lisp/net/shr.el

   1 ;;; shr.el --- Simple HTML Renderer
   2
   3 ;; Copyright (C) 2010-2015 Free Software Foundation, Inc.
   4
   5 ;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
   6 ;; Keywords: html
   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 ;; This package takes a HTML parse tree (as provided by
  26 ;; libxml-parse-html-region) and renders it in the current buffer.  It
  27 ;; does not do CSS, JavaScript or anything advanced: It's geared
  28 ;; towards rendering typical short snippets of HTML, like what you'd
  29 ;; find in HTML email and the like.
  30
  31 ;;; Code:
  32
  33 (eval-when-compile (require 'cl))
  34 (eval-when-compile (require 'url))      ;For url-filename's setf handler.
  35 (require 'browse-url)
  36 (require 'subr-x)
  37 (require 'dom)
  38
  39 (defgroup shr nil
  40   "Simple HTML Renderer"
  41   :version "25.1"
  42   :group 'web)
  43
  44 (defcustom shr-max-image-proportion 0.9
  45   "How big pictures displayed are in relation to the window they're in.
  46 A value of 0.7 means that they are allowed to take up 70% of the
  47 width and height of the window.  If they are larger than this,
  48 and Emacs supports it, then the images will be rescaled down to
  49 fit these criteria."
  50   :version "24.1"
  51   :group 'shr
  52   :type 'float)
  53
  54 (defcustom shr-blocked-images nil
  55   "Images that have URLs matching this regexp will be blocked."
  56   :version "24.1"
  57   :group 'shr
  58   :type '(choice (const nil) regexp))
  59
  60 (defcustom shr-use-fonts t
  61   "If non-nil, use proportional fonts for text."
  62   :version "25.1"
  63   :group 'shr
  64   :type 'boolean)
  65
  66 (defcustom shr-table-horizontal-line nil
  67   "Character used to draw horizontal table lines.
  68 If nil, don't draw horizontal table lines."
  69   :group 'shr
  70   :type '(choice (const nil) character))
  71
  72 (defcustom shr-table-vertical-line ?\s
  73   "Character used to draw vertical table lines."
  74   :group 'shr
  75   :type 'character)
  76
  77 (defcustom shr-table-corner ?\s
  78   "Character used to draw table corners."
  79   :group 'shr
  80   :type 'character)
  81
  82 (defcustom shr-hr-line ?-
  83   "Character used to draw hr lines."
  84   :group 'shr
  85   :type 'character)
  86
  87 (defcustom shr-width nil
  88   "Frame width to use for rendering.
  89 May either be an integer specifying a fixed width in characters,
  90 or nil, meaning that the full width of the window should be
  91 used."
  92   :version "25.1"
  93   :type '(choice (integer :tag "Fixed width in characters")
  94                  (const   :tag "Use the width of the window" nil))
  95   :group 'shr)
  96
  97 (defcustom shr-bullet "* "
  98   "Bullet used for unordered lists.
  99 Alternative suggestions are:
 100 - \"  \"
 101 - \"  \""
 102   :version "24.4"
 103   :type 'string
 104   :group 'shr)
 105
 106 (defcustom shr-external-browser 'browse-url-default-browser
 107   "Function used to launch an external browser."
 108   :version "24.4"
 109   :group 'shr
 110   :type 'function)
 111
 112 (defcustom shr-image-animate t
 113   "Non nil means that images that can be animated will be."
 114   :version "24.4"
 115   :group 'shr
 116   :type 'boolean)
 117
 118 (defvar shr-content-function nil
 119   "If bound, this should be a function that will return the content.
 120 This is used for cid: URLs, and the function is called with the
 121 cid: URL as the argument.")
 122
 123 (defvar shr-put-image-function 'shr-put-image
 124   "Function called to put image and alt string.")
 125
 126 (defface shr-strike-through '((t (:strike-through t)))
 127   "Font for <s> elements."
 128   :group 'shr)
 129
 130 (defface shr-link
 131   '((t (:inherit link)))
 132   "Font for link elements."
 133   :group 'shr)
 134
 135 (defvar shr-inhibit-images nil
 136   "If non-nil, inhibit loading images.")
 137
 138 ;;; Internal variables.
 139
 140 (defvar shr-folding-mode nil)
 141 (defvar shr-start nil)
 142 (defvar shr-indentation 0)
 143 (defvar shr-internal-width nil)
 144 (defvar shr-list-mode nil)
 145 (defvar shr-content-cache nil)
 146 (defvar shr-kinsoku-shorten nil)
 147 (defvar shr-table-depth 0)
 148 (defvar shr-stylesheet nil)
 149 (defvar shr-base nil)
 150 (defvar shr-depth 0)
 151 (defvar shr-warning nil)
 152 (defvar shr-ignore-cache nil)
 153 (defvar shr-external-rendering-functions nil)
 154 (defvar shr-target-id nil)
 155 (defvar shr-table-separator-length 1)
 156 (defvar shr-table-separator-pixel-width 0)
 157 (defvar shr-table-id nil)
 158 (defvar shr-current-font nil)
 159 (defvar shr-internal-bullet nil)
 160
 161 (defvar shr-map
 162   (let ((map (make-sparse-keymap)))
 163     (define-key map "a" 'shr-show-alt-text)
 164     (define-key map "i" 'shr-browse-image)
 165     (define-key map "z" 'shr-zoom-image)
 166     (define-key map [?\t] 'shr-next-link)
 167     (define-key map [?\M-\t] 'shr-previous-link)
 168     (define-key map [follow-link] 'mouse-face)
 169     (define-key map [mouse-2] 'shr-browse-url)
 170     (define-key map "I" 'shr-insert-image)
 171     (define-key map "w" 'shr-copy-url)
 172     (define-key map "u" 'shr-copy-url)
 173     (define-key map "v" 'shr-browse-url)
 174     (define-key map "o" 'shr-save-contents)
 175     (define-key map "\r" 'shr-browse-url)
 176     map))
 177
 178 ;; Public functions and commands.
 179 (declare-function libxml-parse-html-region "xml.c"
 180                   (start end &optional base-url discard-comments))
 181
 182 (defun shr-render-buffer (buffer)
 183   "Display the HTML rendering of the current buffer."
 184   (interactive (list (current-buffer)))
 185   (or (fboundp 'libxml-parse-html-region)
 186       (error "This function requires Emacs to be compiled with libxml2"))
 187   (pop-to-buffer "*html*")
 188   (erase-buffer)
 189   (shr-insert-document
 190    (with-current-buffer buffer
 191      (libxml-parse-html-region (point-min) (point-max))))
 192   (goto-char (point-min)))
 193
 194 ;;;###autoload
 195 (defun shr-render-region (begin end &optional buffer)
 196   "Display the HTML rendering of the region between BEGIN and END."
 197   (interactive "r")
 198   (unless (fboundp 'libxml-parse-html-region)
 199     (error "This function requires Emacs to be compiled with libxml2"))
 200   (with-current-buffer (or buffer (current-buffer))
 201     (let ((dom (libxml-parse-html-region begin end)))
 202       (delete-region begin end)
 203       (goto-char begin)
 204       (shr-insert-document dom))))
 205
 206 ;;;###autoload
 207 (defun shr-insert-document (dom)
 208   "Render the parsed document DOM into the current buffer.
 209 DOM should be a parse tree as generated by
 210 `libxml-parse-html-region' or similar."
 211   (setq shr-content-cache nil)
 212   (let ((start (point))
 213         (shr-start nil)
 214         (shr-base nil)
 215         (shr-depth 0)
 216         (shr-table-id 0)
 217         (shr-warning nil)
 218         (shr-table-separator-pixel-width (shr-string-pixel-width "-"))
 219         (shr-internal-bullet (cons shr-bullet
 220                                    (shr-string-pixel-width shr-bullet)))
 221         (shr-internal-width (or (and shr-width
 222                                      (if (not shr-use-fonts)
 223                                          shr-width
 224                                        (* shr-width (frame-char-width))))
 225                                 ;; We need to adjust the available
 226                                 ;; width for when the user disables
 227                                 ;; the fringes, which will cause the
 228                                 ;; display engine usurp one column for
 229                                 ;; the continuation glyph.
 230                                 (if (not shr-use-fonts)
 231                                     (- (window-body-width) 1
 232                                        (if (and (null shr-width)
 233                                                 (or (zerop
 234                                                      (fringe-columns 'right))
 235                                                     (zerop
 236                                                      (fringe-columns 'left))))
 237                                            0
 238                                          1))
 239                                   (- (window-body-width nil t)
 240                                      (* 2 (frame-char-width))
 241                                      (if (and (null shr-width)
 242                                               (or (zerop
 243                                                    (fringe-columns 'right))
 244                                                   (zerop
 245                                                    (fringe-columns 'left))))
 246                                          (* (frame-char-width) 2)
 247                                        0))))))
 248     (shr-descend dom)
 249     (shr-fill-lines start (point))
 250     (shr-remove-trailing-whitespace start (point))
 251     (when shr-warning
 252       (message "%s" shr-warning))))
 253
 254 (defun shr-remove-trailing-whitespace (start end)
 255   (let ((width (window-width)))
 256     (save-restriction
 257       (narrow-to-region start end)
 258       (goto-char start)
 259       (while (not (eobp))
 260         (end-of-line)
 261         (when (> (shr-previous-newline-padding-width (current-column)) width)
 262           (dolist (overlay (overlays-at (point)))
 263             (when (overlay-get overlay 'before-string)
 264               (overlay-put overlay 'before-string nil))))
 265         (forward-line 1)))))
 266
 267 (defun shr-copy-url (&optional image-url)
 268   "Copy the URL under point to the kill ring.
 269 If IMAGE-URL (the prefix) is non-nil, or there is no link under
 270 point, but there is an image under point then copy the URL of the
 271 image under point instead.
 272 If called twice, then try to fetch the URL and see whether it
 273 redirects somewhere else."
 274   (interactive "P")
 275   (let ((url (or (get-text-property (point) 'shr-url)
 276                  (get-text-property (point) 'image-url))))
 277     (cond
 278      ((not url)
 279       (message "No URL under point"))
 280      ;; Resolve redirected URLs.
 281      ((equal url (car kill-ring))
 282       (url-retrieve
 283        url
 284        (lambda (a)
 285          (when (and (consp a)
 286                     (eq (car a) :redirect))
 287            (with-temp-buffer
 288              (insert (cadr a))
 289              (goto-char (point-min))
 290              ;; Remove common tracking junk from the URL.
 291              (when (re-search-forward ".utm_.*" nil t)
 292                (replace-match "" t t))
 293              (message "Copied %s" (buffer-string))
 294              (copy-region-as-kill (point-min) (point-max)))))
 295        nil t))
 296      ;; Copy the URL to the kill ring.
 297      (t
 298       (with-temp-buffer
 299         (insert (url-encode-url url))
 300         (copy-region-as-kill (point-min) (point-max))
 301         (message "Copied %s" (buffer-string)))))))
 302
 303 (defun shr-next-link ()
 304   "Skip to the next link."
 305   (interactive)
 306   (let ((skip (text-property-any (point) (point-max) 'help-echo nil)))
 307     (if (or (eobp)
 308             (not (setq skip (text-property-not-all skip (point-max)
 309                                                    'help-echo nil))))
 310         (message "No next link")
 311       (goto-char skip)
 312       (message "%s" (get-text-property (point) 'help-echo)))))
 313
 314 (defun shr-previous-link ()
 315   "Skip to the previous link."
 316   (interactive)
 317   (let ((start (point))
 318         (found nil))
 319     ;; Skip past the current link.
 320     (while (and (not (bobp))
 321                 (get-text-property (point) 'help-echo))
 322       (forward-char -1))
 323     ;; Find the previous link.
 324     (while (and (not (bobp))
 325                 (not (setq found (get-text-property (point) 'help-echo))))
 326       (forward-char -1))
 327     (if (not found)
 328         (progn
 329           (message "No previous link")
 330           (goto-char start))
 331       ;; Put point at the start of the link.
 332       (while (and (not (bobp))
 333                   (get-text-property (point) 'help-echo))
 334         (forward-char -1))
 335       (forward-char 1)
 336       (message "%s" (get-text-property (point) 'help-echo)))))
 337
 338 (defun shr-show-alt-text ()
 339   "Show the ALT text of the image under point."
 340   (interactive)
 341   (let ((text (get-text-property (point) 'shr-alt)))
 342     (if (not text)
 343         (message "No image under point")
 344       (message "%s" (shr-fill-text text)))))
 345
 346 (defun shr-browse-image (&optional copy-url)
 347   "Browse the image under point.
 348 If COPY-URL (the prefix if called interactively) is non-nil, copy
 349 the URL of the image to the kill buffer instead."
 350   (interactive "P")
 351   (let ((url (get-text-property (point) 'image-url)))
 352     (cond
 353      ((not url)
 354       (message "No image under point"))
 355      (copy-url
 356       (with-temp-buffer
 357         (insert url)
 358         (copy-region-as-kill (point-min) (point-max))
 359         (message "Copied %s" url)))
 360      (t
 361       (message "Browsing %s..." url)
 362       (browse-url url)))))
 363
 364 (defun shr-insert-image ()
 365   "Insert the image under point into the buffer."
 366   (interactive)
 367   (let ((url (get-text-property (point) 'image-url)))
 368     (if (not url)
 369         (message "No image under point")
 370       (message "Inserting %s..." url)
 371       (url-retrieve url 'shr-image-fetched
 372                     (list (current-buffer) (1- (point)) (point-marker))
 373                     t t))))
 374
 375 (defun shr-zoom-image ()
 376   "Toggle the image size.
 377 The size will be rotated between the default size, the original
 378 size, and full-buffer size."
 379   (interactive)
 380   (let ((url (get-text-property (point) 'image-url))
 381         (size (get-text-property (point) 'image-size))
 382         (buffer-read-only nil))
 383     (if (not url)
 384         (message "No image under point")
 385       ;; Delete the old picture.
 386       (while (get-text-property (point) 'image-url)
 387         (forward-char -1))
 388       (forward-char 1)
 389       (let ((start (point)))
 390         (while (get-text-property (point) 'image-url)
 391           (forward-char 1))
 392         (forward-char -1)
 393         (put-text-property start (point) 'display nil)
 394         (when (> (- (point) start) 2)
 395           (delete-region start (1- (point)))))
 396       (message "Inserting %s..." url)
 397       (url-retrieve url 'shr-image-fetched
 398                     (list (current-buffer) (1- (point)) (point-marker)
 399                           (list (cons 'size
 400                                       (cond ((or (eq size 'default)
 401                                                  (null size))
 402                                              'original)
 403                                             ((eq size 'original)
 404                                              'full)
 405                                             ((eq size 'full)
 406                                              'default)))))
 407                     t))))
 408
 409 ;;; Utility functions.
 410
 411 (defsubst shr-generic (dom)
 412   (dolist (sub (dom-children dom))
 413     (if (stringp sub)
 414         (shr-insert sub)
 415       (shr-descend sub))))
 416
 417 (defun shr-descend (dom)
 418   (let ((function
 419          (or
 420           ;; Allow other packages to override (or provide) rendering
 421           ;; of elements.
 422           (cdr (assq (dom-tag dom) shr-external-rendering-functions))
 423           (intern (concat "shr-tag-" (symbol-name (dom-tag dom))) obarray)))
 424         (style (dom-attr dom 'style))
 425         (shr-stylesheet shr-stylesheet)
 426         (shr-depth (1+ shr-depth))
 427         (start (point)))
 428     ;; shr uses about 12 frames per nested node.
 429     (if (> shr-depth (/ max-specpdl-size 12))
 430         (setq shr-warning "Too deeply nested to render properly; consider increasing `max-specpdl-size'")
 431       (when style
 432         (if (string-match "color\\|display\\|border-collapse" style)
 433             (setq shr-stylesheet (nconc (shr-parse-style style)
 434                                         shr-stylesheet))
 435           (setq style nil)))
 436       ;; If we have a display:none, then just ignore this part of the DOM.
 437       (unless (equal (cdr (assq 'display shr-stylesheet)) "none")
 438         (if (fboundp function)
 439             (funcall function dom)
 440           (shr-generic dom))
 441         (when (and shr-target-id
 442                    (equal (dom-attr dom 'id) shr-target-id))
 443           ;; If the element was empty, we don't have anything to put the
 444           ;; anchor on.  So just insert a dummy character.
 445           (when (= start (point))
 446             (insert "*"))
 447           (put-text-property start (1+ start) 'shr-target-id shr-target-id))
 448         ;; If style is set, then this node has set the color.
 449         (when style
 450           (shr-colorize-region
 451            start (point)
 452            (cdr (assq 'color shr-stylesheet))
 453            (cdr (assq 'background-color shr-stylesheet))))))))
 454
 455 (defun shr-fill-text (text)
 456   (if (zerop (length text))
 457       text
 458     (with-temp-buffer
 459       (let ((shr-indentation 0)
 460             (shr-start nil)
 461             (shr-internal-width (- (window-body-width nil t)
 462                                    (* 2 (frame-char-width))
 463                                    ;; Adjust the window width for when
 464                                    ;; the user disables the fringes,
 465                                    ;; which causes the display engine
 466                                    ;; usurp one coplumn for the
 467                                    ;; continuation glyph.
 468                                    (if (and (null shr-width)
 469                                             (or (zerop (fringe-columns 'right))
 470                                                 (zerop (fringe-columns 'left))))
 471                                        (* (frame-char-width) 2)
 472                                      0))))
 473         (shr-insert text)
 474         (buffer-string)))))
 475
 476 (define-inline shr-char-breakable-p (char)
 477   "Return non-nil if a line can be broken before and after CHAR."
 478   (inline-quote (aref fill-find-break-point-function-table ,char)))
 479 (define-inline shr-char-nospace-p (char)
 480   "Return non-nil if no space is required before and after CHAR."
 481   (inline-quote (aref fill-nospace-between-words-table ,char)))
 482
 483 ;; KINSOKU is a Japanese word meaning a rule that should not be violated.
 484 ;; In Emacs, it is a term used for characters, e.g. punctuation marks,
 485 ;; parentheses, and so on, that should not be placed in the beginning
 486 ;; of a line or the end of a line.
 487 (define-inline shr-char-kinsoku-bol-p (char)
 488   "Return non-nil if a line ought not to begin with CHAR."
 489   (inline-letevals (char)
 490     (inline-quote (and (not (eq ,char ?'))
 491                        (aref (char-category-set ,char) ?>)))))
 492 (define-inline shr-char-kinsoku-eol-p (char)
 493   "Return non-nil if a line ought not to end with CHAR."
 494   (inline-quote (aref (char-category-set ,char) ?<)))
 495 (unless (shr-char-kinsoku-bol-p (make-char 'japanese-jisx0208 33 35))
 496   (load "kinsoku" nil t))
 497
 498 (defun shr-pixel-column ()
 499   (if (not shr-use-fonts)
 500       (current-column)
 501     (if (not (get-buffer-window (current-buffer)))
 502         (save-window-excursion
 503           (set-window-buffer nil (current-buffer))
 504           (car (window-text-pixel-size nil (line-beginning-position) (point))))
 505       (car (window-text-pixel-size nil (line-beginning-position) (point))))))
 506
 507 (defun shr-pixel-region ()
 508   (- (shr-pixel-column)
 509      (save-excursion
 510        (goto-char (mark))
 511        (shr-pixel-column))))
 512
 513 (defun shr-string-pixel-width (string)
 514   (if (not shr-use-fonts)
 515       (length string)
 516     (with-temp-buffer
 517       (insert string)
 518       (shr-pixel-column))))
 519
 520 (defun shr-insert (text)
 521   (when (and (not (bolp))
 522              (get-text-property (1- (point)) 'image-url))
 523     (insert "\n"))
 524   (cond
 525    ((eq shr-folding-mode 'none)
 526     (let ((start (point)))
 527       (insert text)
 528       (save-restriction
 529         (narrow-to-region start (point))
 530         ;; Remove soft hyphens.
 531         (goto-char (point-min))
 532         (while (search-forward "" nil t)
 533           (replace-match "" t t))
 534         (goto-char (point-max)))))
 535    (t
 536     (let ((font-start (point)))
 537       (when (and (string-match "\\`[ \t\n\r ]" text)
 538                  (not (bolp))
 539                  (not (eq (char-after (1- (point))) ? )))
 540         (insert " "))
 541       (let ((start (point))
 542             (bolp (bolp)))
 543         (insert text)
 544         (save-restriction
 545           (narrow-to-region start (point))
 546           (goto-char start)
 547           (when (looking-at "[ \t\n\r ]+")
 548             (replace-match "" t t))
 549           (while (re-search-forward "[ \t\n\r ]+" nil t)
 550             (replace-match " " t t))
 551           ;; Remove soft hyphens.
 552           (goto-char (point-min))
 553           (while (search-forward "" nil t)
 554             (replace-match "" t t))
 555           (goto-char (point-max)))
 556         ;; We may have removed everything we inserted if if was just
 557         ;; spaces.
 558         (unless (= font-start (point))
 559           ;; Mark all lines that should possibly be folded afterwards.
 560           (when bolp
 561             (shr-mark-fill start))
 562           (when shr-use-fonts
 563             (put-text-property font-start (point)
 564                                'face
 565                                (or shr-current-font 'variable-pitch)))))))))
 566
 567 (defun shr-fill-lines (start end)
 568   (if (<= shr-internal-width 0)
 569       nil
 570     (save-restriction
 571       (narrow-to-region start end)
 572       (goto-char start)
 573       (when (get-text-property (point) 'shr-indentation)
 574         (shr-fill-line))
 575       (while (setq start (next-single-property-change start 'shr-indentation))
 576         (goto-char start)
 577         (when (bolp)
 578           (shr-fill-line)))
 579       (goto-char (point-max)))))
 580
 581 (defun shr-vertical-motion (column)
 582   (if (not shr-use-fonts)
 583       (move-to-column column)
 584     (unless (eolp)
 585       (forward-char 1))
 586     (vertical-motion (cons (/ column (frame-char-width)) 0))
 587     (unless (eolp)
 588       (forward-char 1))))
 589
 590 (defun shr-fill-line ()
 591   (let ((shr-indentation (get-text-property (point) 'shr-indentation))
 592         (continuation (get-text-property
 593                        (point) 'shr-continuation-indentation))
 594         start)
 595     (put-text-property (point) (1+ (point)) 'shr-indentation nil)
 596     (let ((face (get-text-property (point) 'face))
 597           (background-start (point)))
 598       (shr-indent)
 599       (when face
 600         (put-text-property background-start (point) 'face
 601                            `,(shr-face-background face))))
 602     (setq start (point))
 603     (setq shr-indentation (or continuation shr-indentation))
 604     (shr-vertical-motion shr-internal-width)
 605     (when (looking-at " $")
 606       (delete-region (point) (line-end-position)))
 607     (while (not (eolp))
 608       ;; We have to do some folding.  First find the first
 609       ;; previous point suitable for folding.
 610       (if (or (not (shr-find-fill-point (line-beginning-position)))
 611               (= (point) start))
 612           ;; We had unbreakable text (for this width), so just go to
 613           ;; the first space and carry on.
 614           (progn
 615             (beginning-of-line)
 616             (skip-chars-forward " ")
 617             (search-forward " " (line-end-position) 'move)))
 618       ;; Success; continue.
 619       (when (= (preceding-char) ?\s)
 620         (delete-char -1))
 621       (let ((face (get-text-property (point) 'face))
 622             (background-start (point)))
 623         (insert "\n")
 624         (shr-indent)
 625         (when face
 626           (put-text-property background-start (point) 'face
 627                              `,(shr-face-background face))))
 628       (setq start (point))
 629       (shr-vertical-motion shr-internal-width)
 630       (when (looking-at " $")
 631         (delete-region (point) (line-end-position))))))
 632
 633 (defun shr-find-fill-point (start)
 634   (let ((bp (point))
 635         (end (point))
 636         failed)
 637     (while (not (or (setq failed (<= (point) start))
 638                     (eq (preceding-char) ? )
 639                     (eq (following-char) ? )
 640                     (shr-char-breakable-p (preceding-char))
 641                     (shr-char-breakable-p (following-char))
 642                     (and (shr-char-kinsoku-bol-p (preceding-char))
 643                          (shr-char-breakable-p (following-char))
 644                          (not (shr-char-kinsoku-bol-p (following-char))))
 645                     (shr-char-kinsoku-eol-p (following-char))
 646                     (bolp)))
 647       (backward-char 1))
 648     (if failed
 649         ;; There's no breakable point, so we give it up.
 650         (let (found)
 651           (goto-char bp)
 652           ;; Don't overflow the window edge, even if
 653           ;; shr-kinsoku-shorten is nil.
 654           (unless (or shr-kinsoku-shorten (null shr-width))
 655             (while (setq found (re-search-forward
 656                                 "\\(\\c>\\)\\| \\|\\c<\\|\\c|"
 657                                 (line-end-position) 'move)))
 658             (if (and found
 659                      (not (match-beginning 1)))
 660                 (goto-char (match-beginning 0)))))
 661       (or
 662        (eolp)
 663        ;; Don't put kinsoku-bol characters at the beginning of a line,
 664        ;; or kinsoku-eol characters at the end of a line.
 665        (cond
 666         ;; Don't overflow the window edge, even if shr-kinsoku-shorten
 667         ;; is nil.
 668         ((or shr-kinsoku-shorten (null shr-width))
 669          (while (and (not (memq (preceding-char) (list ?\C-@ ?\n ? )))
 670                      (or (shr-char-kinsoku-eol-p (preceding-char))
 671                          (shr-char-kinsoku-bol-p (following-char))))
 672            (backward-char 1))
 673          (when (setq failed (<= (point) start))
 674            ;; There's no breakable point that doesn't violate kinsoku,
 675            ;; so we look for the second best position.
 676            (while (and (progn
 677                          (forward-char 1)
 678                          (<= (point) end))
 679                        (progn
 680                          (setq bp (point))
 681                          (shr-char-kinsoku-eol-p (following-char)))))
 682            (goto-char bp)))
 683         ((shr-char-kinsoku-eol-p (preceding-char))
 684          ;; Find backward the point where kinsoku-eol characters begin.
 685          (let ((count 4))
 686            (while
 687                (progn
 688                  (backward-char 1)
 689                  (and (> (setq count (1- count)) 0)
 690                       (not (memq (preceding-char) (list ?\C-@ ?\n ? )))
 691                       (or (shr-char-kinsoku-eol-p (preceding-char))
 692                           (shr-char-kinsoku-bol-p (following-char)))))))
 693          (when (setq failed (<= (point) start))
 694            ;; There's no breakable point that doesn't violate kinsoku,
 695            ;; so we go to the second best position.
 696            (if (looking-at "\\(\\c<+\\)\\c<")
 697                (goto-char (match-end 1))
 698              (forward-char 1))))
 699         ((shr-char-kinsoku-bol-p (following-char))
 700          ;; Find forward the point where kinsoku-bol characters end.
 701          (let ((count 4))
 702            (while (progn
 703                     (forward-char 1)
 704                     (and (>= (setq count (1- count)) 0)
 705                          (shr-char-kinsoku-bol-p (following-char))
 706                          (shr-char-breakable-p (following-char))))))))
 707        (when (eq (following-char) ? )
 708          (forward-char 1))))
 709     (not failed)))
 710
 711 (defun shr-parse-base (url)
 712   ;; Always chop off anchors.
 713   (when (string-match "#.*" url)
 714     (setq url (substring url 0 (match-beginning 0))))
 715   ;; NB: <base href="" > URI may itself be relative to the document s URI
 716   (setq url (shr-expand-url url))
 717   (let* ((parsed (url-generic-parse-url url))
 718          (local (url-filename parsed)))
 719     (setf (url-filename parsed) "")
 720     ;; Chop off the bit after the last slash.
 721     (when (string-match "\\`\\(.*/\\)[^/]+\\'" local)
 722       (setq local (match-string 1 local)))
 723     ;; Always make the local bit end with a slash.
 724     (when (and (not (zerop (length local)))
 725                (not (eq (aref local (1- (length local))) ?/)))
 726       (setq local (concat local "/")))
 727     (list (url-recreate-url parsed)
 728           local
 729           (url-type parsed)
 730           url)))
 731
 732 (autoload 'url-expand-file-name "url-expand")
 733
 734 ;; FIXME This needs some tests writing.
 735 ;; Does it even need to exist, given that url-expand-file-name does?
 736 (defun shr-expand-url (url &optional base)
 737   (setq base
 738         (if base
 739             ;; shr-parse-base should never call this with non-nil base!
 740             (shr-parse-base base)
 741           ;; Bound by the parser.
 742           shr-base))
 743   (when (zerop (length url))
 744     (setq url nil))
 745   ;; Strip leading whitespace
 746   (and url (string-match "\\`\\s-+" url)
 747        (setq url (substring url (match-end 0))))
 748   (cond ((or (not url)
 749              (not base)
 750              (string-match "\\`[a-z]*:" url))
 751          ;; Absolute or empty URI
 752          (or url (nth 3 base)))
 753         ((eq (aref url 0) ?/)
 754          (if (and (> (length url) 1)
 755                   (eq (aref url 1) ?/))
 756              ;; //host...; just use the protocol
 757              (concat (nth 2 base) ":" url)
 758            ;; Just use the host name part.
 759            (concat (car base) url)))
 760         ((eq (aref url 0) ?#)
 761          ;; A link to an anchor.
 762          (concat (nth 3 base) url))
 763         (t
 764          ;; Totally relative.
 765          (url-expand-file-name url (concat (car base) (cadr base))))))
 766
 767 (defun shr-ensure-newline ()
 768   (unless (zerop (current-column))
 769     (insert "\n")))
 770
 771 (defun shr-ensure-paragraph ()
 772   (unless (bobp)
 773     (let ((prefix (get-text-property (line-beginning-position)
 774                                      'shr-prefix-length)))
 775       (cond
 776        ((and (bolp)
 777              (save-excursion
 778                (forward-line -1)
 779                (looking-at " *$")))
 780         ;; We're already at a new paragraph; do nothing.
 781         )
 782        ((and prefix
 783              (= prefix (- (point) (line-beginning-position))))
 784         ;; Do nothing; we're at the start of a <li>.
 785         )
 786        ((save-excursion
 787           (beginning-of-line)
 788           ;; If the current line is totally blank, and doesn't even
 789           ;; have any face properties set, then delete the blank
 790           ;; space.
 791           (and (looking-at " *$")
 792                (not (get-text-property (point) 'face))
 793                (not (= (next-single-property-change (point) 'face nil
 794                                                     (line-end-position))
 795                        (line-end-position)))))
 796         (delete-region (match-beginning 0) (match-end 0)))
 797        (t
 798         (insert "\n\n"))))))
 799
 800 (defun shr-indent ()
 801   (when (> shr-indentation 0)
 802     (insert
 803      (if (not shr-use-fonts)
 804          (make-string shr-indentation ?\s)
 805        (propertize " "
 806                    'display
 807                    `(space :width (,shr-indentation)))))))
 808
 809 (defun shr-fontize-dom (dom &rest types)
 810   (let ((start (point)))
 811     (shr-generic dom)
 812     (dolist (type types)
 813       (shr-add-font start (point) type))))
 814
 815 ;; Add face to the region, but avoid putting the font properties on
 816 ;; blank text at the start of the line, and the newline at the end, to
 817 ;; avoid ugliness.
 818 (defun shr-add-font (start end type)
 819   (save-excursion
 820     (goto-char start)
 821     (while (< (point) end)
 822       (when (bolp)
 823         (skip-chars-forward " "))
 824       (add-face-text-property (point) (min (line-end-position) end) type t)
 825       (if (< (line-end-position) end)
 826           (forward-line 1)
 827         (goto-char end)))))
 828
 829 (defun shr-mouse-browse-url (ev)
 830   "Browse the URL under the mouse cursor."
 831   (interactive "e")
 832   (mouse-set-point ev)
 833   (shr-browse-url))
 834
 835 (defun shr-browse-url (&optional external mouse-event)
 836   "Browse the URL under point.
 837 If EXTERNAL, browse the URL using `shr-external-browser'."
 838   (interactive (list current-prefix-arg last-nonmenu-event))
 839   (mouse-set-point mouse-event)
 840   (let ((url (get-text-property (point) 'shr-url)))
 841     (cond
 842      ((not url)
 843       (message "No link under point"))
 844      ((string-match "^mailto:" url)
 845       (browse-url-mail url))
 846      (t
 847       (if external
 848           (funcall shr-external-browser url)
 849         (browse-url url))))))
 850
 851 (defun shr-save-contents (directory)
 852   "Save the contents from URL in a file."
 853   (interactive "DSave contents of URL to directory: ")
 854   (let ((url (get-text-property (point) 'shr-url)))
 855     (if (not url)
 856         (message "No link under point")
 857       (url-retrieve (shr-encode-url url)
 858                     'shr-store-contents (list url directory)
 859                     nil t))))
 860
 861 (defun shr-store-contents (status url directory)
 862   (unless (plist-get status :error)
 863     (when (or (search-forward "\n\n" nil t)
 864               (search-forward "\r\n\r\n" nil t))
 865       (write-region (point) (point-max)
 866                     (expand-file-name (file-name-nondirectory url)
 867                                       directory)))))
 868
 869 (defun shr-image-fetched (status buffer start end &optional flags)
 870   (let ((image-buffer (current-buffer)))
 871     (when (and (buffer-name buffer)
 872                (not (plist-get status :error)))
 873       (url-store-in-cache image-buffer)
 874       (when (or (search-forward "\n\n" nil t)
 875                 (search-forward "\r\n\r\n" nil t))
 876         (let ((data (shr-parse-image-data)))
 877           (with-current-buffer buffer
 878             (save-excursion
 879               (let ((alt (buffer-substring start end))
 880                     (properties (text-properties-at start))
 881                     (inhibit-read-only t))
 882                 (delete-region start end)
 883                 (goto-char start)
 884                 (funcall shr-put-image-function data alt flags)
 885                 (while properties
 886                   (let ((type (pop properties))
 887                         (value (pop properties)))
 888                     (unless (memq type '(display image-size))
 889                       (put-text-property start (point) type value))))))))))
 890     (kill-buffer image-buffer)))
 891
 892 (defun shr-image-from-data (data)
 893   "Return an image from the data: URI content DATA."
 894   (when (string-match
 895          "\\(\\([^/;,]+\\(/[^;,]+\\)?\\)\\(;[^;,]+\\)*\\)?,\\(.*\\)"
 896          data)
 897     (let ((param (match-string 4 data))
 898           (payload (url-unhex-string (match-string 5 data))))
 899       (when (string-match "^.*\\(;[ \t]*base64\\)$" param)
 900         (setq payload (base64-decode-string payload)))
 901       payload)))
 902
 903 ;; Behind display-graphic-p test.
 904 (declare-function image-size "image.c" (spec &optional pixels frame))
 905 (declare-function image-animate "image" (image &optional index limit))
 906
 907 (defun shr-put-image (spec alt &optional flags)
 908   "Insert image SPEC with a string ALT.  Return image.
 909 SPEC is either an image data blob, or a list where the first
 910 element is the data blob and the second element is the content-type."
 911   (if (display-graphic-p)
 912       (let* ((size (cdr (assq 'size flags)))
 913              (data (if (consp spec)
 914                        (car spec)
 915                      spec))
 916              (content-type (and (consp spec)
 917                                 (cadr spec)))
 918              (start (point))
 919              (image (cond
 920                      ((eq size 'original)
 921                       (create-image data nil t :ascent 100
 922                                     :format content-type))
 923                      ((eq content-type 'image/svg+xml)
 924                       (create-image data 'svg t :ascent 100))
 925                      ((eq size 'full)
 926                       (ignore-errors
 927                         (shr-rescale-image data content-type)))
 928                      (t
 929                       (ignore-errors
 930                         (shr-rescale-image data content-type))))))
 931         (when image
 932           ;; When inserting big-ish pictures, put them at the
 933           ;; beginning of the line.
 934           (when (and (> (current-column) 0)
 935                      (> (car (image-size image t)) 400))
 936             (insert "\n"))
 937           (if (eq size 'original)
 938               (insert-sliced-image image (or alt "*") nil 20 1)
 939             (insert-image image (or alt "*")))
 940           (put-text-property start (point) 'image-size size)
 941           (when (and shr-image-animate
 942                      (cond ((fboundp 'image-multi-frame-p)
 943                        ;; Only animate multi-frame things that specify a
 944                        ;; delay; eg animated gifs as opposed to
 945                        ;; multi-page tiffs.  FIXME?
 946                             (cdr (image-multi-frame-p image)))
 947                            ((fboundp 'image-animated-p)
 948                             (image-animated-p image))))
 949             (image-animate image nil 60)))
 950         image)
 951     (insert alt)))
 952
 953 (defun shr-rescale-image (data &optional content-type)
 954   "Rescale DATA, if too big, to fit the current buffer."
 955   (if (not (and (fboundp 'imagemagick-types)
 956                 (get-buffer-window (current-buffer))))
 957       (create-image data nil t :ascent 100)
 958     (let ((edges (window-inside-pixel-edges
 959                   (get-buffer-window (current-buffer)))))
 960       (create-image
 961        data 'imagemagick t
 962        :ascent 100
 963        :max-width (truncate (* shr-max-image-proportion
 964                                (- (nth 2 edges) (nth 0 edges))))
 965        :max-height (truncate (* shr-max-image-proportion
 966                                 (- (nth 3 edges) (nth 1 edges))))
 967        :format content-type))))
 968
 969 ;; url-cache-extract autoloads url-cache.
 970 (declare-function url-cache-create-filename "url-cache" (url))
 971 (autoload 'mm-disable-multibyte "mm-util")
 972 (autoload 'browse-url-mail "browse-url")
 973
 974 (defun shr-get-image-data (url)
 975   "Get image data for URL.
 976 Return a string with image data."
 977   (with-temp-buffer
 978     (mm-disable-multibyte)
 979     (when (ignore-errors
 980             (url-cache-extract (url-cache-create-filename (shr-encode-url url)))
 981             t)
 982       (when (or (search-forward "\n\n" nil t)
 983                 (search-forward "\r\n\r\n" nil t))
 984         (shr-parse-image-data)))))
 985
 986 (declare-function libxml-parse-xml-region "xml.c"
 987                   (start end &optional base-url discard-comments))
 988
 989 (defun shr-parse-image-data ()
 990   (let ((data (buffer-substring (point) (point-max)))
 991         (content-type
 992          (save-excursion
 993            (save-restriction
 994              (narrow-to-region (point-min) (point))
 995              (let ((content-type (mail-fetch-field "content-type")))
 996                (and content-type
 997                     ;; Remove any comments in the type string.
 998                     (intern (replace-regexp-in-string ";.*" "" content-type)
 999                             obarray)))))))
1000     ;; SVG images may contain references to further images that we may
1001     ;; want to block.  So special-case these by parsing the XML data
1002     ;; and remove the blocked bits.
1003     (when (eq content-type 'image/svg+xml)
1004       (setq data
1005             (shr-dom-to-xml
1006              (libxml-parse-xml-region (point) (point-max)))))
1007     (list data content-type)))
1008
1009 (defun shr-image-displayer (content-function)
1010   "Return a function to display an image.
1011 CONTENT-FUNCTION is a function to retrieve an image for a cid url that
1012 is an argument.  The function to be returned takes three arguments URL,
1013 START, and END.  Note that START and END should be markers."
1014   `(lambda (url start end)
1015      (when url
1016        (if (string-match "\\`cid:" url)
1017            ,(when content-function
1018               `(let ((image (funcall ,content-function
1019                                      (substring url (match-end 0)))))
1020                  (when image
1021                    (goto-char start)
1022                    (funcall shr-put-image-function
1023                             image (buffer-substring start end))
1024                    (delete-region (point) end))))
1025          (url-retrieve url 'shr-image-fetched
1026                        (list (current-buffer) start end)
1027                        t t)))))
1028
1029 (defun shr-heading (dom &rest types)
1030   (shr-ensure-paragraph)
1031   (apply #'shr-fontize-dom dom types)
1032   (shr-ensure-paragraph))
1033
1034 (defun shr-urlify (start url &optional title)
1035   (shr-add-font start (point) 'shr-link)
1036   (add-text-properties
1037    start (point)
1038    (list 'shr-url url
1039          'help-echo (let ((iri (or (ignore-errors
1040                                      (decode-coding-string
1041                                       (url-unhex-string url)
1042                                       'utf-8 t))
1043                                    url)))
1044                       (if title (format "%s (%s)" iri title) iri))
1045          'follow-link t
1046          'mouse-face 'highlight
1047          'keymap shr-map)))
1048
1049 (defun shr-encode-url (url)
1050   "Encode URL."
1051   (browse-url-url-encode-chars url "[)$ ]"))
1052
1053 (autoload 'shr-color-visible "shr-color")
1054 (autoload 'shr-color->hexadecimal "shr-color")
1055
1056 (defun shr-color-check (fg bg)
1057   "Check that FG is visible on BG.
1058 Returns (fg bg) with corrected values.
1059 Returns nil if the colors that would be used are the default
1060 ones, in case fg and bg are nil."
1061   (when (or fg bg)
1062     (let ((fixed (cond ((null fg) 'fg)
1063                        ((null bg) 'bg))))
1064       ;; Convert colors to hexadecimal, or set them to default.
1065       (let ((fg (or (shr-color->hexadecimal fg)
1066                     (frame-parameter nil 'foreground-color)))
1067             (bg (or (shr-color->hexadecimal bg)
1068                     (frame-parameter nil 'background-color))))
1069         (cond ((eq fixed 'bg)
1070                ;; Only return the new fg
1071                (list nil (cadr (shr-color-visible bg fg t))))
1072               ((eq fixed 'fg)
1073                ;; Invert args and results and return only the new bg
1074                (list (cadr (shr-color-visible fg bg t)) nil))
1075               (t
1076                (shr-color-visible bg fg)))))))
1077
1078 (defun shr-colorize-region (start end fg &optional bg)
1079   (when (and (or fg bg) (>= (display-color-cells) 88))
1080     (let ((new-colors (shr-color-check fg bg)))
1081       (when new-colors
1082         (when fg
1083           (add-face-text-property start end
1084                                   (list :foreground (cadr new-colors))
1085                                   t))
1086         (when bg
1087           (add-face-text-property start end
1088                                   (list :background (car new-colors))
1089                                   t)))
1090       new-colors)))
1091
1092 (defun shr-previous-newline-padding-width (width)
1093   (let ((overlays (overlays-at (point)))
1094         (previous-width 0))
1095     (if (null overlays)
1096         width
1097       (dolist (overlay overlays)
1098         (setq previous-width
1099               (+ previous-width
1100                  (length (plist-get (overlay-properties overlay)
1101                                     'before-string)))))
1102       (+ width previous-width))))
1103
1104 ;;; Tag-specific rendering rules.
1105
1106 (defun shr-tag-body (dom)
1107   (let* ((start (point))
1108          (fgcolor (or (dom-attr dom 'fgcolor) (dom-attr dom 'text)))
1109          (bgcolor (dom-attr dom 'bgcolor))
1110          (shr-stylesheet (list (cons 'color fgcolor)
1111                                (cons 'background-color bgcolor))))
1112     (shr-generic dom)
1113     (shr-colorize-region start (point) fgcolor bgcolor)))
1114
1115 (defun shr-tag-style (_dom)
1116   )
1117
1118 (defun shr-tag-script (_dom)
1119   )
1120
1121 (defun shr-tag-comment (_dom)
1122   )
1123
1124 (defun shr-dom-to-xml (dom)
1125   (with-temp-buffer
1126     (shr-dom-print dom)
1127     (buffer-string)))
1128
1129 (defun shr-dom-print (dom)
1130   "Convert DOM into a string containing the xml representation."
1131   (insert (format "<%s" (dom-tag dom)))
1132   (dolist (attr (dom-attributes dom))
1133     ;; Ignore attributes that start with a colon because they are
1134     ;; private elements.
1135     (unless (= (aref (format "%s" (car attr)) 0) ?:)
1136       (insert (format " %s=\"%s\"" (car attr) (cdr attr)))))
1137   (insert ">")
1138   (let (url)
1139     (dolist (elem (dom-children dom))
1140       (cond
1141        ((stringp elem)
1142         (insert elem))
1143        ((eq (dom-tag elem) 'comment)
1144         )
1145        ((or (not (eq (dom-tag elem) 'image))
1146             ;; Filter out blocked elements inside the SVG image.
1147             (not (setq url (dom-attr elem ':xlink:href)))
1148             (not shr-blocked-images)
1149             (not (string-match shr-blocked-images url)))
1150         (insert " ")
1151         (shr-dom-print elem)))))
1152   (insert (format "</%s>" (dom-tag dom))))
1153
1154 (defun shr-tag-svg (dom)
1155   (when (and (image-type-available-p 'svg)
1156              (not shr-inhibit-images))
1157     (funcall shr-put-image-function (list (shr-dom-to-xml dom) 'image/svg+xml)
1158              "SVG Image")))
1159
1160 (defun shr-tag-sup (dom)
1161   (let ((start (point)))
1162     (shr-generic dom)
1163     (put-text-property start (point) 'display '(raise 0.5))))
1164
1165 (defun shr-tag-sub (dom)
1166   (let ((start (point)))
1167     (shr-generic dom)
1168     (put-text-property start (point) 'display '(raise -0.5))))
1169
1170 (defun shr-tag-label (dom)
1171   (shr-generic dom)
1172   (shr-ensure-paragraph))
1173
1174 (defun shr-tag-p (dom)
1175   (shr-ensure-paragraph)
1176   (shr-generic dom)
1177   (shr-ensure-paragraph))
1178
1179 (defun shr-tag-div (dom)
1180   (shr-ensure-newline)
1181   (shr-generic dom)
1182   (shr-ensure-newline))
1183
1184 (defun shr-tag-s (dom)
1185   (shr-fontize-dom dom 'shr-strike-through))
1186
1187 (defun shr-tag-del (dom)
1188   (shr-fontize-dom dom 'shr-strike-through))
1189
1190 (defun shr-tag-b (dom)
1191   (shr-fontize-dom dom 'bold))
1192
1193 (defun shr-tag-i (dom)
1194   (shr-fontize-dom dom 'italic))
1195
1196 (defun shr-tag-em (dom)
1197   (shr-fontize-dom dom 'italic))
1198
1199 (defun shr-tag-strong (dom)
1200   (shr-fontize-dom dom 'bold))
1201
1202 (defun shr-tag-u (dom)
1203   (shr-fontize-dom dom 'underline))
1204
1205 (defun shr-tag-tt (dom)
1206   (let ((shr-current-font 'default))
1207     (shr-generic dom)))
1208
1209 (defun shr-parse-style (style)
1210   (when style
1211     (save-match-data
1212       (when (string-match "\n" style)
1213         (setq style (replace-match " " t t style))))
1214     (let ((plist nil))
1215       (dolist (elem (split-string style ";"))
1216         (when elem
1217           (setq elem (split-string elem ":"))
1218           (when (and (car elem)
1219                      (cadr elem))
1220             (let ((name (replace-regexp-in-string "^ +\\| +$" "" (car elem)))
1221                   (value (replace-regexp-in-string "^ +\\| +$" "" (cadr elem))))
1222               (when (string-match " *!important\\'" value)
1223                 (setq value (substring value 0 (match-beginning 0))))
1224               (unless (equal value "inherit")
1225                 (push (cons (intern name obarray)
1226                             value)
1227                       plist))))))
1228       plist)))
1229
1230 (defun shr-tag-base (dom)
1231   (when-let (base (dom-attr dom 'href))
1232     (setq shr-base (shr-parse-base base)))
1233   (shr-generic dom))
1234
1235 (defun shr-tag-a (dom)
1236   (let ((url (dom-attr dom 'href))
1237         (title (dom-attr dom 'title))
1238         (start (point))
1239         shr-start)
1240     (shr-generic dom)
1241     (when (and shr-target-id
1242                (equal (dom-attr dom 'name) shr-target-id))
1243       ;; We have a zero-length <a name="foo"> element, so just
1244       ;; insert...  something.
1245       (when (= start (point))
1246         (shr-ensure-newline)
1247         (insert " "))
1248       (put-text-property start (1+ start) 'shr-target-id shr-target-id))
1249     (when url
1250       (shr-urlify (or shr-start start) (shr-expand-url url) title))))
1251
1252 (defun shr-tag-object (dom)
1253   (unless shr-inhibit-images
1254     (let ((start (point))
1255           url multimedia image)
1256       (when-let (type (dom-attr dom 'type))
1257         (when (string-match "\\`image/svg" type)
1258           (setq url (dom-attr dom 'data)
1259                 image t)))
1260       (dolist (child (dom-non-text-children dom))
1261         (cond
1262          ((eq (dom-tag child) 'embed)
1263           (setq url (or url (dom-attr child 'src))
1264                 multimedia t))
1265          ((and (eq (dom-tag child) 'param)
1266                (equal (dom-attr child 'name) "movie"))
1267           (setq url (or url (dom-attr child 'value))
1268                 multimedia t))))
1269       (when url
1270         (cond
1271          (image
1272           (shr-tag-img dom url)
1273           (setq dom nil))
1274          (multimedia
1275           (shr-insert " [multimedia] ")
1276           (shr-urlify start (shr-expand-url url)))))
1277       (when dom
1278         (shr-generic dom)))))
1279
1280 (defcustom shr-prefer-media-type-alist '(("webm" . 1.0)
1281                                          ("ogv"  . 1.0)
1282                                          ("ogg"  . 1.0)
1283                                          ("opus" . 1.0)
1284                                          ("flac" . 0.9)
1285                                          ("wav"  . 0.5))
1286   "Preferences for media types.
1287 The key element should be a regexp matched against the type of the source or
1288 url if no type is specified.  The value should be a float in the range 0.0 to
1289 1.0.  Media elements with higher value are preferred."
1290   :version "24.4"
1291   :group 'shr
1292   :type '(alist :key-type regexp :value-type float))
1293
1294 (defun shr--get-media-pref (elem)
1295   "Determine the preference for ELEM.
1296 The preference is a float determined from `shr-prefer-media-type'."
1297   (let ((type (dom-attr elem 'type))
1298         (p 0.0))
1299     (unless type
1300       (setq type (dom-attr elem 'src)))
1301     (when type
1302       (dolist (pref shr-prefer-media-type-alist)
1303         (when (and
1304                (> (cdr pref) p)
1305                (string-match-p (car pref) type))
1306           (setq p (cdr pref)))))
1307     p))
1308
1309 (defun shr--extract-best-source (dom &optional url pref)
1310   "Extract the best `:src' property from <source> blocks in DOM."
1311   (setq pref (or pref -1.0))
1312   (let (new-pref)
1313     (dolist (elem (dom-non-text-children dom))
1314       (when (and (eq (dom-tag elem) 'source)
1315                  (< pref
1316                     (setq new-pref
1317                           (shr--get-media-pref elem))))
1318         (setq pref new-pref
1319               url (dom-attr elem 'src))
1320         ;; libxml's html parser isn't HTML5 compliant and non terminated
1321         ;; source tags might end up as children.  So recursion it is...
1322         (dolist (child (dom-non-text-children elem))
1323           (when (eq (dom-tag child) 'source)
1324             (let ((ret (shr--extract-best-source (list child) url pref)))
1325               (when (< pref (cdr ret))
1326                 (setq url (car ret)
1327                       pref (cdr ret)))))))))
1328   (cons url pref))
1329
1330 (defun shr-tag-video (dom)
1331   (let ((image (dom-attr dom 'poster))
1332         (url (dom-attr dom 'src))
1333         (start (point)))
1334     (unless url
1335       (setq url (car (shr--extract-best-source dom))))
1336     (if image
1337         (shr-tag-img nil image)
1338       (shr-insert " [video] "))
1339     (shr-urlify start (shr-expand-url url))))
1340
1341 (defun shr-tag-audio (dom)
1342   (let ((url (dom-attr dom 'src))
1343         (start (point)))
1344     (unless url
1345       (setq url (car (shr--extract-best-source dom))))
1346     (shr-insert " [audio] ")
1347     (shr-urlify start (shr-expand-url url))))
1348
1349 (defun shr-tag-img (dom &optional url)
1350   (when (or url
1351             (and dom
1352                  (> (length (dom-attr dom 'src)) 0)))
1353     (when (> (current-column) 0)
1354       (insert "\n"))
1355     (let ((alt (dom-attr dom 'alt))
1356           (url (shr-expand-url (or url (dom-attr dom 'src)))))
1357       (let ((start (point-marker)))
1358         (when (zerop (length alt))
1359           (setq alt "*"))
1360         (cond
1361          ((or (member (dom-attr dom 'height) '("0" "1"))
1362               (member (dom-attr dom 'width) '("0" "1")))
1363           ;; Ignore zero-sized or single-pixel images.
1364           )
1365          ((and (not shr-inhibit-images)
1366                (string-match "\\`data:" url))
1367           (let ((image (shr-image-from-data (substring url (match-end 0)))))
1368             (if image
1369                 (funcall shr-put-image-function image alt)
1370               (insert alt))))
1371          ((and (not shr-inhibit-images)
1372                (string-match "\\`cid:" url))
1373           (let ((url (substring url (match-end 0)))
1374                 image)
1375             (if (or (not shr-content-function)
1376                     (not (setq image (funcall shr-content-function url))))
1377                 (insert alt)
1378               (funcall shr-put-image-function image alt))))
1379          ((or shr-inhibit-images
1380               (and shr-blocked-images
1381                    (string-match shr-blocked-images url)))
1382           (setq shr-start (point))
1383           (if (> (string-width alt) 8)
1384               (shr-insert (truncate-string-to-width alt 8))
1385             (shr-insert alt)))
1386          ((and (not shr-ignore-cache)
1387                (url-is-cached (shr-encode-url url)))
1388           (funcall shr-put-image-function (shr-get-image-data url) alt))
1389          (t
1390           (insert alt " ")
1391           (when (and shr-ignore-cache
1392                      (url-is-cached (shr-encode-url url)))
1393             (let ((file (url-cache-create-filename (shr-encode-url url))))
1394               (when (file-exists-p file)
1395                 (delete-file file))))
1396           (url-queue-retrieve
1397            (shr-encode-url url) 'shr-image-fetched
1398            (list (current-buffer) start (set-marker (make-marker) (1- (point))))
1399            t t)))
1400         (when (zerop shr-table-depth) ;; We are not in a table.
1401           (put-text-property start (point) 'keymap shr-map)
1402           (put-text-property start (point) 'shr-alt alt)
1403           (put-text-property start (point) 'image-url url)
1404           (put-text-property start (point) 'image-displayer
1405                              (shr-image-displayer shr-content-function))
1406           (put-text-property start (point) 'help-echo
1407                              (shr-fill-text
1408                               (or (dom-attr dom 'title) alt))))))))
1409
1410 (defun shr-tag-pre (dom)
1411   (let ((shr-folding-mode 'none)
1412         (shr-current-font 'default))
1413     (shr-ensure-newline)
1414     (shr-generic dom)
1415     (shr-ensure-newline)))
1416
1417 (defun shr-tag-blockquote (dom)
1418   (shr-ensure-paragraph)
1419   (let ((start (point))
1420         (shr-indentation (+ shr-indentation
1421                             (* 4 shr-table-separator-pixel-width))))
1422     (shr-generic dom)
1423     (shr-ensure-paragraph)
1424     (shr-mark-fill start)))
1425
1426 (defun shr-tag-dl (dom)
1427   (shr-ensure-paragraph)
1428   (shr-generic dom)
1429   (shr-ensure-paragraph))
1430
1431 (defun shr-tag-dt (dom)
1432   (shr-ensure-newline)
1433   (shr-generic dom)
1434   (shr-ensure-newline))
1435
1436 (defun shr-tag-dd (dom)
1437   (shr-ensure-newline)
1438   (let ((shr-indentation (+ shr-indentation
1439                             (* 4 shr-table-separator-pixel-width))))
1440     (shr-generic dom)))
1441
1442 (defun shr-tag-ul (dom)
1443   (shr-ensure-paragraph)
1444   (let ((shr-list-mode 'ul))
1445     (shr-generic dom))
1446   (shr-ensure-paragraph))
1447
1448 (defun shr-tag-ol (dom)
1449   (shr-ensure-paragraph)
1450   (let ((shr-list-mode 1))
1451     (shr-generic dom))
1452   (shr-ensure-paragraph))
1453
1454 (defun shr-tag-li (dom)
1455   (shr-ensure-newline)
1456   (let ((start (point)))
1457     (let* ((bullet
1458             (if (numberp shr-list-mode)
1459                 (prog1
1460                     (format "%d " shr-list-mode)
1461                   (setq shr-list-mode (1+ shr-list-mode)))
1462               (car shr-internal-bullet)))
1463            (width (if (numberp shr-list-mode)
1464                       (shr-string-pixel-width bullet)
1465                     (cdr shr-internal-bullet))))
1466       (insert bullet)
1467       (shr-mark-fill start)
1468       (let ((shr-indentation (+ shr-indentation width)))
1469         (put-text-property start (1+ start)
1470                            'shr-continuation-indentation shr-indentation)
1471         (put-text-property start (1+ start) 'shr-prefix-length (length bullet))
1472         (shr-generic dom)))))
1473
1474 (defun shr-mark-fill (start)
1475   ;; We may not have inserted any text to fill.
1476   (unless (= start (point))
1477     (put-text-property start (1+ start)
1478                        'shr-indentation shr-indentation)))
1479
1480 (defun shr-tag-br (dom)
1481   (when (and (not (bobp))
1482              ;; Only add a newline if we break the current line, or
1483              ;; the previous line isn't a blank line.
1484              (or (not (bolp))
1485                  (and (> (- (point) 2) (point-min))
1486                       (not (= (char-after (- (point) 2)) ?\n)))))
1487     (insert "\n"))
1488   (shr-generic dom))
1489
1490 (defun shr-tag-span (dom)
1491   (shr-generic dom))
1492
1493 (defun shr-tag-h1 (dom)
1494   (shr-heading dom (if shr-use-fonts
1495                        '(variable-pitch (:height 1.3 :weight bold))
1496                      'bold)))
1497
1498 (defun shr-tag-h2 (dom)
1499   (shr-heading dom 'bold))
1500
1501 (defun shr-tag-h3 (dom)
1502   (shr-heading dom 'italic))
1503
1504 (defun shr-tag-h4 (dom)
1505   (shr-heading dom))
1506
1507 (defun shr-tag-h5 (dom)
1508   (shr-heading dom))
1509
1510 (defun shr-tag-h6 (dom)
1511   (shr-heading dom))
1512
1513 (defun shr-tag-hr (_dom)
1514   (shr-ensure-newline)
1515   (insert (make-string (if (not shr-use-fonts)
1516                            shr-internal-width
1517                          (1+ (/ shr-internal-width
1518                                 shr-table-separator-pixel-width)))
1519                        shr-hr-line)
1520           "\n"))
1521
1522 (defun shr-tag-title (dom)
1523   (shr-heading dom 'bold 'underline))
1524
1525 (defun shr-tag-font (dom)
1526   (let* ((start (point))
1527          (color (dom-attr dom 'color))
1528          (shr-stylesheet (nconc (list (cons 'color color))
1529                                 shr-stylesheet)))
1530     (shr-generic dom)
1531     (when color
1532       (shr-colorize-region start (point) color
1533                            (cdr (assq 'background-color shr-stylesheet))))))
1534
1535 ;;; Table rendering algorithm.
1536
1537 ;; Table rendering is the only complicated thing here.  We do this by
1538 ;; first counting how many TDs there are in each TR, and registering
1539 ;; how wide they think they should be ("width=45%", etc).  Then we
1540 ;; render each TD separately (this is done in temporary buffers, so
1541 ;; that we can use all the rendering machinery as if we were in the
1542 ;; main buffer).  Now we know how much space each TD really takes, so
1543 ;; we then render everything again with the new widths, and finally
1544 ;; insert all these boxes into the main buffer.
1545 (defun shr-tag-table-1 (dom)
1546   (setq dom (or (dom-child-by-tag dom 'tbody) dom))
1547   (let* ((shr-inhibit-images t)
1548          (shr-table-depth (1+ shr-table-depth))
1549          (shr-kinsoku-shorten t)
1550          ;; Find all suggested widths.
1551          (columns (shr-column-specs dom))
1552          ;; Compute how many pixels wide each TD should be.
1553          (suggested-widths (shr-pro-rate-columns columns))
1554          ;; Do a "test rendering" to see how big each TD is (this can
1555          ;; be smaller (if there's little text) or bigger (if there's
1556          ;; unbreakable text).
1557          (elems (or (dom-attr dom 'shr-suggested-widths)
1558                     (shr-make-table dom suggested-widths nil
1559                                     'shr-suggested-widths)))
1560          (sketch (loop for line in elems
1561                        collect (mapcar #'car line)))
1562          (natural (loop for line in elems
1563                         collect (mapcar #'cdr line)))
1564          (sketch-widths (shr-table-widths sketch natural suggested-widths)))
1565     ;; This probably won't work very well.
1566     (when (> (+ (loop for width across sketch-widths
1567                       summing (1+ width))
1568                 shr-indentation shr-table-separator-pixel-width)
1569              (frame-width))
1570       (setq truncate-lines t))
1571     ;; Then render the table again with these new "hard" widths.
1572     (shr-insert-table (shr-make-table dom sketch-widths t) sketch-widths)))
1573
1574 (defun shr-tag-table (dom)
1575   (shr-ensure-paragraph)
1576   (let* ((caption (dom-children (dom-child-by-tag dom 'caption)))
1577          (header (dom-non-text-children (dom-child-by-tag dom 'thead)))
1578          (body (dom-non-text-children (or (dom-child-by-tag dom 'tbody)
1579                                           dom)))
1580          (footer (dom-non-text-children (dom-child-by-tag dom 'tfoot)))
1581          (bgcolor (dom-attr dom 'bgcolor))
1582          (start (point))
1583          (shr-stylesheet (nconc (list (cons 'background-color bgcolor))
1584                                 shr-stylesheet))
1585          (nheader (if header (shr-max-columns header)))
1586          (nbody (if body (shr-max-columns body)))
1587          (nfooter (if footer (shr-max-columns footer))))
1588     (if (and (not caption)
1589              (not header)
1590              (not (dom-child-by-tag dom 'tbody))
1591              (not (dom-child-by-tag dom 'tr))
1592              (not footer))
1593         ;; The table is totally invalid and just contains random junk.
1594         ;; Try to output it anyway.
1595         (shr-generic dom)
1596       ;; It's a real table, so render it.
1597       (if (dom-attr dom 'shr-fixed-table)
1598           (shr-tag-table-1 dom)
1599         ;; Only fix up the table once.
1600         (let ((table
1601                (nconc
1602                 (list 'table nil)
1603                 (if caption `((tr nil (td nil ,@caption))))
1604                 (cond
1605                  (header
1606                   (if footer
1607                       ;; header + body + footer
1608                       (if (= nheader nbody)
1609                           (if (= nbody nfooter)
1610                               `((tr nil (td nil (table nil
1611                                                        (tbody nil ,@header
1612                                                               ,@body ,@footer)))))
1613                             (nconc `((tr nil (td nil (table nil
1614                                                             (tbody nil ,@header
1615                                                                    ,@body)))))
1616                                    (if (= nfooter 1)
1617                                        footer
1618                                      `((tr nil (td nil (table
1619                                                         nil (tbody
1620                                                              nil ,@footer))))))))
1621                         (nconc `((tr nil (td nil (table nil (tbody
1622                                                              nil ,@header)))))
1623                                (if (= nbody nfooter)
1624                                    `((tr nil (td nil (table
1625                                                       nil (tbody nil ,@body
1626                                                                  ,@footer)))))
1627                                  (nconc `((tr nil (td nil (table
1628                                                            nil (tbody nil
1629                                                                       ,@body)))))
1630                                         (if (= nfooter 1)
1631                                             footer
1632                                           `((tr nil (td nil (table
1633                                                              nil
1634                                                              (tbody
1635                                                               nil
1636                                                               ,@footer))))))))))
1637                     ;; header + body
1638                     (if (= nheader nbody)
1639                         `((tr nil (td nil (table nil (tbody nil ,@header
1640                                                             ,@body)))))
1641                       (if (= nheader 1)
1642                           `(,@header (tr nil (td nil (table
1643                                                       nil (tbody nil ,@body)))))
1644                         `((tr nil (td nil (table nil (tbody nil ,@header))))
1645                           (tr nil (td nil (table nil (tbody nil ,@body)))))))))
1646                  (footer
1647                   ;; body + footer
1648                   (if (= nbody nfooter)
1649                       `((tr nil (td nil (table
1650                                          nil (tbody nil ,@body ,@footer)))))
1651                     (nconc `((tr nil (td nil (table nil (tbody nil ,@body)))))
1652                            (if (= nfooter 1)
1653                                footer
1654                              `((tr nil (td nil (table
1655                                                 nil (tbody nil ,@footer)))))))))
1656                  (caption
1657                   `((tr nil (td nil (table nil (tbody nil ,@body))))))
1658                  (body)))))
1659           (dom-set-attribute table 'shr-fixed-table t)
1660           (setcdr dom (cdr table))
1661           (shr-tag-table-1 dom))))
1662     (when bgcolor
1663       (shr-colorize-region start (point) (cdr (assq 'color shr-stylesheet))
1664                            bgcolor))
1665     ;; Finally, insert all the images after the table.  The Emacs buffer
1666     ;; model isn't strong enough to allow us to put the images actually
1667     ;; into the tables.
1668     (when (zerop shr-table-depth)
1669       (save-excursion
1670         (shr-expand-alignments start (point)))
1671       (dolist (elem (dom-by-tag dom 'object))
1672         (shr-tag-object elem))
1673       (dolist (elem (dom-by-tag dom 'img))
1674         (shr-tag-img elem)))))
1675
1676 (defun shr-insert-table (table widths)
1677   (let* ((collapse (equal (cdr (assq 'border-collapse shr-stylesheet))
1678                           "collapse"))
1679          (shr-table-separator-length (if collapse 0 1))
1680          (shr-table-vertical-line (if collapse "" shr-table-vertical-line))
1681          (start (point)))
1682     (setq shr-table-id (1+ shr-table-id))
1683     (unless collapse
1684       (shr-insert-table-ruler widths))
1685     (dolist (row table)
1686       (let ((start (point))
1687             (align 0)
1688             (column-number 0)
1689             (height (let ((max 0))
1690                       (dolist (column row)
1691                         (setq max (max max (nth 2 column))))
1692                       max)))
1693         (dotimes (i (max height 1))
1694           (shr-indent)
1695           (insert shr-table-vertical-line "\n"))
1696         (dolist (column row)
1697           (when (> (nth 2 column) -1)
1698             (goto-char start)
1699             ;; Sum up all the widths from the column.  (There may be
1700             ;; more than one if this is a "colspan" column.)
1701             (dotimes (i (nth 4 column))
1702               ;; The colspan directive may be wrong and there may not be
1703               ;; that number of columns.
1704               (when (<= column-number (1- (length widths)))
1705                 (setq align (+ align
1706                                (aref widths column-number)
1707                                (* 2 shr-table-separator-pixel-width))))
1708               (setq column-number (1+ column-number)))
1709             (let ((lines (nth 3 column))
1710                   (pixel-align (if (not shr-use-fonts)
1711                                    (* align (frame-char-width))
1712                                  align)))
1713               (dolist (line lines)
1714                 (end-of-line)
1715                 (let ((start (point)))
1716                   (insert
1717                    line
1718                    (propertize " "
1719                                'display `(space :align-to (,pixel-align))
1720                                'face (and (> (length line) 0)
1721                                           (shr-face-background
1722                                            (get-text-property
1723                                             (1- (length line)) 'face line)))
1724                                'shr-table-indent shr-table-id)
1725                    shr-table-vertical-line)
1726                   (shr-colorize-region
1727                    start (1- (point)) (nth 5 column) (nth 6 column)))
1728                 (forward-line 1))
1729               ;; Add blank lines at padding at the bottom of the TD,
1730               ;; possibly.
1731               (dotimes (i (- height (length lines)))
1732                 (end-of-line)
1733                 (let ((start (point)))
1734                   (insert (propertize " "
1735                                       'display `(space :align-to (,pixel-align))
1736                                       'shr-table-indent shr-table-id)
1737                           shr-table-vertical-line)
1738                   (shr-colorize-region
1739                    start (1- (point)) (nth 5 column) (nth 6 column)))
1740                 (forward-line 1))))))
1741       (unless collapse
1742         (shr-insert-table-ruler widths)))
1743     (unless (= start (point))
1744       (put-text-property start (1+ start) 'shr-table-id shr-table-id))))
1745
1746 (defun shr-face-background (face)
1747   (and (consp face)
1748        (let ((background nil))
1749          (dolist (elem face)
1750            (when (and (consp elem)
1751                       (eq (car elem) :background))
1752              (setq background (cadr elem))))
1753          (and background
1754               (list :background background)))))
1755
1756 (defun shr-expand-alignments (start end)
1757   (while (< (setq start (next-single-property-change
1758                          start 'shr-table-id nil end))
1759             end)
1760     (goto-char start)
1761     (let* ((shr-use-fonts t)
1762            (id (get-text-property (point) 'shr-table-id))
1763            (base (shr-pixel-column))
1764            elem)
1765       (when id
1766         (save-excursion
1767           (while (setq elem (text-property-any
1768                              (point) end 'shr-table-indent id))
1769             (goto-char elem)
1770             (let ((align (get-text-property (point) 'display)))
1771               (put-text-property (point) (1+ (point)) 'display
1772                                  `(space :align-to (,(+ (car (nth 2 align))
1773                                                         base)))))
1774             (forward-char 1)))))
1775     (setq start (1+ start))))
1776
1777 (defun shr-insert-table-ruler (widths)
1778   (when shr-table-horizontal-line
1779     (when (and (bolp)
1780                (> shr-indentation 0))
1781       (shr-indent))
1782     (insert shr-table-corner)
1783     (let ((total-width 0))
1784       (dotimes (i (length widths))
1785         (setq total-width (+ total-width (aref widths i)
1786                              (* shr-table-separator-pixel-width 2)))
1787         (insert (make-string (1+ (/ (aref widths i)
1788                                     shr-table-separator-pixel-width))
1789                              shr-table-horizontal-line)
1790                 (propertize " "
1791                             'display `(space :align-to (,total-width))
1792                             'shr-table-indent shr-table-id)
1793                 shr-table-corner)))
1794     (insert "\n")))
1795
1796 (defun shr-table-widths (table natural-table suggested-widths)
1797   (let* ((length (length suggested-widths))
1798          (widths (make-vector length 0))
1799          (natural-widths (make-vector length 0)))
1800     (dolist (row table)
1801       (let ((i 0))
1802         (dolist (column row)
1803           (aset widths i (max (aref widths i) column))
1804           (setq i (1+ i)))))
1805     (dolist (row natural-table)
1806       (let ((i 0))
1807         (dolist (column row)
1808           (aset natural-widths i (max (aref natural-widths i) column))
1809           (setq i (1+ i)))))
1810     (let ((extra (- (apply '+ (append suggested-widths nil))
1811                     (apply '+ (append widths nil))
1812                     (* shr-table-separator-pixel-width (1+ (length widths)))))
1813           (expanded-columns 0))
1814       ;; We have extra, unused space, so divide this space amongst the
1815       ;; columns.
1816       (when (> extra 0)
1817         ;; If the natural width is wider than the rendered width, we
1818         ;; want to allow the column to expand.
1819         (dotimes (i length)
1820           (when (> (aref natural-widths i) (aref widths i))
1821             (setq expanded-columns (1+ expanded-columns))))
1822         (dotimes (i length)
1823           (when (> (aref natural-widths i) (aref widths i))
1824             (aset widths i (min
1825                             (aref natural-widths i)
1826                             (+ (/ extra expanded-columns)
1827                                (aref widths i))))))))
1828     widths))
1829
1830 (defun shr-make-table (dom widths &optional fill storage-attribute)
1831   (or (cadr (assoc (list dom widths fill) shr-content-cache))
1832       (let ((data (shr-make-table-1 dom widths fill)))
1833         (push (list (list dom widths fill) data)
1834               shr-content-cache)
1835         (when storage-attribute
1836           (dom-set-attribute dom storage-attribute data))
1837         data)))
1838
1839 (defun shr-make-table-1 (dom widths &optional fill)
1840   (let ((trs nil)
1841         (rowspans (make-vector (length widths) 0))
1842         (colspan-remaining 0)
1843         colspan-width colspan-count
1844         width colspan)
1845     (dolist (row (dom-non-text-children dom))
1846       (when (eq (dom-tag row) 'tr)
1847         (let ((tds nil)
1848               (columns (dom-non-text-children row))
1849               (i 0)
1850               (width-column 0)
1851               column)
1852           (while (< i (length widths))
1853             ;; If we previously had a rowspan definition, then that
1854             ;; means that we now have a "missing" td/th element here.
1855             ;; So just insert a dummy, empty one to (sort of) emulate
1856             ;; rowspan.
1857             (setq column
1858                   (if (zerop (aref rowspans i))
1859                       (pop columns)
1860                     (aset rowspans i (1- (aref rowspans i)))
1861                     '(td)))
1862             (when (and (not (stringp column))
1863                        (or (memq (dom-tag column) '(td th))
1864                            (not column)))
1865               (when-let (span (dom-attr column 'rowspan))
1866                 (aset rowspans i (+ (aref rowspans i)
1867                                     (1- (string-to-number span)))))
1868               ;; Sanity check for invalid column-spans.
1869               (when (>= width-column (length widths))
1870                 (setq width-column 0))
1871               (setq width
1872                     (if column
1873                         (aref widths width-column)
1874                       (* 10 shr-table-separator-pixel-width)))
1875               (when (setq colspan (dom-attr column 'colspan))
1876                 (setq colspan (min (string-to-number colspan)
1877                                    ;; The colspan may be wrong, so
1878                                    ;; truncate it to the length of the
1879                                    ;; remaining columns.
1880                                    (- (length widths) i)))
1881                 (dotimes (j (1- colspan))
1882                   (setq width
1883                         (if (> (+ i 1 j) (1- (length widths)))
1884                             ;; If we have a colspan spec that's longer
1885                             ;; than the table is wide, just use the last
1886                             ;; width as the width.
1887                             (aref widths (1- (length widths)))
1888                           ;; Sum up the widths of the columns we're
1889                           ;; spanning.
1890                           (+ width
1891                              shr-table-separator-length
1892                              (aref widths (+ i 1 j))))))
1893                 (setq width-column (+ width-column (1- colspan))
1894                       colspan-count colspan
1895                       colspan-remaining colspan))
1896               (when column
1897                 (let ((data (shr-render-td column width fill)))
1898                   (if (and (not fill)
1899                            (> colspan-remaining 0))
1900                       (progn
1901                         (setq colspan-width (car data))
1902                         (let ((this-width (/ colspan-width colspan-count)))
1903                           (push (cons this-width (cadr data)) tds)
1904                           (setq colspan-remaining (1- colspan-remaining))))
1905                     (if (not fill)
1906                         (push (cons (car data) (cadr data)) tds)
1907                       (push data tds)))))
1908               (when (and colspan
1909                          (> colspan 1))
1910                 (dotimes (c (1- colspan))
1911                   (setq i (1+ i))
1912                   (push
1913                    (if fill
1914                        (list 0 0 -1 nil 1 nil nil)
1915                      '(0 . 0))
1916                    tds)))
1917               (setq i (1+ i)
1918                     width-column (1+ width-column))))
1919           (push (nreverse tds) trs))))
1920     (nreverse trs)))
1921
1922 (defun shr-pixel-buffer-width ()
1923   (if (not shr-use-fonts)
1924       (save-excursion
1925         (goto-char (point-min))
1926         (let ((max 0))
1927           (while (not (eobp))
1928             (end-of-line)
1929             (setq max (max max (current-column)))
1930             (forward-line 1))
1931           max))
1932     (if (get-buffer-window)
1933         (car (window-text-pixel-size nil (point-min) (point-max)))
1934       (save-window-excursion
1935         (set-window-buffer nil (current-buffer))
1936         (car (window-text-pixel-size nil (point-min) (point-max)))))))
1937
1938 (defun shr-render-td (dom width fill)
1939   (let ((cache (intern (format "shr-td-cache-%s-%s" width fill))))
1940     (or (dom-attr dom cache)
1941         (and fill
1942              (let (result)
1943                (dolist (attr (dom-attributes dom))
1944                  (let ((name (symbol-name (car attr))))
1945                    (when (string-match "shr-td-cache-\\([0-9]+\\)-nil" name)
1946                      (let ((cache-width (string-to-number
1947                                          (match-string 1 name))))
1948                        (when (and (>= cache-width width)
1949                                   (<= (car (cdr attr)) width))
1950                          (setq result (cdr attr)))))))
1951                result))
1952         (let ((result (shr-render-td-1 dom width fill)))
1953           (dom-set-attribute dom cache result)
1954           result))))
1955
1956 (defun shr-render-td-1 (dom width fill)
1957   (with-temp-buffer
1958     (let ((bgcolor (dom-attr dom 'bgcolor))
1959           (fgcolor (dom-attr dom 'fgcolor))
1960           (style (dom-attr dom 'style))
1961           (shr-stylesheet shr-stylesheet)
1962           (max-width 0)
1963           natural-width)
1964       (when style
1965         (setq style (and (string-match "color" style)
1966                          (shr-parse-style style))))
1967       (when bgcolor
1968         (setq style (nconc (list (cons 'background-color bgcolor))
1969                            style)))
1970       (when fgcolor
1971         (setq style (nconc (list (cons 'color fgcolor)) style)))
1972       (when style
1973         (setq shr-stylesheet (append style shr-stylesheet)))
1974       (let ((shr-internal-width width)
1975             (shr-indentation 0))
1976         (shr-descend dom))
1977       (save-window-excursion
1978         (set-window-buffer nil (current-buffer))
1979         (unless fill
1980           (setq natural-width
1981                 (or (dom-attr dom 'shr-td-cache-natural)
1982                     (let ((natural (max (shr-pixel-buffer-width)
1983                                         (shr-dom-max-natural-width dom 0))))
1984                       (dom-set-attribute dom 'shr-td-cache-natural natural)
1985                       natural))))
1986         (if (and natural-width
1987                  (<= natural-width width))
1988             (setq max-width natural-width)
1989           (let ((shr-internal-width width))
1990             (shr-fill-lines (point-min) (point-max))
1991             (setq max-width (shr-pixel-buffer-width)))))
1992       (goto-char (point-max))
1993       ;; Delete padding at the bottom of the TDs.
1994       (delete-region
1995        (point)
1996        (progn
1997          (skip-chars-backward " \t\n")
1998          (end-of-line)
1999          (point)))
2000       (goto-char (point-min))
2001       (list max-width
2002             natural-width
2003             (count-lines (point-min) (point-max))
2004             (split-string (buffer-string) "\n")
2005             (if (dom-attr dom 'colspan)
2006                 (string-to-number (dom-attr dom 'colspan))
2007               1)
2008             (cdr (assq 'color shr-stylesheet))
2009             (cdr (assq 'background-color shr-stylesheet))))))
2010
2011 (defun shr-dom-max-natural-width (dom max)
2012   (if (eq (dom-tag dom) 'table)
2013       (max max (or
2014                 (loop for line in (dom-attr dom 'shr-suggested-widths)
2015                       maximize (+
2016                                 shr-table-separator-length
2017                                 (loop for elem in line
2018                                       summing
2019                                       (+ (cdr elem)
2020                                          (* 2 shr-table-separator-length)))))
2021                 0))
2022     (dolist (child (dom-children dom))
2023       (unless (stringp child)
2024         (setq max (max (shr-dom-max-natural-width child max)))))
2025     max))
2026
2027 (defun shr-buffer-width ()
2028   (goto-char (point-min))
2029   (let ((max 0))
2030     (while (not (eobp))
2031       (end-of-line)
2032       (setq max (max max (current-column)))
2033       (forward-line 1))
2034     max))
2035
2036 (defun shr-pro-rate-columns (columns)
2037   (let ((total-percentage 0)
2038         (widths (make-vector (length columns) 0)))
2039     (dotimes (i (length columns))
2040       (setq total-percentage (+ total-percentage (aref columns i))))
2041     (setq total-percentage (/ 1.0 total-percentage))
2042     (dotimes (i (length columns))
2043       (aset widths i (max (truncate (* (aref columns i)
2044                                        total-percentage
2045                                        (- shr-internal-width
2046                                           (* (1+ (length columns))
2047                                              shr-table-separator-pixel-width))))
2048                           10)))
2049     widths))
2050
2051 ;; Return a summary of the number and shape of the TDs in the table.
2052 (defun shr-column-specs (dom)
2053   (let ((columns (make-vector (shr-max-columns dom) 1)))
2054     (dolist (row (dom-non-text-children dom))
2055       (when (eq (dom-tag row) 'tr)
2056         (let ((i 0))
2057           (dolist (column (dom-non-text-children row))
2058             (when (memq (dom-tag column) '(td th))
2059               (let ((width (dom-attr column 'width)))
2060                 (when (and width
2061                            (string-match "\\([0-9]+\\)%" width)
2062                            (not (zerop (setq width (string-to-number
2063                                                     (match-string 1 width))))))
2064                   (aset columns i (/ width 100.0))))
2065               (setq i (1+ i)))))))
2066     columns))
2067
2068 (defun shr-count (dom elem)
2069   (let ((i 0))
2070     (dolist (sub (dom-children dom))
2071       (when (and (not (stringp sub))
2072                  (eq (dom-tag sub) elem))
2073         (setq i (1+ i))))
2074     i))
2075
2076 (defun shr-max-columns (dom)
2077   (let ((max 0))
2078     (dolist (row (dom-children dom))
2079       (when (and (not (stringp row))
2080                  (eq (dom-tag row) 'tr))
2081         (setq max (max max (+ (shr-count row 'td)
2082                               (shr-count row 'th))))))
2083     max))
2084
2085 (provide 'shr)
2086
2087 ;;; shr.el ends here