lisp/gnus/gnus-html.el

   1 ;;; gnus-html.el --- Render HTML in a buffer.
   2
   3 ;; Copyright (C) 2010-2018 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 <https://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 (require 'help-fns)
  43 (require 'url-queue)
  44
  45 (defcustom gnus-html-image-cache-ttl (days-to-time 7)
  46   "Time used to determine if we should use images from the cache."
  47   :version "24.1"
  48   :group 'gnus-art
  49   ;; FIXME hardly the friendliest type.  The allowed value is actually
  50   ;; any time value, but we are assuming no-one cares about USEC and
  51   ;; PSEC here.  It would be better to eg make it a number of minutes.
  52   :type '(list integer integer))
  53
  54 (defcustom gnus-html-image-automatic-caching t
  55   "Whether automatically cache retrieve images."
  56   :version "24.1"
  57   :group 'gnus-art
  58   :type 'boolean)
  59
  60 (defcustom gnus-html-frame-width 70
  61   "What width to use when rendering HTML."
  62   :version "24.1"
  63   :group 'gnus-art
  64   :type 'integer)
  65
  66 (defcustom gnus-max-image-proportion 0.9
  67   "How big pictures displayed are in relation to the window they're in.
  68 A value of 0.7 means that they are allowed to take up 70% of the
  69 width and height of the window.  If they are larger than this,
  70 and Emacs supports it, then the images will be rescaled down to
  71 fit these criteria."
  72   :version "24.1"
  73   :group 'gnus-art
  74   :type 'float)
  75
  76 (defvar gnus-html-image-map
  77   (let ((map (make-sparse-keymap)))
  78     (define-key map "u" 'gnus-article-copy-string)
  79     (define-key map "i" 'gnus-html-insert-image)
  80     (define-key map "v" 'gnus-html-browse-url)
  81     map))
  82
  83 (defvar gnus-html-displayed-image-map
  84   (let ((map (make-sparse-keymap)))
  85     (define-key map "a" 'gnus-html-show-alt-text)
  86     (define-key map "i" 'gnus-html-browse-image)
  87     (define-key map "\r" 'gnus-html-browse-url)
  88     (define-key map "u" 'gnus-article-copy-string)
  89     (define-key map [tab] 'widget-forward)
  90     map))
  91
  92 (defun gnus-html-encode-url (url)
  93   "Encode URL."
  94   (browse-url-url-encode-chars url "[)$ ]"))
  95
  96 (defun gnus-html-cache-expired (url ttl)
  97   "Check if URL is cached for more than TTL."
  98   (cond (url-standalone-mode
  99          (not (file-exists-p (url-cache-create-filename url))))
 100         (t (let ((cache-time (url-is-cached url)))
 101              (if cache-time
 102                  (time-less-p (time-add cache-time ttl) nil)
 103                t)))))
 104
 105 ;;;###autoload
 106 (defun gnus-article-html (&optional handle)
 107   (let ((article-buffer (current-buffer)))
 108     (unless handle
 109       (setq handle (mm-dissect-buffer t)))
 110     (save-restriction
 111       (narrow-to-region (point) (point))
 112       (save-excursion
 113         (mm-with-part handle
 114           (let* ((coding-system-for-read 'utf-8)
 115                  (coding-system-for-write 'utf-8)
 116                  (default-process-coding-system
 117                    (cons coding-system-for-read coding-system-for-write))
 118                  (charset (mail-content-type-get (mm-handle-type handle)
 119                                                  'charset)))
 120             (when (and charset
 121                        (setq charset (mm-charset-to-coding-system
 122                                       charset nil t))
 123                        (not (eq charset 'ascii)))
 124               (insert (prog1
 125                           (decode-coding-string (buffer-string) charset)
 126                         (erase-buffer)
 127                         (mm-enable-multibyte))))
 128             (call-process-region (point-min) (point-max)
 129                                  "w3m"
 130                                  nil article-buffer nil
 131                                  "-halfdump"
 132                                  "-no-cookie"
 133                                  "-I" "UTF-8"
 134                                  "-O" "UTF-8"
 135                                  "-o" "ext_halfdump=1"
 136                                  "-o" "display_ins_del=2"
 137                                  "-o" "pre_conv=1"
 138                                  "-t" (format "%s" tab-width)
 139                                  "-cols" (format "%s" gnus-html-frame-width)
 140                                  "-o" "display_image=on"
 141                                  "-T" "text/html"))))
 142       (gnus-html-wash-tags))))
 143
 144 (defvar gnus-article-mouse-face)
 145
 146 (defun gnus-html-pre-wash ()
 147   (goto-char (point-min))
 148   (while (re-search-forward " *<pre_int> *</pre_int> *\n" nil t)
 149     (replace-match "" t t))
 150   (goto-char (point-min))
 151   (while (re-search-forward "<a name[^\n>]+>" nil t)
 152     (replace-match "" t t)))
 153
 154 (defun gnus-html-wash-images ()
 155   "Run through current buffer and replace img tags by images."
 156   (let (tag parameters string start end images url alt-text
 157             inhibit-images blocked-images)
 158     (if (buffer-live-p gnus-summary-buffer)
 159         (with-current-buffer gnus-summary-buffer
 160           (setq inhibit-images gnus-inhibit-images
 161                 blocked-images (gnus-blocked-images)))
 162       (setq inhibit-images gnus-inhibit-images
 163             blocked-images (gnus-blocked-images)))
 164     (goto-char (point-min))
 165     ;; Search for all the images first.
 166     (while (re-search-forward "<img_alt \\([^>]*\\)>" nil t)
 167       (setq parameters (match-string 1)
 168             start (match-beginning 0))
 169       (delete-region start (point))
 170       (when (search-forward "</img_alt>" (line-end-position) t)
 171         (delete-region (match-beginning 0) (match-end 0)))
 172       (setq end (point))
 173       (when (string-match "src=\"\\([^\"]+\\)" parameters)
 174         (gnus-message 8 "gnus-html-wash-tags: fetching image URL %s" url)
 175         (setq url (gnus-html-encode-url (match-string 1 parameters))
 176               alt-text (when (string-match "\\(alt\\|title\\)=\"\\([^\"]+\\)"
 177                                            parameters)
 178                          (xml-substitute-special (match-string 2 parameters))))
 179         (add-text-properties
 180          start end
 181          (list 'image-url url
 182                'image-displayer `(lambda (url start end)
 183                                    (gnus-html-display-image url start end
 184                                                             ,alt-text))
 185                'gnus-image (list url start end alt-text)))
 186         (widget-convert-button
 187          'url-link start (point)
 188          :help-echo alt-text
 189          :keymap gnus-html-image-map
 190          url)
 191         (if (string-match "\\`cid:" url)
 192             ;; URLs with cid: have their content stashed in other
 193             ;; parts of the MIME structure, so just insert them
 194             ;; immediately.
 195             (let* ((handle (mm-get-content-id (substring url (match-end 0))))
 196                    (image (when (and handle
 197                                      (not inhibit-images))
 198                             (gnus-create-image
 199                              (mm-with-part handle (buffer-string))
 200                              nil t))))
 201               (if image
 202                   (gnus-add-image
 203                    'cid
 204                    (gnus-put-image
 205                     (gnus-rescale-image
 206                      image (gnus-html-maximum-image-size))
 207                     (gnus-string-or (prog1
 208                                         (buffer-substring start end)
 209                                       (delete-region start end))
 210                                     "*")
 211                     'cid))
 212                 (widget-convert-button
 213                  'link start end
 214                  :action 'gnus-html-insert-image
 215                  :help-echo url
 216                  :keymap gnus-html-image-map
 217                  :button-keymap gnus-html-image-map)))
 218           ;; Normal, external URL.
 219           (if (or inhibit-images
 220                   (gnus-html-image-url-blocked-p url blocked-images))
 221               (widget-convert-button
 222                'link start end
 223                :action 'gnus-html-insert-image
 224                :help-echo url
 225                :keymap gnus-html-image-map
 226                :button-keymap gnus-html-image-map)
 227             ;; Non-blocked url
 228             (let ((width
 229                    (when (string-match "width=\"?\\([0-9]+\\)" parameters)
 230                      (string-to-number (match-string 1 parameters))))
 231                   (height
 232                    (when (string-match "height=\"?\\([0-9]+\\)" parameters)
 233                      (string-to-number (match-string 1 parameters)))))
 234               ;; Don't fetch images that are really small.  They're
 235               ;; probably tracking pictures.
 236               (when (and (or (null height)
 237                              (> height 4))
 238                          (or (null width)
 239                              (> width 4)))
 240                 (gnus-html-display-image url start end alt-text)))))))))
 241
 242 (defun gnus-html-display-image (url start end &optional alt-text)
 243   "Display image at URL on text from START to END.
 244 Use ALT-TEXT for the image string."
 245   (or alt-text (setq alt-text "*"))
 246   (if (string-match "\\`cid:" url)
 247       (let ((handle (mm-get-content-id (substring url (match-end 0)))))
 248         (when handle
 249           (gnus-html-put-image (mm-with-part handle (buffer-string))
 250                                url alt-text)))
 251     (if (gnus-html-cache-expired url gnus-html-image-cache-ttl)
 252         ;; We don't have it, so schedule it for fetching
 253         ;; asynchronously.
 254         (gnus-html-schedule-image-fetching
 255          (current-buffer)
 256          (list url alt-text))
 257       ;; It's already cached, so just insert it.
 258       (gnus-html-put-image (gnus-html-get-image-data url) url alt-text))))
 259
 260 (defun gnus-html-wash-tags ()
 261   (let (tag parameters string start end images url)
 262     (gnus-html-pre-wash)
 263     (gnus-html-wash-images)
 264
 265     (goto-char (point-min))
 266     ;; Then do the other tags.
 267     (while (re-search-forward "<\\([^ />]+\\)\\([^>]*\\)>" nil t)
 268       (setq tag (match-string 1)
 269             parameters (match-string 2)
 270             start (match-beginning 0))
 271       (when (> (length parameters) 0)
 272         (set-text-properties 0 (1- (length parameters)) nil parameters))
 273       (delete-region start (point))
 274       (when (search-forward (concat "</" tag ">") nil t)
 275         (delete-region (match-beginning 0) (match-end 0)))
 276       (setq end (point))
 277       (cond
 278        ;; Fetch and insert a picture.
 279        ((equal tag "img_alt"))
 280        ;; Add a link.
 281        ((or (equal tag "a")
 282             (equal tag "A"))
 283         (when (string-match "href=\"\\([^\"]+\\)" parameters)
 284           (setq url (match-string 1 parameters))
 285           (gnus-message 8 "gnus-html-wash-tags: fetching link URL %s" url)
 286           (gnus-article-add-button start end
 287                                    'browse-url (mm-url-decode-entities-string url)
 288                                    url)
 289           (let ((overlay (make-overlay start end)))
 290             (overlay-put overlay 'evaporate t)
 291             (overlay-put overlay 'gnus-button-url url)
 292             (put-text-property start end 'gnus-string url)
 293             (when gnus-article-mouse-face
 294               (overlay-put overlay 'mouse-face gnus-article-mouse-face)))))
 295        ;; The upper-case IMG_ALT is apparently just an artifact that
 296        ;; should be deleted.
 297        ((equal tag "IMG_ALT")
 298         (delete-region start end))
 299        ;; w3m does not normalize the case
 300        ((or (equal tag "b")
 301             (equal tag "B"))
 302         (overlay-put (make-overlay start end) 'face 'gnus-emphasis-bold))
 303        ((or (equal tag "u")
 304             (equal tag "U"))
 305         (overlay-put (make-overlay start end) 'face 'gnus-emphasis-underline))
 306        ((or (equal tag "i")
 307             (equal tag "I"))
 308         (overlay-put (make-overlay start end) 'face 'gnus-emphasis-italic))
 309        ((or (equal tag "s")
 310             (equal tag "S"))
 311         (overlay-put (make-overlay start end) 'face 'gnus-emphasis-strikethru))
 312        ((or (equal tag "ins")
 313             (equal tag "INS"))
 314         (overlay-put (make-overlay start end) 'face 'gnus-emphasis-underline))
 315        ;; Handle different UL types
 316        ((equal tag "_SYMBOL")
 317         (when (string-match "TYPE=\\(.+\\)" parameters)
 318           (let ((type (string-to-number (match-string 1 parameters))))
 319             (delete-region start end)
 320             (cond ((= type 33) (insert " "))
 321                   ((= type 34) (insert " "))
 322                   ((= type 35) (insert " "))
 323                   ((= type 36) (insert " "))
 324                   ((= type 37) (insert " "))
 325                   ((= type 38) (insert " "))
 326                   ((= type 39) (insert " "))
 327                   ((= type 40) (insert " "))
 328                   ((= type 42) (insert " "))
 329                   ((= type 43) (insert " "))
 330                   (t (insert " "))))))
 331        ;; Whatever.  Just ignore the tag.
 332        (t
 333         ))
 334       (goto-char start))
 335     (goto-char (point-min))
 336     ;; The output from -halfdump isn't totally regular, so strip
 337     ;; off any </pre_int>s that were left over.
 338     (while (re-search-forward "</pre_int>\\|</internal>" nil t)
 339       (replace-match "" t t))
 340     (mm-url-decode-entities)))
 341
 342 (defun gnus-html-insert-image (&rest args)
 343   "Fetch and insert the image under point."
 344   (interactive)
 345   (apply 'gnus-html-display-image (get-text-property (point) 'gnus-image)))
 346
 347 (defun gnus-html-show-alt-text ()
 348   "Show the ALT text of the image under point."
 349   (interactive)
 350   (message "%s" (get-text-property (point) 'gnus-alt-text)))
 351
 352 (defun gnus-html-browse-image ()
 353   "Browse the image under point."
 354   (interactive)
 355   (browse-url (get-text-property (point) 'image-url)))
 356
 357 (defun gnus-html-browse-url ()
 358   "Browse the image under point."
 359   (interactive)
 360   (let ((url (get-text-property (point) 'gnus-string)))
 361     (cond
 362      ((not url)
 363       (message "No link under point"))
 364      ((string-match "^mailto:" url)
 365       (gnus-url-mailto url))
 366      (t
 367       (browse-url url)))))
 368
 369 (defun gnus-html-schedule-image-fetching (buffer image)
 370   "Retrieve IMAGE, and place it into BUFFER on arrival."
 371   (gnus-message 8 "gnus-html-schedule-image-fetching: buffer %s, image %s"
 372                 buffer image)
 373   (url-queue-retrieve (car image)
 374                       'gnus-html-image-fetched
 375                       (list buffer image) t t))
 376
 377 (defun gnus-html-image-fetched (status buffer image)
 378   "Callback function called when image has been fetched."
 379   (unless (plist-get status :error)
 380     (when (and (or (search-forward "\n\n" nil t)
 381                    (search-forward "\r\n\r\n" nil t))
 382                (not (eobp)))
 383       (when gnus-html-image-automatic-caching
 384         (url-store-in-cache (current-buffer)))
 385       (when (buffer-live-p buffer)
 386         (let ((data (buffer-substring (point) (point-max))))
 387           (with-current-buffer buffer
 388             (let ((inhibit-read-only t))
 389               (gnus-html-put-image data (car image) (cadr image))))))))
 390   (kill-buffer (current-buffer)))
 391
 392 (defun gnus-html-get-image-data (url)
 393   "Get image data for URL.
 394 Return a string with image data."
 395   (with-temp-buffer
 396     (mm-disable-multibyte)
 397     (url-cache-extract (url-cache-create-filename url))
 398     (when (or (search-forward "\n\n" nil t)
 399               (search-forward "\r\n\r\n" nil t))
 400       (buffer-substring (point) (point-max)))))
 401
 402 (defun gnus-html-maximum-image-size ()
 403   "Return the maximum size of an image according to `gnus-max-image-proportion'."
 404   (let ((edges (window-inside-pixel-edges
 405                 (get-buffer-window (current-buffer)))))
 406     ;; (width . height)
 407     (cons
 408      ;; Aimed width
 409      (truncate
 410       (* gnus-max-image-proportion
 411          (- (nth 2 edges) (nth 0 edges))))
 412      ;; Aimed height
 413      (truncate (* gnus-max-image-proportion
 414                   (- (nth 3 edges) (nth 1 edges)))))))
 415
 416 ;; Behind display-graphic-p test.
 417 (declare-function image-size "image.c" (spec &optional pixels frame))
 418
 419 (defun gnus-html-put-image (data url &optional alt-text)
 420   "Put an image with DATA from URL and optional ALT-TEXT."
 421   (when (display-graphic-p)
 422     (let* ((start (text-property-any (point-min) (point-max)
 423                                      'image-url url))
 424            (end (when start
 425                   (next-single-property-change start 'image-url))))
 426       ;; Image found?
 427       (when start
 428         (let* ((image
 429                 (ignore-errors
 430                   (gnus-create-image data nil t)))
 431                (size (and image (image-size image t))))
 432           (save-excursion
 433             (goto-char start)
 434             (let ((alt-text (or alt-text
 435                                 (buffer-substring-no-properties start end)))
 436                   (inhibit-read-only t))
 437               (if (and image
 438                        ;; Kludge to avoid displaying 30x30 gif images, which
 439                        ;; seems to be a signal of a broken image.
 440                        (not (and (listp image)
 441                                  (eq (plist-get (cdr image) :type)
 442                                      'gif)
 443                                  (= (car size) 30)
 444                                  (= (cdr size) 30))))
 445                   ;; Good image, add it!
 446                   (let ((image (gnus-rescale-image image (gnus-html-maximum-image-size))))
 447                     (delete-region start end)
 448                     (gnus-put-image image alt-text 'external)
 449                     (widget-convert-button
 450                      'url-link start (point)
 451                      :help-echo alt-text
 452                      :keymap gnus-html-displayed-image-map
 453                      url)
 454                     (put-text-property start (point) 'gnus-alt-text alt-text)
 455                     (when url
 456                       (add-text-properties
 457                        start (point)
 458                        `(image-url
 459                          ,url
 460                          image-displayer
 461                          (lambda (url start end)
 462                            (gnus-html-display-image url start end
 463                                                     ,alt-text)))))
 464                     (gnus-add-image 'external image)
 465                     t)
 466                 ;; Bad image, try to show something else
 467                 (when (fboundp 'find-image)
 468                   (delete-region start end)
 469                   (setq image (find-image
 470                                '((:type xpm :file "lock-broken.xpm"))))
 471                   (gnus-put-image image alt-text 'internal)
 472                   (gnus-add-image 'internal image))
 473                 nil))))))))
 474
 475 (defun gnus-html-image-url-blocked-p (url blocked-images)
 476   "Find out if URL is blocked by BLOCKED-IMAGES."
 477   (let ((ret (and blocked-images
 478                   (string-match blocked-images url))))
 479     (if ret
 480         (gnus-message 8 "gnus-html-image-url-blocked-p: %s blocked by regex %s"
 481                       url blocked-images)
 482       (gnus-message 9 "gnus-html-image-url-blocked-p: %s passes regex %s"
 483                     url blocked-images))
 484     ret))
 485
 486 ;;;###autoload
 487 (defun gnus-html-prefetch-images (summary)
 488   (when (buffer-live-p summary)
 489     (let (inhibit-images blocked-images)
 490       (with-current-buffer summary
 491         (setq inhibit-images gnus-inhibit-images
 492               blocked-images (gnus-blocked-images)))
 493       (save-match-data
 494         (while (re-search-forward "<img[^>]+src=[\"']\\(http[^\"']+\\)" nil t)
 495           (let ((url (gnus-html-encode-url
 496                       (mm-url-decode-entities-string (match-string 1)))))
 497             (unless (or inhibit-images
 498                         (gnus-html-image-url-blocked-p url blocked-images))
 499               (when (gnus-html-cache-expired url gnus-html-image-cache-ttl)
 500                 (gnus-html-schedule-image-fetching nil
 501                                                    (list url))))))))))
 502
 503 (provide 'gnus-html)
 504
 505 ;;; gnus-html.el ends here