head-function . nndoc-generate-lanl-gov-head)
139 (article-transform-function . nndoc-transform-lanl-gov-announce)
140 (subtype preprints guess))
142 (article-begin . "^\n+")
143 (body-end-function . nndoc-rfc822-forward-body-end-function)
144 (generate-head-function . nndoc-rfc822-forward-generate-head)
145 (generate-article-function . nndoc-rfc822-forward-generate-article))
147 (article-begin-function . nndoc-outlook-article-begin)
149 (oe-dbx ;; Outlook Express DBX format
150 (dissection-function . nndoc-oe-dbx-dissection)
151 (generate-head-function . nndoc-oe-dbx-generate-head)
152 (generate-article-function . nndoc-oe-dbx-generate-article))
154 (article-begin . "^-+ \\(Start of \\)?forwarded message.*\n+")
155 (body-end . "^-+ End \\(of \\)?forwarded message.*$")
156 (prepare-body-function . nndoc-unquote-dashes))
157 (mail-in-mail ;; Wild guess on mailer daemon's messages or others
158 (article-begin-function . nndoc-mail-in-mail-article-begin))
169 (defvar nndoc-binary-file-names ".[Dd][Bb][Xx]$"
170 "Regexp for binary nndoc file names.")
173 (defvoo nndoc-file-begin nil)
174 (defvoo nndoc-first-article nil)
175 (defvoo nndoc-article-begin nil)
176 (defvoo nndoc-head-begin nil)
177 (defvoo nndoc-head-end nil)
178 (defvoo nndoc-file-end nil)
179 (defvoo nndoc-body-begin nil)
180 (defvoo nndoc-body-end-function nil)
181 (defvoo nndoc-body-begin-function nil)
182 (defvoo nndoc-head-begin-function nil)
183 (defvoo nndoc-body-end nil)
184 ;; nndoc-dissection-alist is a list of sublists. Each sublist holds the
185 ;; following items. ARTICLE acts as the association key and is an ordinal
186 ;; starting at 1. HEAD-BEGIN [0], HEAD-END [1], BODY-BEGIN [2] and BODY-END
187 ;; [3] are positions in the `nndoc' buffer. LINE-COUNT [4] is a count of
188 ;; lines in the body. For MIME dissections only, ARTICLE-INSERT [5] and
189 ;; SUMMARY-INSERT [6] give headers to insert for full article or summary line
190 ;; generation, respectively. Other headers usually follow directly from the
191 ;; buffer. Value `nil' means no insert.
192 (defvoo nndoc-dissection-alist nil)
193 (defvoo nndoc-prepare-body-function nil)
194 (defvoo nndoc-generate-head-function nil)
195 (defvoo nndoc-article-transform-function nil)
196 (defvoo nndoc-article-begin-function nil)
197 (defvoo nndoc-generate-article-function nil)
198 (defvoo nndoc-dissection-function nil)
199 (defvoo nndoc-pre-dissection-function nil)
201 (defvoo nndoc-status-string "")
202 (defvoo nndoc-group-alist nil)
203 (defvoo nndoc-current-buffer nil
204 "Current nndoc news buffer.")
205 (defvoo nndoc-address nil)
207 (defconst nndoc-version "nndoc 1.0"
212 ;;; Interface functions
214 (nnoo-define-basics nndoc)
216 (deffoo nndoc-retrieve-headers (articles &optional newsgroup server fetch-old)
217 (when (nndoc-possibly-change-buffer newsgroup server)
218 (with-current-buffer nntp-server-buffer
221 (if (stringp (car articles))
224 (when (setq entry (cdr (assq (setq article (pop articles))
225 nndoc-dissection-alist)))
226 (insert (format "221 %d Article retrieved.\n" article))
227 (if nndoc-generate-head-function
228 (funcall nndoc-generate-head-function article)
229 (insert-buffer-substring
230 nndoc-current-buffer (car entry) (nth 1 entry)))
231 (goto-char (point-max))
232 (unless (eq (char-after (1- (point))) ?\n)
234 (insert (format "Lines: %d\n" (nth 4 entry)))
237 (nnheader-fold-continuation-lines)
240 (deffoo nndoc-request-article (article &optional newsgroup server buffer)
241 (nndoc-possibly-change-buffer newsgroup server)
243 (let ((buffer (or buffer nntp-server-buffer))
244 (entry (cdr (assq article nndoc-dissection-alist)))
250 ((stringp article) nil)
251 (nndoc-generate-article-function
252 (funcall nndoc-generate-article-function article))
254 (insert-buffer-substring
255 nndoc-current-buffer (car entry) (nth 1 entry))
258 (insert-buffer-substring
259 nndoc-current-buffer (nth 2 entry) (nth 3 entry))
261 (when nndoc-prepare-body-function
262 (funcall nndoc-prepare-body-function))
263 (when nndoc-article-transform-function
264 (funcall nndoc-article-transform-function article))
267 (deffoo nndoc-request-group (group &optional server dont-check info)
271 ((not (nndoc-possibly-change-buffer group server))
272 (nnheader-report 'nndoc "No such file or buffer: %s"
275 (nnheader-report 'nndoc "Selected group %s" group)
277 ((zerop (setq number (length nndoc-dissection-alist)))
278 (nndoc-close-group group)
279 (nnheader-report 'nndoc "No articles in group %s" group))
281 (nnheader-insert "211 %d %d %d %s\n" number 1 number group)))))
283 (deffoo nndoc-request-type (group &optional article)
284 (cond ((not article) 'unknown)
285 (nndoc-post-type nndoc-post-type)
288 (deffoo nndoc-close-group (group &optional server)
289 (nndoc-possibly-change-buffer group server)
290 (and nndoc-current-buffer
291 (buffer-name nndoc-current-buffer)
292 (kill-buffer nndoc-current-buffer))
293 (setq nndoc-group-alist (delq (assoc group nndoc-group-alist)
295 (setq nndoc-current-buffer nil)
296 (nnoo-close-server 'nndoc server)
297 (setq nndoc-dissection-alist nil)
300 (deffoo nndoc-request-list (&optional server)
303 (deffoo nndoc-request-newgroups (date &optional server)
306 (deffoo nndoc-request-list-newsgroups (&optional server)
310 ;;; Internal functions.
312 (defun nndoc-possibly-change-buffer (group source)
315 ;; The current buffer is this group's buffer.
316 ((and nndoc-current-buffer
317 (buffer-name nndoc-current-buffer)
318 (eq nndoc-current-buffer
319 (setq buf (cdr (assoc group nndoc-group-alist))))))
320 ;; We change buffers by taking an old from the group alist.
321 ;; `source' is either a string (a file name) or a buffer object.
323 (setq nndoc-current-buffer buf))
324 ;; It's a totally new group.
325 ((or (and (bufferp nndoc-address)
326 (buffer-name nndoc-address))
327 (and (stringp nndoc-address)
328 (file-exists-p nndoc-address)
329 (not (file-directory-p nndoc-address))))
330 (push (cons group (setq nndoc-current-buffer
332 (concat " *nndoc " group "*"))))
334 (setq nndoc-dissection-alist nil)
335 (with-current-buffer nndoc-current-buffer
337 (if (and (stringp nndoc-address)
338 (string-match nndoc-binary-file-names nndoc-address))
339 (let ((coding-system-for-read 'binary))
340 (mm-insert-file-contents nndoc-address))
341 (if (stringp nndoc-address)
342 (nnheader-insert-file-contents nndoc-address)
343 (insert-buffer-substring nndoc-address))
344 (run-hooks 'nndoc-open-document-hook)))))
345 ;; Initialize the nndoc structures according to this new document.
346 (when (and nndoc-current-buffer
347 (not nndoc-dissection-alist))
348 (with-current-buffer nndoc-current-buffer
350 (if (eq nndoc-article-type 'mime-parts)
351 (nndoc-dissect-mime-parts)
352 (nndoc-dissect-buffer))))
353 (unless nndoc-current-buffer
354 (nndoc-close-server))
355 ;; Return whether we managed to select a file.
356 nndoc-current-buffer))
359 ;;; Deciding what document type we have
362 (defun nndoc-set-delims ()
363 "Set the nndoc delimiter variables according to the type of the document."
364 (let ((vars '(nndoc-file-begin
366 nndoc-article-begin-function
367 nndoc-head-begin nndoc-head-end
368 nndoc-file-end nndoc-article-begin
369 nndoc-body-begin nndoc-body-end-function nndoc-body-end
370 nndoc-prepare-body-function nndoc-article-transform-function
371 nndoc-generate-head-function nndoc-body-begin-function
372 nndoc-head-begin-function
373 nndoc-generate-article-function
374 nndoc-dissection-function
375 nndoc-pre-dissection-function)))
377 (set (pop vars) nil)))
379 ;; Guess away until we find the real file type.
380 (while (assq 'guess (setq defs (cdr (assq nndoc-article-type
382 (setq nndoc-article-type (nndoc-guess-type nndoc-article-type)))
383 ;; Set the nndoc variables.
385 (set (intern (format "nndoc-%s" (caar defs)))
388 (defun nndoc-guess-type (subtype)
389 (let ((alist nndoc-type-alist)
390 results result entry)
391 (while (and (not result)
392 (setq entry (pop alist)))
393 (when (memq subtype (or (cdr (assq 'subtype entry)) '(guess)))
394 (goto-char (point-min))
395 ;; Remove blank lines.
396 (while (eq (following-char) ?\n)
398 (when (numberp (setq result (funcall (intern
399 (format "nndoc-%s-type-p"
401 (push (cons result entry) results)
403 (unless (or result results)
404 (error "Document is not of any recognized type"))
407 (cadar (last (sort results 'car-less-than-car))))))
410 ;;; Built-in type predicates and functions
413 (defun nndoc-mbox-type-p ()
414 (when (looking-at message-unix-mail-delimiter)
417 (defun nndoc-mbox-article-begin ()
418 (when (re-search-forward (concat "^" message-unix-mail-delimiter) nil t)
419 (goto-char (match-beginning 0))))
421 (defun nndoc-mbox-body-end ()
426 (and (re-search-backward
427 (concat "^" message-unix-mail-delimiter) nil t)
429 (search-forward "\n\n" beg t)
431 "^Content-Length:[ \t]*\\([0-9]+\\) *$" end t)
432 (setq len (string-to-number (match-string 1)))
433 (search-forward "\n\n" beg t)
434 (unless (= (setq len (+ (point) len)) (point-max))
435 (and (< len (point-max))
437 (looking-at message-unix-mail-delimiter)))))
440 (defun nndoc-mmdf-type-p ()
441 (when (looking-at "\^A\^A\^A\^A$")
444 (defun nndoc-news-type-p ()
445 (when (looking-at "^Path:.*\n")
448 (defun nndoc-rnews-type-p ()
449 (when (looking-at "#! *rnews")
452 (defun nndoc-rnews-body-end ()
453 (and (re-search-backward nndoc-article-begin nil t)
455 (goto-char (+ (point) (string-to-number (match-string 1))))))
457 (defun nndoc-google-type-p ()
458 (when (re-search-forward "^=3D=3D 1 of [0-9]+ =3D=3D$" nil t)
461 (defun nndoc-decode-content-transfer-encoding ()
464 (message-narrow-to-head)
465 (message-fetch-field "content-transfer-encoding"))))
467 (search-forward "\n\n" nil t))
469 (narrow-to-region (point) (point-max))
470 (mm-decode-content-transfer-encoding
471 (intern (downcase (mail-header-strip encoding))))))))
473 (defun nndoc-babyl-type-p ()
474 (when (re-search-forward "\^_\^L *\n" nil t)
477 (defun nndoc-babyl-body-begin ()
478 (re-search-forward "^\n" nil t)
479 (when (looking-at "\\*\\*\\* EOOH \\*\\*\\*")
480 (let ((next (or (save-excursion
481 (re-search-forward nndoc-article-begin nil t))
483 (unless (re-search-forward "^\n" next t)
487 (forward-line -1)))))
489 (defun nndoc-babyl-head-begin ()
490 (when (re-search-forward "^[0-9].*\n" nil t)
491 (when (looking-at "\\*\\*\\* EOOH \\*\\*\\*")
495 (defun nndoc-forward-type-p ()
496 (when (and (re-search-forward "^-+ \\(Start of \\)?forwarded message.*\n+"
498 (looking-at "[\r\n]*[a-zA-Z][a-zA-Z0-9-]*:\\|^>?From "))
501 (defun nndoc-rfc934-type-p ()
502 (when (and (re-search-forward "^-+ Start of forwarded.*\n+" nil t)
503 (not (re-search-forward "^Subject:.*digest" nil t))
504 (not (re-search-backward "^From:" nil t 2))
505 (not (re-search-forward "^From:" nil t 2)))
508 (defun nndoc-mailman-type-p ()
509 (when (re-search-forward "^--__--__--\n+" nil t)
512 (defun nndoc-rfc822-forward-type-p ()
514 (message-narrow-to-head)
515 (when (re-search-forward "^Content-Type: *message/rfc822" nil t)
518 (defun nndoc-rfc822-forward-body-end-function ()
519 (goto-char (point-max)))
521 (defun nndoc-rfc822-forward-generate-article (article &optional head)
522 (let ((entry (cdr (assq article nndoc-dissection-alist)))
525 (with-current-buffer nndoc-current-buffer
527 (message-narrow-to-head)
528 (setq encoding (message-fetch-field "content-transfer-encoding"))))
529 (insert-buffer-substring nndoc-current-buffer (car entry) (nth 3 entry))
532 (narrow-to-region begin (point-max))
533 (mm-decode-content-transfer-encoding
534 (intern (downcase (mail-header-strip encoding))))))
537 (when (search-forward "\n\n" nil t)
538 (delete-region (1- (point)) (point-max)))))
541 (defun nndoc-rfc822-forward-generate-head (article)
542 (nndoc-rfc822-forward-generate-article article 'head))
544 (defun nndoc-mime-parts-type-p ()
545 (let ((case-fold-search t)
546 (limit (search-forward "\n\n" nil t)))
547 (goto-char (point-min))
551 ^Content-Type:[ \t]*multipart/[a-z]+ *; *\\(\\(\n[ \t]\\)?.*;\\)*"
552 "\\(\n[ \t]\\)?[ \t]*boundary=\"?[^\"\n]*[^\" \t\n]")
556 (defun nndoc-transform-mime-parts (article)
557 (let* ((entry (cdr (assq article nndoc-dissection-alist)))
558 (headers (nth 5 entry)))
560 (goto-char (point-min))
563 (defun nndoc-generate-mime-parts-head (article)
564 (let* ((entry (cdr (assq article nndoc-dissection-alist)))
565 (headers (nth 6 entry)))
567 (narrow-to-region (point) (point))
568 (insert-buffer-substring
569 nndoc-current-buffer (car entry) (nth 1 entry))
570 (goto-char (point-max)))
574 (defun nndoc-clari-briefs-type-p ()
575 (when (let ((case-fold-search nil))
576 (re-search-forward "^\t[^a-z]+ ([^a-z]+) --" nil t))
579 (defun nndoc-transform-clari-briefs (article)
580 (goto-char (point-min))
581 (when (looking-at " *\\*\\(.*\\)\n")
582 (replace-match "" t t))
583 (nndoc-generate-clari-briefs-head article))
585 (defun nndoc-generate-clari-briefs-head (article)
586 (let ((entry (cdr (assq article nndoc-dissection-alist)))
588 (with-current-buffer nndoc-current-buffer
590 (narrow-to-region (car entry) (nth 3 entry))
591 (goto-char (point-min))
592 (when (looking-at " *\\*\\(.*\\)$")
593 (setq subject (match-string 1))
594 (when (string-match "[ \t]+$" subject)
595 (setq subject (substring subject 0 (match-beginning 0)))))
597 (let ((case-fold-search nil))
599 "^\t\\([^a-z]+\\(,[^(]+\\)? ([^a-z]+)\\) --" nil t))
600 (setq from (match-string 1)))))
601 (insert "From: " "clari@clari.net (" (or from "unknown") ")"
602 "\nSubject: " (or subject "(no subject)") "\n")))
604 (defun nndoc-exim-bounce-type-p ()
605 (and (re-search-forward "^------ This is a copy of the message, including all the headers. ------" nil t)
608 (defun nndoc-exim-bounce-body-end-function ()
609 (goto-char (point-max)))
612 (defun nndoc-mime-digest-type-p ()
613 (let ((case-fold-search t)
614 boundary-id b-delimiter entry)
617 (concat "^Content-Type: *multipart/digest;[ \t\n]*[ \t]"
618 "boundary=\"?\\([^\"\n]*[^\" \t\n]\\)")
621 (setq boundary-id (match-string 1)
622 b-delimiter (concat "\n--" boundary-id "[ \t]*$"))
623 (setq entry (assq 'mime-digest nndoc-type-alist))
626 (cons 'head-begin "^ ?\n")
627 (cons 'head-end "^ ?$")
628 (cons 'body-begin "^ ?\n")
629 (cons 'article-begin b-delimiter)
630 (cons 'body-end-function 'nndoc-digest-body-end)
631 (cons 'file-end (concat "^--" boundary-id "--[ \t]*$"))))
634 (defun nndoc-standard-digest-type-p ()
635 (when (and (re-search-forward (concat "^" (make-string 70 ?-) "\n\n") nil t)
637 (concat "\n\n" (make-string 30 ?-) "\n\n") nil t))
640 (defun nndoc-digest-body-end ()
641 (and (re-search-forward nndoc-article-begin nil t)
642 (goto-char (match-beginning 0))))
644 (defun nndoc-slack-digest-type-p ()
647 (defun nndoc-lanl-gov-announce-type-p ()
648 (when (let ((case-fold-search nil))
649 (re-search-forward "^\\\\\\\\\n\\(Paper\\( (\\*cross-listing\\*)\\)?: [a-zA-Z-\\.]+/[0-9]+\\|arXiv:\\)" nil t))
652 (defun nndoc-transform-lanl-gov-announce (article)
653 (let ((case-fold-search nil))
654 (goto-char (point-max))
655 (when (re-search-backward "^\\\\\\\\ +( *\\([^ ]*\\) , *\\([^ ]*\\))" nil t)
656 (replace-match "\n\nGet it at \\1 (\\2)" t nil))
657 (goto-char (point-min))
658 (while (re-search-forward "^\\\\\\\\$" nil t)
659 (replace-match "" t nil))
660 (goto-char (point-min))
661 (when (re-search-forward "^replaced with revised version +\\(.*[^ ]\\) +" nil t)
662 (replace-match "Date: \\1 (revised) " t nil))
663 (goto-char (point-min))
664 (unless (re-search-forward "^From" nil t)
665 (goto-char (point-min))
666 (when (re-search-forward "^Authors?: \\(.*\\)" nil t)
667 (goto-char (point-min))
668 (insert "From: " (match-string 1) "\n")))
669 (when (re-search-forward "^arXiv:" nil t)
670 (replace-match "Paper: arXiv:" t nil))))
672 (defun nndoc-generate-lanl-gov-head (article)
673 (let ((entry (cdr (assq article nndoc-dissection-alist)))
674 (from "<no address given>")
676 (with-current-buffer nndoc-current-buffer
678 (narrow-to-region (car entry) (nth 1 entry))
679 (goto-char (point-min))
680 (when (looking-at "^\\(Paper.*: \\|arXiv:\\)\\([0-9a-zA-Z-\\./]+\\)")
681 (setq subject (concat " (" (match-string 2) ")"))
682 (when (re-search-forward "^From: \\(.*\\)" nil t)
683 (setq from (concat "<"
684 (cadr (funcall gnus-extract-address-components
685 (match-string 1))) ">")))
686 (if (re-search-forward "^Date: +\\([^(]*\\)" nil t)
687 (setq date (match-string 1))
688 (when (re-search-forward "^replaced with revised version +\\([^(]*\\)" nil t)
689 (setq date (match-string 1))))
690 (when (re-search-forward "^Title: \\([^\f]*\\)\nAuthors?: \\(.*\\)"
692 (setq subject (concat (match-string 1) subject))
693 (setq from (concat (match-string 2) " " from))))))
694 (while (and from (string-match "(\[^)\]*)" from))
695 (setq from (replace-match "" t t from)))
696 (insert "From: " (or from "unknown")
697 "\nSubject: " (or subject "(no subject)") "\n")
698 (if date (insert "Date: " date))))
700 (defun nndoc-nsmail-type-p ()
701 (when (looking-at "From - ")
704 (defun nndoc-outlook-article-begin ()
705 (prog1 (re-search-forward "From:\\|Received:" nil t)
706 (goto-char (match-beginning 0))))
708 (defun nndoc-outlook-type-p ()
709 ;; FIXME: Is JMF the magic of outlook mailbox? -- ShengHuo.
712 (defun nndoc-oe-dbx-type-p ()
713 (looking-at (mm-string-to-multibyte "\317\255\022\376")))
715 (defun nndoc-read-little-endian ()
716 (+ (prog1 (char-after) (forward-char 1))
717 (lsh (prog1 (char-after) (forward-char 1)) 8)
718 (lsh (prog1 (char-after) (forward-char 1)) 16)
719 (lsh (prog1 (char-after) (forward-char 1)) 24)))
721 (defun nndoc-oe-dbx-decode-block ()
723 (nndoc-read-little-endian) ;; this address
724 (nndoc-read-little-endian) ;; next address offset
725 (nndoc-read-little-endian) ;; blocksize
726 (nndoc-read-little-endian))) ;; next address
728 (defun nndoc-oe-dbx-dissection ()
729 (let ((i 0) blk p tp)
730 (goto-char 60117) ;; 0x0000EAD4+1
733 (setq blk (nndoc-oe-dbx-decode-block)))
734 (while (and blk (> (car blk) 0) (or (zerop (nth 3 blk))
736 (push (list (incf i) p nil nil nil 0) nndoc-dissection-alist)
737 (while (and (> (car blk) 0) (> (nth 3 blk) p))
738 (goto-char (1+ (nth 3 blk)))
739 (setq blk (nndoc-oe-dbx-decode-block)))
740 (if (or (<= (car blk) p)
742 (not (zerop (nth 3 blk))))
744 (setq tp (+ (car blk) (nth 1 blk) 17))
745 (if (or (<= tp p) (>= tp (point-max)))
749 blk (nndoc-oe-dbx-decode-block)))))))
751 (defun nndoc-oe-dbx-generate-article (article &optional head)
752 (let ((entry (cdr (assq article nndoc-dissection-alist)))
753 (cur (current-buffer))
756 (with-current-buffer nndoc-current-buffer
758 (while (> p (point-min))
760 (setq blk (nndoc-oe-dbx-decode-block))
762 (with-current-buffer cur
763 (insert-buffer-substring nndoc-current-buffer p (+ p (nth 2 blk))))
764 (setq p (1+ (nth 3 blk)))))
766 (while (re-search-forward "\r$" nil t)
770 (when (search-forward "\n\n" nil t)
771 (setcar (cddddr entry) (count-lines (point) (point-max)))
772 (delete-region (1- (point)) (point-max))))
775 (defun nndoc-oe-dbx-generate-head (article)
776 (nndoc-oe-dbx-generate-article article 'head))
778 (defun nndoc-mail-in-mail-type-p ()
782 (while (re-search-forward "\n\n[-A-Za-z0-9]+:" nil t)
785 (while (looking-at "[ \t]\\|[-A-Za-z0-9]+:")
786 (if (looking-at "[-A-Za-z0-9]+:")
787 (setq found (1+ found)))
789 (if (and (> found 0) (looking-at "\n"))
793 (defun nndoc-mail-in-mail-article-begin ()
796 (while (re-search-forward "\n\n\\([-A-Za-z0-9]+:\\)" nil t)
798 (setq point (match-beginning 1))
800 (while (looking-at "[ \t]\\|[-A-Za-z0-9]+:")
801 (if (looking-at "[-A-Za-z0-9]+:")
802 (setq found (1+ found)))
804 (if (and (> found 0) (looking-at "\n"))
809 (deffoo nndoc-request-accept-article (group &optional server last)
813 ;;; Functions for dissecting the documents
816 (defun nndoc-search (regexp)
818 (re-search-forward regexp nil t)
819 (beginning-of-line)))
821 (defun nndoc-dissect-buffer ()
822 "Go through the document and partition it into heads/bodies/articles."
825 art-begin head-begin head-end body-begin body-end)
826 (setq nndoc-dissection-alist nil)
827 (with-current-buffer nndoc-current-buffer
828 (goto-char (point-min))
829 ;; Remove blank lines.
830 (while (eq (following-char) ?\n)
832 (when nndoc-pre-dissection-function
834 (funcall nndoc-pre-dissection-function)))
835 (if nndoc-dissection-function
836 (funcall nndoc-dissection-function)
837 ;; Find the beginning of the file.
838 (when nndoc-file-begin
839 (nndoc-search nndoc-file-begin))
840 ;; Go through the file.
841 (while (if (and first nndoc-first-article)
842 (nndoc-search nndoc-first-article)
844 (goto-char art-begin)
845 (nndoc-article-begin)))
848 (cond (nndoc-head-begin-function
849 (funcall nndoc-head-begin-function))
851 (nndoc-search nndoc-head-begin)))
854 (looking-at nndoc-file-end)))
855 (goto-char (point-max))
856 (setq head-begin (point))
857 (nndoc-search (or nndoc-head-end "^$"))
858 (setq head-end (point))
859 (if nndoc-body-begin-function
860 (funcall nndoc-body-begin-function)
861 (nndoc-search (or nndoc-body-begin "^\n")))
862 (setq body-begin (point))
863 (or (and nndoc-body-end-function
864 (funcall nndoc-body-end-function))
866 (nndoc-search nndoc-body-end))
867 (and (nndoc-article-begin)
868 (setq art-begin (point)))
870 (goto-char (point-max))
872 (and (re-search-backward nndoc-file-end nil t)
873 (beginning-of-line)))))
874 (setq body-end (point))
875 (push (list (incf i) head-begin head-end body-begin body-end
876 (count-lines body-begin body-end))
877 nndoc-dissection-alist)))))))
879 (defun nndoc-article-begin ()
880 (if nndoc-article-begin-function
881 (funcall nndoc-article-begin-function)
883 (nndoc-search nndoc-article-begin))))
885 (defun nndoc-unquote-dashes ()
886 "Unquote quoted non-separators in digests."
887 (while (re-search-forward "^- -"nil t)
888 (replace-match "-" t t)))
890 ;; Against compiler warnings.
891 (defvar nndoc-mime-split-ordinal)
893 (defun nndoc-dissect-mime-parts ()
894 "Go through a MIME composite article and partition it into sub-articles.
895 When a MIME entity contains sub-entities, dissection produces one article for
896 the header of this entity, and one article per sub-entity."
897 (setq nndoc-dissection-alist nil
898 nndoc-mime-split-ordinal 0)
899 (with-current-buffer nndoc-current-buffer
900 (nndoc-dissect-mime-parts-sub (point-min) (point-max) nil nil nil)))
902 (defun nndoc-dissect-mime-parts-sub (head-begin body-end article-insert
904 "Dissect an entity, within a composite MIME message.
905 The complete message or MIME entity extends from HEAD-BEGIN to BODY-END.
906 ARTICLE-INSERT should be added at beginning for generating a full article.
907 The string POSITION holds a dotted decimal representation of the article
908 position in the hierarchical structure, it is nil for the outer entity.
909 PARENT is the message-ID of the parent summary line, or nil for none."
910 (let ((case-fold-search t)
911 (message-id (nnmail-message-id))
912 head-end body-begin summary-insert message-rfc822 multipart-any
913 subject content-type type subtype boundary-regexp)
914 ;; Gracefully handle a missing body.
915 (goto-char head-begin)
916 (if (or (and (eq (char-after) ?\n) (or (forward-char 1) t))
917 (search-forward "\n\n" body-end t))
918 (setq head-end (1- (point))
920 (setq head-end body-end
921 body-begin body-end))
922 (narrow-to-region head-begin head-end)
923 ;; Save MIME attributes.
924 (goto-char head-begin)
925 (setq content-type (message-fetch-field "Content-Type"))
928 "^ *\\([^ \t\n/;]+\\)/\\([^ \t\n/;]+\\)" content-type)
929 (setq type (downcase (match-string 1 content-type))
930 subtype (downcase (match-string 2 content-type))
931 message-rfc822 (and (string= type "message")
932 (string= subtype "rfc822"))
933 multipart-any (string= type "multipart")))
934 (when (string-match ";[ \t\n]*name=\\([^ \t\n;]+\\)" content-type)
935 (setq subject (match-string 1 content-type)))
936 (when (string-match "boundary=\"?\\([^\"\n]*[^\" \t\n]\\)" content-type)
937 (setq boundary-regexp (concat "^--"
939 (match-string 1 content-type))
940 "\\(--\\)?[ \t]*\n"))))
942 (when (or multipart-any (not article-insert))
943 (setq subject (message-fetch-field "Subject"))))
947 ;; Prepare the article and summary inserts.
948 (unless article-insert
949 (setq article-insert (buffer-string)
950 head-end head-begin))
952 (unless (string-match "MIME-Version:" article-insert)
954 (concat article-insert "MIME-Version: 1.0\n")))
955 (setq summary-insert article-insert)
956 ;; - summary Subject.
958 (let ((line (concat "Subject: <" position
959 (and position multipart-any ".")
960 (and multipart-any "*")
961 (and (or position multipart-any) " ")
962 (cond ((string= subtype "plain") type)
963 ((string= subtype "basic") type)
969 (if (string-match "Subject:.*\n\\([ \t].*\n\\)*" summary-insert)
970 (replace-match line t t summary-insert)
971 (concat summary-insert line))))
972 ;; - summary Message-ID.
974 (let ((line (concat "Message-ID: " message-id "\n")))
975 (if (string-match "Message-ID:.*\n\\([ \t].*\n\\)*" summary-insert)
976 (replace-match line t t summary-insert)
977 (concat summary-insert line))))
978 ;; - summary References.
981 (let ((line (concat "References: " parent "\n")))
982 (if (string-match "References:.*\n\\([ \t].*\n\\)*"
984 (replace-match line t t summary-insert)
985 (concat summary-insert line)))))
986 ;; Generate dissection information for this entity.
987 (push (list (incf nndoc-mime-split-ordinal)
988 head-begin head-end body-begin body-end
989 (count-lines body-begin body-end)
990 article-insert summary-insert)
991 nndoc-dissection-alist)
992 ;; Recurse for all sub-entities, if any.
997 (nndoc-dissect-mime-parts-sub body-begin body-end nil
998 position message-id)))
999 ((and multipart-any boundary-regexp)
1000 (let ((part-counter 0)
1001 part-begin part-end eof-flag)
1002 (while (string-match "\
1003 ^\\(Lines\\|Content-\\(Type\\|Transfer-Encoding\\|Disposition\\)\\):.*\n\\([ \t].*\n\\)*"
1005 (setq article-insert (replace-match "" t t article-insert)))
1006 (let ((case-fold-search nil))
1007 (goto-char body-begin)
1008 (setq eof-flag (not (re-search-forward boundary-regexp body-end t)))
1009 (while (not eof-flag)
1010 (setq part-begin (point))
1011 (cond ((re-search-forward boundary-regexp body-end t)
1012 (or (not (match-string 1))
1013 (string= (match-string 1) "")
1016 (setq part-end (point))
1018 (t (setq part-end body-end
1021 (nndoc-dissect-mime-parts-sub
1022 part-begin part-end article-insert
1025 (format "%d" (incf part-counter)))
1029 (defun nndoc-add-type (definition &optional position)
1030 "Add document DEFINITION to the list of nndoc document definitions.
1031 If POSITION is nil or `last', the definition will be added
1032 as the last checked definition, if t or `first', add as the
1033 first definition, and if any other symbol, add after that
1034 symbol in the alist."
1035 ;; First remove any old instances.
1036 (gnus-pull (car definition) nndoc-type-alist)
1037 ;; Then enter the new definition in the proper place.
1039 ((or (null position) (eq position 'last))
1040 (setq nndoc-type-alist (nconc nndoc-type-alist (list definition))))
1041 ((or (eq position t) (eq position 'first))
1042 (push definition nndoc-type-alist))
1044 (let ((list (memq (assq position nndoc-type-alist)
1047 (error "No such position: %s" position))
1048 (setcdr list (cons definition (cdr list)))))))
1052 ;;; nndoc.el ends here