lisp/gnus/html2text.el

   1 ;;; html2text.el --- a simple html to plain text converter
   2
   3 ;; Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
   4
   5 ;; Author: Joakim Hove <hove@phys.ntnu.no>
   6
   7 ;; This file is part of GNU Emacs.
   8
   9 ;; GNU Emacs is free software; you can redistribute it and/or modify
  10 ;; it under the terms of the GNU General Public License as published by
  11 ;; the Free Software Foundation; either version 3, or (at your option)
  12 ;; any later version.
  13
  14 ;; GNU Emacs is distributed in the hope that it will be useful,
  15 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  16 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17 ;; GNU General Public License for more details.
  18
  19 ;; You should have received a copy of the GNU General Public License
  20 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  21 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  22 ;; Boston, MA 02110-1301, USA.
  23
  24 ;;; Commentary:
  25
  26 ;; These functions provide a simple way to wash/clean html infected
  27 ;; mails.  Definitely do not work in all cases, but some improvement
  28 ;; in readability is generally obtained.  Formatting is only done in
  29 ;; the buffer, so the next time you enter the article it will be
  30 ;; "re-htmlized".
  31 ;;
  32 ;; The main function is `html2text'.
  33
  34 ;;; Code:
  35
  36 ;;
  37 ;; <Global variables>
  38 ;;
  39
  40 (eval-when-compile
  41   (require 'cl))
  42
  43 (defvar html2text-format-single-element-list '(("hr" . html2text-clean-hr)))
  44
  45 (defvar html2text-replace-list
  46   '(("&acute;" . "`")
  47     ("&amp;" . "&")
  48     ("&apos;" . "'")
  49     ("&brvbar;" . "|")
  50     ("&cent;" . "c")
  51     ("&circ;" . "^")
  52     ("&copy;" . "(C)")
  53     ("&curren;" . "(#)")
  54     ("&deg;" . "degree")
  55     ("&divide;" . "/")
  56     ("&euro;" . "e")
  57     ("&frac12;" . "1/2")
  58     ("&gt;" . ">")
  59     ("&iquest;" . "?")
  60     ("&laquo;" . "<<")
  61     ("&ldquo" . "\"")
  62     ("&lsaquo;" . "(")
  63     ("&lsquo;" . "`")
  64     ("&lt;" . "<")
  65     ("&mdash;" . "--")
  66     ("&nbsp;" . " ")
  67     ("&ndash;" . "-")
  68     ("&permil;" . "%%")
  69     ("&plusmn;" . "+-")
  70     ("&pound;" . "£")
  71     ("&quot;" . "\"")
  72     ("&raquo;" . ">>")
  73     ("&rdquo" . "\"")
  74     ("&reg;" . "(R)")
  75     ("&rsaquo;" . ")")
  76     ("&rsquo;" . "'")
  77     ("&sect;" . "§")
  78     ("&sup1;" . "^1")
  79     ("&sup2;" . "^2")
  80     ("&sup3;" . "^3")
  81     ("&tilde;" . "~"))
  82   "The map of entity to text.
  83
  84 This is an alist were each element is a dotted pair consisting of an
  85 old string, and a replacement string.  This replacement is done by the
  86 function `html2text-substitute' which basically performs a
  87 `replace-string' operation for every element in the list.  This is
  88 completely verbatim - without any use of REGEXP.")
  89
  90 (defvar html2text-remove-tag-list
  91   '("html" "body" "p" "img" "dir" "head" "div" "br" "font" "title" "meta")
  92   "A list of removable tags.
  93
  94 This is a list of tags which should be removed, without any
  95 formatting.  Note that tags in the list are presented *without*
  96 any \"<\" or \">\".  All occurrences of a tag appearing in this
  97 list are removed, irrespective of whether it is a closing or
  98 opening tag, or if the tag has additional attributes.  The
  99 deletion is done by the function `html2text-remove-tags'.
 100
 101 For instance the text:
 102
 103 \"Here comes something <font size\"+3\" face=\"Helvetica\"> big </font>.\"
 104
 105 will be reduced to:
 106
 107 \"Here comes something big.\"
 108
 109 If this list contains the element \"font\".")
 110
 111 (defvar html2text-format-tag-list
 112   '(("b"          . html2text-clean-bold)
 113     ("strong"     . html2text-clean-bold)
 114     ("u"          . html2text-clean-underline)
 115     ("i"          . html2text-clean-italic)
 116     ("em"         . html2text-clean-italic)
 117     ("blockquote" . html2text-clean-blockquote)
 118     ("a"          . html2text-clean-anchor)
 119     ("ul"         . html2text-clean-ul)
 120     ("ol"         . html2text-clean-ol)
 121     ("dl"         . html2text-clean-dl)
 122     ("center"     . html2text-clean-center))
 123   "An alist of tags and processing functions.
 124
 125 This is an alist where each dotted pair consists of a tag, and then
 126 the name of a function to be called when this tag is found.  The
 127 function is called with the arguments p1, p2, p3 and p4. These are
 128 demontrated below:
 129
 130 \"<b> This is bold text </b>\"
 131  ^   ^                 ^    ^
 132  |   |                 |    |
 133 p1  p2                p3   p4
 134
 135 Then the called function will typically format the text somewhat and
 136 remove the tags.")
 137
 138 (defvar html2text-remove-tag-list2  '("li" "dt" "dd" "meta")
 139   "Another list of removable tags.
 140
 141 This is a list of tags which are removed similarly to the list
 142 `html2text-remove-tag-list' - but these tags are retained for the
 143 formatting, and then moved afterward.")
 144
 145 ;;
 146 ;; </Global variables>
 147 ;;
 148
 149 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 150 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 151
 152 ;;
 153 ;; <Utility functions>
 154 ;;
 155
 156
 157 (defun html2text-replace-string (from-string to-string min max)
 158   "Replace FROM-STRING with TO-STRING in region from MIN to MAX."
 159   (goto-char min)
 160   (let ((delta (- (string-width to-string) (string-width from-string)))
 161         (change 0))
 162     (while (search-forward from-string max t)
 163       (replace-match to-string)
 164       (setq change (+ change delta)))
 165     change))
 166
 167 ;;
 168 ;; </Utility functions>
 169 ;;
 170
 171 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 172 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 173
 174 ;;
 175 ;; <Functions related to attributes> i.e. <font size=+3>
 176 ;;
 177
 178 (defun html2text-attr-value (list attribute)
 179   "Get value of ATTRIBUTE from LIST."
 180   (nth 1 (assoc attribute list)))
 181
 182 (defun html2text-get-attr (p1 p2)
 183   (goto-char p1)
 184   (re-search-forward " +[^ ]" p2 t)
 185   (let* ((attr-string (buffer-substring-no-properties (1- (point)) (1- p2)))
 186          (tmp-list (split-string attr-string))
 187          (attr-list)
 188          (counter 0)
 189          (prev (car tmp-list))
 190          (this (nth 1 tmp-list))
 191          (next (nth 2 tmp-list))
 192          (index 1))
 193
 194     (cond
 195      ;; size=3
 196      ((string-match "[^ ]=[^ ]" prev)
 197       (let ((attr  (nth 0 (split-string prev "=")))
 198             (value (nth 1 (split-string prev "="))))
 199         (setq attr-list (cons (list attr value) attr-list))))
 200      ;; size= 3
 201      ((string-match "[^ ]=\\'" prev)
 202       (setq attr-list (cons (list (substring prev 0 -1) this) attr-list))))
 203
 204     (while (< index (length tmp-list))
 205       (cond
 206        ;; size=3
 207        ((string-match "[^ ]=[^ ]" this)
 208         (let ((attr  (nth 0 (split-string this "=")))
 209               (value (nth 1 (split-string this "="))))
 210           (setq attr-list (cons (list attr value) attr-list))))
 211        ;; size =3
 212        ((string-match "\\`=[^ ]" this)
 213         (setq attr-list (cons (list prev (substring this 1)) attr-list)))
 214        ;; size= 3
 215        ((string-match "[^ ]=\\'" this)
 216         (setq attr-list (cons (list (substring this 0 -1) next) attr-list)))
 217        ;; size = 3
 218        ((string= "=" this)
 219         (setq attr-list (cons (list prev next) attr-list))))
 220       (setq index (1+ index))
 221       (setq prev this)
 222       (setq this next)
 223       (setq next (nth (1+ index) tmp-list)))
 224     ;;
 225     ;; Tags with no accompanying "=" i.e. value=nil
 226     ;;
 227     (setq prev (car tmp-list))
 228     (setq this (nth 1 tmp-list))
 229     (setq next (nth 2 tmp-list))
 230     (setq index 1)
 231
 232     (when (and (not (string-match "=" prev))
 233                (not (string= (substring this 0 1) "=")))
 234       (setq attr-list (cons (list prev nil) attr-list)))
 235     (while (< index (1- (length tmp-list)))
 236       (when (and (not (string-match "=" this))
 237                  (not (or (string= (substring next 0 1) "=")
 238                           (string= (substring prev -1) "="))))
 239         (setq attr-list (cons (list this nil) attr-list)))
 240       (setq index (1+ index))
 241       (setq prev this)
 242       (setq this next)
 243       (setq next (nth (1+ index) tmp-list)))
 244
 245     (when (and this
 246                (not (string-match "=" this))
 247                (not (string= (substring prev -1) "=")))
 248       (setq attr-list (cons (list this nil) attr-list)))
 249     ;; return - value
 250     attr-list))
 251
 252 ;;
 253 ;; </Functions related to attributes>
 254 ;;
 255
 256 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 257 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 258
 259 ;;
 260 ;; <Functions to be called to format a tag-pair>
 261 ;;
 262 (defun html2text-clean-list-items (p1 p2 list-type)
 263   (goto-char p1)
 264   (let ((item-nr 0)
 265         (items   0))
 266     (while (search-forward "<li>" p2 t)
 267       (setq items (1+ items)))
 268     (goto-char p1)
 269     (while (< item-nr items)
 270       (setq item-nr (1+ item-nr))
 271       (search-forward "<li>" (point-max) t)
 272       (cond
 273        ((string= list-type "ul") (insert " o "))
 274        ((string= list-type "ol") (insert (format " %s: " item-nr)))
 275        (t (insert " x "))))))
 276
 277 (defun html2text-clean-dtdd (p1 p2)
 278   (goto-char p1)
 279   (let ((items   0)
 280         (item-nr 0))
 281     (while (search-forward "<dt>" p2 t)
 282       (setq items (1+ items)))
 283     (goto-char p1)
 284     (while (< item-nr items)
 285       (setq item-nr (1+ item-nr))
 286       (re-search-forward "<dt>\\([ ]*\\)" (point-max) t)
 287       (when (match-string 1)
 288         (delete-region (point) (- (point) (string-width (match-string 1)))))
 289       (let ((def-p1 (point))
 290             (def-p2 0))
 291         (re-search-forward "\\([ ]*\\)\\(</dt>\\|<dd>\\)" (point-max) t)
 292         (if (match-string 1)
 293             (progn
 294               (let* ((mw1 (string-width (match-string 1)))
 295                      (mw2 (string-width (match-string 2)))
 296                      (mw  (+ mw1 mw2)))
 297                 (goto-char (- (point) mw))
 298                 (delete-region (point) (+ (point) mw1))
 299                 (setq def-p2 (point))))
 300           (setq def-p2 (- (point) (string-width (match-string 2)))))
 301         (put-text-property def-p1 def-p2 'face 'bold)))))
 302
 303 (defun html2text-delete-tags (p1 p2 p3 p4)
 304   (delete-region p1 p2)
 305   (delete-region (- p3 (- p2 p1)) (- p4 (- p2 p1))))
 306
 307 (defun html2text-delete-single-tag (p1 p2)
 308   (delete-region p1 p2))
 309
 310 (defun html2text-clean-hr (p1 p2)
 311   (html2text-delete-single-tag p1 p2)
 312   (goto-char p1)
 313   (newline 1)
 314   (insert (make-string fill-column ?-)))
 315
 316 (defun html2text-clean-ul (p1 p2 p3 p4)
 317   (html2text-delete-tags p1 p2 p3 p4)
 318   (html2text-clean-list-items p1 (- p3 (- p1 p2)) "ul"))
 319
 320 (defun html2text-clean-ol (p1 p2 p3 p4)
 321   (html2text-delete-tags p1 p2 p3 p4)
 322   (html2text-clean-list-items p1 (- p3 (- p1 p2)) "ol"))
 323
 324 (defun html2text-clean-dl (p1 p2 p3 p4)
 325   (html2text-delete-tags p1 p2 p3 p4)
 326   (html2text-clean-dtdd p1 (- p3 (- p1 p2))))
 327
 328 (defun html2text-clean-center (p1 p2 p3 p4)
 329   (html2text-delete-tags p1 p2 p3 p4)
 330   (center-region p1 (- p3 (- p2 p1))))
 331
 332 (defun html2text-clean-bold (p1 p2 p3 p4)
 333   (put-text-property p2 p3 'face 'bold)
 334   (html2text-delete-tags p1 p2 p3 p4))
 335
 336 (defun html2text-clean-title (p1 p2 p3 p4)
 337   (put-text-property p2 p3 'face 'bold)
 338   (html2text-delete-tags p1 p2 p3 p4))
 339
 340 (defun html2text-clean-underline (p1 p2 p3 p4)
 341   (put-text-property p2 p3 'face 'underline)
 342   (html2text-delete-tags p1 p2 p3 p4))
 343
 344 (defun html2text-clean-italic (p1 p2 p3 p4)
 345   (put-text-property p2 p3 'face 'italic)
 346   (html2text-delete-tags p1 p2 p3 p4))
 347
 348 (defun html2text-clean-font (p1 p2 p3 p4)
 349   (html2text-delete-tags p1 p2 p3 p4))
 350
 351 (defun html2text-clean-blockquote (p1 p2 p3 p4)
 352   (html2text-delete-tags p1 p2 p3 p4))
 353
 354 (defun html2text-clean-anchor (p1 p2 p3 p4)
 355   ;; If someone can explain how to make the URL clickable I will surely
 356   ;; improve upon this.
 357   ;; Maybe `goto-addr.el' can be used here.
 358   (let* ((attr-list (html2text-get-attr p1 p2))
 359          (href (html2text-attr-value attr-list "href")))
 360     (delete-region p1 p4)
 361     (when href
 362       (goto-char p1)
 363       (insert (substring href 1 -1 ))
 364       (put-text-property p1 (point) 'face 'bold))))
 365
 366 ;;
 367 ;; </Functions to be called to format a tag-pair>
 368 ;;
 369
 370 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 371 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 372
 373 ;;
 374 ;; <Functions to be called to fix up paragraphs>
 375 ;;
 376
 377 (defun html2text-fix-paragraph (p1 p2)
 378   (goto-char p1)
 379   (let ((refill-start)
 380         (refill-stop))
 381     (when (re-search-forward "<br>$" p2 t)
 382       (goto-char p1)
 383       (when (re-search-forward ".+[^<][^b][^r][^>]$" p2 t)
 384         (beginning-of-line)
 385         (setq refill-start (point))
 386         (goto-char p2)
 387         (re-search-backward ".+[^<][^b][^r][^>]$" refill-start t)
 388         (forward-line 1)
 389         (end-of-line)
 390         ;; refill-stop should ideally be adjusted to
 391         ;; accomodate the "<br>" strings which are removed
 392         ;; between refill-start and refill-stop.  Can simply
 393         ;; be returned from my-replace-string
 394         (setq refill-stop (+ (point)
 395                              (html2text-replace-string
 396                               "<br>" ""
 397                               refill-start (point))))
 398         ;; (message "Point = %s  refill-stop = %s" (point) refill-stop)
 399         ;; (sleep-for 4)
 400         (fill-region refill-start refill-stop))))
 401   (html2text-replace-string "<br>" "" p1 p2))
 402
 403 ;;
 404 ;; This one is interactive ...
 405 ;;
 406 (defun html2text-fix-paragraphs ()
 407   "This _tries_ to fix up the paragraphs - this is done in quite a ad-hook
 408 fashion, quite close to pure guess-work. It does work in some cases though."
 409   (interactive)
 410   (goto-char (point-min))
 411   (while (re-search-forward "^<br>$" nil t)
 412     (delete-region (match-beginning 0) (match-end 0)))
 413   ;; Removing lonely <br> on a single line, if they are left intact we
 414   ;; dont have any paragraphs at all.
 415   (goto-char (point-min))
 416   (while (not (eobp))
 417     (let ((p1 (point)))
 418       (forward-paragraph 1)
 419       ;;(message "Kaller fix med p1=%s  p2=%s " p1 (1- (point))) (sleep-for 5)
 420       (html2text-fix-paragraph p1 (1- (point)))
 421       (goto-char p1)
 422       (when (not (eobp))
 423         (forward-paragraph 1)))))
 424
 425 ;;
 426 ;; </Functions to be called to fix up paragraphs>
 427 ;;
 428
 429 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 430 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 431
 432 ;;
 433 ;; <Interactive functions>
 434 ;;
 435
 436 (defun html2text-remove-tags (tag-list)
 437   "Removes the tags listed in the list `html2text-remove-tag-list'.
 438 See the documentation for that variable."
 439   (interactive)
 440   (dolist (tag tag-list)
 441     (goto-char (point-min))
 442     (while (re-search-forward (format "\\(</?%s[^>]*>\\)" tag) (point-max) t)
 443       (delete-region (match-beginning 0) (match-end 0)))))
 444
 445 (defun html2text-format-tags ()
 446   "See the variable `html2text-format-tag-list' for documentation."
 447   (interactive)
 448   (dolist (tag-and-function html2text-format-tag-list)
 449     (let ((tag      (car tag-and-function))
 450           (function (cdr tag-and-function)))
 451       (goto-char (point-min))
 452       (while (re-search-forward (format "\\(<%s\\( [^>]*\\)?>\\)" tag)
 453                                 (point-max) t)
 454         (let ((p1)
 455               (p2 (point))
 456               (p3) (p4))
 457           (search-backward "<" (point-min) t)
 458           (setq p1 (point))
 459           (unless (search-forward (format "</%s>" tag) (point-max) t)
 460             (goto-char p2)
 461             (insert (format "</%s>" tag)))
 462           (setq p4 (point))
 463           (search-backward "</" (point-min) t)
 464           (setq p3 (point))
 465           (funcall function p1 p2 p3 p4)
 466           (goto-char p1))))))
 467
 468 (defun html2text-substitute ()
 469   "See the variable `html2text-replace-list' for documentation."
 470   (interactive)
 471   (dolist (e html2text-replace-list)
 472     (goto-char (point-min))
 473     (let ((old-string (car e))
 474           (new-string (cdr e)))
 475       (html2text-replace-string old-string new-string (point-min) (point-max)))))
 476
 477 (defun html2text-format-single-elements ()
 478   (interactive)
 479   (dolist (tag-and-function html2text-format-single-element-list)
 480     (let ((tag      (car tag-and-function))
 481           (function (cdr tag-and-function)))
 482       (goto-char (point-min))
 483       (while (re-search-forward (format "\\(<%s\\( [^>]*\\)?>\\)" tag)
 484                                 (point-max) t)
 485         (let ((p1)
 486               (p2 (point)))
 487           (search-backward "<" (point-min) t)
 488           (setq p1 (point))
 489           (funcall function p1 p2))))))
 490
 491 ;;
 492 ;; Main function
 493 ;;
 494
 495 ;;;###autoload
 496 (defun html2text ()
 497   "Convert HTML to plain text in the current buffer."
 498   (interactive)
 499   (save-excursion
 500     (let ((case-fold-search t)
 501           (buffer-read-only))
 502       (html2text-remove-tags html2text-remove-tag-list)
 503       (html2text-format-tags)
 504       (html2text-remove-tags html2text-remove-tag-list2)
 505       (html2text-substitute)
 506       (html2text-format-single-elements)
 507       (html2text-fix-paragraphs))))
 508
 509 ;;
 510 ;; </Interactive functions>
 511 ;;
 512 (provide 'html2text)
 513 ;; arch-tag: e9e57b79-35d4-4de1-a647-e7e01fe56d1e
 514 ;;; html2text.el ends here