lisp/gnus/html2text.el

   1 ;;; html2text.el --- a simple html to plain text converter -*- coding: utf-8 -*-
   2
   3 ;; Copyright (C) 2002-2014 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 of the License, or
  12 ;; (at your option) 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.  If not, see <http://www.gnu.org/licenses/>.
  21
  22 ;;; Commentary:
  23
  24 ;; These functions provide a simple way to wash/clean html infected
  25 ;; mails.  Definitely do not work in all cases, but some improvement
  26 ;; in readability is generally obtained.  Formatting is only done in
  27 ;; the buffer, so the next time you enter the article it will be
  28 ;; "re-htmlized".
  29 ;;
  30 ;; The main function is `html2text'.
  31
  32 ;;; Code:
  33
  34 ;;
  35 ;; <Global variables>
  36 ;;
  37
  38 (eval-when-compile
  39   (require 'cl))
  40
  41 (defvar html2text-format-single-element-list '(("hr" . html2text-clean-hr)))
  42
  43 (defvar html2text-replace-list
  44   '(("&acute;" . "`")
  45     ("&amp;" . "&")
  46     ("&apos;" . "'")
  47     ("&brvbar;" . "|")
  48     ("&cent;" . "c")
  49     ("&circ;" . "^")
  50     ("&copy;" . "(C)")
  51     ("&curren;" . "(#)")
  52     ("&deg;" . "degree")
  53     ("&divide;" . "/")
  54     ("&euro;" . "e")
  55     ("&frac12;" . "1/2")
  56     ("&gt;" . ">")
  57     ("&iquest;" . "?")
  58     ("&laquo;" . "<<")
  59     ("&ldquo" . "\"")
  60     ("&lsaquo;" . "(")
  61     ("&lsquo;" . "`")
  62     ("&lt;" . "<")
  63     ("&mdash;" . "--")
  64     ("&nbsp;" . " ")
  65     ("&ndash;" . "-")
  66     ("&permil;" . "%%")
  67     ("&plusmn;" . "+-")
  68     ("&pound;" . "£")
  69     ("&quot;" . "\"")
  70     ("&raquo;" . ">>")
  71     ("&rdquo" . "\"")
  72     ("&reg;" . "(R)")
  73     ("&rsaquo;" . ")")
  74     ("&rsquo;" . "'")
  75     ("&sect;" . "§")
  76     ("&sup1;" . "^1")
  77     ("&sup2;" . "^2")
  78     ("&sup3;" . "^3")
  79     ("&tilde;" . "~"))
  80   "The map of entity to text.
  81
  82 This is an alist were each element is a dotted pair consisting of an
  83 old string, and a replacement string.  This replacement is done by the
  84 function `html2text-substitute' which basically performs a
  85 `replace-string' operation for every element in the list.  This is
  86 completely verbatim - without any use of REGEXP.")
  87
  88 (defvar html2text-remove-tag-list
  89   '("html" "body" "p" "img" "dir" "head" "div" "br" "font" "title" "meta")
  90   "A list of removable tags.
  91
  92 This is a list of tags which should be removed, without any
  93 formatting.  Note that tags in the list are presented *without*
  94 any \"<\" or \">\".  All occurrences of a tag appearing in this
  95 list are removed, irrespective of whether it is a closing or
  96 opening tag, or if the tag has additional attributes.  The
  97 deletion is done by the function `html2text-remove-tags'.
  98
  99 For instance the text:
 100
 101 \"Here comes something <font size\"+3\" face=\"Helvetica\"> big </font>.\"
 102
 103 will be reduced to:
 104
 105 \"Here comes something big.\"
 106
 107 If this list contains the element \"font\".")
 108
 109 (defvar html2text-format-tag-list
 110   '(("b"          . html2text-clean-bold)
 111     ("strong"     . html2text-clean-bold)
 112     ("u"          . html2text-clean-underline)
 113     ("i"          . html2text-clean-italic)
 114     ("em"         . html2text-clean-italic)
 115     ("blockquote" . html2text-clean-blockquote)
 116     ("a"          . html2text-clean-anchor)
 117     ("ul"         . html2text-clean-ul)
 118     ("ol"         . html2text-clean-ol)
 119     ("dl"         . html2text-clean-dl)
 120     ("center"     . html2text-clean-center))
 121   "An alist of tags and processing functions.
 122
 123 This is an alist where each dotted pair consists of a tag, and then
 124 the name of a function to be called when this tag is found.  The
 125 function is called with the arguments p1, p2, p3 and p4. These are
 126 demonstrated below:
 127
 128 \"<b> This is bold text </b>\"
 129  ^   ^                 ^    ^
 130  |   |                 |    |
 131 p1  p2                p3   p4
 132
 133 Then the called function will typically format the text somewhat and
 134 remove the tags.")
 135
 136 (defvar html2text-remove-tag-list2  '("li" "dt" "dd" "meta")
 137   "Another list of removable tags.
 138
 139 This is a list of tags which are removed similarly to the list
 140 `html2text-remove-tag-list' - but these tags are retained for the
 141 formatting, and then moved afterward.")
 142
 143 ;;
 144 ;; </Global variables>
 145 ;;
 146
 147 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 148 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 149
 150 ;;
 151 ;; <Utility functions>
 152 ;;
 153
 154
 155 (defun html2text-replace-string (from-string to-string min max)
 156   "Replace FROM-STRING with TO-STRING in region from MIN to MAX."
 157   (goto-char min)
 158   (let ((delta (- (string-width to-string) (string-width from-string)))
 159         (change 0))
 160     (while (search-forward from-string max t)
 161       (replace-match to-string)
 162       (setq change (+ change delta)))
 163     change))
 164
 165 ;;
 166 ;; </Utility functions>
 167 ;;
 168
 169 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 170 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 171
 172 ;;
 173 ;; <Functions related to attributes> i.e. <font size=+3>
 174 ;;
 175
 176 (defun html2text-attr-value (list attribute)
 177   "Get value of ATTRIBUTE from LIST."
 178   (nth 1 (assoc attribute list)))
 179
 180 (defun html2text-get-attr (p1 p2)
 181   (goto-char p1)
 182   (re-search-forward " +[^ ]" p2 t)
 183   (let* ((attr-string (buffer-substring-no-properties (1- (point)) (1- p2)))
 184          (tmp-list (split-string attr-string))
 185          (attr-list)
 186          (counter 0)
 187          (prev (car tmp-list))
 188          (this (nth 1 tmp-list))
 189          (next (nth 2 tmp-list))
 190          (index 1))
 191
 192     (cond
 193      ;; size=3
 194      ((string-match "[^ ]=[^ ]" prev)
 195       (let ((attr  (nth 0 (split-string prev "=")))
 196             (value (substring prev (1+ (string-match "=" prev)))))
 197         (setq attr-list (cons (list attr value) attr-list))))
 198      ;; size= 3
 199      ((string-match "[^ ]=\\'" prev)
 200       (setq attr-list (cons (list (substring prev 0 -1) this) attr-list))))
 201
 202     (while (< index (length tmp-list))
 203       (cond
 204        ;; size=3
 205        ((string-match "[^ ]=[^ ]" this)
 206         (let ((attr  (nth 0 (split-string this "=")))
 207               (value (substring prev (1+ (string-match "=" this)))))
 208           (setq attr-list (cons (list attr value) attr-list))))
 209        ;; size =3
 210        ((string-match "\\`=[^ ]" this)
 211         (setq attr-list (cons (list prev (substring this 1)) attr-list)))
 212        ;; size= 3
 213        ((string-match "[^ ]=\\'" this)
 214         (setq attr-list (cons (list (substring this 0 -1) next) attr-list)))
 215        ;; size = 3
 216        ((string= "=" this)
 217         (setq attr-list (cons (list prev next) attr-list))))
 218       (setq index (1+ index))
 219       (setq prev this)
 220       (setq this next)
 221       (setq next (nth (1+ index) tmp-list)))
 222     ;;
 223     ;; Tags with no accompanying "=" i.e. value=nil
 224     ;;
 225     (setq prev (car tmp-list))
 226     (setq this (nth 1 tmp-list))
 227     (setq next (nth 2 tmp-list))
 228     (setq index 1)
 229
 230     (when (and (not (string-match "=" prev))
 231                (not (string= (substring this 0 1) "=")))
 232       (setq attr-list (cons (list prev nil) attr-list)))
 233     (while (< index (1- (length tmp-list)))
 234       (when (and (not (string-match "=" this))
 235                  (not (or (string= (substring next 0 1) "=")
 236                           (string= (substring prev -1) "="))))
 237         (setq attr-list (cons (list this nil) attr-list)))
 238       (setq index (1+ index))
 239       (setq prev this)
 240       (setq this next)
 241       (setq next (nth (1+ index) tmp-list)))
 242
 243     (when (and this
 244                (not (string-match "=" this))
 245                (not (string= (substring prev -1) "=")))
 246       (setq attr-list (cons (list this nil) attr-list)))
 247     ;; return - value
 248     attr-list))
 249
 250 ;;
 251 ;; </Functions related to attributes>
 252 ;;
 253
 254 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 255 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 256
 257 ;;
 258 ;; <Functions to be called to format a tag-pair>
 259 ;;
 260 (defun html2text-clean-list-items (p1 p2 list-type)
 261   (goto-char p1)
 262   (let ((item-nr 0)
 263         (items   0))
 264     (while (search-forward "<li>" p2 t)
 265       (setq items (1+ items)))
 266     (goto-char p1)
 267     (while (< item-nr items)
 268       (setq item-nr (1+ item-nr))
 269       (search-forward "<li>" (point-max) t)
 270       (cond
 271        ((string= list-type "ul") (insert " o "))
 272        ((string= list-type "ol") (insert (format " %s: " item-nr)))
 273        (t (insert " x "))))))
 274
 275 (defun html2text-clean-dtdd (p1 p2)
 276   (goto-char p1)
 277   (let ((items   0)
 278         (item-nr 0))
 279     (while (search-forward "<dt>" p2 t)
 280       (setq items (1+ items)))
 281     (goto-char p1)
 282     (while (< item-nr items)
 283       (setq item-nr (1+ item-nr))
 284       (re-search-forward "<dt>\\([ ]*\\)" (point-max) t)
 285       (when (match-string 1)
 286         (delete-region (point) (- (point) (string-width (match-string 1)))))
 287       (let ((def-p1 (point))
 288             (def-p2 0))
 289         (re-search-forward "\\([ ]*\\)\\(</dt>\\|<dd>\\)" (point-max) t)
 290         (if (match-string 1)
 291             (progn
 292               (let* ((mw1 (string-width (match-string 1)))
 293                      (mw2 (string-width (match-string 2)))
 294                      (mw  (+ mw1 mw2)))
 295                 (goto-char (- (point) mw))
 296                 (delete-region (point) (+ (point) mw1))
 297                 (setq def-p2 (point))))
 298           (setq def-p2 (- (point) (string-width (match-string 2)))))
 299         (put-text-property def-p1 def-p2 'face 'bold)))))
 300
 301 (defun html2text-delete-tags (p1 p2 p3 p4)
 302   (delete-region p1 p2)
 303   (delete-region (- p3 (- p2 p1)) (- p4 (- p2 p1))))
 304
 305 (defun html2text-delete-single-tag (p1 p2)
 306   (delete-region p1 p2))
 307
 308 (defun html2text-clean-hr (p1 p2)
 309   (html2text-delete-single-tag p1 p2)
 310   (goto-char p1)
 311   (newline 1)
 312   (insert (make-string fill-column ?-)))
 313
 314 (defun html2text-clean-ul (p1 p2 p3 p4)
 315   (html2text-delete-tags p1 p2 p3 p4)
 316   (html2text-clean-list-items p1 (- p3 (- p1 p2)) "ul"))
 317
 318 (defun html2text-clean-ol (p1 p2 p3 p4)
 319   (html2text-delete-tags p1 p2 p3 p4)
 320   (html2text-clean-list-items p1 (- p3 (- p1 p2)) "ol"))
 321
 322 (defun html2text-clean-dl (p1 p2 p3 p4)
 323   (html2text-delete-tags p1 p2 p3 p4)
 324   (html2text-clean-dtdd p1 (- p3 (- p1 p2))))
 325
 326 (defun html2text-clean-center (p1 p2 p3 p4)
 327   (html2text-delete-tags p1 p2 p3 p4)
 328   (center-region p1 (- p3 (- p2 p1))))
 329
 330 (defun html2text-clean-bold (p1 p2 p3 p4)
 331   (put-text-property p2 p3 'face 'bold)
 332   (html2text-delete-tags p1 p2 p3 p4))
 333
 334 (defun html2text-clean-title (p1 p2 p3 p4)
 335   (put-text-property p2 p3 'face 'bold)
 336   (html2text-delete-tags p1 p2 p3 p4))
 337
 338 (defun html2text-clean-underline (p1 p2 p3 p4)
 339   (put-text-property p2 p3 'face 'underline)
 340   (html2text-delete-tags p1 p2 p3 p4))
 341
 342 (defun html2text-clean-italic (p1 p2 p3 p4)
 343   (put-text-property p2 p3 'face 'italic)
 344   (html2text-delete-tags p1 p2 p3 p4))
 345
 346 (defun html2text-clean-font (p1 p2 p3 p4)
 347   (html2text-delete-tags p1 p2 p3 p4))
 348
 349 (defun html2text-clean-blockquote (p1 p2 p3 p4)
 350   (html2text-delete-tags p1 p2 p3 p4))
 351
 352 (defun html2text-clean-anchor (p1 p2 p3 p4)
 353   ;; If someone can explain how to make the URL clickable I will surely
 354   ;; improve upon this.
 355   ;; Maybe `goto-addr.el' can be used here.
 356   (let* ((attr-list (html2text-get-attr p1 p2))
 357          (href (html2text-attr-value attr-list "href")))
 358     (delete-region p1 p4)
 359     (when href
 360       (goto-char p1)
 361       (insert (if (string-match "\\`['\"].*['\"]\\'" href)
 362                   (substring href 1 -1) href))
 363       (put-text-property p1 (point) 'face 'bold))))
 364
 365 ;;
 366 ;; </Functions to be called to format a tag-pair>
 367 ;;
 368
 369 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 370 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 371
 372 ;;
 373 ;; <Functions to be called to fix up paragraphs>
 374 ;;
 375
 376 (defun html2text-fix-paragraph (p1 p2)
 377   (goto-char p1)
 378   (let ((refill-start)
 379         (refill-stop))
 380     (when (re-search-forward "<br>$" p2 t)
 381       (goto-char p1)
 382       (when (re-search-forward ".+[^<][^b][^r][^>]$" p2 t)
 383         (beginning-of-line)
 384         (setq refill-start (point))
 385         (goto-char p2)
 386         (re-search-backward ".+[^<][^b][^r][^>]$" refill-start t)
 387         (forward-line 1)
 388         (end-of-line)
 389         ;; refill-stop should ideally be adjusted to
 390         ;; accommodate the "<br>" strings which are removed
 391         ;; between refill-start and refill-stop.  Can simply
 392         ;; be returned from my-replace-string
 393         (setq refill-stop (+ (point)
 394                              (html2text-replace-string
 395                               "<br>" ""
 396                               refill-start (point))))
 397         ;; (message "Point = %s  refill-stop = %s" (point) refill-stop)
 398         ;; (sleep-for 4)
 399         (fill-region refill-start refill-stop))))
 400   (html2text-replace-string "<br>" "" p1 p2))
 401
 402 ;;
 403 ;; This one is interactive ...
 404 ;;
 405 (defun html2text-fix-paragraphs ()
 406   "This _tries_ to fix up the paragraphs - this is done in quite a ad-hook
 407 fashion, quite close to pure guess-work. It does work in some cases though."
 408   (interactive)
 409   (goto-char (point-min))
 410   (while (re-search-forward "^<br>$" nil t)
 411     (delete-region (match-beginning 0) (match-end 0)))
 412   ;; Removing lonely <br> on a single line, if they are left intact we
 413   ;; don't have any paragraphs at all.
 414   (goto-char (point-min))
 415   (while (not (eobp))
 416     (let ((p1 (point)))
 417       (forward-paragraph 1)
 418       ;;(message "Kaller fix med p1=%s  p2=%s " p1 (1- (point))) (sleep-for 5)
 419       (html2text-fix-paragraph p1 (1- (point)))
 420       (goto-char p1)
 421       (when (not (eobp))
 422         (forward-paragraph 1)))))
 423
 424 ;;
 425 ;; </Functions to be called to fix up paragraphs>
 426 ;;
 427
 428 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 429 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 430
 431 ;;
 432 ;; <Interactive functions>
 433 ;;
 434
 435 (defun html2text-remove-tags (tag-list)
 436   "Removes the tags listed in the list `html2text-remove-tag-list'.
 437 See the documentation for that variable."
 438   (interactive)
 439   (dolist (tag tag-list)
 440     (goto-char (point-min))
 441     (while (re-search-forward (format "\\(</?%s[^>]*>\\)" tag) (point-max) t)
 442       (delete-region (match-beginning 0) (match-end 0)))))
 443
 444 (defun html2text-format-tags ()
 445   "See the variable `html2text-format-tag-list' for documentation."
 446   (interactive)
 447   (dolist (tag-and-function html2text-format-tag-list)
 448     (let ((tag      (car tag-and-function))
 449           (function (cdr tag-and-function)))
 450       (goto-char (point-min))
 451       (while (re-search-forward (format "\\(<%s\\( [^>]*\\)?>\\)" tag)
 452                                 (point-max) t)
 453         (let ((p1)
 454               (p2 (point))
 455               (p3) (p4))
 456           (search-backward "<" (point-min) t)
 457           (setq p1 (point))
 458           (unless (search-forward (format "</%s>" tag) (point-max) t)
 459             (goto-char p2)
 460             (insert (format "</%s>" tag)))
 461           (setq p4 (point))
 462           (search-backward "</" (point-min) t)
 463           (setq p3 (point))
 464           (funcall function p1 p2 p3 p4)
 465           (goto-char p1))))))
 466
 467 (defun html2text-substitute ()
 468   "See the variable `html2text-replace-list' for documentation."
 469   (interactive)
 470   (dolist (e html2text-replace-list)
 471     (goto-char (point-min))
 472     (let ((old-string (car e))
 473           (new-string (cdr e)))
 474       (html2text-replace-string old-string new-string (point-min) (point-max)))))
 475
 476 (defun html2text-format-single-elements ()
 477   (interactive)
 478   (dolist (tag-and-function html2text-format-single-element-list)
 479     (let ((tag      (car tag-and-function))
 480           (function (cdr tag-and-function)))
 481       (goto-char (point-min))
 482       (while (re-search-forward (format "\\(<%s\\( [^>]*\\)?>\\)" tag)
 483                                 (point-max) t)
 484         (let ((p1)
 485               (p2 (point)))
 486           (search-backward "<" (point-min) t)
 487           (setq p1 (point))
 488           (funcall function p1 p2))))))
 489
 490 ;;
 491 ;; Main function
 492 ;;
 493
 494 ;;;###autoload
 495 (defun html2text ()
 496   "Convert HTML to plain text in the current buffer."
 497   (interactive)
 498   (save-excursion
 499     (let ((case-fold-search t)
 500           (buffer-read-only))
 501       (html2text-remove-tags html2text-remove-tag-list)
 502       (html2text-format-tags)
 503       (html2text-remove-tags html2text-remove-tag-list2)
 504       (html2text-substitute)
 505       (html2text-format-single-elements)
 506       (html2text-fix-paragraphs))))
 507
 508 ;;
 509 ;; </Interactive functions>
 510 ;;
 511 (provide 'html2text)
 512
 513 ;;; html2text.el ends here