lisp/obsolete/html2text.el

   1 ;;; html2text.el --- a simple html to plain text converter -*- coding: utf-8 -*-
   2
   3 ;; Copyright (C) 2002-2018 Free Software Foundation, Inc.
   4
   5 ;; Author: Joakim Hove <hove@phys.ntnu.no>
   6 ;; Obsolete-since: 26.1
   7
   8 ;; This file is part of GNU Emacs.
   9
  10 ;; GNU Emacs is free software: you can redistribute it and/or modify
  11 ;; it under the terms of the GNU General Public License as published by
  12 ;; the Free Software Foundation, either version 3 of the License, or
  13 ;; (at your option) any later version.
  14
  15 ;; GNU Emacs is distributed in the hope that it will be useful,
  16 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  18 ;; GNU General Public License for more details.
  19
  20 ;; You should have received a copy of the GNU General Public License
  21 ;; along with GNU Emacs.  If not, see <https://www.gnu.org/licenses/>.
  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 ;; This package was obsoleted by shr.el.
  34
  35 ;;; Code:
  36
  37 ;;
  38 ;; <Global variables>
  39 ;;
  40
  41
  42 (defvar html2text-format-single-element-list '(("hr" . html2text-clean-hr)))
  43
  44 (defvar html2text-replace-list
  45   '(("&acute;" . "`")
  46     ("&amp;" . "&")
  47     ("&apos;" . "'")
  48     ("&brvbar;" . "|")
  49     ("&cent;" . "c")
  50     ("&circ;" . "^")
  51     ("&copy;" . "(C)")
  52     ("&curren;" . "(#)")
  53     ("&deg;" . "degree")
  54     ("&divide;" . "/")
  55     ("&euro;" . "e")
  56     ("&frac12;" . "1/2")
  57     ("&gt;" . ">")
  58     ("&iquest;" . "?")
  59     ("&laquo;" . "<<")
  60     ("&ldquo" . "\"")
  61     ("&lsaquo;" . "(")
  62     ("&lsquo;" . "`")
  63     ("&lt;" . "<")
  64     ("&mdash;" . "--")
  65     ("&nbsp;" . " ")
  66     ("&ndash;" . "-")
  67     ("&permil;" . "%%")
  68     ("&plusmn;" . "+-")
  69     ("&pound;" . "£")
  70     ("&quot;" . "\"")
  71     ("&raquo;" . ">>")
  72     ("&rdquo" . "\"")
  73     ("&reg;" . "(R)")
  74     ("&rsaquo;" . ")")
  75     ("&rsquo;" . "'")
  76     ("&sect;" . "§")
  77     ("&sup1;" . "^1")
  78     ("&sup2;" . "^2")
  79     ("&sup3;" . "^3")
  80     ("&tilde;" . "~"))
  81   "The map of entity to text.
  82
  83 This is an alist were each element is a dotted pair consisting of an
  84 old string, and a replacement string.  This replacement is done by the
  85 function `html2text-substitute' which basically performs a
  86 `replace-string' operation for every element in the list.  This is
  87 completely verbatim - without any use of REGEXP.")
  88
  89 (defvar html2text-remove-tag-list
  90   '("html" "body" "p" "img" "dir" "head" "div" "br" "font" "title" "meta")
  91   "A list of removable tags.
  92
  93 This is a list of tags which should be removed, without any
  94 formatting.  Note that tags in the list are presented *without*
  95 any \"<\" or \">\".  All occurrences of a tag appearing in this
  96 list are removed, irrespective of whether it is a closing or
  97 opening tag, or if the tag has additional attributes.  The
  98 deletion is done by the function `html2text-remove-tags'.
  99
 100 For instance the text:
 101
 102 \"Here comes something <font size\"+3\" face=\"Helvetica\"> big </font>.\"
 103
 104 will be reduced to:
 105
 106 \"Here comes something big.\"
 107
 108 If this list contains the element \"font\".")
 109
 110 (defvar html2text-format-tag-list
 111   '(("b"          . html2text-clean-bold)
 112     ("strong"     . html2text-clean-bold)
 113     ("u"          . html2text-clean-underline)
 114     ("i"          . html2text-clean-italic)
 115     ("em"         . html2text-clean-italic)
 116     ("blockquote" . html2text-clean-blockquote)
 117     ("a"          . html2text-clean-anchor)
 118     ("ul"         . html2text-clean-ul)
 119     ("ol"         . html2text-clean-ol)
 120     ("dl"         . html2text-clean-dl)
 121     ("center"     . html2text-clean-center))
 122   "An alist of tags and processing functions.
 123
 124 This is an alist where each dotted pair consists of a tag, and then
 125 the name of a function to be called when this tag is found.  The
 126 function is called with the arguments p1, p2, p3 and p4. These are
 127 demonstrated below:
 128
 129 \"<b> This is bold text </b>\"
 130  ^   ^                 ^    ^
 131  |   |                 |    |
 132 p1  p2                p3   p4
 133
 134 Then the called function will typically format the text somewhat and
 135 remove the tags.")
 136
 137 (defvar html2text-remove-tag-list2  '("li" "dt" "dd" "meta")
 138   "Another list of removable tags.
 139
 140 This is a list of tags which are removed similarly to the list
 141 `html2text-remove-tag-list' - but these tags are retained for the
 142 formatting, and then moved afterward.")
 143
 144 ;;
 145 ;; </Global variables>
 146 ;;
 147
 148 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 149 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 150
 151 ;;
 152 ;; <Utility functions>
 153 ;;
 154
 155
 156 (defun html2text-replace-string (from-string to-string min max)
 157   "Replace FROM-STRING with TO-STRING in region from MIN to MAX."
 158   (goto-char min)
 159   (let ((delta (- (string-width to-string) (string-width from-string)))
 160         (change 0))
 161     (while (search-forward from-string max t)
 162       (replace-match to-string)
 163       (setq change (+ change delta)))
 164     change))
 165
 166 ;;
 167 ;; </Utility functions>
 168 ;;
 169
 170 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 171 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 172
 173 ;;
 174 ;; <Functions related to attributes> i.e. <font size=+3>
 175 ;;
 176
 177 (defun html2text-attr-value (list attribute)
 178   "Get value of ATTRIBUTE from LIST."
 179   (nth 1 (assoc attribute list)))
 180
 181 (defun html2text-get-attr (p1 p2)
 182   (goto-char p1)
 183   (re-search-forward "\\s-+" p2 t)
 184   (let (attr-list)
 185     (while (re-search-forward "[-a-z0-9._]+" p2 t)
 186       (setq attr-list
 187             (cons
 188              (list (match-string 0)
 189                    (when (looking-at "\\s-*=")
 190                      (goto-char (match-end 0))
 191                      (skip-chars-forward "[:space:]")
 192                      (when (or (looking-at "\"[^\"]*\"\\|'[^']*'")
 193                                (looking-at "[-a-z0-9._:]+"))
 194                        (goto-char (match-end 0))
 195                        (match-string 0))))
 196              attr-list)))
 197     attr-list))
 198
 199 ;;
 200 ;; </Functions related to attributes>
 201 ;;
 202
 203 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 204 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 205
 206 ;;
 207 ;; <Functions to be called to format a tag-pair>
 208 ;;
 209 (defun html2text-clean-list-items (p1 p2 list-type)
 210   (goto-char p1)
 211   (let ((item-nr 0)
 212         (items   0))
 213     (while (search-forward "<li>" p2 t)
 214       (setq items (1+ items)))
 215     (goto-char p1)
 216     (while (< item-nr items)
 217       (setq item-nr (1+ item-nr))
 218       (search-forward "<li>" (point-max) t)
 219       (cond
 220        ((string= list-type "ul") (insert " o "))
 221        ((string= list-type "ol") (insert (format " %s: " item-nr)))
 222        (t (insert " x "))))))
 223
 224 (defun html2text-clean-dtdd (p1 p2)
 225   (goto-char p1)
 226   (let ((items   0)
 227         (item-nr 0))
 228     (while (search-forward "<dt>" p2 t)
 229       (setq items (1+ items)))
 230     (goto-char p1)
 231     (while (< item-nr items)
 232       (setq item-nr (1+ item-nr))
 233       (re-search-forward "<dt>\\([ ]*\\)" (point-max) t)
 234       (when (match-string 1)
 235         (delete-region (point) (- (point) (string-width (match-string 1)))))
 236       (let ((def-p1 (point))
 237             (def-p2 0))
 238         (re-search-forward "\\([ ]*\\)\\(</dt>\\|<dd>\\)" (point-max) t)
 239         (if (match-string 1)
 240             (progn
 241               (let* ((mw1 (string-width (match-string 1)))
 242                      (mw2 (string-width (match-string 2)))
 243                      (mw  (+ mw1 mw2)))
 244                 (goto-char (- (point) mw))
 245                 (delete-region (point) (+ (point) mw1))
 246                 (setq def-p2 (point))))
 247           (setq def-p2 (- (point) (string-width (match-string 2)))))
 248         (put-text-property def-p1 def-p2 'face 'bold)))))
 249
 250 (defun html2text-delete-tags (p1 p2 p3 p4)
 251   (delete-region p1 p2)
 252   (delete-region (- p3 (- p2 p1)) (- p4 (- p2 p1))))
 253
 254 (defun html2text-delete-single-tag (p1 p2)
 255   (delete-region p1 p2))
 256
 257 (defun html2text-clean-hr (p1 p2)
 258   (html2text-delete-single-tag p1 p2)
 259   (goto-char p1)
 260   (newline 1)
 261   (insert (make-string fill-column ?-)))
 262
 263 (defun html2text-clean-ul (p1 p2 p3 p4)
 264   (html2text-delete-tags p1 p2 p3 p4)
 265   (html2text-clean-list-items p1 (- p3 (- p1 p2)) "ul"))
 266
 267 (defun html2text-clean-ol (p1 p2 p3 p4)
 268   (html2text-delete-tags p1 p2 p3 p4)
 269   (html2text-clean-list-items p1 (- p3 (- p1 p2)) "ol"))
 270
 271 (defun html2text-clean-dl (p1 p2 p3 p4)
 272   (html2text-delete-tags p1 p2 p3 p4)
 273   (html2text-clean-dtdd p1 (- p3 (- p1 p2))))
 274
 275 (defun html2text-clean-center (p1 p2 p3 p4)
 276   (html2text-delete-tags p1 p2 p3 p4)
 277   (center-region p1 (- p3 (- p2 p1))))
 278
 279 (defun html2text-clean-bold (p1 p2 p3 p4)
 280   (put-text-property p2 p3 'face 'bold)
 281   (html2text-delete-tags p1 p2 p3 p4))
 282
 283 (defun html2text-clean-title (p1 p2 p3 p4)
 284   (put-text-property p2 p3 'face 'bold)
 285   (html2text-delete-tags p1 p2 p3 p4))
 286
 287 (defun html2text-clean-underline (p1 p2 p3 p4)
 288   (put-text-property p2 p3 'face 'underline)
 289   (html2text-delete-tags p1 p2 p3 p4))
 290
 291 (defun html2text-clean-italic (p1 p2 p3 p4)
 292   (put-text-property p2 p3 'face 'italic)
 293   (html2text-delete-tags p1 p2 p3 p4))
 294
 295 (defun html2text-clean-font (p1 p2 p3 p4)
 296   (html2text-delete-tags p1 p2 p3 p4))
 297
 298 (defun html2text-clean-blockquote (p1 p2 p3 p4)
 299   (html2text-delete-tags p1 p2 p3 p4))
 300
 301 (defun html2text-clean-anchor (p1 p2 _p3 p4)
 302   ;; If someone can explain how to make the URL clickable I will surely
 303   ;; improve upon this.
 304   ;; Maybe `goto-addr.el' can be used here.
 305   (let* ((attr-list (html2text-get-attr p1 p2))
 306          (href (html2text-attr-value attr-list "href")))
 307     (delete-region p1 p4)
 308     (when href
 309       (goto-char p1)
 310       (insert (if (string-match "\\`['\"].*['\"]\\'" href)
 311                   (substring href 1 -1) href))
 312       (put-text-property p1 (point) 'face 'bold))))
 313
 314 ;;
 315 ;; </Functions to be called to format a tag-pair>
 316 ;;
 317
 318 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 319 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 320
 321 ;;
 322 ;; <Functions to be called to fix up paragraphs>
 323 ;;
 324
 325 (defun html2text-fix-paragraph (p1 p2)
 326   (goto-char p1)
 327   (let ((refill-start)
 328         (refill-stop))
 329     (when (re-search-forward "<br>$" p2 t)
 330       (goto-char p1)
 331       (when (re-search-forward ".+[^<][^b][^r][^>]$" p2 t)
 332         (beginning-of-line)
 333         (setq refill-start (point))
 334         (goto-char p2)
 335         (re-search-backward ".+[^<][^b][^r][^>]$" refill-start t)
 336         (forward-line 1)
 337         (end-of-line)
 338         ;; refill-stop should ideally be adjusted to
 339         ;; accommodate the "<br>" strings which are removed
 340         ;; between refill-start and refill-stop.  Can simply
 341         ;; be returned from my-replace-string
 342         (setq refill-stop (+ (point)
 343                              (html2text-replace-string
 344                               "<br>" ""
 345                               refill-start (point))))
 346         ;; (message "Point = %s  refill-stop = %s" (point) refill-stop)
 347         ;; (sleep-for 4)
 348         (fill-region refill-start refill-stop))))
 349   (html2text-replace-string "<br>" "" p1 p2))
 350
 351 ;;
 352 ;; This one is interactive ...
 353 ;;
 354 (defun html2text-fix-paragraphs ()
 355   "This _tries_ to fix up the paragraphs - this is done in quite a ad-hook
 356 fashion, quite close to pure guess-work. It does work in some cases though."
 357   (interactive)
 358   (goto-char (point-min))
 359   (while (re-search-forward "^<br>$" nil t)
 360     (delete-region (match-beginning 0) (match-end 0)))
 361   ;; Removing lonely <br> on a single line, if they are left intact we
 362   ;; don't have any paragraphs at all.
 363   (goto-char (point-min))
 364   (while (not (eobp))
 365     (let ((p1 (point)))
 366       (forward-paragraph 1)
 367       ;;(message "Kaller fix med p1=%s  p2=%s " p1 (1- (point))) (sleep-for 5)
 368       (html2text-fix-paragraph p1 (1- (point)))
 369       (goto-char p1)
 370       (when (not (eobp))
 371         (forward-paragraph 1)))))
 372
 373 ;;
 374 ;; </Functions to be called to fix up paragraphs>
 375 ;;
 376
 377 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 378 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
 379
 380 ;;
 381 ;; <Interactive functions>
 382 ;;
 383
 384 (defun html2text-remove-tags (tag-list)
 385   "Removes the tags listed in the list `html2text-remove-tag-list'.
 386 See the documentation for that variable."
 387   (interactive)
 388   (dolist (tag tag-list)
 389     (goto-char (point-min))
 390     (while (re-search-forward (format "\\(</?%s[^>]*>\\)" tag) (point-max) t)
 391       (delete-region (match-beginning 0) (match-end 0)))))
 392
 393 (defun html2text-format-tags ()
 394   "See the variable `html2text-format-tag-list' for documentation."
 395   (interactive)
 396   (dolist (tag-and-function html2text-format-tag-list)
 397     (let ((tag      (car tag-and-function))
 398           (function (cdr tag-and-function)))
 399       (goto-char (point-min))
 400       (while (re-search-forward (format "\\(<%s\\( [^>]*\\)?>\\)" tag)
 401                                 (point-max) t)
 402         (let ((p1)
 403               (p2 (point))
 404               (p3) (p4))
 405           (search-backward "<" (point-min) t)
 406           (setq p1 (point))
 407           (unless (search-forward (format "</%s>" tag) (point-max) t)
 408             (goto-char p2)
 409             (insert (format "</%s>" tag)))
 410           (setq p4 (point))
 411           (search-backward "</" (point-min) t)
 412           (setq p3 (point))
 413           (funcall function p1 p2 p3 p4)
 414           (goto-char p1))))))
 415
 416 (defun html2text-substitute ()
 417   "See the variable `html2text-replace-list' for documentation."
 418   (interactive)
 419   (dolist (e html2text-replace-list)
 420     (goto-char (point-min))
 421     (let ((old-string (car e))
 422           (new-string (cdr e)))
 423       (html2text-replace-string old-string new-string (point-min) (point-max)))))
 424
 425 (defun html2text-format-single-elements ()
 426   (interactive)
 427   (dolist (tag-and-function html2text-format-single-element-list)
 428     (let ((tag      (car tag-and-function))
 429           (function (cdr tag-and-function)))
 430       (goto-char (point-min))
 431       (while (re-search-forward (format "\\(<%s\\( [^>]*\\)?>\\)" tag)
 432                                 (point-max) t)
 433         (let ((p1)
 434               (p2 (point)))
 435           (search-backward "<" (point-min) t)
 436           (setq p1 (point))
 437           (funcall function p1 p2))))))
 438
 439 ;;
 440 ;; Main function
 441 ;;
 442
 443 ;;;###autoload
 444 (defun html2text ()
 445   "Convert HTML to plain text in the current buffer."
 446   (interactive)
 447   (save-excursion
 448     (let ((case-fold-search t)
 449           (buffer-read-only))
 450       (html2text-remove-tags html2text-remove-tag-list)
 451       (html2text-format-tags)
 452       (html2text-remove-tags html2text-remove-tag-list2)
 453       (html2text-substitute)
 454       (html2text-format-single-elements)
 455       (html2text-fix-paragraphs))))
 456
 457 ;;
 458 ;; </Interactive functions>
 459 ;;
 460 (provide 'html2text)
 461
 462 ;;; html2text.el ends here