lisp/gnus/html2text.el

   1 ;;; html2text.el --- a simple html to plain text converter
   2 ;; Copyright (C) 2002, 2003, 2004 Free Software Foundation, Inc.
   3
   4 ;; Author: Joakim Hove <hove@phys.ntnu.no>
   5
   6 ;; This file is part of GNU Emacs.
   7
   8 ;; GNU Emacs is free software; you can redistribute it and/or modify
   9 ;; it under the terms of the GNU General Public License as published by
  10 ;; the Free Software Foundation; either version 2, or (at your option)
  11 ;; any later version.
  12
  13 ;; GNU Emacs is distributed in the hope that it will be useful,
  14 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  15 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16 ;; GNU General Public License for more details.
  17
  18 ;; You should have received a copy of the GNU General Public License
  19 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  20 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  21 ;; Boston, MA 02111-1307, USA.
  22
  23 ;;; Commentary:
  24
  25 ;; These functions provide a simple way to wash/clean html infected
  26 ;; mails.  Definitely do not work in all cases, but some improvement
  27 ;; in readability is generally obtained.  Formatting is only done in
  28 ;; the buffer, so the next time you enter the article it will be
  29 ;; "re-htmlized".
  30 ;;
  31 ;; The main function is `html2text'.
  32
  33 ;;; Code:
  34
  35 ;;
  36 ;; <Global variables>
  37 ;;
  38
  39 (eval-when-compile
  40   (require 'cl))
  41
  42 (defvar html2text-format-single-element-list '(("hr" . html2text-clean-hr)))
  43
  44 (defvar html2text-replace-list
  45   '(("&nbsp;" . " ") ("&gt;" . ">") ("&lt;" . "<") ("&quot;" . "\"")
  46     ("&amp;" . "&") ("&apos;" . "'"))
  47   "The map of entity to text.
  48
  49 This is an alist were each element is a dotted pair consisting of an
  50 old string, and a replacement string.  This replacement is done by the
  51 function `html2text-substitute' which basically performs a
  52 `replace-string' operation for every element in the list.  This is
  53 completely verbatim - without any use of REGEXP.")
  54
  55 (defvar html2text-remove-tag-list
  56   '("html" "body" "p" "img" "dir" "head" "div" "br" "font" "title" "meta")
  57   "A list of removable tags.
  58
  59 This is a list of tags which should be removed, without any
  60 formatting.  Note that tags in the list are presented *without*
  61 any \"<\" or \">\".  All occurences of a tag appearing in this
  62 list are removed, irrespective of whether it is a closing or
  63 opening tag, or if the tag has additional attributes.  The
  64 deletion is done by the function `html2text-remove-tags'.
  65
  66 For instance the text:
  67
  68 \"Here comes something <font size\"+3\" face=\"Helvetica\"> big </font>.\"
  69
  70 will be reduced to:
  71
  72 \"Here comes something big.\"
  73
  74 If this list contains the element \"font\".")
  75
  76 (defvar html2text-format-tag-list
  77   '(("b"          . html2text-clean-bold)
  78     ("strong"     . html2text-clean-bold)
  79     ("u"          . html2text-clean-underline)
  80     ("i"          . html2text-clean-italic)
  81     ("em"         . html2text-clean-italic)
  82     ("blockquote" . html2text-clean-blockquote)
  83     ("a"          . html2text-clean-anchor)
  84     ("ul"         . html2text-clean-ul)
  85     ("ol"         . html2text-clean-ol)
  86     ("dl"         . html2text-clean-dl)
  87     ("center"     . html2text-clean-center))
  88   "An alist of tags and processing functions.
  89
  90 This is an alist where each dotted pair consists of a tag, and then
  91 the name of a function to be called when this tag is found.  The
  92 function is called with the arguments p1, p2, p3 and p4. These are
  93 demontrated below:
  94
  95 \"<b> This is bold text </b>\"
  96  ^   ^                 ^    ^
  97  |   |                 |    |
  98 p1  p2                p3   p4
  99
 100 Then the called function will typically format the text somewhat and
 101 remove the tags.")
 102
 103 (defvar html2text-remove-tag-list2  '("li" "dt" "dd" "meta")
 104   "Another list of removable tags.
 105
 106 This is a list of tags which are removed similarly to the list
 107 `html2text-remove-tag-list' - but these tags are retained for the
 108 formatting, and then moved afterward.")
 109
 110 ;;
 111 ;; </Global variables>
 112 ;;
 113
 114 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 115 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 116
 117 ;;
 118 ;; <Utility functions>
 119 ;;
 120
 121
 122 (defun html2text-replace-string (from-string to-string min max)
 123   "Replace FROM-STRING with TO-STRING in region from MIN to MAX."
 124   (goto-char min)
 125   (let ((delta (- (string-width to-string) (string-width from-string)))
 126         (change 0))
 127     (while (search-forward from-string max t)
 128       (replace-match to-string)
 129       (setq change (+ change delta)))
 130     change))
 131
 132 ;;
 133 ;; </Utility functions>
 134 ;;
 135
 136 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 137 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 138
 139 ;;
 140 ;; <Functions related to attributes> i.e. <font size=+3>
 141 ;;
 142
 143 (defun html2text-attr-value (list attribute)
 144   "Get value of ATTRIBUTE from LIST."
 145   (nth 1 (assoc attribute list)))
 146
 147 (defun html2text-get-attr (p1 p2)
 148   (goto-char p1)
 149   (re-search-forward " +[^ ]" p2 t)
 150   (let* ((attr-string (buffer-substring-no-properties (1- (point)) (1- p2)))
 151          (tmp-list (split-string attr-string))
 152          (attr-list)
 153          (counter 0)
 154          (prev (car tmp-list))
 155          (this (nth 1 tmp-list))
 156          (next (nth 2 tmp-list))
 157          (index 1))
 158
 159     (cond
 160      ;; size=3
 161      ((string-match "[^ ]=[^ ]" prev)
 162       (let ((attr  (nth 0 (split-string prev "=")))
 163             (value (nth 1 (split-string prev "="))))
 164         (setq attr-list (cons (list attr value) attr-list))))
 165      ;; size= 3
 166      ((string-match "[^ ]=\\'" prev)
 167       (setq attr-list (cons (list (substring prev 0 -1) this) attr-list))))
 168
 169     (while (< index (length tmp-list))
 170       (cond
 171        ;; size=3
 172        ((string-match "[^ ]=[^ ]" this)
 173         (let ((attr  (nth 0 (split-string this "=")))
 174               (value (nth 1 (split-string this "="))))
 175           (setq attr-list (cons (list attr value) attr-list))))
 176        ;; size =3
 177        ((string-match "\\`=[^ ]" this)
 178         (setq attr-list (cons (list prev (substring this 1)) attr-list)))
 179        ;; size= 3
 180        ((string-match "[^ ]=\\'" this)
 181         (setq attr-list (cons (list (substring this 0 -1) next) attr-list)))
 182        ;; size = 3
 183        ((string= "=" this)
 184         (setq attr-list (cons (list prev next) attr-list))))
 185       (setq index (1+ index))
 186       (setq prev this)
 187       (setq this next)
 188       (setq next (nth (1+ index) tmp-list)))
 189     ;;
 190     ;; Tags with no accompanying "=" i.e. value=nil
 191     ;;
 192     (setq prev (car tmp-list))
 193     (setq this (nth 1 tmp-list))
 194     (setq next (nth 2 tmp-list))
 195     (setq index 1)
 196
 197     (when (and (not (string-match "=" prev))
 198                (not (string= (substring this 0 1) "=")))
 199       (setq attr-list (cons (list prev nil) attr-list)))
 200     (while (< index (1- (length tmp-list)))
 201       (when (and (not (string-match "=" this))
 202                  (not (or (string= (substring next 0 1) "=")
 203                           (string= (substring prev -1) "="))))
 204         (setq attr-list (cons (list this nil) attr-list)))
 205       (setq index (1+ index))
 206       (setq prev this)
 207       (setq this next)
 208       (setq next (nth (1+ index) tmp-list)))
 209
 210     (when (and this
 211                (not (string-match "=" this))
 212                (not (string= (substring prev -1) "=")))
 213       (setq attr-list (cons (list this nil) attr-list)))
 214     ;; return - value
 215     attr-list))
 216
 217 ;;
 218 ;; </Functions related to attributes>
 219 ;;
 220
 221 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 222 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 223
 224 ;;
 225 ;; <Functions to be called to format a tag-pair>
 226 ;;
 227 (defun html2text-clean-list-items (p1 p2 list-type)
 228   (goto-char p1)
 229   (let ((item-nr 0)
 230         (items   0))
 231     (while (re-search-forward "<li>" p2 t)
 232       (setq items (1+ items)))
 233     (goto-char p1)
 234     (while (< item-nr items)
 235       (setq item-nr (1+ item-nr))
 236       (re-search-forward "<li>" (point-max) t)
 237       (cond
 238        ((string= list-type "ul") (insert " o "))
 239        ((string= list-type "ol") (insert (format " %s: " item-nr)))
 240        (t (insert " x "))))))
 241
 242 (defun html2text-clean-dtdd (p1 p2)
 243   (goto-char p1)
 244   (let ((items   0)
 245         (item-nr 0))
 246     (while (re-search-forward "<dt>" p2 t)
 247       (setq items (1+ items)))
 248     (goto-char p1)
 249     (while (< item-nr items)
 250       (setq item-nr (1+ item-nr))
 251       (re-search-forward "<dt>\\([ ]*\\)" (point-max) t)
 252       (when (match-string 1)
 253         (delete-region (point) (- (point) (string-width (match-string 1)))))
 254       (let ((def-p1 (point))
 255             (def-p2 0))
 256         (re-search-forward "\\([ ]*\\)\\(</dt>\\|<dd>\\)" (point-max) t)
 257         (if (match-string 1)
 258             (progn
 259               (let* ((mw1 (string-width (match-string 1)))
 260                      (mw2 (string-width (match-string 2)))
 261                      (mw  (+ mw1 mw2)))
 262                 (goto-char (- (point) mw))
 263                 (delete-region (point) (+ (point) mw1))
 264                 (setq def-p2 (point))))
 265           (setq def-p2 (- (point) (string-width (match-string 2)))))
 266         (put-text-property def-p1 def-p2 'face 'bold)))))
 267
 268 (defun html2text-delete-tags (p1 p2 p3 p4)
 269   (delete-region p1 p2)
 270   (delete-region (- p3 (- p2 p1)) (- p4 (- p2 p1))))
 271
 272 (defun html2text-delete-single-tag (p1 p2)
 273   (delete-region p1 p2))
 274
 275 (defun html2text-clean-hr (p1 p2)
 276   (html2text-delete-single-tag p1 p2)
 277   (goto-char p1)
 278   (newline 1)
 279   (insert (make-string fill-column ?-)))
 280
 281 (defun html2text-clean-ul (p1 p2 p3 p4)
 282   (html2text-delete-tags p1 p2 p3 p4)
 283   (html2text-clean-list-items p1 (- p3 (- p1 p2)) "ul"))
 284
 285 (defun html2text-clean-ol (p1 p2 p3 p4)
 286   (html2text-delete-tags p1 p2 p3 p4)
 287   (html2text-clean-list-items p1 (- p3 (- p1 p2)) "ol"))
 288
 289 (defun html2text-clean-dl (p1 p2 p3 p4)
 290   (html2text-delete-tags p1 p2 p3 p4)
 291   (html2text-clean-dtdd p1 (- p3 (- p1 p2))))
 292
 293 (defun html2text-clean-center (p1 p2 p3 p4)
 294   (html2text-delete-tags p1 p2 p3 p4)
 295   (center-region p1 (- p3 (- p2 p1))))
 296
 297 (defun html2text-clean-bold (p1 p2 p3 p4)
 298   (put-text-property p2 p3 'face 'bold)
 299   (html2text-delete-tags p1 p2 p3 p4))
 300
 301 (defun html2text-clean-title (p1 p2 p3 p4)
 302   (put-text-property p2 p3 'face 'bold)
 303   (html2text-delete-tags p1 p2 p3 p4))
 304
 305 (defun html2text-clean-underline (p1 p2 p3 p4)
 306   (put-text-property p2 p3 'face 'underline)
 307   (html2text-delete-tags p1 p2 p3 p4))
 308
 309 (defun html2text-clean-italic (p1 p2 p3 p4)
 310   (put-text-property p2 p3 'face 'italic)
 311   (html2text-delete-tags p1 p2 p3 p4))
 312
 313 (defun html2text-clean-font (p1 p2 p3 p4)
 314   (html2text-delete-tags p1 p2 p3 p4))
 315
 316 (defun html2text-clean-blockquote (p1 p2 p3 p4)
 317   (html2text-delete-tags p1 p2 p3 p4))
 318
 319 (defun html2text-clean-anchor (p1 p2 p3 p4)
 320   ;; If someone can explain how to make the URL clickable I will surely
 321   ;; improve upon this.
 322   ;; Maybe `goto-addr.el' can be used here.
 323   (let* ((attr-list (html2text-get-attr p1 p2))
 324          (href (html2text-attr-value attr-list "href")))
 325     (delete-region p1 p4)
 326     (when href
 327       (goto-char p1)
 328       (insert (substring href 1 -1 ))
 329       (put-text-property p1 (point) 'face 'bold))))
 330
 331 ;;
 332 ;; </Functions to be called to format a tag-pair>
 333 ;;
 334
 335 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 336 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 337
 338 ;;
 339 ;; <Functions to be called to fix up paragraphs>
 340 ;;
 341
 342 (defun html2text-fix-paragraph (p1 p2)
 343   (goto-char p1)
 344   (let ((has-br-line)
 345         (refill-start)
 346         (refill-stop))
 347     (when (re-search-forward "<br>$" p2 t)
 348       (goto-char p1)
 349       (when (re-search-forward ".+[^<][^b][^r][^>]$" p2 t)
 350         (beginning-of-line)
 351         (setq refill-start (point))
 352         (goto-char p2)
 353         (re-search-backward ".+[^<][^b][^r][^>]$" refill-start t)
 354         (next-line 1)
 355         (end-of-line)
 356         ;; refill-stop should ideally be adjusted to
 357         ;; accomodate the "<br>" strings which are removed
 358         ;; between refill-start and refill-stop.  Can simply
 359         ;; be returned from my-replace-string
 360         (setq refill-stop (+ (point)
 361                              (html2text-replace-string
 362                               "<br>" ""
 363                               refill-start (point))))
 364         ;; (message "Point = %s  refill-stop = %s" (point) refill-stop)
 365         ;; (sleep-for 4)
 366         (fill-region refill-start refill-stop))))
 367   (html2text-replace-string "<br>" "" p1 p2))
 368
 369 ;;
 370 ;; This one is interactive ...
 371 ;;
 372 (defun html2text-fix-paragraphs ()
 373   "This _tries_ to fix up the paragraphs - this is done in quite a ad-hook
 374 fashion, quite close to pure guess-work. It does work in some cases though."
 375   (interactive)
 376   (goto-char (point-min))
 377   (replace-regexp "^<br>$" "")
 378   ;; Removing lonely <br> on a single line, if they are left intact we
 379   ;; dont have any paragraphs at all.
 380   (goto-char (point-min))
 381   (while (not (eobp))
 382     (let ((p1 (point)))
 383       (forward-paragraph 1)
 384       ;;(message "Kaller fix med p1=%s  p2=%s " p1 (1- (point))) (sleep-for 5)
 385       (html2text-fix-paragraph p1 (1- (point)))
 386       (goto-char p1)
 387       (when (not (eobp))
 388         (forward-paragraph 1)))))
 389
 390 ;;
 391 ;; </Functions to be called to fix up paragraphs>
 392 ;;
 393
 394 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 395 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 396
 397 ;;
 398 ;; <Interactive functions>
 399 ;;
 400
 401 (defun html2text-remove-tags (tag-list)
 402   "Removes the tags listed in the list `html2text-remove-tag-list'.
 403 See the documentation for that variable."
 404   (interactive)
 405   (dolist (tag tag-list)
 406     (goto-char (point-min))
 407     (while (re-search-forward (format "\\(</?%s[^>]*>\\)" tag) (point-max) t)
 408       (delete-region (match-beginning 0) (match-end 0)))))
 409
 410 (defun html2text-format-tags ()
 411   "See the variable `html2text-format-tag-list' for documentation."
 412   (interactive)
 413   (dolist (tag-and-function html2text-format-tag-list)
 414     (let ((tag      (car tag-and-function))
 415           (function (cdr tag-and-function)))
 416       (goto-char (point-min))
 417       (while (re-search-forward (format "\\(<%s\\( [^>]*\\)?>\\)" tag)
 418                                 (point-max) t)
 419         (let ((p1)
 420               (p2 (point))
 421               (p3) (p4))
 422           (search-backward "<" (point-min) t)
 423           (setq p1 (point))
 424           (re-search-forward (format "</%s>" tag) (point-max) t)
 425           (setq p4 (point))
 426           (search-backward "</" (point-min) t)
 427           (setq p3 (point))
 428           (funcall function p1 p2 p3 p4)
 429           (goto-char p1))))))
 430
 431 (defun html2text-substitute ()
 432   "See the variable `html2text-replace-list' for documentation."
 433   (interactive)
 434   (dolist (e html2text-replace-list)
 435     (goto-char (point-min))
 436     (let ((old-string (car e))
 437           (new-string (cdr e)))
 438       (html2text-replace-string old-string new-string (point-min) (point-max)))))
 439
 440 (defun html2text-format-single-elements ()
 441   (interactive)
 442   (dolist (tag-and-function html2text-format-single-element-list)
 443     (let ((tag      (car tag-and-function))
 444           (function (cdr tag-and-function)))
 445       (goto-char (point-min))
 446       (while (re-search-forward (format "\\(<%s\\( [^>]*\\)?>\\)" tag)
 447                                 (point-max) t)
 448         (let ((p1)
 449               (p2 (point)))
 450           (search-backward "<" (point-min) t)
 451           (setq p1 (point))
 452           (funcall function p1 p2))))))
 453
 454 ;;
 455 ;; Main function
 456 ;;
 457
 458 ;;;###autoload
 459 (defun html2text ()
 460   "Convert HTML to plain text in the current buffer."
 461   (interactive)
 462   (save-excursion
 463     (let ((case-fold-search t)
 464           (buffer-read-only))
 465       (html2text-remove-tags html2text-remove-tag-list)
 466       (html2text-format-tags)
 467       (html2text-remove-tags html2text-remove-tag-list2)
 468       (html2text-substitute)
 469       (html2text-format-single-elements)
 470       (html2text-fix-paragraphs))))
 471
 472 ;;
 473 ;; </Interactive functions>
 474 ;;
 475 (provide 'html2text)
 476 ;;; arch-tag: e9e57b79-35d4-4de1-a647-e7e01fe56d1e
 477 ;;; html2text.el ends here