lisp/gnus/gnus-html.el

   1 ;;; gnus-html.el --- Render HTML in a buffer.
   2
   3 ;; Copyright (C) 2010-2013 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 ;; For Emacs <22.2 and XEmacs.
  32 (eval-and-compile
  33   (unless (fboundp 'declare-function) (defmacro declare-function (&rest r))))
  34
  35 (eval-when-compile (require 'cl))
  36
  37 (require 'gnus-art)
  38 (eval-when-compile (require 'mm-decode))
  39
  40 (require 'mm-url)
  41 (require 'url)
  42 (require 'url-cache)
  43 (require 'xml)
  44 (require 'browse-url)
  45 (require 'mm-util)
  46 (eval-and-compile (unless (featurep 'xemacs) (require 'help-fns)))
  47
  48 (defcustom gnus-html-image-cache-ttl (days-to-time 7)
  49   "Time used to determine if we should use images from the cache."
  50   :version "24.1"
  51   :group 'gnus-art
  52   ;; FIXME hardly the friendliest type.  The allowed value is actually
  53   ;; any time value, but we are assuming no-one cares about USEC and
  54   ;; PSEC here.  It would be better to eg make it a number of minutes.
  55   :type '(list integer integer))
  56
  57 (defcustom gnus-html-image-automatic-caching t
  58   "Whether automatically cache retrieve images."
  59   :version "24.1"
  60   :group 'gnus-art
  61   :type 'boolean)
  62
  63 (defcustom gnus-html-frame-width 70
  64   "What width to use when rendering HTML."
  65   :version "24.1"
  66   :group 'gnus-art
  67   :type 'integer)
  68
  69 (defcustom gnus-max-image-proportion 0.9
  70   "How big pictures displayed are in relation to the window they're in.
  71 A value of 0.7 means that they are allowed to take up 70% of the
  72 width and height of the window.  If they are larger than this,
  73 and Emacs supports it, then the images will be rescaled down to
  74 fit these criteria."
  75   :version "24.1"
  76   :group 'gnus-art
  77   :type 'float)
  78
  79 (defvar gnus-html-image-map
  80   (let ((map (make-sparse-keymap)))
  81     (define-key map "u" 'gnus-article-copy-string)
  82     (define-key map "i" 'gnus-html-insert-image)
  83     (define-key map "v" 'gnus-html-browse-url)
  84     map))
  85
  86 (defvar gnus-html-displayed-image-map
  87   (let ((map (make-sparse-keymap)))
  88     (define-key map "a" 'gnus-html-show-alt-text)
  89     (define-key map "i" 'gnus-html-browse-image)
  90     (define-key map "\r" 'gnus-html-browse-url)
  91     (define-key map "u" 'gnus-article-copy-string)
  92     (define-key map [tab] 'widget-forward)
  93     map))
  94
  95 (eval-and-compile
  96   (defalias 'gnus-html-encode-url-chars
  97     (if (fboundp 'browse-url-url-encode-chars)
  98         'browse-url-url-encode-chars
  99       (lambda (text chars)
 100         "URL-encode the chars in TEXT that match CHARS.
 101 CHARS is a regexp-like character alternative (e.g., \"[)$]\")."
 102         (let ((encoded-text (copy-sequence text))
 103               (s 0))
 104           (while (setq s (string-match chars encoded-text s))
 105             (setq encoded-text
 106                   (replace-match (format "%%%x"
 107                                          (string-to-char
 108                                           (match-string 0 encoded-text)))
 109                                  t t encoded-text)
 110                   s (1+ s)))
 111           encoded-text)))))
 112
 113 (defun gnus-html-encode-url (url)
 114   "Encode URL."
 115   (gnus-html-encode-url-chars url "[)$ ]"))
 116
 117 (defun gnus-html-cache-expired (url ttl)
 118   "Check if URL is cached for more than TTL."
 119   (cond (url-standalone-mode
 120          (not (file-exists-p (url-cache-create-filename url))))
 121         (t (let ((cache-time (url-is-cached url)))
 122              (if cache-time
 123                  (time-less-p
 124                   (time-add
 125                    cache-time
 126                    ttl)
 127                   (current-time))
 128                t)))))
 129
 130 ;;;###autoload
 131 (defun gnus-article-html (&optional handle)
 132   (let ((article-buffer (current-buffer)))
 133     (unless handle
 134       (setq handle (mm-dissect-buffer t)))
 135     (save-restriction
 136       (narrow-to-region (point) (point))
 137       (save-excursion
 138         (mm-with-part handle
 139           (let* ((coding-system-for-read 'utf-8)
 140                  (coding-system-for-write 'utf-8)
 141                  (default-process-coding-system
 142                    (cons coding-system-for-read coding-system-for-write))
 143                  (charset (mail-content-type-get (mm-handle-type handle)
 144                                                  'charset)))
 145             (when (and charset
 146                        (setq charset (mm-charset-to-coding-system charset))
 147                        (not (eq charset 'ascii)))
 148               (insert (prog1
 149                           (mm-decode-coding-string (buffer-string) charset)
 150                         (erase-buffer)
 151                         (mm-enable-multibyte))))
 152             (call-process-region (point-min) (point-max)
 153                                  "w3m"
 154                                  nil article-buffer nil
 155                                  "-halfdump"
 156                                  "-no-cookie"
 157                                  "-I" "UTF-8"
 158                                  "-O" "UTF-8"
 159                                  "-o" "ext_halfdump=1"
 160                                  "-o" "display_ins_del=2"
 161                                  "-o" "pre_conv=1"
 162                                  "-t" (format "%s" tab-width)
 163                                  "-cols" (format "%s" gnus-html-frame-width)
 164                                  "-o" "display_image=on"
 165                                  "-T" "text/html"))))
 166       (gnus-html-wash-tags))))
 167
 168 (defvar gnus-article-mouse-face)
 169
 170 (defun gnus-html-pre-wash ()
 171   (goto-char (point-min))
 172   (while (re-search-forward " *<pre_int> *</pre_int> *\n" nil t)
 173     (replace-match "" t t))
 174   (goto-char (point-min))
 175   (while (re-search-forward "<a name[^\n>]+>" nil t)
 176     (replace-match "" t t)))
 177
 178 (defun gnus-html-wash-images ()
 179   "Run through current buffer and replace img tags by images."
 180   (let (tag parameters string start end images url alt-text
 181             inhibit-images blocked-images)
 182     (if (buffer-live-p gnus-summary-buffer)
 183         (with-current-buffer gnus-summary-buffer
 184           (setq inhibit-images gnus-inhibit-images
 185                 blocked-images (gnus-blocked-images)))
 186       (setq inhibit-images gnus-inhibit-images
 187             blocked-images (gnus-blocked-images)))
 188     (goto-char (point-min))
 189     ;; Search for all the images first.
 190     (while (re-search-forward "<img_alt \\([^>]*\\)>" nil t)
 191       (setq parameters (match-string 1)
 192             start (match-beginning 0))
 193       (delete-region start (point))
 194       (when (search-forward "</img_alt>" (line-end-position) t)
 195         (delete-region (match-beginning 0) (match-end 0)))
 196       (setq end (point))
 197       (when (string-match "src=\"\\([^\"]+\\)" parameters)
 198         (gnus-message 8 "gnus-html-wash-tags: fetching image URL %s" url)
 199         (setq url (gnus-html-encode-url (match-string 1 parameters))
 200               alt-text (when (string-match "\\(alt\\|title\\)=\"\\([^\"]+\\)"
 201                                            parameters)
 202                          (xml-substitute-special (match-string 2 parameters))))
 203         (gnus-add-text-properties
 204          start end
 205          (list 'image-url url
 206                'image-displayer `(lambda (url start end)
 207                                    (gnus-html-display-image url start end
 208                                                             ,alt-text))
 209                'gnus-image (list url start end alt-text)))
 210         (widget-convert-button
 211          'url-link start (point)
 212          :help-echo alt-text
 213          :keymap gnus-html-image-map
 214          url)
 215         (if (string-match "\\`cid:" url)
 216             ;; URLs with cid: have their content stashed in other
 217             ;; parts of the MIME structure, so just insert them
 218             ;; immediately.
 219             (let* ((handle (mm-get-content-id (substring url (match-end 0))))
 220                    (image (when (and handle
 221                                      (not inhibit-images))
 222                             (gnus-create-image
 223                              (mm-with-part handle (buffer-string))
 224                              nil t))))
 225               (if image
 226                   (gnus-add-image
 227                    'cid
 228                    (gnus-put-image
 229                     (gnus-rescale-image
 230                      image (gnus-html-maximum-image-size))
 231                     (gnus-string-or (prog1
 232                                         (buffer-substring start end)
 233                                       (delete-region start end))
 234                                     "*")
 235                     'cid))
 236                 (widget-convert-button
 237                  'link start end
 238                  :action 'gnus-html-insert-image
 239                  :help-echo url
 240                  :keymap gnus-html-image-map
 241                  :button-keymap gnus-html-image-map)))
 242           ;; Normal, external URL.
 243           (if (or inhibit-images
 244                   (gnus-html-image-url-blocked-p url blocked-images))
 245               (widget-convert-button
 246                'link start end
 247                :action 'gnus-html-insert-image
 248                :help-echo url
 249                :keymap gnus-html-image-map
 250                :button-keymap gnus-html-image-map)
 251             ;; Non-blocked url
 252             (let ((width
 253                    (when (string-match "width=\"?\\([0-9]+\\)" parameters)
 254                      (string-to-number (match-string 1 parameters))))
 255                   (height
 256                    (when (string-match "height=\"?\\([0-9]+\\)" parameters)
 257                      (string-to-number (match-string 1 parameters)))))
 258               ;; Don't fetch images that are really small.  They're
 259               ;; probably tracking pictures.
 260               (when (and (or (null height)
 261                              (> height 4))
 262                          (or (null width)
 263                              (> width 4)))
 264                 (gnus-html-display-image url start end alt-text)))))))))
 265
 266 (defun gnus-html-display-image (url start end &optional alt-text)
 267   "Display image at URL on text from START to END.
 268 Use ALT-TEXT for the image string."
 269   (or alt-text (setq alt-text "*"))
 270   (if (string-match "\\`cid:" url)
 271       (let ((handle (mm-get-content-id (substring url (match-end 0)))))
 272         (when handle
 273           (gnus-html-put-image (mm-with-part handle (buffer-string))
 274                                url alt-text)))
 275     (if (gnus-html-cache-expired url gnus-html-image-cache-ttl)
 276         ;; We don't have it, so schedule it for fetching
 277         ;; asynchronously.
 278         (gnus-html-schedule-image-fetching
 279          (current-buffer)
 280          (list url alt-text))
 281       ;; It's already cached, so just insert it.
 282       (gnus-html-put-image (gnus-html-get-image-data url) url alt-text))))
 283
 284 (defun gnus-html-wash-tags ()
 285   (let (tag parameters string start end images url)
 286     (gnus-html-pre-wash)
 287     (gnus-html-wash-images)
 288
 289     (goto-char (point-min))
 290     ;; Then do the other tags.
 291     (while (re-search-forward "<\\([^ />]+\\)\\([^>]*\\)>" nil t)
 292       (setq tag (match-string 1)
 293             parameters (match-string 2)
 294             start (match-beginning 0))
 295       (when (> (length parameters) 0)
 296         (set-text-properties 0 (1- (length parameters)) nil parameters))
 297       (delete-region start (point))
 298       (when (search-forward (concat "</" tag ">") nil t)
 299         (delete-region (match-beginning 0) (match-end 0)))
 300       (setq end (point))
 301       (cond
 302        ;; Fetch and insert a picture.
 303        ((equal tag "img_alt"))
 304        ;; Add a link.
 305        ((or (equal tag "a")
 306             (equal tag "A"))
 307         (when (string-match "href=\"\\([^\"]+\\)" parameters)
 308           (setq url (match-string 1 parameters))
 309           (gnus-message 8 "gnus-html-wash-tags: fetching link URL %s" url)
 310           (gnus-article-add-button start end
 311                                    'browse-url (mm-url-decode-entities-string url)
 312                                    url)
 313           (let ((overlay (gnus-make-overlay start end)))
 314             (gnus-overlay-put overlay 'evaporate t)
 315             (gnus-overlay-put overlay 'gnus-button-url url)
 316             (gnus-put-text-property start end 'gnus-string url)
 317             (when gnus-article-mouse-face
 318               (gnus-overlay-put overlay 'mouse-face gnus-article-mouse-face)))))
 319        ;; The upper-case IMG_ALT is apparently just an artifact that
 320        ;; should be deleted.
 321        ((equal tag "IMG_ALT")
 322         (delete-region start end))
 323        ;; w3m does not normalize the case
 324        ((or (equal tag "b")
 325             (equal tag "B"))
 326         (gnus-overlay-put (gnus-make-overlay start end) 'face 'gnus-emphasis-bold))
 327        ((or (equal tag "u")
 328             (equal tag "U"))
 329         (gnus-overlay-put (gnus-make-overlay start end) 'face 'gnus-emphasis-underline))
 330        ((or (equal tag "i")
 331             (equal tag "I"))
 332         (gnus-overlay-put (gnus-make-overlay start end) 'face 'gnus-emphasis-italic))
 333        ((or (equal tag "s")
 334             (equal tag "S"))
 335         (gnus-overlay-put (gnus-make-overlay start end) 'face 'gnus-emphasis-strikethru))
 336        ((or (equal tag "ins")
 337             (equal tag "INS"))
 338         (gnus-overlay-put (gnus-make-overlay start end) 'face 'gnus-emphasis-underline))
 339        ;; Handle different UL types
 340        ((equal tag "_SYMBOL")
 341         (when (string-match "TYPE=\\(.+\\)" parameters)
 342           (let ((type (string-to-number (match-string 1 parameters))))
 343             (delete-region start end)
 344             (cond ((= type 33) (insert " "))
 345                   ((= type 34) (insert " "))
 346                   ((= type 35) (insert " "))
 347                   ((= type 36) (insert " "))
 348                   ((= type 37) (insert " "))
 349                   ((= type 38) (insert " "))
 350                   ((= type 39) (insert " "))
 351                   ((= type 40) (insert " "))
 352                   ((= type 42) (insert " "))
 353                   ((= type 43) (insert " "))
 354                   (t (insert " "))))))
 355        ;; Whatever.  Just ignore the tag.
 356        (t
 357         ))
 358       (goto-char start))
 359     (goto-char (point-min))
 360     ;; The output from -halfdump isn't totally regular, so strip
 361     ;; off any </pre_int>s that were left over.
 362     (while (re-search-forward "</pre_int>\\|</internal>" nil t)
 363       (replace-match "" t t))
 364     (mm-url-decode-entities)))
 365
 366 (defun gnus-html-insert-image (&rest args)
 367   "Fetch and insert the image under point."
 368   (interactive)
 369   (apply 'gnus-html-display-image (get-text-property (point) 'gnus-image)))
 370
 371 (defun gnus-html-show-alt-text ()
 372   "Show the ALT text of the image under point."
 373   (interactive)
 374   (message "%s" (get-text-property (point) 'gnus-alt-text)))
 375
 376 (defun gnus-html-browse-image ()
 377   "Browse the image under point."
 378   (interactive)
 379   (browse-url (get-text-property (point) 'image-url)))
 380
 381 (defun gnus-html-browse-url ()
 382   "Browse the image under point."
 383   (interactive)
 384   (let ((url (get-text-property (point) 'gnus-string)))
 385     (cond
 386      ((not url)
 387       (message "No link under point"))
 388      ((string-match "^mailto:" url)
 389       (gnus-url-mailto url))
 390      (t
 391       (browse-url url)))))
 392
 393 (defun gnus-html-schedule-image-fetching (buffer image)
 394   "Retrieve IMAGE, and place it into BUFFER on arrival."
 395   (gnus-message 8 "gnus-html-schedule-image-fetching: buffer %s, image %s"
 396                 buffer image)
 397   (if (fboundp 'url-queue-retrieve)
 398       (url-queue-retrieve (car image)
 399                           'gnus-html-image-fetched
 400                           (list buffer image) t t)
 401     (ignore-errors
 402       (url-retrieve (car image)
 403                     'gnus-html-image-fetched
 404                     (list buffer image)))))
 405
 406 (defun gnus-html-image-fetched (status buffer image)
 407   "Callback function called when image has been fetched."
 408   (unless (plist-get status :error)
 409     (when (and (or (search-forward "\n\n" nil t)
 410                    (search-forward "\r\n\r\n" nil t))
 411                (not (eobp)))
 412       (when gnus-html-image-automatic-caching
 413         (url-store-in-cache (current-buffer)))
 414       (when (buffer-live-p buffer)
 415         (let ((data (buffer-substring (point) (point-max))))
 416           (with-current-buffer buffer
 417             (let ((inhibit-read-only t))
 418               (gnus-html-put-image data (car image) (cadr image))))))))
 419   (kill-buffer (current-buffer)))
 420
 421 (defun gnus-html-get-image-data (url)
 422   "Get image data for URL.
 423 Return a string with image data."
 424   (with-temp-buffer
 425     (mm-disable-multibyte)
 426     (url-cache-extract (url-cache-create-filename url))
 427     (when (or (search-forward "\n\n" nil t)
 428               (search-forward "\r\n\r\n" nil t))
 429       (buffer-substring (point) (point-max)))))
 430
 431 (defun gnus-html-maximum-image-size ()
 432   "Return the maximum size of an image according to `gnus-max-image-proportion'."
 433   (let ((edges (gnus-window-inside-pixel-edges
 434                 (get-buffer-window (current-buffer)))))
 435     ;; (width . height)
 436     (cons
 437      ;; Aimed width
 438      (truncate
 439       (* gnus-max-image-proportion
 440          (- (nth 2 edges) (nth 0 edges))))
 441      ;; Aimed height
 442      (truncate (* gnus-max-image-proportion
 443                   (- (nth 3 edges) (nth 1 edges)))))))
 444
 445 ;; Behind display-graphic-p test.
 446 (declare-function image-size "image.c" (spec &optional pixels frame))
 447
 448 (defun gnus-html-put-image (data url &optional alt-text)
 449   "Put an image with DATA from URL and optional ALT-TEXT."
 450   (when (gnus-graphic-display-p)
 451     (let* ((start (text-property-any (point-min) (point-max)
 452                                      'image-url url))
 453            (end (when start
 454                   (next-single-property-change start 'image-url))))
 455       ;; Image found?
 456       (when start
 457         (let* ((image
 458                 (ignore-errors
 459                   (gnus-create-image data nil t)))
 460                (size (and image
 461                           (if (featurep 'xemacs)
 462                               (cons (glyph-width image) (glyph-height image))
 463                             (image-size image t)))))
 464           (save-excursion
 465             (goto-char start)
 466             (let ((alt-text (or alt-text
 467                                 (buffer-substring-no-properties start end)))
 468                   (inhibit-read-only t))
 469               (if (and image
 470                        ;; Kludge to avoid displaying 30x30 gif images, which
 471                        ;; seems to be a signal of a broken image.
 472                        (not (and (if (featurep 'xemacs)
 473                                      (glyphp image)
 474                                    (listp image))
 475                                  (eq (if (featurep 'xemacs)
 476                                          (let ((d (cdadar
 477                                                    (specifier-spec-list
 478                                                     (glyph-image image)))))
 479                                            (and (vectorp d)
 480                                                 (aref d 0)))
 481                                        (plist-get (cdr image) :type))
 482                                      'gif)
 483                                  (= (car size) 30)
 484                                  (= (cdr size) 30))))
 485                   ;; Good image, add it!
 486                   (let ((image (gnus-rescale-image image (gnus-html-maximum-image-size))))
 487                     (delete-region start end)
 488                     (gnus-put-image image alt-text 'external)
 489                     (widget-convert-button
 490                      'url-link start (point)
 491                      :help-echo alt-text
 492                      :keymap gnus-html-displayed-image-map
 493                      url)
 494                     (gnus-put-text-property start (point)
 495                                             'gnus-alt-text alt-text)
 496                     (when url
 497                       (gnus-add-text-properties
 498                        start (point)
 499                        `(image-url
 500                          ,url
 501                          image-displayer
 502                          (lambda (url start end)
 503                            (gnus-html-display-image url start end
 504                                                     ,alt-text)))))
 505                     (gnus-add-image 'external image)
 506                     t)
 507                 ;; Bad image, try to show something else
 508                 (when (fboundp 'find-image)
 509                   (delete-region start end)
 510                   (setq image (find-image
 511                                '((:type xpm :file "lock-broken.xpm"))))
 512                   (gnus-put-image image alt-text 'internal)
 513                   (gnus-add-image 'internal image))
 514                 nil))))))))
 515
 516 (defun gnus-html-image-url-blocked-p (url blocked-images)
 517   "Find out if URL is blocked by BLOCKED-IMAGES."
 518   (let ((ret (and blocked-images
 519                   (string-match blocked-images url))))
 520     (if ret
 521         (gnus-message 8 "gnus-html-image-url-blocked-p: %s blocked by regex %s"
 522                       url blocked-images)
 523       (gnus-message 9 "gnus-html-image-url-blocked-p: %s passes regex %s"
 524                     url blocked-images))
 525     ret))
 526
 527 ;;;###autoload
 528 (defun gnus-html-prefetch-images (summary)
 529   (when (buffer-live-p summary)
 530     (let (inhibit-images blocked-images)
 531       (with-current-buffer summary
 532         (setq inhibit-images gnus-inhibit-images
 533               blocked-images (gnus-blocked-images)))
 534       (save-match-data
 535         (while (re-search-forward "<img[^>]+src=[\"']\\(http[^\"']+\\)" nil t)
 536           (let ((url (gnus-html-encode-url
 537                       (mm-url-decode-entities-string (match-string 1)))))
 538             (unless (or inhibit-images
 539                         (gnus-html-image-url-blocked-p url blocked-images))
 540               (when (gnus-html-cache-expired url gnus-html-image-cache-ttl)
 541                 (gnus-html-schedule-image-fetching nil
 542                                                    (list url))))))))))
 543
 544 (provide 'gnus-html)
 545
 546 ;;; gnus-html.el ends here