]> code.delx.au - gnu-emacs/blob - lisp/net/shr.el
; Merge from origin/emacs-25
[gnu-emacs] / lisp / net / shr.el
1 ;;; shr.el --- Simple HTML Renderer
2
3 ;; Copyright (C) 2010-2016 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 (require 'seq)
39 (require 'svg)
40
41 (defgroup shr nil
42 "Simple HTML Renderer"
43 :version "25.1"
44 :group 'web)
45
46 (defcustom shr-max-image-proportion 0.9
47 "How big pictures displayed are in relation to the window they're in.
48 A value of 0.7 means that they are allowed to take up 70% of the
49 width and height of the window. If they are larger than this,
50 and Emacs supports it, then the images will be rescaled down to
51 fit these criteria."
52 :version "24.1"
53 :group 'shr
54 :type 'float)
55
56 (defcustom shr-blocked-images nil
57 "Images that have URLs matching this regexp will be blocked."
58 :version "24.1"
59 :group 'shr
60 :type '(choice (const nil) regexp))
61
62 (defcustom shr-use-fonts t
63 "If non-nil, use proportional fonts for text."
64 :version "25.1"
65 :group 'shr
66 :type 'boolean)
67
68 (defcustom shr-use-colors t
69 "If non-nil, respect color specifications in the HTML."
70 :version "25.2"
71 :group 'shr
72 :type 'boolean)
73
74 (defcustom shr-table-horizontal-line nil
75 "Character used to draw horizontal table lines.
76 If nil, don't draw horizontal table lines."
77 :group 'shr
78 :type '(choice (const nil) character))
79
80 (defcustom shr-table-vertical-line ?\s
81 "Character used to draw vertical table lines."
82 :group 'shr
83 :type 'character)
84
85 (defcustom shr-table-corner ?\s
86 "Character used to draw table corners."
87 :group 'shr
88 :type 'character)
89
90 (defcustom shr-hr-line ?-
91 "Character used to draw hr lines."
92 :group 'shr
93 :type 'character)
94
95 (defcustom shr-width nil
96 "Frame width to use for rendering.
97 May either be an integer specifying a fixed width in characters,
98 or nil, meaning that the full width of the window should be
99 used."
100 :version "25.1"
101 :type '(choice (integer :tag "Fixed width in characters")
102 (const :tag "Use the width of the window" nil))
103 :group 'shr)
104
105 (defcustom shr-bullet "* "
106 "Bullet used for unordered lists.
107 Alternative suggestions are:
108 - \" \"
109 - \" \""
110 :version "24.4"
111 :type 'string
112 :group 'shr)
113
114 (defcustom shr-external-browser 'browse-url-default-browser
115 "Function used to launch an external browser."
116 :version "24.4"
117 :group 'shr
118 :type 'function)
119
120 (defcustom shr-image-animate t
121 "Non nil means that images that can be animated will be."
122 :version "24.4"
123 :group 'shr
124 :type 'boolean)
125
126 (defvar shr-content-function nil
127 "If bound, this should be a function that will return the content.
128 This is used for cid: URLs, and the function is called with the
129 cid: URL as the argument.")
130
131 (defvar shr-put-image-function 'shr-put-image
132 "Function called to put image and alt string.")
133
134 (defface shr-strike-through '((t (:strike-through t)))
135 "Font for <s> elements."
136 :group 'shr)
137
138 (defface shr-link
139 '((t (:inherit link)))
140 "Font for link elements."
141 :group 'shr)
142
143 (defvar shr-inhibit-images nil
144 "If non-nil, inhibit loading images.")
145
146 (defvar shr-external-rendering-functions nil
147 "Alist of tag/function pairs used to alter how shr renders certain tags.
148 For instance, eww uses this to alter rendering of title, forms
149 and other things:
150 ((title . eww-tag-title)
151 (form . eww-tag-form)
152 ...)")
153
154 ;;; Internal variables.
155
156 (defvar shr-folding-mode nil)
157 (defvar shr-start nil)
158 (defvar shr-indentation 0)
159 (defvar shr-internal-width nil)
160 (defvar shr-list-mode nil)
161 (defvar shr-content-cache nil)
162 (defvar shr-kinsoku-shorten nil)
163 (defvar shr-table-depth 0)
164 (defvar shr-stylesheet nil)
165 (defvar shr-base nil)
166 (defvar shr-depth 0)
167 (defvar shr-warning nil)
168 (defvar shr-ignore-cache nil)
169 (defvar shr-target-id nil)
170 (defvar shr-table-separator-length 1)
171 (defvar shr-table-separator-pixel-width 0)
172 (defvar shr-table-id nil)
173 (defvar shr-current-font nil)
174 (defvar shr-internal-bullet nil)
175
176 (defvar shr-map
177 (let ((map (make-sparse-keymap)))
178 (define-key map "a" 'shr-show-alt-text)
179 (define-key map "i" 'shr-browse-image)
180 (define-key map "z" 'shr-zoom-image)
181 (define-key map [?\t] 'shr-next-link)
182 (define-key map [?\M-\t] 'shr-previous-link)
183 (define-key map [follow-link] 'mouse-face)
184 (define-key map [mouse-2] 'shr-browse-url)
185 (define-key map "I" 'shr-insert-image)
186 (define-key map "w" 'shr-copy-url)
187 (define-key map "u" 'shr-copy-url)
188 (define-key map "v" 'shr-browse-url)
189 (define-key map "O" 'shr-save-contents)
190 (define-key map "\r" 'shr-browse-url)
191 map))
192
193 (defvar shr-image-map
194 (let ((map (copy-keymap shr-map)))
195 (when (boundp 'image-map)
196 (set-keymap-parent map image-map))
197 map))
198
199 ;; Public functions and commands.
200 (declare-function libxml-parse-html-region "xml.c"
201 (start end &optional base-url discard-comments))
202
203 (defun shr-render-buffer (buffer)
204 "Display the HTML rendering of the current buffer."
205 (interactive (list (current-buffer)))
206 (or (fboundp 'libxml-parse-html-region)
207 (error "This function requires Emacs to be compiled with libxml2"))
208 (pop-to-buffer "*html*")
209 (erase-buffer)
210 (shr-insert-document
211 (with-current-buffer buffer
212 (libxml-parse-html-region (point-min) (point-max))))
213 (goto-char (point-min)))
214
215 ;;;###autoload
216 (defun shr-render-region (begin end &optional buffer)
217 "Display the HTML rendering of the region between BEGIN and END."
218 (interactive "r")
219 (unless (fboundp 'libxml-parse-html-region)
220 (error "This function requires Emacs to be compiled with libxml2"))
221 (with-current-buffer (or buffer (current-buffer))
222 (let ((dom (libxml-parse-html-region begin end)))
223 (delete-region begin end)
224 (goto-char begin)
225 (shr-insert-document dom))))
226
227 (defun shr--have-one-fringe-p ()
228 "Return non-nil if we know at least one of the fringes has non-zero width."
229 (and (fboundp 'fringe-columns)
230 (or (not (zerop (fringe-columns 'right)))
231 (not (zerop (fringe-columns 'left))))))
232
233 ;;;###autoload
234 (defun shr-insert-document (dom)
235 "Render the parsed document DOM into the current buffer.
236 DOM should be a parse tree as generated by
237 `libxml-parse-html-region' or similar."
238 (setq shr-content-cache nil)
239 (let ((start (point))
240 (shr-start nil)
241 (shr-base nil)
242 (shr-depth 0)
243 (shr-table-id 0)
244 (shr-warning nil)
245 (shr-table-separator-pixel-width (shr-string-pixel-width "-"))
246 (shr-internal-bullet (cons shr-bullet
247 (shr-string-pixel-width shr-bullet)))
248 (shr-internal-width (or (and shr-width
249 (if (not shr-use-fonts)
250 shr-width
251 (* shr-width (frame-char-width))))
252 ;; We need to adjust the available
253 ;; width for when the user disables
254 ;; the fringes, which will cause the
255 ;; display engine usurp one column for
256 ;; the continuation glyph.
257 (if (not shr-use-fonts)
258 (- (window-body-width) 1
259 (if (and (null shr-width)
260 (not (shr--have-one-fringe-p)))
261 0
262 1))
263 (- (window-body-width nil t)
264 (* 2 (frame-char-width))
265 (if (and (null shr-width)
266 (not (shr--have-one-fringe-p)))
267 (* (frame-char-width) 2)
268 0)))))
269 bidi-display-reordering)
270 ;; If the window was hscrolled for some reason, shr-fill-lines
271 ;; below will misbehave, because it silently assumes that it
272 ;; starts with a non-hscrolled window (vertical-motion will move
273 ;; to a wrong place otherwise).
274 (set-window-hscroll nil 0)
275 (shr-descend dom)
276 (shr-fill-lines start (point))
277 (shr-remove-trailing-whitespace start (point))
278 (when shr-warning
279 (message "%s" shr-warning))))
280
281 (defun shr-remove-trailing-whitespace (start end)
282 (let ((width (window-width)))
283 (save-restriction
284 (narrow-to-region start end)
285 (goto-char start)
286 (while (not (eobp))
287 (end-of-line)
288 (when (> (shr-previous-newline-padding-width (current-column)) width)
289 (dolist (overlay (overlays-at (point)))
290 (when (overlay-get overlay 'before-string)
291 (overlay-put overlay 'before-string nil))))
292 (forward-line 1)))))
293
294 (defun shr-copy-url (&optional image-url)
295 "Copy the URL under point to the kill ring.
296 If IMAGE-URL (the prefix) is non-nil, or there is no link under
297 point, but there is an image under point then copy the URL of the
298 image under point instead.
299 If called twice, then try to fetch the URL and see whether it
300 redirects somewhere else."
301 (interactive "P")
302 (let ((url (or (get-text-property (point) 'shr-url)
303 (get-text-property (point) 'image-url))))
304 (cond
305 ((not url)
306 (message "No URL under point"))
307 ;; Resolve redirected URLs.
308 ((equal url (car kill-ring))
309 (url-retrieve
310 url
311 (lambda (a)
312 (when (and (consp a)
313 (eq (car a) :redirect))
314 (with-temp-buffer
315 (insert (cadr a))
316 (goto-char (point-min))
317 ;; Remove common tracking junk from the URL.
318 (when (re-search-forward ".utm_.*" nil t)
319 (replace-match "" t t))
320 (message "Copied %s" (buffer-string))
321 (copy-region-as-kill (point-min) (point-max)))))
322 nil t))
323 ;; Copy the URL to the kill ring.
324 (t
325 (with-temp-buffer
326 (insert (url-encode-url url))
327 (copy-region-as-kill (point-min) (point-max))
328 (message "Copied %s" (buffer-string)))))))
329
330 (defun shr-next-link ()
331 "Skip to the next link."
332 (interactive)
333 (let ((current (get-text-property (point) 'shr-url))
334 (start (point))
335 skip)
336 (while (and (not (eobp))
337 (equal (get-text-property (point) 'shr-url) current))
338 (forward-char 1))
339 (cond
340 ((and (not (eobp))
341 (get-text-property (point) 'shr-url))
342 ;; The next link is adjacent.
343 (message "%s" (get-text-property (point) 'help-echo)))
344 ((or (eobp)
345 (not (setq skip (text-property-not-all (point) (point-max)
346 'shr-url nil))))
347 (goto-char start)
348 (message "No next link"))
349 (t
350 (goto-char skip)
351 (message "%s" (get-text-property (point) 'help-echo))))))
352
353 (defun shr-previous-link ()
354 "Skip to the previous link."
355 (interactive)
356 (let ((start (point))
357 (found nil))
358 ;; Skip past the current link.
359 (while (and (not (bobp))
360 (get-text-property (point) 'help-echo))
361 (forward-char -1))
362 ;; Find the previous link.
363 (while (and (not (bobp))
364 (not (setq found (get-text-property (point) 'help-echo))))
365 (forward-char -1))
366 (if (not found)
367 (progn
368 (message "No previous link")
369 (goto-char start))
370 ;; Put point at the start of the link.
371 (while (and (not (bobp))
372 (get-text-property (point) 'help-echo))
373 (forward-char -1))
374 (forward-char 1)
375 (message "%s" (get-text-property (point) 'help-echo)))))
376
377 (defun shr-show-alt-text ()
378 "Show the ALT text of the image under point."
379 (interactive)
380 (let ((text (get-text-property (point) 'shr-alt)))
381 (if (not text)
382 (message "No image under point")
383 (message "%s" (shr-fill-text text)))))
384
385 (defun shr-browse-image (&optional copy-url)
386 "Browse the image under point.
387 If COPY-URL (the prefix if called interactively) is non-nil, copy
388 the URL of the image to the kill buffer instead."
389 (interactive "P")
390 (let ((url (get-text-property (point) 'image-url)))
391 (cond
392 ((not url)
393 (message "No image under point"))
394 (copy-url
395 (with-temp-buffer
396 (insert url)
397 (copy-region-as-kill (point-min) (point-max))
398 (message "Copied %s" url)))
399 (t
400 (message "Browsing %s..." url)
401 (browse-url url)))))
402
403 (defun shr-insert-image ()
404 "Insert the image under point into the buffer."
405 (interactive)
406 (let ((url (get-text-property (point) 'image-url)))
407 (if (not url)
408 (message "No image under point")
409 (message "Inserting %s..." url)
410 (url-retrieve url 'shr-image-fetched
411 (list (current-buffer) (1- (point)) (point-marker))
412 t t))))
413
414 (defun shr-zoom-image ()
415 "Toggle the image size.
416 The size will be rotated between the default size, the original
417 size, and full-buffer size."
418 (interactive)
419 (let ((url (get-text-property (point) 'image-url))
420 (size (get-text-property (point) 'image-size))
421 (buffer-read-only nil))
422 (if (not url)
423 (message "No image under point")
424 ;; Delete the old picture.
425 (while (get-text-property (point) 'image-url)
426 (forward-char -1))
427 (forward-char 1)
428 (let ((start (point)))
429 (while (get-text-property (point) 'image-url)
430 (forward-char 1))
431 (forward-char -1)
432 (put-text-property start (point) 'display nil)
433 (when (> (- (point) start) 2)
434 (delete-region start (1- (point)))))
435 (message "Inserting %s..." url)
436 (url-retrieve url 'shr-image-fetched
437 (list (current-buffer) (1- (point)) (point-marker)
438 (list (cons 'size
439 (cond ((or (eq size 'default)
440 (null size))
441 'original)
442 ((eq size 'original)
443 'full)
444 ((eq size 'full)
445 'default)))))
446 t))))
447
448 ;;; Utility functions.
449
450 (defsubst shr-generic (dom)
451 (dolist (sub (dom-children dom))
452 (if (stringp sub)
453 (shr-insert sub)
454 (shr-descend sub))))
455
456 (defun shr-descend (dom)
457 (let ((function
458 (intern (concat "shr-tag-" (symbol-name (dom-tag dom))) obarray))
459 ;; Allow other packages to override (or provide) rendering
460 ;; of elements.
461 (external (cdr (assq (dom-tag dom) shr-external-rendering-functions)))
462 (style (dom-attr dom 'style))
463 (shr-stylesheet shr-stylesheet)
464 (shr-depth (1+ shr-depth))
465 (start (point)))
466 ;; shr uses many frames per nested node.
467 (if (> shr-depth (/ max-specpdl-size 15))
468 (setq shr-warning "Too deeply nested to render properly; consider increasing `max-specpdl-size'")
469 (when style
470 (if (string-match "color\\|display\\|border-collapse" style)
471 (setq shr-stylesheet (nconc (shr-parse-style style)
472 shr-stylesheet))
473 (setq style nil)))
474 ;; If we have a display:none, then just ignore this part of the DOM.
475 (unless (equal (cdr (assq 'display shr-stylesheet)) "none")
476 (cond (external
477 (funcall external dom))
478 ((fboundp function)
479 (funcall function dom))
480 (t
481 (shr-generic dom)))
482 (when (and shr-target-id
483 (equal (dom-attr dom 'id) shr-target-id))
484 ;; If the element was empty, we don't have anything to put the
485 ;; anchor on. So just insert a dummy character.
486 (when (= start (point))
487 (insert "*"))
488 (put-text-property start (1+ start) 'shr-target-id shr-target-id))
489 ;; If style is set, then this node has set the color.
490 (when style
491 (shr-colorize-region
492 start (point)
493 (cdr (assq 'color shr-stylesheet))
494 (cdr (assq 'background-color shr-stylesheet))))))))
495
496 (defun shr-fill-text (text)
497 (if (zerop (length text))
498 text
499 (with-temp-buffer
500 (let ((shr-indentation 0)
501 (shr-start nil)
502 (shr-internal-width (- (window-body-width nil t)
503 (* 2 (frame-char-width))
504 ;; Adjust the window width for when
505 ;; the user disables the fringes,
506 ;; which causes the display engine
507 ;; to usurp one column for the
508 ;; continuation glyph.
509 (if (and (null shr-width)
510 (not (shr--have-one-fringe-p)))
511 (* (frame-char-width) 2)
512 0))))
513 (shr-insert text)
514 (buffer-string)))))
515
516 (define-inline shr-char-breakable-p (char)
517 "Return non-nil if a line can be broken before and after CHAR."
518 (inline-quote (aref fill-find-break-point-function-table ,char)))
519 (define-inline shr-char-nospace-p (char)
520 "Return non-nil if no space is required before and after CHAR."
521 (inline-quote (aref fill-nospace-between-words-table ,char)))
522
523 ;; KINSOKU is a Japanese word meaning a rule that should not be violated.
524 ;; In Emacs, it is a term used for characters, e.g. punctuation marks,
525 ;; parentheses, and so on, that should not be placed in the beginning
526 ;; of a line or the end of a line.
527 (define-inline shr-char-kinsoku-bol-p (char)
528 "Return non-nil if a line ought not to begin with CHAR."
529 (inline-letevals (char)
530 (inline-quote (and (not (eq ,char ?'))
531 (aref (char-category-set ,char) ?>)))))
532 (define-inline shr-char-kinsoku-eol-p (char)
533 "Return non-nil if a line ought not to end with CHAR."
534 (inline-quote (aref (char-category-set ,char) ?<)))
535 (unless (shr-char-kinsoku-bol-p (make-char 'japanese-jisx0208 33 35))
536 (load "kinsoku" nil t))
537
538 (defun shr-pixel-column ()
539 (if (not shr-use-fonts)
540 (current-column)
541 (if (not (get-buffer-window (current-buffer)))
542 (save-window-excursion
543 (set-window-buffer nil (current-buffer))
544 (car (window-text-pixel-size nil (line-beginning-position) (point))))
545 (car (window-text-pixel-size nil (line-beginning-position) (point))))))
546
547 (defun shr-pixel-region ()
548 (- (shr-pixel-column)
549 (save-excursion
550 (goto-char (mark))
551 (shr-pixel-column))))
552
553 (defun shr-string-pixel-width (string)
554 (if (not shr-use-fonts)
555 (length string)
556 (with-temp-buffer
557 (insert string)
558 (shr-pixel-column))))
559
560 (defun shr-insert (text)
561 (when (and (not (bolp))
562 (get-text-property (1- (point)) 'image-url))
563 (insert "\n"))
564 (cond
565 ((eq shr-folding-mode 'none)
566 (let ((start (point)))
567 (insert text)
568 (save-restriction
569 (narrow-to-region start (point))
570 ;; Remove soft hyphens.
571 (goto-char (point-min))
572 (while (search-forward "­" nil t)
573 (replace-match "" t t))
574 (goto-char (point-max)))))
575 (t
576 (let ((font-start (point)))
577 (when (and (string-match "\\`[ \t\n\r ]" text)
578 (not (bolp))
579 (not (eq (char-after (1- (point))) ? )))
580 (insert " "))
581 (let ((start (point))
582 (bolp (bolp)))
583 (insert text)
584 (save-restriction
585 (narrow-to-region start (point))
586 (goto-char start)
587 (when (looking-at "[ \t\n\r ]+")
588 (replace-match "" t t))
589 (while (re-search-forward "[ \t\n\r ]+" nil t)
590 (replace-match " " t t))
591 ;; Remove soft hyphens.
592 (goto-char (point-min))
593 (while (search-forward "­" nil t)
594 (replace-match "" t t))
595 (goto-char (point-max)))
596 ;; We may have removed everything we inserted if if was just
597 ;; spaces.
598 (unless (= font-start (point))
599 ;; Mark all lines that should possibly be folded afterwards.
600 (when bolp
601 (shr-mark-fill start))
602 (when shr-use-fonts
603 (put-text-property font-start (point)
604 'face
605 (or shr-current-font 'variable-pitch)))))))))
606
607 (defun shr-fill-lines (start end)
608 (if (<= shr-internal-width 0)
609 nil
610 (save-restriction
611 (narrow-to-region start end)
612 (goto-char start)
613 (when (get-text-property (point) 'shr-indentation)
614 (shr-fill-line))
615 (while (setq start (next-single-property-change start 'shr-indentation))
616 (goto-char start)
617 (when (bolp)
618 (shr-fill-line)))
619 (goto-char (point-max)))))
620
621 (defun shr-vertical-motion (column)
622 (if (not shr-use-fonts)
623 (move-to-column column)
624 (unless (eolp)
625 (forward-char 1))
626 (vertical-motion (cons (/ column (frame-char-width)) 0))
627 (unless (eolp)
628 (forward-char 1))))
629
630 (defun shr-fill-line ()
631 (let ((shr-indentation (get-text-property (point) 'shr-indentation))
632 (continuation (get-text-property
633 (point) 'shr-continuation-indentation))
634 start)
635 (put-text-property (point) (1+ (point)) 'shr-indentation nil)
636 (let ((face (get-text-property (point) 'face))
637 (background-start (point)))
638 (shr-indent)
639 (when face
640 (put-text-property background-start (point) 'face
641 `,(shr-face-background face))))
642 (setq start (point))
643 (setq shr-indentation (or continuation shr-indentation))
644 (shr-vertical-motion shr-internal-width)
645 (when (looking-at " $")
646 (delete-region (point) (line-end-position)))
647 (while (not (eolp))
648 ;; We have to do some folding. First find the first
649 ;; previous point suitable for folding.
650 (if (or (not (shr-find-fill-point (line-beginning-position)))
651 (= (point) start))
652 ;; We had unbreakable text (for this width), so just go to
653 ;; the first space and carry on.
654 (progn
655 (beginning-of-line)
656 (skip-chars-forward " ")
657 (search-forward " " (line-end-position) 'move)))
658 ;; Success; continue.
659 (when (= (preceding-char) ?\s)
660 (delete-char -1))
661 (let ((face (get-text-property (point) 'face))
662 (background-start (point)))
663 (insert "\n")
664 (shr-indent)
665 (when face
666 (put-text-property background-start (point) 'face
667 `,(shr-face-background face))))
668 (setq start (point))
669 (shr-vertical-motion shr-internal-width)
670 (when (looking-at " $")
671 (delete-region (point) (line-end-position))))))
672
673 (defun shr-find-fill-point (start)
674 (let ((bp (point))
675 (end (point))
676 failed)
677 (while (not (or (setq failed (<= (point) start))
678 (eq (preceding-char) ? )
679 (eq (following-char) ? )
680 (shr-char-breakable-p (preceding-char))
681 (shr-char-breakable-p (following-char))
682 (and (shr-char-kinsoku-bol-p (preceding-char))
683 (shr-char-breakable-p (following-char))
684 (not (shr-char-kinsoku-bol-p (following-char))))
685 (shr-char-kinsoku-eol-p (following-char))
686 (bolp)))
687 (backward-char 1))
688 (if failed
689 ;; There's no breakable point, so we give it up.
690 (let (found)
691 (goto-char bp)
692 ;; Don't overflow the window edge, even if
693 ;; shr-kinsoku-shorten is nil.
694 (unless (or shr-kinsoku-shorten (null shr-width))
695 (while (setq found (re-search-forward
696 "\\(\\c>\\)\\| \\|\\c<\\|\\c|"
697 (line-end-position) 'move)))
698 (if (and found
699 (not (match-beginning 1)))
700 (goto-char (match-beginning 0)))))
701 (or
702 (eolp)
703 ;; Don't put kinsoku-bol characters at the beginning of a line,
704 ;; or kinsoku-eol characters at the end of a line.
705 (cond
706 ;; Don't overflow the window edge, even if shr-kinsoku-shorten
707 ;; is nil.
708 ((or shr-kinsoku-shorten (null shr-width))
709 (while (and (not (memq (preceding-char) (list ?\C-@ ?\n ? )))
710 (or (shr-char-kinsoku-eol-p (preceding-char))
711 (shr-char-kinsoku-bol-p (following-char))))
712 (backward-char 1))
713 (when (setq failed (<= (point) start))
714 ;; There's no breakable point that doesn't violate kinsoku,
715 ;; so we look for the second best position.
716 (while (and (progn
717 (forward-char 1)
718 (<= (point) end))
719 (progn
720 (setq bp (point))
721 (shr-char-kinsoku-eol-p (following-char)))))
722 (goto-char bp)))
723 ((shr-char-kinsoku-eol-p (preceding-char))
724 ;; Find backward the point where kinsoku-eol characters begin.
725 (let ((count 4))
726 (while
727 (progn
728 (backward-char 1)
729 (and (> (setq count (1- count)) 0)
730 (not (memq (preceding-char) (list ?\C-@ ?\n ? )))
731 (or (shr-char-kinsoku-eol-p (preceding-char))
732 (shr-char-kinsoku-bol-p (following-char)))))))
733 (when (setq failed (<= (point) start))
734 ;; There's no breakable point that doesn't violate kinsoku,
735 ;; so we go to the second best position.
736 (if (looking-at "\\(\\c<+\\)\\c<")
737 (goto-char (match-end 1))
738 (forward-char 1))))
739 ((shr-char-kinsoku-bol-p (following-char))
740 ;; Find forward the point where kinsoku-bol characters end.
741 (let ((count 4))
742 (while (progn
743 (forward-char 1)
744 (and (>= (setq count (1- count)) 0)
745 (shr-char-kinsoku-bol-p (following-char))
746 (shr-char-breakable-p (following-char))))))))
747 (when (eq (following-char) ? )
748 (forward-char 1))))
749 (not failed)))
750
751 (defun shr-parse-base (url)
752 ;; Always chop off anchors.
753 (when (string-match "#.*" url)
754 (setq url (substring url 0 (match-beginning 0))))
755 ;; NB: <base href="" > URI may itself be relative to the document s URI
756 (setq url (shr-expand-url url))
757 (let* ((parsed (url-generic-parse-url url))
758 (local (url-filename parsed)))
759 (setf (url-filename parsed) "")
760 ;; Chop off the bit after the last slash.
761 (when (string-match "\\`\\(.*/\\)[^/]+\\'" local)
762 (setq local (match-string 1 local)))
763 ;; Always make the local bit end with a slash.
764 (when (and (not (zerop (length local)))
765 (not (eq (aref local (1- (length local))) ?/)))
766 (setq local (concat local "/")))
767 (list (url-recreate-url parsed)
768 local
769 (url-type parsed)
770 url)))
771
772 (autoload 'url-expand-file-name "url-expand")
773
774 ;; FIXME This needs some tests writing.
775 ;; Does it even need to exist, given that url-expand-file-name does?
776 (defun shr-expand-url (url &optional base)
777 (setq base
778 (if base
779 ;; shr-parse-base should never call this with non-nil base!
780 (shr-parse-base base)
781 ;; Bound by the parser.
782 shr-base))
783 (when (zerop (length url))
784 (setq url nil))
785 ;; Strip leading whitespace
786 (and url (string-match "\\`\\s-+" url)
787 (setq url (substring url (match-end 0))))
788 (cond ((or (not url)
789 (not base)
790 (string-match "\\`[a-z]*:" url))
791 ;; Absolute or empty URI
792 (or url (nth 3 base)))
793 ((eq (aref url 0) ?/)
794 (if (and (> (length url) 1)
795 (eq (aref url 1) ?/))
796 ;; //host...; just use the protocol
797 (concat (nth 2 base) ":" url)
798 ;; Just use the host name part.
799 (concat (car base) url)))
800 ((eq (aref url 0) ?#)
801 ;; A link to an anchor.
802 (concat (nth 3 base) url))
803 (t
804 ;; Totally relative.
805 (url-expand-file-name url (concat (car base) (cadr base))))))
806
807 (defun shr-ensure-newline ()
808 (unless (zerop (current-column))
809 (insert "\n")))
810
811 (defun shr-ensure-paragraph ()
812 (unless (bobp)
813 (let ((prefix (get-text-property (line-beginning-position)
814 'shr-prefix-length)))
815 (cond
816 ((and (bolp)
817 (save-excursion
818 (forward-line -1)
819 (looking-at " *$")))
820 ;; We're already at a new paragraph; do nothing.
821 )
822 ((and prefix
823 (= prefix (- (point) (line-beginning-position))))
824 ;; Do nothing; we're at the start of a <li>.
825 )
826 ((save-excursion
827 (beginning-of-line)
828 ;; If the current line is totally blank, and doesn't even
829 ;; have any face properties set, then delete the blank
830 ;; space.
831 (and (looking-at " *$")
832 (not (get-text-property (point) 'face))
833 (not (= (next-single-property-change (point) 'face nil
834 (line-end-position))
835 (line-end-position)))))
836 (delete-region (match-beginning 0) (match-end 0)))
837 (t
838 (insert "\n\n"))))))
839
840 (defun shr-indent ()
841 (when (> shr-indentation 0)
842 (insert
843 (if (not shr-use-fonts)
844 (make-string shr-indentation ?\s)
845 (propertize " "
846 'display
847 `(space :width (,shr-indentation)))))))
848
849 (defun shr-fontize-dom (dom &rest types)
850 (let ((start (point)))
851 (shr-generic dom)
852 (dolist (type types)
853 (shr-add-font start (point) type))))
854
855 ;; Add face to the region, but avoid putting the font properties on
856 ;; blank text at the start of the line, and the newline at the end, to
857 ;; avoid ugliness.
858 (defun shr-add-font (start end type)
859 (save-excursion
860 (goto-char start)
861 (while (< (point) end)
862 (when (bolp)
863 (skip-chars-forward " "))
864 (add-face-text-property (point) (min (line-end-position) end) type t)
865 (if (< (line-end-position) end)
866 (forward-line 1)
867 (goto-char end)))))
868
869 (defun shr-mouse-browse-url (ev)
870 "Browse the URL under the mouse cursor."
871 (interactive "e")
872 (mouse-set-point ev)
873 (shr-browse-url))
874
875 (defun shr-browse-url (&optional external mouse-event)
876 "Browse the URL under point.
877 If EXTERNAL, browse the URL using `shr-external-browser'."
878 (interactive (list current-prefix-arg last-nonmenu-event))
879 (mouse-set-point mouse-event)
880 (let ((url (get-text-property (point) 'shr-url)))
881 (cond
882 ((not url)
883 (message "No link under point"))
884 ((string-match "^mailto:" url)
885 (browse-url-mail url))
886 (t
887 (if external
888 (funcall shr-external-browser url)
889 (browse-url url))))))
890
891 (defun shr-save-contents (directory)
892 "Save the contents from URL in a file."
893 (interactive "DSave contents of URL to directory: ")
894 (let ((url (get-text-property (point) 'shr-url)))
895 (if (not url)
896 (message "No link under point")
897 (url-retrieve (shr-encode-url url)
898 'shr-store-contents (list url directory)
899 nil t))))
900
901 (defun shr-store-contents (status url directory)
902 (unless (plist-get status :error)
903 (when (or (search-forward "\n\n" nil t)
904 (search-forward "\r\n\r\n" nil t))
905 (write-region (point) (point-max)
906 (expand-file-name (file-name-nondirectory url)
907 directory)))))
908
909 (defun shr-image-fetched (status buffer start end &optional flags)
910 (let ((image-buffer (current-buffer)))
911 (when (and (buffer-name buffer)
912 (not (plist-get status :error)))
913 (url-store-in-cache image-buffer)
914 (when (or (search-forward "\n\n" nil t)
915 (search-forward "\r\n\r\n" nil t))
916 (let ((data (shr-parse-image-data)))
917 (with-current-buffer buffer
918 (save-excursion
919 (let ((alt (buffer-substring start end))
920 (properties (text-properties-at start))
921 (inhibit-read-only t))
922 (delete-region start end)
923 (goto-char start)
924 (funcall shr-put-image-function data alt flags)
925 (while properties
926 (let ((type (pop properties))
927 (value (pop properties)))
928 (unless (memq type '(display image-size))
929 (put-text-property start (point) type value))))))))))
930 (kill-buffer image-buffer)))
931
932 (defun shr-image-from-data (data)
933 "Return an image from the data: URI content DATA."
934 (when (string-match
935 "\\(\\([^/;,]+\\(/[^;,]+\\)?\\)\\(;[^;,]+\\)*\\)?,\\(.*\\)"
936 data)
937 (let ((param (match-string 4 data))
938 (payload (url-unhex-string (match-string 5 data))))
939 (when (string-match "^.*\\(;[ \t]*base64\\)$" param)
940 (setq payload (base64-decode-string payload)))
941 payload)))
942
943 ;; Behind display-graphic-p test.
944 (declare-function image-size "image.c" (spec &optional pixels frame))
945 (declare-function image-animate "image" (image &optional index limit))
946
947 (defun shr-put-image (spec alt &optional flags)
948 "Insert image SPEC with a string ALT. Return image.
949 SPEC is either an image data blob, or a list where the first
950 element is the data blob and the second element is the content-type."
951 (if (display-graphic-p)
952 (let* ((size (cdr (assq 'size flags)))
953 (data (if (consp spec)
954 (car spec)
955 spec))
956 (content-type (and (consp spec)
957 (cadr spec)))
958 (start (point))
959 (image (cond
960 ((eq size 'original)
961 (create-image data nil t :ascent 100
962 :format content-type))
963 ((eq content-type 'image/svg+xml)
964 (create-image data 'svg t :ascent 100))
965 ((eq size 'full)
966 (ignore-errors
967 (shr-rescale-image data content-type
968 (plist-get flags :width)
969 (plist-get flags :height))))
970 (t
971 (ignore-errors
972 (shr-rescale-image data content-type
973 (plist-get flags :width)
974 (plist-get flags :height)))))))
975 (when image
976 ;; When inserting big-ish pictures, put them at the
977 ;; beginning of the line.
978 (when (and (> (current-column) 0)
979 (> (car (image-size image t)) 400))
980 (insert "\n"))
981 (if (eq size 'original)
982 (insert-sliced-image image (or alt "*") nil 20 1)
983 (insert-image image (or alt "*")))
984 (put-text-property start (point) 'image-size size)
985 (when (and shr-image-animate
986 (cond ((fboundp 'image-multi-frame-p)
987 ;; Only animate multi-frame things that specify a
988 ;; delay; eg animated gifs as opposed to
989 ;; multi-page tiffs. FIXME?
990 (cdr (image-multi-frame-p image)))
991 ((fboundp 'image-animated-p)
992 (image-animated-p image))))
993 (image-animate image nil 60)))
994 image)
995 (insert (or alt ""))))
996
997 (defun shr-rescale-image (data content-type width height)
998 "Rescale DATA, if too big, to fit the current buffer.
999 WIDTH and HEIGHT are the sizes given in the HTML data, if any."
1000 (if (not (and (fboundp 'imagemagick-types)
1001 (get-buffer-window (current-buffer))))
1002 (create-image data nil t :ascent 100)
1003 (let* ((edges (window-inside-pixel-edges
1004 (get-buffer-window (current-buffer))))
1005 (max-width (truncate (* shr-max-image-proportion
1006 (- (nth 2 edges) (nth 0 edges)))))
1007 (max-height (truncate (* shr-max-image-proportion
1008 (- (nth 3 edges) (nth 1 edges))))))
1009 (when (or (and width
1010 (> width max-width))
1011 (and height
1012 (> height max-height)))
1013 (setq width nil
1014 height nil))
1015 (if (and width height)
1016 (create-image
1017 data 'imagemagick t
1018 :ascent 100
1019 :width width
1020 :height height
1021 :format content-type)
1022 (create-image
1023 data 'imagemagick t
1024 :ascent 100
1025 :max-width max-width
1026 :max-height max-height
1027 :format content-type)))))
1028
1029 ;; url-cache-extract autoloads url-cache.
1030 (declare-function url-cache-create-filename "url-cache" (url))
1031 (autoload 'mm-disable-multibyte "mm-util")
1032 (autoload 'browse-url-mail "browse-url")
1033
1034 (defun shr-get-image-data (url)
1035 "Get image data for URL.
1036 Return a string with image data."
1037 (with-temp-buffer
1038 (mm-disable-multibyte)
1039 (when (ignore-errors
1040 (url-cache-extract (url-cache-create-filename (shr-encode-url url)))
1041 t)
1042 (when (or (search-forward "\n\n" nil t)
1043 (search-forward "\r\n\r\n" nil t))
1044 (shr-parse-image-data)))))
1045
1046 (declare-function libxml-parse-xml-region "xml.c"
1047 (start end &optional base-url discard-comments))
1048
1049 (defun shr-parse-image-data ()
1050 (let ((data (buffer-substring (point) (point-max)))
1051 (content-type
1052 (save-excursion
1053 (save-restriction
1054 (narrow-to-region (point-min) (point))
1055 (let ((content-type (mail-fetch-field "content-type")))
1056 (and content-type
1057 ;; Remove any comments in the type string.
1058 (intern (replace-regexp-in-string ";.*" "" content-type)
1059 obarray)))))))
1060 ;; SVG images may contain references to further images that we may
1061 ;; want to block. So special-case these by parsing the XML data
1062 ;; and remove the blocked bits.
1063 (when (eq content-type 'image/svg+xml)
1064 (setq data
1065 (shr-dom-to-xml
1066 (libxml-parse-xml-region (point) (point-max)))))
1067 (list data content-type)))
1068
1069 (defun shr-image-displayer (content-function)
1070 "Return a function to display an image.
1071 CONTENT-FUNCTION is a function to retrieve an image for a cid url that
1072 is an argument. The function to be returned takes three arguments URL,
1073 START, and END. Note that START and END should be markers."
1074 `(lambda (url start end)
1075 (when url
1076 (if (string-match "\\`cid:" url)
1077 ,(when content-function
1078 `(let ((image (funcall ,content-function
1079 (substring url (match-end 0)))))
1080 (when image
1081 (goto-char start)
1082 (funcall shr-put-image-function
1083 image (buffer-substring start end))
1084 (delete-region (point) end))))
1085 (url-retrieve url 'shr-image-fetched
1086 (list (current-buffer) start end)
1087 t t)))))
1088
1089 (defun shr-heading (dom &rest types)
1090 (shr-ensure-paragraph)
1091 (apply #'shr-fontize-dom dom types)
1092 (shr-ensure-paragraph))
1093
1094 (defun shr-urlify (start url &optional title)
1095 (shr-add-font start (point) 'shr-link)
1096 (add-text-properties
1097 start (point)
1098 (list 'shr-url url
1099 'help-echo (let ((iri (or (ignore-errors
1100 (decode-coding-string
1101 (url-unhex-string url)
1102 'utf-8 t))
1103 url)))
1104 (if title (format "%s (%s)" iri title) iri))
1105 'follow-link t
1106 'mouse-face 'highlight))
1107 ;; Don't overwrite any keymaps that are already in the buffer (i.e.,
1108 ;; image keymaps).
1109 (while (and start
1110 (< start (point)))
1111 (let ((next (next-single-property-change start 'keymap nil (point))))
1112 (if (get-text-property start 'keymap)
1113 (setq start next)
1114 (put-text-property start (or next (point)) 'keymap shr-map)))))
1115
1116 (defun shr-encode-url (url)
1117 "Encode URL."
1118 (browse-url-url-encode-chars url "[)$ ]"))
1119
1120 (autoload 'shr-color-visible "shr-color")
1121 (autoload 'shr-color->hexadecimal "shr-color")
1122
1123 (defun shr-color-check (fg bg)
1124 "Check that FG is visible on BG.
1125 Returns (fg bg) with corrected values.
1126 Returns nil if the colors that would be used are the default
1127 ones, in case fg and bg are nil."
1128 (when (or fg bg)
1129 (let ((fixed (cond ((null fg) 'fg)
1130 ((null bg) 'bg))))
1131 ;; Convert colors to hexadecimal, or set them to default.
1132 (let ((fg (or (shr-color->hexadecimal fg)
1133 (frame-parameter nil 'foreground-color)))
1134 (bg (or (shr-color->hexadecimal bg)
1135 (frame-parameter nil 'background-color))))
1136 (cond ((eq fixed 'bg)
1137 ;; Only return the new fg
1138 (list nil (cadr (shr-color-visible bg fg t))))
1139 ((eq fixed 'fg)
1140 ;; Invert args and results and return only the new bg
1141 (list (cadr (shr-color-visible fg bg t)) nil))
1142 (t
1143 (shr-color-visible bg fg)))))))
1144
1145 (defun shr-colorize-region (start end fg &optional bg)
1146 (when (and shr-use-colors
1147 (or fg bg)
1148 (>= (display-color-cells) 88))
1149 (let ((new-colors (shr-color-check fg bg)))
1150 (when new-colors
1151 (when fg
1152 (add-face-text-property start end
1153 (list :foreground (cadr new-colors))
1154 t))
1155 (when bg
1156 (add-face-text-property start end
1157 (list :background (car new-colors))
1158 t)))
1159 new-colors)))
1160
1161 (defun shr-previous-newline-padding-width (width)
1162 (let ((overlays (overlays-at (point)))
1163 (previous-width 0))
1164 (if (null overlays)
1165 width
1166 (dolist (overlay overlays)
1167 (setq previous-width
1168 (+ previous-width
1169 (length (plist-get (overlay-properties overlay)
1170 'before-string)))))
1171 (+ width previous-width))))
1172
1173 ;;; Tag-specific rendering rules.
1174
1175 (defun shr-tag-html (dom)
1176 (let ((dir (dom-attr dom 'dir)))
1177 (cond
1178 ((equal dir "ltr")
1179 (setq bidi-paragraph-direction 'left-to-right))
1180 ((equal dir "rtl")
1181 (setq bidi-paragraph-direction 'right-to-left))
1182 ((equal dir "auto")
1183 (setq bidi-paragraph-direction nil))))
1184 (shr-generic dom))
1185
1186 (defun shr-tag-body (dom)
1187 (let* ((start (point))
1188 (fgcolor (or (dom-attr dom 'fgcolor) (dom-attr dom 'text)))
1189 (bgcolor (dom-attr dom 'bgcolor))
1190 (shr-stylesheet (list (cons 'color fgcolor)
1191 (cons 'background-color bgcolor))))
1192 (shr-generic dom)
1193 (shr-colorize-region start (point) fgcolor bgcolor)))
1194
1195 (defun shr-tag-style (_dom)
1196 )
1197
1198 (defun shr-tag-script (_dom)
1199 )
1200
1201 (defun shr-tag-comment (_dom)
1202 )
1203
1204 (defun shr-dom-to-xml (dom)
1205 (with-temp-buffer
1206 (shr-dom-print dom)
1207 (buffer-string)))
1208
1209 (defun shr-dom-print (dom)
1210 "Convert DOM into a string containing the xml representation."
1211 (insert (format "<%s" (dom-tag dom)))
1212 (dolist (attr (dom-attributes dom))
1213 ;; Ignore attributes that start with a colon because they are
1214 ;; private elements.
1215 (unless (= (aref (format "%s" (car attr)) 0) ?:)
1216 (insert (format " %s=\"%s\"" (car attr) (cdr attr)))))
1217 (insert ">")
1218 (let (url)
1219 (dolist (elem (dom-children dom))
1220 (cond
1221 ((stringp elem)
1222 (insert elem))
1223 ((eq (dom-tag elem) 'comment)
1224 )
1225 ((or (not (eq (dom-tag elem) 'image))
1226 ;; Filter out blocked elements inside the SVG image.
1227 (not (setq url (dom-attr elem ':xlink:href)))
1228 (not shr-blocked-images)
1229 (not (string-match shr-blocked-images url)))
1230 (insert " ")
1231 (shr-dom-print elem)))))
1232 (insert (format "</%s>" (dom-tag dom))))
1233
1234 (defun shr-tag-svg (dom)
1235 (when (and (image-type-available-p 'svg)
1236 (not shr-inhibit-images)
1237 (dom-attr dom 'width)
1238 (dom-attr dom 'height))
1239 (funcall shr-put-image-function (list (shr-dom-to-xml dom) 'image/svg+xml)
1240 "SVG Image")))
1241
1242 (defun shr-tag-sup (dom)
1243 (let ((start (point)))
1244 (shr-generic dom)
1245 (put-text-property start (point) 'display '(raise 0.5))))
1246
1247 (defun shr-tag-sub (dom)
1248 (let ((start (point)))
1249 (shr-generic dom)
1250 (put-text-property start (point) 'display '(raise -0.5))))
1251
1252 (defun shr-tag-label (dom)
1253 (shr-generic dom)
1254 (shr-ensure-paragraph))
1255
1256 (defun shr-tag-p (dom)
1257 (shr-ensure-paragraph)
1258 (shr-generic dom)
1259 (shr-ensure-paragraph))
1260
1261 (defun shr-tag-div (dom)
1262 (shr-ensure-paragraph)
1263 (shr-generic dom)
1264 (shr-ensure-newline))
1265
1266 (defun shr-tag-s (dom)
1267 (shr-fontize-dom dom 'shr-strike-through))
1268
1269 (defun shr-tag-b (dom)
1270 (shr-fontize-dom dom 'bold))
1271
1272 (defun shr-tag-i (dom)
1273 (shr-fontize-dom dom 'italic))
1274
1275 (defun shr-tag-em (dom)
1276 (shr-fontize-dom dom 'italic))
1277
1278 (defun shr-tag-strong (dom)
1279 (shr-fontize-dom dom 'bold))
1280
1281 (defun shr-tag-u (dom)
1282 (shr-fontize-dom dom 'underline))
1283
1284 (defun shr-tag-tt (dom)
1285 (let ((shr-current-font 'default))
1286 (shr-generic dom)))
1287
1288 (defun shr-tag-ins (cont)
1289 (let* ((start (point))
1290 (color "green")
1291 (shr-stylesheet (nconc (list (cons 'color color))
1292 shr-stylesheet)))
1293 (shr-generic cont)
1294 (shr-colorize-region start (point) color
1295 (cdr (assq 'background-color shr-stylesheet)))))
1296
1297 (defun shr-tag-del (cont)
1298 (let* ((start (point))
1299 (color "red")
1300 (shr-stylesheet (nconc (list (cons 'color color))
1301 shr-stylesheet)))
1302 (shr-fontize-dom cont 'shr-strike-through)
1303 (shr-colorize-region start (point) color
1304 (cdr (assq 'background-color shr-stylesheet)))))
1305
1306 (defun shr-parse-style (style)
1307 (when style
1308 (save-match-data
1309 (when (string-match "\n" style)
1310 (setq style (replace-match " " t t style))))
1311 (let ((plist nil))
1312 (dolist (elem (split-string style ";"))
1313 (when elem
1314 (setq elem (split-string elem ":"))
1315 (when (and (car elem)
1316 (cadr elem))
1317 (let ((name (replace-regexp-in-string "^ +\\| +$" "" (car elem)))
1318 (value (replace-regexp-in-string "^ +\\| +$" "" (cadr elem))))
1319 (when (string-match " *!important\\'" value)
1320 (setq value (substring value 0 (match-beginning 0))))
1321 (unless (equal value "inherit")
1322 (push (cons (intern name obarray)
1323 value)
1324 plist))))))
1325 plist)))
1326
1327 (defun shr-tag-base (dom)
1328 (when-let (base (dom-attr dom 'href))
1329 (setq shr-base (shr-parse-base base)))
1330 (shr-generic dom))
1331
1332 (defun shr-tag-a (dom)
1333 (let ((url (dom-attr dom 'href))
1334 (title (dom-attr dom 'title))
1335 (start (point))
1336 shr-start)
1337 (shr-generic dom)
1338 (when (and shr-target-id
1339 (equal (dom-attr dom 'name) shr-target-id))
1340 ;; We have a zero-length <a name="foo"> element, so just
1341 ;; insert... something.
1342 (when (= start (point))
1343 (shr-ensure-newline)
1344 (insert " "))
1345 (put-text-property start (1+ start) 'shr-target-id shr-target-id))
1346 (when url
1347 (shr-urlify (or shr-start start) (shr-expand-url url) title))))
1348
1349 (defun shr-tag-object (dom)
1350 (unless shr-inhibit-images
1351 (let ((start (point))
1352 url multimedia image)
1353 (when-let (type (dom-attr dom 'type))
1354 (when (string-match "\\`image/svg" type)
1355 (setq url (dom-attr dom 'data)
1356 image t)))
1357 (dolist (child (dom-non-text-children dom))
1358 (cond
1359 ((eq (dom-tag child) 'embed)
1360 (setq url (or url (dom-attr child 'src))
1361 multimedia t))
1362 ((and (eq (dom-tag child) 'param)
1363 (equal (dom-attr child 'name) "movie"))
1364 (setq url (or url (dom-attr child 'value))
1365 multimedia t))))
1366 (when url
1367 (cond
1368 (image
1369 (shr-tag-img dom url)
1370 (setq dom nil))
1371 (multimedia
1372 (shr-insert " [multimedia] ")
1373 (shr-urlify start (shr-expand-url url)))))
1374 (when dom
1375 (shr-generic dom)))))
1376
1377 (defcustom shr-prefer-media-type-alist '(("webm" . 1.0)
1378 ("ogv" . 1.0)
1379 ("ogg" . 1.0)
1380 ("opus" . 1.0)
1381 ("flac" . 0.9)
1382 ("wav" . 0.5))
1383 "Preferences for media types.
1384 The key element should be a regexp matched against the type of the source or
1385 url if no type is specified. The value should be a float in the range 0.0 to
1386 1.0. Media elements with higher value are preferred."
1387 :version "24.4"
1388 :group 'shr
1389 :type '(alist :key-type regexp :value-type float))
1390
1391 (defun shr--get-media-pref (elem)
1392 "Determine the preference for ELEM.
1393 The preference is a float determined from `shr-prefer-media-type'."
1394 (let ((type (dom-attr elem 'type))
1395 (p 0.0))
1396 (unless type
1397 (setq type (dom-attr elem 'src)))
1398 (when type
1399 (dolist (pref shr-prefer-media-type-alist)
1400 (when (and
1401 (> (cdr pref) p)
1402 (string-match-p (car pref) type))
1403 (setq p (cdr pref)))))
1404 p))
1405
1406 (defun shr--extract-best-source (dom &optional url pref)
1407 "Extract the best `:src' property from <source> blocks in DOM."
1408 (setq pref (or pref -1.0))
1409 (let (new-pref)
1410 (dolist (elem (dom-non-text-children dom))
1411 (when (and (eq (dom-tag elem) 'source)
1412 (< pref
1413 (setq new-pref
1414 (shr--get-media-pref elem))))
1415 (setq pref new-pref
1416 url (dom-attr elem 'src))
1417 ;; libxml's html parser isn't HTML5 compliant and non terminated
1418 ;; source tags might end up as children. So recursion it is...
1419 (dolist (child (dom-non-text-children elem))
1420 (when (eq (dom-tag child) 'source)
1421 (let ((ret (shr--extract-best-source (list child) url pref)))
1422 (when (< pref (cdr ret))
1423 (setq url (car ret)
1424 pref (cdr ret)))))))))
1425 (cons url pref))
1426
1427 (defun shr-tag-video (dom)
1428 (let ((image (dom-attr dom 'poster))
1429 (url (dom-attr dom 'src))
1430 (start (point)))
1431 (unless url
1432 (setq url (car (shr--extract-best-source dom))))
1433 (if (> (length image) 0)
1434 (shr-tag-img nil image)
1435 (shr-insert " [video] "))
1436 (shr-urlify start (shr-expand-url url))))
1437
1438 (defun shr-tag-audio (dom)
1439 (let ((url (dom-attr dom 'src))
1440 (start (point)))
1441 (unless url
1442 (setq url (car (shr--extract-best-source dom))))
1443 (shr-insert " [audio] ")
1444 (shr-urlify start (shr-expand-url url))))
1445
1446 (defun shr-tag-img (dom &optional url)
1447 (when (or url
1448 (and dom
1449 (> (length (dom-attr dom 'src)) 0)))
1450 (when (> (current-column) 0)
1451 (insert "\n"))
1452 (let ((alt (dom-attr dom 'alt))
1453 (width (shr-string-number (dom-attr dom 'width)))
1454 (height (shr-string-number (dom-attr dom 'height)))
1455 (url (shr-expand-url (or url (dom-attr dom 'src)))))
1456 (let ((start (point-marker)))
1457 (when (zerop (length alt))
1458 (setq alt "*"))
1459 (cond
1460 ((or (member (dom-attr dom 'height) '("0" "1"))
1461 (member (dom-attr dom 'width) '("0" "1")))
1462 ;; Ignore zero-sized or single-pixel images.
1463 )
1464 ((and (not shr-inhibit-images)
1465 (string-match "\\`data:" url))
1466 (let ((image (shr-image-from-data (substring url (match-end 0)))))
1467 (if image
1468 (funcall shr-put-image-function image alt
1469 (list :width width :height height))
1470 (insert alt))))
1471 ((and (not shr-inhibit-images)
1472 (string-match "\\`cid:" url))
1473 (let ((url (substring url (match-end 0)))
1474 image)
1475 (if (or (not shr-content-function)
1476 (not (setq image (funcall shr-content-function url))))
1477 (insert alt)
1478 (funcall shr-put-image-function image alt
1479 (list :width width :height height)))))
1480 ((or shr-inhibit-images
1481 (and shr-blocked-images
1482 (string-match shr-blocked-images url)))
1483 (setq shr-start (point))
1484 (shr-insert alt))
1485 ((and (not shr-ignore-cache)
1486 (url-is-cached (shr-encode-url url)))
1487 (funcall shr-put-image-function (shr-get-image-data url) alt
1488 (list :width width :height height)))
1489 (t
1490 (when (and shr-ignore-cache
1491 (url-is-cached (shr-encode-url url)))
1492 (let ((file (url-cache-create-filename (shr-encode-url url))))
1493 (when (file-exists-p file)
1494 (delete-file file))))
1495 (when (image-type-available-p 'svg)
1496 (insert-image
1497 (shr-make-placeholder-image dom)
1498 (or alt "")))
1499 (insert " ")
1500 (url-queue-retrieve
1501 (shr-encode-url url) 'shr-image-fetched
1502 (list (current-buffer) start (set-marker (make-marker) (point))
1503 (list :width width :height height))
1504 t t)))
1505 (when (zerop shr-table-depth) ;; We are not in a table.
1506 (put-text-property start (point) 'keymap shr-image-map)
1507 (put-text-property start (point) 'shr-alt alt)
1508 (put-text-property start (point) 'image-url url)
1509 (put-text-property start (point) 'image-displayer
1510 (shr-image-displayer shr-content-function))
1511 (put-text-property start (point) 'help-echo
1512 (shr-fill-text
1513 (or (dom-attr dom 'title) alt))))))))
1514
1515 (defun shr-string-number (string)
1516 (if (null string)
1517 nil
1518 (setq string (replace-regexp-in-string "[^0-9]" "" string))
1519 (if (zerop (length string))
1520 nil
1521 (string-to-number string))))
1522
1523 (defun shr-make-placeholder-image (dom)
1524 (let* ((edges (and
1525 (get-buffer-window (current-buffer))
1526 (window-inside-pixel-edges
1527 (get-buffer-window (current-buffer)))))
1528 (scaling (image-compute-scaling-factor image-scaling-factor))
1529 (width (truncate
1530 (* (or (shr-string-number (dom-attr dom 'width)) 100)
1531 scaling)))
1532 (height (truncate
1533 (* (or (shr-string-number (dom-attr dom 'height)) 100)
1534 scaling)))
1535 (max-width
1536 (and edges
1537 (truncate (* shr-max-image-proportion
1538 (- (nth 2 edges) (nth 0 edges))))))
1539 (max-height (and edges
1540 (truncate (* shr-max-image-proportion
1541 (- (nth 3 edges) (nth 1 edges))))))
1542 svg image)
1543 (when (and max-width
1544 (> width max-width))
1545 (setq height (truncate (* (/ (float max-width) width) height))
1546 width max-width))
1547 (when (and max-height
1548 (> height max-height))
1549 (setq width (truncate (* (/ (float max-height) height) width))
1550 height max-height))
1551 (setq svg (svg-create width height))
1552 (svg-gradient svg "background" 'linear '((0 . "#b0b0b0") (100 . "#808080")))
1553 (svg-rectangle svg 0 0 width height :gradient "background"
1554 :stroke-width 2 :stroke-color "black")
1555 (let ((image (svg-image svg)))
1556 (setf (image-property image :ascent) 100)
1557 image)))
1558
1559 (defun shr-tag-pre (dom)
1560 (let ((shr-folding-mode 'none)
1561 (shr-current-font 'default))
1562 (shr-ensure-newline)
1563 (shr-generic dom)
1564 (shr-ensure-newline)))
1565
1566 (defun shr-tag-blockquote (dom)
1567 (shr-ensure-paragraph)
1568 (let ((start (point))
1569 (shr-indentation (+ shr-indentation
1570 (* 4 shr-table-separator-pixel-width))))
1571 (shr-generic dom)
1572 (shr-ensure-paragraph)
1573 (shr-mark-fill start)))
1574
1575 (defun shr-tag-dl (dom)
1576 (shr-ensure-paragraph)
1577 (shr-generic dom)
1578 (shr-ensure-paragraph))
1579
1580 (defun shr-tag-dt (dom)
1581 (shr-ensure-newline)
1582 (shr-generic dom)
1583 (shr-ensure-newline))
1584
1585 (defun shr-tag-dd (dom)
1586 (shr-ensure-newline)
1587 (let ((shr-indentation (+ shr-indentation
1588 (* 4 shr-table-separator-pixel-width))))
1589 (shr-generic dom)))
1590
1591 (defun shr-tag-ul (dom)
1592 (shr-ensure-paragraph)
1593 (let ((shr-list-mode 'ul))
1594 (shr-generic dom))
1595 (shr-ensure-paragraph))
1596
1597 (defun shr-tag-ol (dom)
1598 (shr-ensure-paragraph)
1599 (let ((shr-list-mode 1))
1600 (shr-generic dom))
1601 (shr-ensure-paragraph))
1602
1603 (defun shr-tag-li (dom)
1604 (shr-ensure-newline)
1605 (let ((start (point)))
1606 (let* ((bullet
1607 (if (numberp shr-list-mode)
1608 (prog1
1609 (format "%d " shr-list-mode)
1610 (setq shr-list-mode (1+ shr-list-mode)))
1611 (car shr-internal-bullet)))
1612 (width (if (numberp shr-list-mode)
1613 (shr-string-pixel-width bullet)
1614 (cdr shr-internal-bullet))))
1615 (insert bullet)
1616 (shr-mark-fill start)
1617 (let ((shr-indentation (+ shr-indentation width)))
1618 (put-text-property start (1+ start)
1619 'shr-continuation-indentation shr-indentation)
1620 (put-text-property start (1+ start) 'shr-prefix-length (length bullet))
1621 (shr-generic dom)))))
1622
1623 (defun shr-mark-fill (start)
1624 ;; We may not have inserted any text to fill.
1625 (unless (= start (point))
1626 (put-text-property start (1+ start)
1627 'shr-indentation shr-indentation)))
1628
1629 (defun shr-tag-br (dom)
1630 (when (and (not (bobp))
1631 ;; Only add a newline if we break the current line, or
1632 ;; the previous line isn't a blank line.
1633 (or (not (bolp))
1634 (and (> (- (point) 2) (point-min))
1635 (not (= (char-after (- (point) 2)) ?\n)))))
1636 (insert "\n"))
1637 (shr-generic dom))
1638
1639 (defun shr-tag-span (dom)
1640 (shr-generic dom))
1641
1642 (defun shr-tag-h1 (dom)
1643 (shr-heading dom (if shr-use-fonts
1644 '(variable-pitch (:height 1.3 :weight bold))
1645 'bold)))
1646
1647 (defun shr-tag-h2 (dom)
1648 (shr-heading dom 'bold))
1649
1650 (defun shr-tag-h3 (dom)
1651 (shr-heading dom 'italic))
1652
1653 (defun shr-tag-h4 (dom)
1654 (shr-heading dom))
1655
1656 (defun shr-tag-h5 (dom)
1657 (shr-heading dom))
1658
1659 (defun shr-tag-h6 (dom)
1660 (shr-heading dom))
1661
1662 (defun shr-tag-hr (_dom)
1663 (shr-ensure-newline)
1664 (insert (make-string (if (not shr-use-fonts)
1665 shr-internal-width
1666 (1+ (/ shr-internal-width
1667 shr-table-separator-pixel-width)))
1668 shr-hr-line)
1669 "\n"))
1670
1671 (defun shr-tag-title (dom)
1672 (shr-heading dom 'bold 'underline))
1673
1674 (defun shr-tag-font (dom)
1675 (let* ((start (point))
1676 (color (dom-attr dom 'color))
1677 (shr-stylesheet (nconc (list (cons 'color color))
1678 shr-stylesheet)))
1679 (shr-generic dom)
1680 (when color
1681 (shr-colorize-region start (point) color
1682 (cdr (assq 'background-color shr-stylesheet))))))
1683
1684 (defun shr-tag-bdo (dom)
1685 (let* ((direction (dom-attr dom 'dir))
1686 (char (cond
1687 ((equal direction "ltr")
1688 #x202d) ; LRO
1689 ((equal direction "rtl")
1690 #x202e)))) ; RLO
1691 (when char
1692 (insert char))
1693 (shr-generic dom)
1694 (when char
1695 (insert #x202c)))) ; PDF
1696
1697 (defun shr-tag-bdi (dom)
1698 (insert #x2068) ; FSI
1699 (shr-generic dom)
1700 (insert #x2069)) ; PDI
1701
1702 ;;; Table rendering algorithm.
1703
1704 ;; Table rendering is the only complicated thing here. We do this by
1705 ;; first counting how many TDs there are in each TR, and registering
1706 ;; how wide they think they should be ("width=45%", etc). Then we
1707 ;; render each TD separately (this is done in temporary buffers, so
1708 ;; that we can use all the rendering machinery as if we were in the
1709 ;; main buffer). Now we know how much space each TD really takes, so
1710 ;; we then render everything again with the new widths, and finally
1711 ;; insert all these boxes into the main buffer.
1712 (defun shr-tag-table-1 (dom)
1713 (setq dom (or (dom-child-by-tag dom 'tbody) dom))
1714 (let* ((shr-inhibit-images t)
1715 (shr-table-depth (1+ shr-table-depth))
1716 (shr-kinsoku-shorten t)
1717 ;; Find all suggested widths.
1718 (columns (shr-column-specs dom))
1719 ;; Compute how many pixels wide each TD should be.
1720 (suggested-widths (shr-pro-rate-columns columns))
1721 ;; Do a "test rendering" to see how big each TD is (this can
1722 ;; be smaller (if there's little text) or bigger (if there's
1723 ;; unbreakable text).
1724 (elems (or (dom-attr dom 'shr-suggested-widths)
1725 (shr-make-table dom suggested-widths nil
1726 'shr-suggested-widths)))
1727 (sketch (loop for line in elems
1728 collect (mapcar #'car line)))
1729 (natural (loop for line in elems
1730 collect (mapcar #'cdr line)))
1731 (sketch-widths (shr-table-widths sketch natural suggested-widths)))
1732 ;; This probably won't work very well.
1733 (when (> (+ (loop for width across sketch-widths
1734 summing (1+ width))
1735 shr-indentation shr-table-separator-pixel-width)
1736 (frame-width))
1737 (setq truncate-lines t))
1738 ;; Then render the table again with these new "hard" widths.
1739 (shr-insert-table (shr-make-table dom sketch-widths t) sketch-widths)))
1740
1741 (defun shr-table-body (dom)
1742 (let ((tbodies (seq-filter (lambda (child)
1743 (eq (dom-tag child) 'tbody))
1744 (dom-non-text-children dom))))
1745 (cond
1746 ((null tbodies)
1747 dom)
1748 ((= (length tbodies) 1)
1749 (car tbodies))
1750 (t
1751 ;; Table with multiple tbodies. Convert into a single tbody.
1752 `(tbody nil ,@(cl-reduce 'append
1753 (mapcar 'dom-non-text-children tbodies)))))))
1754
1755 (defun shr-tag-table (dom)
1756 (shr-ensure-paragraph)
1757 (let* ((caption (dom-children (dom-child-by-tag dom 'caption)))
1758 (header (dom-non-text-children (dom-child-by-tag dom 'thead)))
1759 (body (dom-non-text-children (shr-table-body dom)))
1760 (footer (dom-non-text-children (dom-child-by-tag dom 'tfoot)))
1761 (bgcolor (dom-attr dom 'bgcolor))
1762 (start (point))
1763 (shr-stylesheet (nconc (list (cons 'background-color bgcolor))
1764 shr-stylesheet))
1765 (nheader (if header (shr-max-columns header)))
1766 (nbody (if body (shr-max-columns body) 0))
1767 (nfooter (if footer (shr-max-columns footer))))
1768 (if (and (not caption)
1769 (not header)
1770 (not (dom-child-by-tag dom 'tbody))
1771 (not (dom-child-by-tag dom 'tr))
1772 (not footer))
1773 ;; The table is totally invalid and just contains random junk.
1774 ;; Try to output it anyway.
1775 (shr-generic dom)
1776 ;; It's a real table, so render it.
1777 (if (dom-attr dom 'shr-fixed-table)
1778 (shr-tag-table-1 dom)
1779 ;; Only fix up the table once.
1780 (let ((table
1781 (nconc
1782 (list 'table nil)
1783 (if caption `((tr nil (td nil ,@caption))))
1784 (cond
1785 (header
1786 (if footer
1787 ;; header + body + footer
1788 (if (= nheader nbody)
1789 (if (= nbody nfooter)
1790 `((tr nil (td nil (table nil
1791 (tbody nil ,@header
1792 ,@body ,@footer)))))
1793 (nconc `((tr nil (td nil (table nil
1794 (tbody nil ,@header
1795 ,@body)))))
1796 (if (= nfooter 1)
1797 footer
1798 `((tr nil (td nil (table
1799 nil (tbody
1800 nil ,@footer))))))))
1801 (nconc `((tr nil (td nil (table nil (tbody
1802 nil ,@header)))))
1803 (if (= nbody nfooter)
1804 `((tr nil (td nil (table
1805 nil (tbody nil ,@body
1806 ,@footer)))))
1807 (nconc `((tr nil (td nil (table
1808 nil (tbody nil
1809 ,@body)))))
1810 (if (= nfooter 1)
1811 footer
1812 `((tr nil (td nil (table
1813 nil
1814 (tbody
1815 nil
1816 ,@footer))))))))))
1817 ;; header + body
1818 (if (= nheader nbody)
1819 `((tr nil (td nil (table nil (tbody nil ,@header
1820 ,@body)))))
1821 (if (= nheader 1)
1822 `(,@header (tr nil (td nil (table
1823 nil (tbody nil ,@body)))))
1824 `((tr nil (td nil (table nil (tbody nil ,@header))))
1825 (tr nil (td nil (table nil (tbody nil ,@body)))))))))
1826 (footer
1827 ;; body + footer
1828 (if (= nbody nfooter)
1829 `((tr nil (td nil (table
1830 nil (tbody nil ,@body ,@footer)))))
1831 (nconc `((tr nil (td nil (table nil (tbody nil ,@body)))))
1832 (if (= nfooter 1)
1833 footer
1834 `((tr nil (td nil (table
1835 nil (tbody nil ,@footer)))))))))
1836 (caption
1837 `((tr nil (td nil (table nil (tbody nil ,@body))))))
1838 (body)))))
1839 (dom-set-attribute table 'shr-fixed-table t)
1840 (setcdr dom (cdr table))
1841 (shr-tag-table-1 dom))))
1842 (when bgcolor
1843 (shr-colorize-region start (point) (cdr (assq 'color shr-stylesheet))
1844 bgcolor))
1845 ;; Finally, insert all the images after the table. The Emacs buffer
1846 ;; model isn't strong enough to allow us to put the images actually
1847 ;; into the tables.
1848 (when (zerop shr-table-depth)
1849 (save-excursion
1850 (shr-expand-alignments start (point)))
1851 (dolist (elem (dom-by-tag dom 'object))
1852 (shr-tag-object elem))
1853 (dolist (elem (dom-by-tag dom 'img))
1854 (shr-tag-img elem)))))
1855
1856 (defun shr-insert-table (table widths)
1857 (let* ((collapse (equal (cdr (assq 'border-collapse shr-stylesheet))
1858 "collapse"))
1859 (shr-table-separator-length (if collapse 0 1))
1860 (shr-table-vertical-line (if collapse "" shr-table-vertical-line))
1861 (start (point)))
1862 (setq shr-table-id (1+ shr-table-id))
1863 (unless collapse
1864 (shr-insert-table-ruler widths))
1865 (dolist (row table)
1866 (let ((start (point))
1867 (align 0)
1868 (column-number 0)
1869 (height (let ((max 0))
1870 (dolist (column row)
1871 (setq max (max max (nth 2 column))))
1872 max)))
1873 (dotimes (i (max height 1))
1874 (shr-indent)
1875 (insert shr-table-vertical-line "\n"))
1876 (dolist (column row)
1877 (when (> (nth 2 column) -1)
1878 (goto-char start)
1879 ;; Sum up all the widths from the column. (There may be
1880 ;; more than one if this is a "colspan" column.)
1881 (dotimes (i (nth 4 column))
1882 ;; The colspan directive may be wrong and there may not be
1883 ;; that number of columns.
1884 (when (<= column-number (1- (length widths)))
1885 (setq align (+ align
1886 (aref widths column-number)
1887 (* 2 shr-table-separator-pixel-width))))
1888 (setq column-number (1+ column-number)))
1889 (let ((lines (nth 3 column))
1890 (pixel-align (if (not shr-use-fonts)
1891 (* align (frame-char-width))
1892 align)))
1893 (dolist (line lines)
1894 (end-of-line)
1895 (let ((start (point))
1896 (background (and (> (length line) 0)
1897 (shr-face-background
1898 (get-text-property
1899 (1- (length line)) 'face line))))
1900 (space (propertize
1901 " "
1902 'display `(space :align-to (,pixel-align))
1903 'shr-table-indent shr-table-id)))
1904 (when background
1905 (setq space (propertize space 'face background)))
1906 (insert line space shr-table-vertical-line)
1907 (shr-colorize-region
1908 start (1- (point)) (nth 5 column) (nth 6 column)))
1909 (forward-line 1))
1910 ;; Add blank lines at padding at the bottom of the TD,
1911 ;; possibly.
1912 (dotimes (i (- height (length lines)))
1913 (end-of-line)
1914 (let ((start (point)))
1915 (insert (propertize " "
1916 'display `(space :align-to (,pixel-align))
1917 'shr-table-indent shr-table-id)
1918 shr-table-vertical-line)
1919 (shr-colorize-region
1920 start (1- (point)) (nth 5 column) (nth 6 column)))
1921 (forward-line 1))))))
1922 (unless collapse
1923 (shr-insert-table-ruler widths)))
1924 (unless (= start (point))
1925 (put-text-property start (1+ start) 'shr-table-id shr-table-id))))
1926
1927 (defun shr-face-background (face)
1928 (and (consp face)
1929 (or (and (plist-get face :background)
1930 (list :background (plist-get face :background)))
1931 (let ((background nil))
1932 (dolist (elem face)
1933 (when (and (consp elem)
1934 (eq (car elem) :background)
1935 (not background))
1936 (setq background (cadr elem))))
1937 (and background
1938 (list :background background))))))
1939
1940 (defun shr-expand-alignments (start end)
1941 (while (< (setq start (next-single-property-change
1942 start 'shr-table-id nil end))
1943 end)
1944 (goto-char start)
1945 (let* ((shr-use-fonts t)
1946 (id (get-text-property (point) 'shr-table-id))
1947 (base (shr-pixel-column))
1948 elem)
1949 (when id
1950 (save-excursion
1951 (while (setq elem (text-property-any
1952 (point) end 'shr-table-indent id))
1953 (goto-char elem)
1954 (let ((align (get-text-property (point) 'display)))
1955 (put-text-property (point) (1+ (point)) 'display
1956 `(space :align-to (,(+ (car (nth 2 align))
1957 base)))))
1958 (forward-char 1)))))
1959 (setq start (1+ start))))
1960
1961 (defun shr-insert-table-ruler (widths)
1962 (when shr-table-horizontal-line
1963 (when (and (bolp)
1964 (> shr-indentation 0))
1965 (shr-indent))
1966 (insert shr-table-corner)
1967 (let ((total-width 0))
1968 (dotimes (i (length widths))
1969 (setq total-width (+ total-width (aref widths i)
1970 (* shr-table-separator-pixel-width 2)))
1971 (insert (make-string (1+ (/ (aref widths i)
1972 shr-table-separator-pixel-width))
1973 shr-table-horizontal-line)
1974 (propertize " "
1975 'display `(space :align-to (,total-width))
1976 'shr-table-indent shr-table-id)
1977 shr-table-corner)))
1978 (insert "\n")))
1979
1980 (defun shr-table-widths (table natural-table suggested-widths)
1981 (let* ((length (length suggested-widths))
1982 (widths (make-vector length 0))
1983 (natural-widths (make-vector length 0)))
1984 (dolist (row table)
1985 (let ((i 0))
1986 (dolist (column row)
1987 (aset widths i (max (aref widths i) column))
1988 (setq i (1+ i)))))
1989 (dolist (row natural-table)
1990 (let ((i 0))
1991 (dolist (column row)
1992 (aset natural-widths i (max (aref natural-widths i) column))
1993 (setq i (1+ i)))))
1994 (let ((extra (- (apply '+ (append suggested-widths nil))
1995 (apply '+ (append widths nil))
1996 (* shr-table-separator-pixel-width (1+ (length widths)))))
1997 (expanded-columns 0))
1998 ;; We have extra, unused space, so divide this space amongst the
1999 ;; columns.
2000 (when (> extra 0)
2001 ;; If the natural width is wider than the rendered width, we
2002 ;; want to allow the column to expand.
2003 (dotimes (i length)
2004 (when (> (aref natural-widths i) (aref widths i))
2005 (setq expanded-columns (1+ expanded-columns))))
2006 (dotimes (i length)
2007 (when (> (aref natural-widths i) (aref widths i))
2008 (aset widths i (min
2009 (aref natural-widths i)
2010 (+ (/ extra expanded-columns)
2011 (aref widths i))))))))
2012 widths))
2013
2014 (defun shr-make-table (dom widths &optional fill storage-attribute)
2015 (or (cadr (assoc (list dom widths fill) shr-content-cache))
2016 (let ((data (shr-make-table-1 dom widths fill)))
2017 (push (list (list dom widths fill) data)
2018 shr-content-cache)
2019 (when storage-attribute
2020 (dom-set-attribute dom storage-attribute data))
2021 data)))
2022
2023 (defun shr-make-table-1 (dom widths &optional fill)
2024 (let ((trs nil)
2025 (rowspans (make-vector (length widths) 0))
2026 (colspan-remaining 0)
2027 colspan-width colspan-count
2028 width colspan)
2029 (dolist (row (dom-non-text-children dom))
2030 (when (eq (dom-tag row) 'tr)
2031 (let ((tds nil)
2032 (columns (dom-non-text-children row))
2033 (i 0)
2034 (width-column 0)
2035 column)
2036 (while (< i (length widths))
2037 ;; If we previously had a rowspan definition, then that
2038 ;; means that we now have a "missing" td/th element here.
2039 ;; So just insert a dummy, empty one to (sort of) emulate
2040 ;; rowspan.
2041 (setq column
2042 (if (zerop (aref rowspans i))
2043 (pop columns)
2044 (aset rowspans i (1- (aref rowspans i)))
2045 '(td)))
2046 (when (and (not (stringp column))
2047 (or (memq (dom-tag column) '(td th))
2048 (not column)))
2049 (when-let (span (dom-attr column 'rowspan))
2050 (aset rowspans i (+ (aref rowspans i)
2051 (1- (string-to-number span)))))
2052 ;; Sanity check for invalid column-spans.
2053 (when (>= width-column (length widths))
2054 (setq width-column 0))
2055 (setq width
2056 (if column
2057 (aref widths width-column)
2058 (* 10 shr-table-separator-pixel-width)))
2059 (when (setq colspan (dom-attr column 'colspan))
2060 (setq colspan (min (string-to-number colspan)
2061 ;; The colspan may be wrong, so
2062 ;; truncate it to the length of the
2063 ;; remaining columns.
2064 (- (length widths) i)))
2065 (dotimes (j (1- colspan))
2066 (setq width
2067 (if (> (+ i 1 j) (1- (length widths)))
2068 ;; If we have a colspan spec that's longer
2069 ;; than the table is wide, just use the last
2070 ;; width as the width.
2071 (aref widths (1- (length widths)))
2072 ;; Sum up the widths of the columns we're
2073 ;; spanning.
2074 (+ width
2075 shr-table-separator-length
2076 (aref widths (+ i 1 j))))))
2077 (setq width-column (+ width-column (1- colspan))
2078 colspan-count colspan
2079 colspan-remaining colspan))
2080 (when column
2081 (let ((data (shr-render-td column width fill)))
2082 (if (and (not fill)
2083 (> colspan-remaining 0))
2084 (progn
2085 (setq colspan-width (car data))
2086 (let ((this-width (/ colspan-width colspan-count)))
2087 (push (cons this-width (cadr data)) tds)
2088 (setq colspan-remaining (1- colspan-remaining))))
2089 (if (not fill)
2090 (push (cons (car data) (cadr data)) tds)
2091 (push data tds)))))
2092 (when (and colspan
2093 (> colspan 1))
2094 (dotimes (c (1- colspan))
2095 (setq i (1+ i))
2096 (push
2097 (if fill
2098 (list 0 0 -1 nil 1 nil nil)
2099 '(0 . 0))
2100 tds)))
2101 (setq i (1+ i)
2102 width-column (1+ width-column))))
2103 (push (nreverse tds) trs))))
2104 (nreverse trs)))
2105
2106 (defun shr-pixel-buffer-width ()
2107 (if (not shr-use-fonts)
2108 (save-excursion
2109 (goto-char (point-min))
2110 (let ((max 0))
2111 (while (not (eobp))
2112 (end-of-line)
2113 (setq max (max max (current-column)))
2114 (forward-line 1))
2115 max))
2116 (if (get-buffer-window)
2117 (car (window-text-pixel-size nil (point-min) (point-max)))
2118 (save-window-excursion
2119 (set-window-buffer nil (current-buffer))
2120 (car (window-text-pixel-size nil (point-min) (point-max)))))))
2121
2122 (defun shr-render-td (dom width fill)
2123 (let ((cache (intern (format "shr-td-cache-%s-%s" width fill))))
2124 (or (dom-attr dom cache)
2125 (and fill
2126 (let (result)
2127 (dolist (attr (dom-attributes dom))
2128 (let ((name (symbol-name (car attr))))
2129 (when (string-match "shr-td-cache-\\([0-9]+\\)-nil" name)
2130 (let ((cache-width (string-to-number
2131 (match-string 1 name))))
2132 (when (and (>= cache-width width)
2133 (<= (car (cdr attr)) width))
2134 (setq result (cdr attr)))))))
2135 result))
2136 (let ((result (shr-render-td-1 dom width fill)))
2137 (dom-set-attribute dom cache result)
2138 result))))
2139
2140 (defun shr-render-td-1 (dom width fill)
2141 (with-temp-buffer
2142 (let ((bgcolor (dom-attr dom 'bgcolor))
2143 (fgcolor (dom-attr dom 'fgcolor))
2144 (style (dom-attr dom 'style))
2145 (shr-stylesheet shr-stylesheet)
2146 (max-width 0)
2147 natural-width)
2148 (when style
2149 (setq style (and (string-match "color" style)
2150 (shr-parse-style style))))
2151 (when bgcolor
2152 (setq style (nconc (list (cons 'background-color bgcolor))
2153 style)))
2154 (when fgcolor
2155 (setq style (nconc (list (cons 'color fgcolor)) style)))
2156 (when style
2157 (setq shr-stylesheet (append style shr-stylesheet)))
2158 (let ((shr-internal-width width)
2159 (shr-indentation 0))
2160 (shr-descend dom))
2161 (save-window-excursion
2162 (set-window-buffer nil (current-buffer))
2163 (unless fill
2164 (setq natural-width
2165 (or (dom-attr dom 'shr-td-cache-natural)
2166 (let ((natural (max (shr-pixel-buffer-width)
2167 (shr-dom-max-natural-width dom 0))))
2168 (dom-set-attribute dom 'shr-td-cache-natural natural)
2169 natural))))
2170 (if (and natural-width
2171 (<= natural-width width))
2172 (setq max-width natural-width)
2173 (let ((shr-internal-width width))
2174 (shr-fill-lines (point-min) (point-max))
2175 (setq max-width (shr-pixel-buffer-width)))))
2176 (goto-char (point-max))
2177 ;; Delete padding at the bottom of the TDs.
2178 (delete-region
2179 (point)
2180 (progn
2181 (skip-chars-backward " \t\n")
2182 (end-of-line)
2183 (point)))
2184 (goto-char (point-min))
2185 (list max-width
2186 natural-width
2187 (count-lines (point-min) (point-max))
2188 (split-string (buffer-string) "\n")
2189 (if (dom-attr dom 'colspan)
2190 (string-to-number (dom-attr dom 'colspan))
2191 1)
2192 (cdr (assq 'color shr-stylesheet))
2193 (cdr (assq 'background-color shr-stylesheet))))))
2194
2195 (defun shr-dom-max-natural-width (dom max)
2196 (if (eq (dom-tag dom) 'table)
2197 (max max (or
2198 (loop for line in (dom-attr dom 'shr-suggested-widths)
2199 maximize (+
2200 shr-table-separator-length
2201 (loop for elem in line
2202 summing
2203 (+ (cdr elem)
2204 (* 2 shr-table-separator-length)))))
2205 0))
2206 (dolist (child (dom-children dom))
2207 (unless (stringp child)
2208 (setq max (max (shr-dom-max-natural-width child max)))))
2209 max))
2210
2211 (defun shr-buffer-width ()
2212 (goto-char (point-min))
2213 (let ((max 0))
2214 (while (not (eobp))
2215 (end-of-line)
2216 (setq max (max max (current-column)))
2217 (forward-line 1))
2218 max))
2219
2220 (defun shr-pro-rate-columns (columns)
2221 (let ((total-percentage 0)
2222 (widths (make-vector (length columns) 0)))
2223 (dotimes (i (length columns))
2224 (setq total-percentage (+ total-percentage (aref columns i))))
2225 (setq total-percentage (/ 1.0 total-percentage))
2226 (dotimes (i (length columns))
2227 (aset widths i (max (truncate (* (aref columns i)
2228 total-percentage
2229 (- shr-internal-width
2230 (* (1+ (length columns))
2231 shr-table-separator-pixel-width))))
2232 10)))
2233 widths))
2234
2235 ;; Return a summary of the number and shape of the TDs in the table.
2236 (defun shr-column-specs (dom)
2237 (let ((columns (make-vector (shr-max-columns dom) 1)))
2238 (dolist (row (dom-non-text-children dom))
2239 (when (eq (dom-tag row) 'tr)
2240 (let ((i 0))
2241 (dolist (column (dom-non-text-children row))
2242 (when (memq (dom-tag column) '(td th))
2243 (let ((width (dom-attr column 'width)))
2244 (when (and width
2245 (string-match "\\([0-9]+\\)%" width)
2246 (not (zerop (setq width (string-to-number
2247 (match-string 1 width))))))
2248 (aset columns i (/ width 100.0))))
2249 (setq i (1+ i)))))))
2250 columns))
2251
2252 (defun shr-count (dom elem)
2253 (let ((i 0))
2254 (dolist (sub (dom-children dom))
2255 (when (and (not (stringp sub))
2256 (eq (dom-tag sub) elem))
2257 (setq i (1+ i))))
2258 i))
2259
2260 (defun shr-max-columns (dom)
2261 (let ((max 0))
2262 (dolist (row (dom-children dom))
2263 (when (and (not (stringp row))
2264 (eq (dom-tag row) 'tr))
2265 (setq max (max max (+ (shr-count row 'td)
2266 (shr-count row 'th))))))
2267 max))
2268
2269 (provide 'shr)
2270
2271 ;;; shr.el ends here