lisp/gnus/gnus-html.el

   1 ;;; gnus-html.el --- Render HTML in a buffer.
   2
   3 ;; Copyright (C) 2010-2014 Free Software Foundation, Inc.
   4
   5 ;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
   6 ;; Keywords: html, web
   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 <http://www.gnu.org/licenses/>.
  22
  23 ;;; Commentary:
  24
  25 ;; The idea is to provide a simple, fast and pretty minimal way to
  26 ;; render HTML (including links and images) in a buffer, based on an
  27 ;; external HTML renderer (i.e., w3m).
  28
  29 ;;; Code:
  30
  31 (eval-when-compile (require 'cl))
  32
  33 (require 'gnus-art)
  34 (eval-when-compile (require 'mm-decode))
  35
  36 (require 'mm-url)
  37 (require 'url)
  38 (require 'url-cache)
  39 (require 'xml)
  40 (require 'browse-url)
  41 (require 'mm-util)
  42 (eval-and-compile (unless (featurep 'xemacs) (require 'help-fns)))
  43
  44 (defcustom gnus-html-image-cache-ttl (days-to-time 7)
  45   "Time used to determine if we should use images from the cache."
  46   :version "24.1"
  47   :group 'gnus-art
  48   ;; FIXME hardly the friendliest type.  The allowed value is actually
  49   ;; any time value, but we are assuming no-one cares about USEC and
  50   ;; PSEC here.  It would be better to eg make it a number of minutes.
  51   :type '(list integer integer))
  52
  53 (defcustom gnus-html-image-automatic-caching t
  54   "Whether automatically cache retrieve images."
  55   :version "24.1"
  56   :group 'gnus-art
  57   :type 'boolean)
  58
  59 (defcustom gnus-html-frame-width 70
  60   "What width to use when rendering HTML."
  61   :version "24.1"
  62   :group 'gnus-art
  63   :type 'integer)
  64
  65 (defcustom gnus-max-image-proportion 0.9
  66   "How big pictures displayed are in relation to the window they're in.
  67 A value of 0.7 means that they are allowed to take up 70% of the
  68 width and height of the window.  If they are larger than this,
  69 and Emacs supports it, then the images will be rescaled down to
  70 fit these criteria."
  71   :version "24.1"
  72   :group 'gnus-art
  73   :type 'float)
  74
  75 (defvar gnus-html-image-map
  76   (let ((map (make-sparse-keymap)))
  77     (define-key map "u" 'gnus-article-copy-string)
  78     (define-key map "i" 'gnus-html-insert-image)
  79     (define-key map "v" 'gnus-html-browse-url)
  80     map))
  81
  82 (defvar gnus-html-displayed-image-map
  83   (let ((map (make-sparse-keymap)))
  84     (define-key map "a" 'gnus-html-show-alt-text)
  85     (define-key map "i" 'gnus-html-browse-image)
  86     (define-key map "\r" 'gnus-html-browse-url)
  87     (define-key map "u" 'gnus-article-copy-string)
  88     (define-key map [tab] 'widget-forward)
  89     map))
  90
  91 (eval-and-compile
  92   (defalias 'gnus-html-encode-url-chars
  93     (if (fboundp 'browse-url-url-encode-chars)
  94         'browse-url-url-encode-chars
  95       (lambda (text chars)
  96         "URL-encode the chars in TEXT that match CHARS.
  97 CHARS is a regexp-like character alternative (e.g., \"[)$]\")."
  98         (let ((encoded-text (copy-sequence text))
  99               (s 0))
 100           (while (setq s (string-match chars encoded-text s))
 101             (setq encoded-text
 102                   (replace-match (format "%%%x"
 103                                          (string-to-char
 104                                           (match-string 0 encoded-text)))
 105                                  t t encoded-text)
 106                   s (1+ s)))
 107           encoded-text)))))
 108
 109 (defun gnus-html-encode-url (url)
 110   "Encode URL."
 111   (gnus-html-encode-url-chars url "[)$ ]"))
 112
 113 (defun gnus-html-cache-expired (url ttl)
 114   "Check if URL is cached for more than TTL."
 115   (cond (url-standalone-mode
 116          (not (file-exists-p (url-cache-create-filename url))))
 117         (t (let ((cache-time (url-is-cached url)))
 118              (if cache-time
 119                  (time-less-p
 120                   (time-add
 121                    cache-time
 122                    ttl)
 123                   (current-time))
 124                t)))))
 125
 126 ;;;###autoload
 127 (defun gnus-article-html (&optional handle)
 128   (let ((article-buffer (current-buffer)))
 129     (unless handle
 130       (setq handle (mm-dissect-buffer t)))
 131     (save-restriction
 132       (narrow-to-region (point) (point))
 133       (save-excursion
 134         (mm-with-part handle
 135           (let* ((coding-system-for-read 'utf-8)
 136                  (coding-system-for-write 'utf-8)
 137                  (default-process-coding-system
 138                    (cons coding-system-for-read coding-system-for-write))
 139                  (charset (mail-content-type-get (mm-handle-type handle)
 140                                                  'charset)))
 141             (when (and charset
 142                        (setq charset (mm-charset-to-coding-system
 143                                       charset nil t))
 144                        (not (eq charset 'ascii)))
 145               (insert (prog1
 146                           (mm-decode-coding-string (buffer-string) charset)
 147                         (erase-buffer)
 148                         (mm-enable-multibyte))))
 149             (call-process-region (point-min) (point-max)
 150                                  "w3m"
 151                                  nil article-buffer nil
 152                                  "-halfdump"
 153                                  "-no-cookie"
 154                                  "-I" "UTF-8"
 155                                  "-O" "UTF-8"
 156                                  "-o" "ext_halfdump=1"
 157                                  "-o" "display_ins_del=2"
 158                                  "-o" "pre_conv=1"
 159                                  "-t" (format "%s" tab-width)
 160                                  "-cols" (format "%s" gnus-html-frame-width)
 161                                  "-o" "display_image=on"
 162                                  "-T" "text/html"))))
 163       (gnus-html-wash-tags))))
 164
 165 (defvar gnus-article-mouse-face)
 166
 167 (defun gnus-html-pre-wash ()
 168   (goto-char (point-min))
 169   (while (re-search-forward " *<pre_int> *</pre_int> *\n" nil t)
 170     (replace-match "" t t))
 171   (goto-char (point-min))
 172   (while (re-search-forward "<a name[^\n>]+>" nil t)
 173     (replace-match "" t t)))
 174
 175 (defun gnus-html-wash-images ()
 176   "Run through current buffer and replace img tags by images."
 177   (let (tag parameters string start end images url alt-text
 178             inhibit-images blocked-images)
 179     (if (buffer-live-p gnus-summary-buffer)
 180         (with-current-buffer gnus-summary-buffer
 181           (setq inhibit-images gnus-inhibit-images
 182                 blocked-images (gnus-blocked-images)))
 183       (setq inhibit-images gnus-inhibit-images
 184             blocked-images (gnus-blocked-images)))
 185     (goto-char (point-min))
 186     ;; Search for all the images first.
 187     (while (re-search-forward "<img_alt \\([^>]*\\)>" nil t)
 188       (setq parameters (match-string 1)
 189             start (match-beginning 0))
 190       (delete-region start (point))
 191       (when (search-forward "</img_alt>" (line-end-position) t)
 192         (delete-region (match-beginning 0) (match-end 0)))
 193       (setq end (point))
 194       (when (string-match "src=\"\\([^\"]+\\)" parameters)
 195         (gnus-message 8 "gnus-html-wash-tags: fetching image URL %s" url)
 196         (setq url (gnus-html-encode-url (match-string 1 parameters))
 197               alt-text (when (string-match "\\(alt\\|title\\)=\"\\([^\"]+\\)"
 198                                            parameters)
 199                          (xml-substitute-special (match-string 2 parameters))))
 200         (gnus-add-text-properties
 201          start end
 202          (list 'image-url url
 203                'image-displayer `(lambda (url start end)
 204                                    (gnus-html-display-image url start end
 205                                                             ,alt-text))
 206                'gnus-image (list url start end alt-text)))
 207         (widget-convert-button
 208          'url-link start (point)
 209          :help-echo alt-text
 210          :keymap gnus-html-image-map
 211          url)
 212         (if (string-match "\\`cid:" url)
 213             ;; URLs with cid: have their content stashed in other
 214             ;; parts of the MIME structure, so just insert them
 215             ;; immediately.
 216             (let* ((handle (mm-get-content-id (substring url (match-end 0))))
 217                    (image (when (and handle
 218                                      (not inhibit-images))
 219                             (gnus-create-image
 220                              (mm-with-part handle (buffer-string))
 221                              nil t))))
 222               (if image
 223                   (gnus-add-image
 224                    'cid
 225                    (gnus-put-image
 226                     (gnus-rescale-image
 227                      image (gnus-html-maximum-image-size))
 228                     (gnus-string-or (prog1
 229                                         (buffer-substring start end)
 230                                       (delete-region start end))
 231                                     "*")
 232                     'cid))
 233                 (widget-convert-button
 234                  'link start end
 235                  :action 'gnus-html-insert-image
 236                  :help-echo url
 237                  :keymap gnus-html-image-map
 238                  :button-keymap gnus-html-image-map)))
 239           ;; Normal, external URL.
 240           (if (or inhibit-images
 241                   (gnus-html-image-url-blocked-p url blocked-images))
 242               (widget-convert-button
 243                'link start end
 244                :action 'gnus-html-insert-image
 245                :help-echo url
 246                :keymap gnus-html-image-map
 247                :button-keymap gnus-html-image-map)
 248             ;; Non-blocked url
 249             (let ((width
 250                    (when (string-match "width=\"?\\([0-9]+\\)" parameters)
 251                      (string-to-number (match-string 1 parameters))))
 252                   (height
 253                    (when (string-match "height=\"?\\([0-9]+\\)" parameters)
 254                      (string-to-number (match-string 1 parameters)))))
 255               ;; Don't fetch images that are really small.  They're
 256               ;; probably tracking pictures.
 257               (when (and (or (null height)
 258                              (> height 4))
 259                          (or (null width)
 260                              (> width 4)))
 261                 (gnus-html-display-image url start end alt-text)))))))))
 262
 263 (defun gnus-html-display-image (url start end &optional alt-text)
 264   "Display image at URL on text from START to END.
 265 Use ALT-TEXT for the image string."
 266   (or alt-text (setq alt-text "*"))
 267   (if (string-match "\\`cid:" url)
 268       (let ((handle (mm-get-content-id (substring url (match-end 0)))))
 269         (when handle
 270           (gnus-html-put-image (mm-with-part handle (buffer-string))
 271                                url alt-text)))
 272     (if (gnus-html-cache-expired url gnus-html-image-cache-ttl)
 273         ;; We don't have it, so schedule it for fetching
 274         ;; asynchronously.
 275         (gnus-html-schedule-image-fetching
 276          (current-buffer)
 277          (list url alt-text))
 278       ;; It's already cached, so just insert it.
 279       (gnus-html-put-image (gnus-html-get-image-data url) url alt-text))))
 280
 281 (defun gnus-html-wash-tags ()
 282   (let (tag parameters string start end images url)
 283     (gnus-html-pre-wash)
 284     (gnus-html-wash-images)
 285
 286     (goto-char (point-min))
 287     ;; Then do the other tags.
 288     (while (re-search-forward "<\\([^ />]+\\)\\([^>]*\\)>" nil t)
 289       (setq tag (match-string 1)
 290             parameters (match-string 2)
 291             start (match-beginning 0))
 292       (when (> (length parameters) 0)
 293         (set-text-properties 0 (1- (length parameters)) nil parameters))
 294       (delete-region start (point))
 295       (when (search-forward (concat "</" tag ">") nil t)
 296         (delete-region (match-beginning 0) (match-end 0)))
 297       (setq end (point))
 298       (cond
 299        ;; Fetch and insert a picture.
 300        ((equal tag "img_alt"))
 301        ;; Add a link.
 302        ((or (equal tag "a")
 303             (equal tag "A"))
 304         (when (string-match "href=\"\\([^\"]+\\)" parameters)
 305           (setq url (match-string 1 parameters))
 306           (gnus-message 8 "gnus-html-wash-tags: fetching link URL %s" url)
 307           (gnus-article-add-button start end
 308                                    'browse-url (mm-url-decode-entities-string url)
 309                                    url)
 310           (let ((overlay (gnus-make-overlay start end)))
 311             (gnus-overlay-put overlay 'evaporate t)
 312             (gnus-overlay-put overlay 'gnus-button-url url)
 313             (gnus-put-text-property start end 'gnus-string url)
 314             (when gnus-article-mouse-face
 315               (gnus-overlay-put overlay 'mouse-face gnus-article-mouse-face)))))
 316        ;; The upper-case IMG_ALT is apparently just an artifact that
 317        ;; should be deleted.
 318        ((equal tag "IMG_ALT")
 319         (delete-region start end))
 320        ;; w3m does not normalize the case
 321        ((or (equal tag "b")
 322             (equal tag "B"))
 323         (gnus-overlay-put (gnus-make-overlay start end) 'face 'gnus-emphasis-bold))
 324        ((or (equal tag "u")
 325             (equal tag "U"))
 326         (gnus-overlay-put (gnus-make-overlay start end) 'face 'gnus-emphasis-underline))
 327        ((or (equal tag "i")
 328             (equal tag "I"))
 329         (gnus-overlay-put (gnus-make-overlay start end) 'face 'gnus-emphasis-italic))
 330        ((or (equal tag "s")
 331             (equal tag "S"))
 332         (gnus-overlay-put (gnus-make-overlay start end) 'face 'gnus-emphasis-strikethru))
 333        ((or (equal tag "ins")
 334             (equal tag "INS"))
 335         (gnus-overlay-put (gnus-make-overlay start end) 'face 'gnus-emphasis-underline))
 336        ;; Handle different UL types
 337        ((equal tag "_SYMBOL")
 338         (when (string-match "TYPE=\\(.+\\)" parameters)
 339           (let ((type (string-to-number (match-string 1 parameters))))
 340             (delete-region start end)
 341             (cond ((= type 33) (insert " "))
 342                   ((= type 34) (insert " "))
 343                   ((= type 35) (insert " "))
 344                   ((= type 36) (insert " "))
 345                   ((= type 37) (insert " "))
 346                   ((= type 38) (insert " "))
 347                   ((= type 39) (insert " "))
 348                   ((= type 40) (insert " "))
 349                   ((= type 42) (insert " "))
 350                   ((= type 43) (insert " "))
 351                   (t (insert " "))))))
 352        ;; Whatever.  Just ignore the tag.
 353        (t
 354         ))
 355       (goto-char start))
 356     (goto-char (point-min))
 357     ;; The output from -halfdump isn't totally regular, so strip
 358     ;; off any </pre_int>s that were left over.
 359     (while (re-search-forward "</pre_int>\\|</internal>" nil t)
 360       (replace-match "" t t))
 361     (mm-url-decode-entities)))
 362
 363 (defun gnus-html-insert-image (&rest args)
 364   "Fetch and insert the image under point."
 365   (interactive)
 366   (apply 'gnus-html-display-image (get-text-property (point) 'gnus-image)))
 367
 368 (defun gnus-html-show-alt-text ()
 369   "Show the ALT text of the image under point."
 370   (interactive)
 371   (message "%s" (get-text-property (point) 'gnus-alt-text)))
 372
 373 (defun gnus-html-browse-image ()
 374   "Browse the image under point."
 375   (interactive)
 376   (browse-url (get-text-property (point) 'image-url)))
 377
 378 (defun gnus-html-browse-url ()
 379   "Browse the image under point."
 380   (interactive)
 381   (let ((url (get-text-property (point) 'gnus-string)))
 382     (cond
 383      ((not url)
 384       (message "No link under point"))
 385      ((string-match "^mailto:" url)
 386       (gnus-url-mailto url))
 387      (t
 388       (browse-url url)))))
 389
 390 (defun gnus-html-schedule-image-fetching (buffer image)
 391   "Retrieve IMAGE, and place it into BUFFER on arrival."
 392   (gnus-message 8 "gnus-html-schedule-image-fetching: buffer %s, image %s"
 393                 buffer image)
 394   (if (fboundp 'url-queue-retrieve)
 395       (url-queue-retrieve (car image)
 396                           'gnus-html-image-fetched
 397                           (list buffer image) t t)
 398     (ignore-errors
 399       (url-retrieve (car image)
 400                     'gnus-html-image-fetched
 401                     (list buffer image)))))
 402
 403 (defun gnus-html-image-fetched (status buffer image)
 404   "Callback function called when image has been fetched."
 405   (unless (plist-get status :error)
 406     (when (and (or (search-forward "\n\n" nil t)
 407                    (search-forward "\r\n\r\n" nil t))
 408                (not (eobp)))
 409       (when gnus-html-image-automatic-caching
 410         (url-store-in-cache (current-buffer)))
 411       (when (buffer-live-p buffer)
 412         (let ((data (buffer-substring (point) (point-max))))
 413           (with-current-buffer buffer
 414             (let ((inhibit-read-only t))
 415               (gnus-html-put-image data (car image) (cadr image))))))))
 416   (kill-buffer (current-buffer)))
 417
 418 (defun gnus-html-get-image-data (url)
 419   "Get image data for URL.
 420 Return a string with image data."
 421   (with-temp-buffer
 422     (mm-disable-multibyte)
 423     (url-cache-extract (url-cache-create-filename url))
 424     (when (or (search-forward "\n\n" nil t)
 425               (search-forward "\r\n\r\n" nil t))
 426       (buffer-substring (point) (point-max)))))
 427
 428 (defun gnus-html-maximum-image-size ()
 429   "Return the maximum size of an image according to `gnus-max-image-proportion'."
 430   (let ((edges (gnus-window-inside-pixel-edges
 431                 (get-buffer-window (current-buffer)))))
 432     ;; (width . height)
 433     (cons
 434      ;; Aimed width
 435      (truncate
 436       (* gnus-max-image-proportion
 437          (- (nth 2 edges) (nth 0 edges))))
 438      ;; Aimed height
 439      (truncate (* gnus-max-image-proportion
 440                   (- (nth 3 edges) (nth 1 edges)))))))
 441
 442 ;; Behind display-graphic-p test.
 443 (declare-function image-size "image.c" (spec &optional pixels frame))
 444
 445 (defun gnus-html-put-image (data url &optional alt-text)
 446   "Put an image with DATA from URL and optional ALT-TEXT."
 447   (when (gnus-graphic-display-p)
 448     (let* ((start (text-property-any (point-min) (point-max)
 449                                      'image-url url))
 450            (end (when start
 451                   (next-single-property-change start 'image-url))))
 452       ;; Image found?
 453       (when start
 454         (let* ((image
 455                 (ignore-errors
 456                   (gnus-create-image data nil t)))
 457                (size (and image
 458                           (if (featurep 'xemacs)
 459                               (cons (glyph-width image) (glyph-height image))
 460                             (image-size image t)))))
 461           (save-excursion
 462             (goto-char start)
 463             (let ((alt-text (or alt-text
 464                                 (buffer-substring-no-properties start end)))
 465                   (inhibit-read-only t))
 466               (if (and image
 467                        ;; Kludge to avoid displaying 30x30 gif images, which
 468                        ;; seems to be a signal of a broken image.
 469                        (not (and (if (featurep 'xemacs)
 470                                      (glyphp image)
 471                                    (listp image))
 472                                  (eq (if (featurep 'xemacs)
 473                                          (let ((d (cdadar
 474                                                    (specifier-spec-list
 475                                                     (glyph-image image)))))
 476                                            (and (vectorp d)
 477                                                 (aref d 0)))
 478                                        (plist-get (cdr image) :type))
 479                                      'gif)
 480                                  (= (car size) 30)
 481                                  (= (cdr size) 30))))
 482                   ;; Good image, add it!
 483                   (let ((image (gnus-rescale-image image (gnus-html-maximum-image-size))))
 484                     (delete-region start end)
 485                     (gnus-put-image image alt-text 'external)
 486                     (widget-convert-button
 487                      'url-link start (point)
 488                      :help-echo alt-text
 489                      :keymap gnus-html-displayed-image-map
 490                      url)
 491                     (gnus-put-text-property start (point)
 492                                             'gnus-alt-text alt-text)
 493                     (when url
 494                       (gnus-add-text-properties
 495                        start (point)
 496                        `(image-url
 497                          ,url
 498                          image-displayer
 499                          (lambda (url start end)
 500                            (gnus-html-display-image url start end
 501                                                     ,alt-text)))))
 502                     (gnus-add-image 'external image)
 503                     t)
 504                 ;; Bad image, try to show something else
 505                 (when (fboundp 'find-image)
 506                   (delete-region start end)
 507                   (setq image (find-image
 508                                '((:type xpm :file "lock-broken.xpm"))))
 509                   (gnus-put-image image alt-text 'internal)
 510                   (gnus-add-image 'internal image))
 511                 nil))))))))
 512
 513 (defun gnus-html-image-url-blocked-p (url blocked-images)
 514   "Find out if URL is blocked by BLOCKED-IMAGES."
 515   (let ((ret (and blocked-images
 516                   (string-match blocked-images url))))
 517     (if ret
 518         (gnus-message 8 "gnus-html-image-url-blocked-p: %s blocked by regex %s"
 519                       url blocked-images)
 520       (gnus-message 9 "gnus-html-image-url-blocked-p: %s passes regex %s"
 521                     url blocked-images))
 522     ret))
 523
 524 ;;;###autoload
 525 (defun gnus-html-prefetch-images (summary)
 526   (when (buffer-live-p summary)
 527     (let (inhibit-images blocked-images)
 528       (with-current-buffer summary
 529         (setq inhibit-images gnus-inhibit-images
 530               blocked-images (gnus-blocked-images)))
 531       (save-match-data
 532         (while (re-search-forward "<img[^>]+src=[\"']\\(http[^\"']+\\)" nil t)
 533           (let ((url (gnus-html-encode-url
 534                       (mm-url-decode-entities-string (match-string 1)))))
 535             (unless (or inhibit-images
 536                         (gnus-html-image-url-blocked-p url blocked-images))
 537               (when (gnus-html-cache-expired url gnus-html-image-cache-ttl)
 538                 (gnus-html-schedule-image-fetching nil
 539                                                    (list url))))))))))
 540
 541 (provide 'gnus-html)
 542
 543 ;;; gnus-html.el ends here