Merge changes made in Gnus trunk.
[emacs.git] / lisp / gnus / shr.el
blob67effc07ee24f035c4c4026918f007719402c648
1 ;;; shr.el --- Simple HTML Renderer
3 ;; Copyright (C) 2010-2011 Free Software Foundation, Inc.
5 ;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
6 ;; Keywords: html
8 ;; This file is part of GNU Emacs.
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.
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.
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/>.
23 ;;; Commentary:
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.
31 ;;; Code:
33 (eval-when-compile (require 'cl))
34 (require 'browse-url)
36 (defgroup shr nil
37 "Simple HTML Renderer"
38 :group 'mail)
40 (defcustom shr-max-image-proportion 0.9
41 "How big pictures displayed are in relation to the window they're in.
42 A value of 0.7 means that they are allowed to take up 70% of the
43 width and height of the window. If they are larger than this,
44 and Emacs supports it, then the images will be rescaled down to
45 fit these criteria."
46 :version "24.1"
47 :group 'shr
48 :type 'float)
50 (defcustom shr-blocked-images nil
51 "Images that have URLs matching this regexp will be blocked."
52 :version "24.1"
53 :group 'shr
54 :type 'regexp)
56 (defcustom shr-table-horizontal-line ?
57 "Character used to draw horizontal table lines."
58 :group 'shr
59 :type 'character)
61 (defcustom shr-table-vertical-line ?
62 "Character used to draw vertical table lines."
63 :group 'shr
64 :type 'character)
66 (defcustom shr-table-corner ?
67 "Character used to draw table corners."
68 :group 'shr
69 :type 'character)
71 (defcustom shr-hr-line ?-
72 "Character used to draw hr lines."
73 :group 'shr
74 :type 'character)
76 (defcustom shr-width fill-column
77 "Frame width to use for rendering.
78 May either be an integer specifying a fixed width in characters,
79 or nil, meaning that the full width of the window should be
80 used."
81 :type '(choice (integer :tag "Fixed width in characters")
82 (const :tag "Use the width of the window" nil))
83 :group 'shr)
85 (defvar shr-content-function nil
86 "If bound, this should be a function that will return the content.
87 This is used for cid: URLs, and the function is called with the
88 cid: URL as the argument.")
90 (defvar shr-put-image-function 'shr-put-image
91 "Function called to put image and alt string.")
93 (defface shr-strike-through '((t (:strike-through t)))
94 "Font for <s> elements."
95 :group 'shr)
97 (defface shr-link
98 '((t (:inherit link)))
99 "Font for link elements."
100 :group 'shr)
102 ;;; Internal variables.
104 (defvar shr-folding-mode nil)
105 (defvar shr-state nil)
106 (defvar shr-start nil)
107 (defvar shr-indentation 0)
108 (defvar shr-inhibit-images nil)
109 (defvar shr-list-mode nil)
110 (defvar shr-content-cache nil)
111 (defvar shr-kinsoku-shorten nil)
112 (defvar shr-table-depth 0)
113 (defvar shr-stylesheet nil)
114 (defvar shr-base nil)
116 (defvar shr-map
117 (let ((map (make-sparse-keymap)))
118 (define-key map "a" 'shr-show-alt-text)
119 (define-key map "i" 'shr-browse-image)
120 (define-key map "I" 'shr-insert-image)
121 (define-key map "u" 'shr-copy-url)
122 (define-key map "v" 'shr-browse-url)
123 (define-key map "o" 'shr-save-contents)
124 (define-key map "\r" 'shr-browse-url)
125 map))
127 ;; Public functions and commands.
129 (defun shr-visit-file (file)
130 (interactive "fHTML file name: ")
131 (pop-to-buffer "*html*")
132 (erase-buffer)
133 (shr-insert-document
134 (with-temp-buffer
135 (insert-file-contents file)
136 (libxml-parse-html-region (point-min) (point-max)))))
138 ;;;###autoload
139 (defun shr-insert-document (dom)
140 (setq shr-content-cache nil)
141 (let ((shr-state nil)
142 (shr-start nil)
143 (shr-base nil)
144 (shr-width (or shr-width (window-width))))
145 (shr-descend (shr-transform-dom dom))))
147 (defun shr-copy-url ()
148 "Copy the URL under point to the kill ring.
149 If called twice, then try to fetch the URL and see whether it
150 redirects somewhere else."
151 (interactive)
152 (let ((url (get-text-property (point) 'shr-url)))
153 (cond
154 ((not url)
155 (message "No URL under point"))
156 ;; Resolve redirected URLs.
157 ((equal url (car kill-ring))
158 (url-retrieve
160 (lambda (a)
161 (when (and (consp a)
162 (eq (car a) :redirect))
163 (with-temp-buffer
164 (insert (cadr a))
165 (goto-char (point-min))
166 ;; Remove common tracking junk from the URL.
167 (when (re-search-forward ".utm_.*" nil t)
168 (replace-match "" t t))
169 (message "Copied %s" (buffer-string))
170 (copy-region-as-kill (point-min) (point-max)))))))
171 ;; Copy the URL to the kill ring.
173 (with-temp-buffer
174 (insert url)
175 (copy-region-as-kill (point-min) (point-max))
176 (message "Copied %s" url))))))
178 (defun shr-show-alt-text ()
179 "Show the ALT text of the image under point."
180 (interactive)
181 (let ((text (get-text-property (point) 'shr-alt)))
182 (if (not text)
183 (message "No image under point")
184 (message "%s" text))))
186 (defun shr-browse-image (&optional copy-url)
187 "Browse the image under point.
188 If COPY-URL (the prefix if called interactively) is non-nil, copy
189 the URL of the image to the kill buffer instead."
190 (interactive "P")
191 (let ((url (get-text-property (point) 'image-url)))
192 (cond
193 ((not url)
194 (message "No image under point"))
195 (copy-url
196 (with-temp-buffer
197 (insert url)
198 (copy-region-as-kill (point-min) (point-max))
199 (message "Copied %s" url)))
201 (message "Browsing %s..." url)
202 (browse-url url)))))
204 (defun shr-insert-image ()
205 "Insert the image under point into the buffer."
206 (interactive)
207 (let ((url (get-text-property (point) 'image-url)))
208 (if (not url)
209 (message "No image under point")
210 (message "Inserting %s..." url)
211 (url-retrieve url 'shr-image-fetched
212 (list (current-buffer) (1- (point)) (point-marker))
213 t))))
215 ;;; Utility functions.
217 (defun shr-transform-dom (dom)
218 (let ((result (list (pop dom))))
219 (dolist (arg (pop dom))
220 (push (cons (intern (concat ":" (symbol-name (car arg))) obarray)
221 (cdr arg))
222 result))
223 (dolist (sub dom)
224 (if (stringp sub)
225 (push (cons 'text sub) result)
226 (push (shr-transform-dom sub) result)))
227 (nreverse result)))
229 (defun shr-descend (dom)
230 (let ((function (intern (concat "shr-tag-" (symbol-name (car dom))) obarray))
231 (style (cdr (assq :style (cdr dom))))
232 (shr-stylesheet shr-stylesheet)
233 (start (point)))
234 (when style
235 (if (string-match "color" style)
236 (setq shr-stylesheet (nconc (shr-parse-style style)
237 shr-stylesheet))
238 (setq style nil)))
239 (if (fboundp function)
240 (funcall function (cdr dom))
241 (shr-generic (cdr dom)))
242 ;; If style is set, then this node has set the color.
243 (when style
244 (shr-colorize-region start (point)
245 (cdr (assq 'color shr-stylesheet))
246 (cdr (assq 'background-color shr-stylesheet))))))
248 (defun shr-generic (cont)
249 (dolist (sub cont)
250 (cond
251 ((eq (car sub) 'text)
252 (shr-insert (cdr sub)))
253 ((listp (cdr sub))
254 (shr-descend sub)))))
256 (defmacro shr-char-breakable-p (char)
257 "Return non-nil if a line can be broken before and after CHAR."
258 `(aref fill-find-break-point-function-table ,char))
259 (defmacro shr-char-nospace-p (char)
260 "Return non-nil if no space is required before and after CHAR."
261 `(aref fill-nospace-between-words-table ,char))
263 ;; KINSOKU is a Japanese word meaning a rule that should not be violated.
264 ;; In Emacs, it is a term used for characters, e.g. punctuation marks,
265 ;; parentheses, and so on, that should not be placed in the beginning
266 ;; of a line or the end of a line.
267 (defmacro shr-char-kinsoku-bol-p (char)
268 "Return non-nil if a line ought not to begin with CHAR."
269 `(aref (char-category-set ,char) ?>))
270 (defmacro shr-char-kinsoku-eol-p (char)
271 "Return non-nil if a line ought not to end with CHAR."
272 `(aref (char-category-set ,char) ?<))
273 (unless (shr-char-kinsoku-bol-p (make-char 'japanese-jisx0208 33 35))
274 (load "kinsoku" nil t))
276 (defun shr-insert (text)
277 (when (and (eq shr-state 'image)
278 (not (string-match "\\`[ \t\n]+\\'" text)))
279 (insert "\n")
280 (setq shr-state nil))
281 (cond
282 ((eq shr-folding-mode 'none)
283 (insert text))
285 (when (and (string-match "\\`[ \t\n]" text)
286 (not (bolp))
287 (not (eq (char-after (1- (point))) ? )))
288 (insert " "))
289 (dolist (elem (split-string text))
290 (when (and (bolp)
291 (> shr-indentation 0))
292 (shr-indent))
293 ;; No space is needed behind a wide character categorized as
294 ;; kinsoku-bol, between characters both categorized as nospace,
295 ;; or at the beginning of a line.
296 (let (prev)
297 (when (and (> (current-column) shr-indentation)
298 (eq (preceding-char) ? )
299 (or (= (line-beginning-position) (1- (point)))
300 (and (shr-char-breakable-p
301 (setq prev (char-after (- (point) 2))))
302 (shr-char-kinsoku-bol-p prev))
303 (and (shr-char-nospace-p prev)
304 (shr-char-nospace-p (aref elem 0)))))
305 (delete-char -1)))
306 ;; The shr-start is a special variable that is used to pass
307 ;; upwards the first point in the buffer where the text really
308 ;; starts.
309 (unless shr-start
310 (setq shr-start (point)))
311 (insert elem)
312 (let (found)
313 (while (and (> (current-column) shr-width)
314 (progn
315 (setq found (shr-find-fill-point))
316 (not (eolp))))
317 (when (eq (preceding-char) ? )
318 (delete-char -1))
319 (insert "\n")
320 (unless found
321 (put-text-property (1- (point)) (point) 'shr-break t)
322 ;; No space is needed at the beginning of a line.
323 (when (eq (following-char) ? )
324 (delete-char 1)))
325 (when (> shr-indentation 0)
326 (shr-indent))
327 (end-of-line))
328 (insert " ")))
329 (unless (string-match "[ \t\n]\\'" text)
330 (delete-char -1)))))
332 (defun shr-find-fill-point ()
333 (when (> (move-to-column shr-width) shr-width)
334 (backward-char 1))
335 (let ((bp (point))
336 failed)
337 (while (not (or (setq failed (= (current-column) shr-indentation))
338 (eq (preceding-char) ? )
339 (eq (following-char) ? )
340 (shr-char-breakable-p (preceding-char))
341 (shr-char-breakable-p (following-char))
342 (if (eq (preceding-char) ?')
343 (not (memq (char-after (- (point) 2))
344 (list nil ?\n ? )))
345 (and (shr-char-kinsoku-bol-p (preceding-char))
346 (shr-char-breakable-p (following-char))
347 (not (shr-char-kinsoku-bol-p (following-char)))))
348 (shr-char-kinsoku-eol-p (following-char))))
349 (backward-char 1))
350 (if (and (not (or failed (eolp)))
351 (eq (preceding-char) ?'))
352 (while (not (or (setq failed (eolp))
353 (eq (following-char) ? )
354 (shr-char-breakable-p (following-char))
355 (shr-char-kinsoku-eol-p (following-char))))
356 (forward-char 1)))
357 (if failed
358 ;; There's no breakable point, so we give it up.
359 (let (found)
360 (goto-char bp)
361 (unless shr-kinsoku-shorten
362 (while (and (setq found (re-search-forward
363 "\\(\\c>\\)\\| \\|\\c<\\|\\c|"
364 (line-end-position) 'move))
365 (eq (preceding-char) ?')))
366 (if (and found (not (match-beginning 1)))
367 (goto-char (match-beginning 0)))))
369 (eolp)
370 ;; Don't put kinsoku-bol characters at the beginning of a line,
371 ;; or kinsoku-eol characters at the end of a line.
372 (cond
373 (shr-kinsoku-shorten
374 (while (and (not (memq (preceding-char) (list ?\C-@ ?\n ? )))
375 (shr-char-kinsoku-eol-p (preceding-char)))
376 (backward-char 1))
377 (when (setq failed (= (current-column) shr-indentation))
378 ;; There's no breakable point that doesn't violate kinsoku,
379 ;; so we look for the second best position.
380 (while (and (progn
381 (forward-char 1)
382 (<= (current-column) shr-width))
383 (progn
384 (setq bp (point))
385 (shr-char-kinsoku-eol-p (following-char)))))
386 (goto-char bp)))
387 ((shr-char-kinsoku-eol-p (preceding-char))
388 (if (shr-char-kinsoku-eol-p (following-char))
389 ;; There are consecutive kinsoku-eol characters.
390 (setq failed t)
391 (let ((count 4))
392 (while
393 (progn
394 (backward-char 1)
395 (and (> (setq count (1- count)) 0)
396 (not (memq (preceding-char) (list ?\C-@ ?\n ? )))
397 (or (shr-char-kinsoku-eol-p (preceding-char))
398 (shr-char-kinsoku-bol-p (following-char)))))))
399 (if (setq failed (= (current-column) shr-indentation))
400 ;; There's no breakable point that doesn't violate kinsoku,
401 ;; so we go to the second best position.
402 (if (looking-at "\\(\\c<+\\)\\c<")
403 (goto-char (match-end 1))
404 (forward-char 1)))))
406 (if (shr-char-kinsoku-bol-p (preceding-char))
407 ;; There are consecutive kinsoku-bol characters.
408 (setq failed t)
409 (let ((count 4))
410 (while (and (>= (setq count (1- count)) 0)
411 (shr-char-kinsoku-bol-p (following-char))
412 (shr-char-breakable-p (following-char)))
413 (forward-char 1))))))
414 (when (eq (following-char) ? )
415 (forward-char 1))))
416 (not failed)))
418 (defun shr-expand-url (url)
419 (cond
420 ;; Absolute URL.
421 ((or (not url)
422 (string-match "\\`[a-z]*:" url)
423 (not shr-base))
424 url)
425 ((and (not (string-match "/\\'" shr-base))
426 (not (string-match "\\`/" url)))
427 (concat shr-base "/" url))
429 (concat shr-base url))))
431 (defun shr-ensure-newline ()
432 (unless (zerop (current-column))
433 (insert "\n")))
435 (defun shr-ensure-paragraph ()
436 (unless (bobp)
437 (if (<= (current-column) shr-indentation)
438 (unless (save-excursion
439 (forward-line -1)
440 (looking-at " *$"))
441 (insert "\n"))
442 (if (save-excursion
443 (beginning-of-line)
444 (looking-at " *$"))
445 (insert "\n")
446 (insert "\n\n")))))
448 (defun shr-indent ()
449 (when (> shr-indentation 0)
450 (insert (make-string shr-indentation ? ))))
452 (defun shr-fontize-cont (cont &rest types)
453 (let (shr-start)
454 (shr-generic cont)
455 (dolist (type types)
456 (shr-add-font (or shr-start (point)) (point) type))))
458 ;; Add an overlay in the region, but avoid putting the font properties
459 ;; on blank text at the start of the line, and the newline at the end,
460 ;; to avoid ugliness.
461 (defun shr-add-font (start end type)
462 (save-excursion
463 (goto-char start)
464 (while (< (point) end)
465 (when (bolp)
466 (skip-chars-forward " "))
467 (let ((overlay (make-overlay (point) (min (line-end-position) end))))
468 (overlay-put overlay 'face type))
469 (if (< (line-end-position) end)
470 (forward-line 1)
471 (goto-char end)))))
473 (defun shr-browse-url ()
474 "Browse the URL under point."
475 (interactive)
476 (let ((url (get-text-property (point) 'shr-url)))
477 (cond
478 ((not url)
479 (message "No link under point"))
480 ((string-match "^mailto:" url)
481 (browse-url-mailto url))
483 (browse-url url)))))
485 (defun shr-save-contents (directory)
486 "Save the contents from URL in a file."
487 (interactive "DSave contents of URL to directory: ")
488 (let ((url (get-text-property (point) 'shr-url)))
489 (if (not url)
490 (message "No link under point")
491 (url-retrieve (shr-encode-url url)
492 'shr-store-contents (list url directory)))))
494 (defun shr-store-contents (status url directory)
495 (unless (plist-get status :error)
496 (when (or (search-forward "\n\n" nil t)
497 (search-forward "\r\n\r\n" nil t))
498 (write-region (point) (point-max)
499 (expand-file-name (file-name-nondirectory url)
500 directory)))))
502 (defun shr-image-fetched (status buffer start end)
503 (when (and (buffer-name buffer)
504 (not (plist-get status :error)))
505 (url-store-in-cache (current-buffer))
506 (when (or (search-forward "\n\n" nil t)
507 (search-forward "\r\n\r\n" nil t))
508 (let ((data (buffer-substring (point) (point-max))))
509 (with-current-buffer buffer
510 (save-excursion
511 (let ((alt (buffer-substring start end))
512 (inhibit-read-only t))
513 (delete-region start end)
514 (goto-char start)
515 (funcall shr-put-image-function data alt)))))))
516 (kill-buffer (current-buffer)))
518 (defun shr-put-image (data alt)
519 "Put image DATA with a string ALT. Return image."
520 (if (display-graphic-p)
521 (let ((image (ignore-errors
522 (shr-rescale-image data))))
523 (when image
524 ;; When inserting big-ish pictures, put them at the
525 ;; beginning of the line.
526 (when (and (> (current-column) 0)
527 (> (car (image-size image t)) 400))
528 (insert "\n"))
529 (insert-image image (or alt "*")))
530 image)
531 (insert alt)))
533 (defun shr-rescale-image (data)
534 (if (or (not (fboundp 'imagemagick-types))
535 (not (get-buffer-window (current-buffer))))
536 (create-image data nil t
537 :ascent 100)
538 (let* ((image (create-image data nil t :ascent 100))
539 (size (image-size image t))
540 (width (car size))
541 (height (cdr size))
542 (edges (window-inside-pixel-edges
543 (get-buffer-window (current-buffer))))
544 (window-width (truncate (* shr-max-image-proportion
545 (- (nth 2 edges) (nth 0 edges)))))
546 (window-height (truncate (* shr-max-image-proportion
547 (- (nth 3 edges) (nth 1 edges)))))
548 scaled-image)
549 (when (> height window-height)
550 (setq image (or (create-image data 'imagemagick t
551 :height window-height)
552 image))
553 (setq size (image-size image t)))
554 (when (> (car size) window-width)
555 (setq image (or
556 (create-image data 'imagemagick t
557 :width window-width
558 :ascent 100)
559 image)))
560 (when (and (fboundp 'create-animated-image)
561 (eq (image-type data nil t) 'gif))
562 (setq image (create-animated-image data 'gif t
563 :ascent 100)))
564 image)))
566 ;; url-cache-extract autoloads url-cache.
567 (declare-function url-cache-create-filename "url-cache" (url))
568 (autoload 'mm-disable-multibyte "mm-util")
569 (autoload 'browse-url-mailto "browse-url")
571 (defun shr-get-image-data (url)
572 "Get image data for URL.
573 Return a string with image data."
574 (with-temp-buffer
575 (mm-disable-multibyte)
576 (when (ignore-errors
577 (url-cache-extract (url-cache-create-filename (shr-encode-url url)))
579 (when (or (search-forward "\n\n" nil t)
580 (search-forward "\r\n\r\n" nil t))
581 (buffer-substring (point) (point-max))))))
583 (defun shr-image-displayer (content-function)
584 "Return a function to display an image.
585 CONTENT-FUNCTION is a function to retrieve an image for a cid url that
586 is an argument. The function to be returned takes three arguments URL,
587 START, and END. Note that START and END should be merkers."
588 `(lambda (url start end)
589 (when url
590 (if (string-match "\\`cid:" url)
591 ,(when content-function
592 `(let ((image (funcall ,content-function
593 (substring url (match-end 0)))))
594 (when image
595 (goto-char start)
596 (funcall shr-put-image-function
597 image (buffer-substring start end))
598 (delete-region (point) end))))
599 (url-retrieve url 'shr-image-fetched
600 (list (current-buffer) start end)
601 t)))))
603 (defun shr-heading (cont &rest types)
604 (shr-ensure-paragraph)
605 (apply #'shr-fontize-cont cont types)
606 (shr-ensure-paragraph))
608 (autoload 'widget-convert-button "wid-edit")
610 (defun shr-urlify (start url &optional title)
611 (widget-convert-button
612 'url-link start (point)
613 :help-echo (if title (format "%s (%s)" url title) url)
614 :keymap shr-map
615 url)
616 (shr-add-font start (point) 'shr-link)
617 (put-text-property start (point) 'shr-url url))
619 (defun shr-encode-url (url)
620 "Encode URL."
621 (browse-url-url-encode-chars url "[)$ ]"))
623 (autoload 'shr-color-visible "shr-color")
624 (autoload 'shr-color->hexadecimal "shr-color")
626 (defun shr-color-check (fg bg)
627 "Check that FG is visible on BG.
628 Returns (fg bg) with corrected values.
629 Returns nil if the colors that would be used are the default
630 ones, in case fg and bg are nil."
631 (when (or fg bg)
632 (let ((fixed (cond ((null fg) 'fg)
633 ((null bg) 'bg))))
634 ;; Convert colors to hexadecimal, or set them to default.
635 (let ((fg (or (shr-color->hexadecimal fg)
636 (frame-parameter nil 'foreground-color)))
637 (bg (or (shr-color->hexadecimal bg)
638 (frame-parameter nil 'background-color))))
639 (cond ((eq fixed 'bg)
640 ;; Only return the new fg
641 (list nil (cadr (shr-color-visible bg fg t))))
642 ((eq fixed 'fg)
643 ;; Invert args and results and return only the new bg
644 (list (cadr (shr-color-visible fg bg t)) nil))
646 (shr-color-visible bg fg)))))))
648 (defun shr-colorize-region (start end fg &optional bg)
649 (when (or fg bg)
650 (let ((new-colors (shr-color-check fg bg)))
651 (when new-colors
652 (when fg
653 (shr-put-color start end :foreground (cadr new-colors)))
654 (when bg
655 (shr-put-color start end :background (car new-colors))))
656 new-colors)))
658 ;; Put a color in the region, but avoid putting colors on blank
659 ;; text at the start of the line, and the newline at the end, to avoid
660 ;; ugliness. Also, don't overwrite any existing color information,
661 ;; since this can be called recursively, and we want the "inner" color
662 ;; to win.
663 (defun shr-put-color (start end type color)
664 (save-excursion
665 (goto-char start)
666 (while (< (point) end)
667 (when (and (bolp)
668 (not (eq type :background)))
669 (skip-chars-forward " "))
670 (when (> (line-end-position) (point))
671 (shr-put-color-1 (point) (min (line-end-position) end) type color))
672 (if (< (line-end-position) end)
673 (forward-line 1)
674 (goto-char end)))
675 (when (and (eq type :background)
676 (= shr-table-depth 0))
677 (shr-expand-newlines start end color))))
679 (defun shr-expand-newlines (start end color)
680 (save-restriction
681 ;; Skip past all white space at the start and ends.
682 (goto-char start)
683 (skip-chars-forward " \t\n")
684 (beginning-of-line)
685 (setq start (point))
686 (goto-char end)
687 (skip-chars-backward " \t\n")
688 (forward-line 1)
689 (setq end (point))
690 (narrow-to-region start end)
691 (let ((width (shr-natural-width))
692 column)
693 (goto-char (point-min))
694 (while (not (eobp))
695 (end-of-line)
696 (when (and (< (setq column (current-column)) width)
697 (< (setq column (shr-previous-newline-padding-width column))
698 width))
699 (let ((overlay (make-overlay (point) (1+ (point)))))
700 (overlay-put overlay 'before-string
701 (concat
702 (mapconcat
703 (lambda (overlay)
704 (let ((string (plist-get
705 (overlay-properties overlay)
706 'before-string)))
707 (if (not string)
709 (overlay-put overlay 'before-string "")
710 string)))
711 (overlays-at (point))
713 (propertize (make-string (- width column) ? )
714 'face (list :background color))))))
715 (forward-line 1)))))
717 (defun shr-previous-newline-padding-width (width)
718 (let ((overlays (overlays-at (point)))
719 (previous-width 0))
720 (if (null overlays)
721 width
722 (dolist (overlay overlays)
723 (setq previous-width
724 (+ previous-width
725 (length (plist-get (overlay-properties overlay)
726 'before-string)))))
727 (+ width previous-width))))
729 (defun shr-put-color-1 (start end type color)
730 (let* ((old-props (get-text-property start 'face))
731 (do-put (and (listp old-props)
732 (not (memq type old-props))))
733 change)
734 (while (< start end)
735 (setq change (next-single-property-change start 'face nil end))
736 (when do-put
737 (put-text-property start change 'face
738 (nconc (list type color) old-props)))
739 (setq old-props (get-text-property change 'face))
740 (setq do-put (and (listp old-props)
741 (not (memq type old-props))))
742 (setq start change))
743 (when (and do-put
744 (> end start))
745 (put-text-property start end 'face
746 (nconc (list type color old-props))))))
748 ;;; Tag-specific rendering rules.
750 (defun shr-tag-body (cont)
751 (let* ((start (point))
752 (fgcolor (cdr (or (assq :fgcolor cont)
753 (assq :text cont))))
754 (bgcolor (cdr (assq :bgcolor cont)))
755 (shr-stylesheet (list (cons 'color fgcolor)
756 (cons 'background-color bgcolor))))
757 (shr-generic cont)
758 (shr-colorize-region start (point) fgcolor bgcolor)))
760 (defun shr-tag-style (cont)
763 (defun shr-tag-script (cont)
766 (defun shr-tag-sup (cont)
767 (let ((start (point)))
768 (shr-generic cont)
769 (put-text-property start (point) 'display '(raise 0.5))))
771 (defun shr-tag-sub (cont)
772 (let ((start (point)))
773 (shr-generic cont)
774 (put-text-property start (point) 'display '(raise -0.5))))
776 (defun shr-tag-label (cont)
777 (shr-generic cont)
778 (shr-ensure-paragraph))
780 (defun shr-tag-p (cont)
781 (shr-ensure-paragraph)
782 (shr-indent)
783 (shr-generic cont)
784 (shr-ensure-paragraph))
786 (defun shr-tag-div (cont)
787 (shr-ensure-newline)
788 (shr-indent)
789 (shr-generic cont)
790 (shr-ensure-newline))
792 (defun shr-tag-s (cont)
793 (shr-fontize-cont cont 'shr-strike-through))
795 (defun shr-tag-del (cont)
796 (shr-fontize-cont cont 'shr-strike-through))
798 (defun shr-tag-b (cont)
799 (shr-fontize-cont cont 'bold))
801 (defun shr-tag-i (cont)
802 (shr-fontize-cont cont 'italic))
804 (defun shr-tag-em (cont)
805 (shr-fontize-cont cont 'bold))
807 (defun shr-tag-strong (cont)
808 (shr-fontize-cont cont 'bold))
810 (defun shr-tag-u (cont)
811 (shr-fontize-cont cont 'underline))
813 (defun shr-parse-style (style)
814 (when style
815 (save-match-data
816 (when (string-match "\n" style)
817 (setq style (replace-match " " t t style))))
818 (let ((plist nil))
819 (dolist (elem (split-string style ";"))
820 (when elem
821 (setq elem (split-string elem ":"))
822 (when (and (car elem)
823 (cadr elem))
824 (let ((name (replace-regexp-in-string "^ +\\| +$" "" (car elem)))
825 (value (replace-regexp-in-string "^ +\\| +$" "" (cadr elem))))
826 (when (string-match " *!important\\'" value)
827 (setq value (substring value 0 (match-beginning 0))))
828 (push (cons (intern name obarray)
829 value)
830 plist)))))
831 plist)))
833 (defun shr-tag-base (cont)
834 (setq shr-base (cdr (assq :href cont))))
836 (defun shr-tag-a (cont)
837 (let ((url (cdr (assq :href cont)))
838 (title (cdr (assq :title cont)))
839 (start (point))
840 shr-start)
841 (shr-generic cont)
842 (shr-urlify (or shr-start start) (shr-expand-url url) title)))
844 (defun shr-tag-object (cont)
845 (let ((start (point))
846 url)
847 (dolist (elem cont)
848 (when (eq (car elem) 'embed)
849 (setq url (or url (cdr (assq :src (cdr elem))))))
850 (when (and (eq (car elem) 'param)
851 (equal (cdr (assq :name (cdr elem))) "movie"))
852 (setq url (or url (cdr (assq :value (cdr elem)))))))
853 (when url
854 (shr-insert " [multimedia] ")
855 (shr-urlify start (shr-expand-url url)))
856 (shr-generic cont)))
858 (defun shr-tag-video (cont)
859 (let ((image (cdr (assq :poster cont)))
860 (url (cdr (assq :src cont)))
861 (start (point)))
862 (shr-tag-img nil image)
863 (shr-urlify start (shr-expand-url url))))
865 (defun shr-tag-img (cont &optional url)
866 (when (or url
867 (and cont
868 (cdr (assq :src cont))))
869 (when (and (> (current-column) 0)
870 (not (eq shr-state 'image)))
871 (insert "\n"))
872 (let ((alt (cdr (assq :alt cont)))
873 (url (shr-expand-url (or url (cdr (assq :src cont))))))
874 (let ((start (point-marker)))
875 (when (zerop (length alt))
876 (setq alt "*"))
877 (cond
878 ((or (member (cdr (assq :height cont)) '("0" "1"))
879 (member (cdr (assq :width cont)) '("0" "1")))
880 ;; Ignore zero-sized or single-pixel images.
882 ((and (not shr-inhibit-images)
883 (string-match "\\`cid:" url))
884 (let ((url (substring url (match-end 0)))
885 image)
886 (if (or (not shr-content-function)
887 (not (setq image (funcall shr-content-function url))))
888 (insert alt)
889 (funcall shr-put-image-function image alt))))
890 ((or shr-inhibit-images
891 (and shr-blocked-images
892 (string-match shr-blocked-images url)))
893 (setq shr-start (point))
894 (let ((shr-state 'space))
895 (if (> (string-width alt) 8)
896 (shr-insert (truncate-string-to-width alt 8))
897 (shr-insert alt))))
898 ((url-is-cached (shr-encode-url url))
899 (funcall shr-put-image-function (shr-get-image-data url) alt))
901 (insert alt)
902 (funcall
903 (if (fboundp 'url-queue-retrieve)
904 'url-queue-retrieve
905 'url-retrieve)
906 (shr-encode-url url) 'shr-image-fetched
907 (list (current-buffer) start (point-marker))
908 t)))
909 (put-text-property start (point) 'keymap shr-map)
910 (put-text-property start (point) 'shr-alt alt)
911 (put-text-property start (point) 'image-url url)
912 (put-text-property start (point) 'image-displayer
913 (shr-image-displayer shr-content-function))
914 (put-text-property start (point) 'help-echo alt)
915 (setq shr-state 'image)))))
917 (defun shr-tag-pre (cont)
918 (let ((shr-folding-mode 'none))
919 (shr-ensure-newline)
920 (shr-indent)
921 (shr-generic cont)
922 (shr-ensure-newline)))
924 (defun shr-tag-blockquote (cont)
925 (shr-ensure-paragraph)
926 (shr-indent)
927 (let ((shr-indentation (+ shr-indentation 4)))
928 (shr-generic cont))
929 (shr-ensure-paragraph))
931 (defun shr-tag-ul (cont)
932 (shr-ensure-paragraph)
933 (let ((shr-list-mode 'ul))
934 (shr-generic cont))
935 (shr-ensure-paragraph))
937 (defun shr-tag-ol (cont)
938 (shr-ensure-paragraph)
939 (let ((shr-list-mode 1))
940 (shr-generic cont))
941 (shr-ensure-paragraph))
943 (defun shr-tag-li (cont)
944 (shr-ensure-paragraph)
945 (shr-indent)
946 (let* ((bullet
947 (if (numberp shr-list-mode)
948 (prog1
949 (format "%d " shr-list-mode)
950 (setq shr-list-mode (1+ shr-list-mode)))
951 "* "))
952 (shr-indentation (+ shr-indentation (length bullet))))
953 (insert bullet)
954 (shr-generic cont)))
956 (defun shr-tag-br (cont)
957 (unless (bobp)
958 (insert "\n")
959 (shr-indent))
960 (shr-generic cont))
962 (defun shr-tag-h1 (cont)
963 (shr-heading cont 'bold 'underline))
965 (defun shr-tag-h2 (cont)
966 (shr-heading cont 'bold))
968 (defun shr-tag-h3 (cont)
969 (shr-heading cont 'italic))
971 (defun shr-tag-h4 (cont)
972 (shr-heading cont))
974 (defun shr-tag-h5 (cont)
975 (shr-heading cont))
977 (defun shr-tag-h6 (cont)
978 (shr-heading cont))
980 (defun shr-tag-hr (cont)
981 (shr-ensure-newline)
982 (insert (make-string shr-width shr-hr-line) "\n"))
984 (defun shr-tag-title (cont)
985 (shr-heading cont 'bold 'underline))
987 (defun shr-tag-font (cont)
988 (let* ((start (point))
989 (color (cdr (assq :color cont)))
990 (shr-stylesheet (nconc (list (cons 'color color))
991 shr-stylesheet)))
992 (shr-generic cont)
993 (when color
994 (shr-colorize-region start (point) color
995 (cdr (assq 'background-color shr-stylesheet))))))
997 ;;; Table rendering algorithm.
999 ;; Table rendering is the only complicated thing here. We do this by
1000 ;; first counting how many TDs there are in each TR, and registering
1001 ;; how wide they think they should be ("width=45%", etc). Then we
1002 ;; render each TD separately (this is done in temporary buffers, so
1003 ;; that we can use all the rendering machinery as if we were in the
1004 ;; main buffer). Now we know how much space each TD really takes, so
1005 ;; we then render everything again with the new widths, and finally
1006 ;; insert all these boxes into the main buffer.
1007 (defun shr-tag-table-1 (cont)
1008 (setq cont (or (cdr (assq 'tbody cont))
1009 cont))
1010 (let* ((shr-inhibit-images t)
1011 (shr-table-depth (1+ shr-table-depth))
1012 (shr-kinsoku-shorten t)
1013 ;; Find all suggested widths.
1014 (columns (shr-column-specs cont))
1015 ;; Compute how many characters wide each TD should be.
1016 (suggested-widths (shr-pro-rate-columns columns))
1017 ;; Do a "test rendering" to see how big each TD is (this can
1018 ;; be smaller (if there's little text) or bigger (if there's
1019 ;; unbreakable text).
1020 (sketch (shr-make-table cont suggested-widths))
1021 (sketch-widths (shr-table-widths sketch suggested-widths)))
1022 ;; This probably won't work very well.
1023 (when (> (+ (loop for width across sketch-widths
1024 summing (1+ width))
1025 shr-indentation 1)
1026 (frame-width))
1027 (setq truncate-lines t))
1028 ;; Then render the table again with these new "hard" widths.
1029 (shr-insert-table (shr-make-table cont sketch-widths t) sketch-widths))
1030 ;; Finally, insert all the images after the table. The Emacs buffer
1031 ;; model isn't strong enough to allow us to put the images actually
1032 ;; into the tables.
1033 (when (zerop shr-table-depth)
1034 (dolist (elem (shr-find-elements cont 'img))
1035 (shr-tag-img (cdr elem)))))
1037 (defun shr-tag-table (cont)
1038 (shr-ensure-paragraph)
1039 (let* ((caption (cdr (assq 'caption cont)))
1040 (header (cdr (assq 'thead cont)))
1041 (body (or (cdr (assq 'tbody cont)) cont))
1042 (footer (cdr (assq 'tfoot cont)))
1043 (bgcolor (cdr (assq :bgcolor cont)))
1044 (start (point))
1045 (shr-stylesheet (nconc (list (cons 'background-color bgcolor))
1046 shr-stylesheet))
1047 (nheader (if header (shr-max-columns header)))
1048 (nbody (if body (shr-max-columns body)))
1049 (nfooter (if footer (shr-max-columns footer))))
1050 (shr-tag-table-1
1051 (nconc
1052 (if caption `((tr (td ,@caption))))
1053 (if header
1054 (if footer
1055 ;; hader + body + footer
1056 (if (= nheader nbody)
1057 (if (= nbody nfooter)
1058 `((tr (td (table (tbody ,@header ,@body ,@footer)))))
1059 (nconc `((tr (td (table (tbody ,@header ,@body)))))
1060 (if (= nfooter 1)
1061 footer
1062 `((tr (td (table (tbody ,@footer))))))))
1063 (nconc `((tr (td (table (tbody ,@header)))))
1064 (if (= nbody nfooter)
1065 `((tr (td (table (tbody ,@body ,@footer)))))
1066 (nconc `((tr (td (table (tbody ,@body)))))
1067 (if (= nfooter 1)
1068 footer
1069 `((tr (td (table (tbody ,@footer))))))))))
1070 ;; header + body
1071 (if (= nheader nbody)
1072 `((tr (td (table (tbody ,@header ,@body)))))
1073 (if (= nheader 1)
1074 `(,@header (tr (td (table (tbody ,@body)))))
1075 `((tr (td (table (tbody ,@header))))
1076 (tr (td (table (tbody ,@body))))))))
1077 (if footer
1078 ;; body + footer
1079 (if (= nbody nfooter)
1080 `((tr (td (table (tbody ,@body ,@footer)))))
1081 (nconc `((tr (td (table (tbody ,@body)))))
1082 (if (= nfooter 1)
1083 footer
1084 `((tr (td (table (tbody ,@footer))))))))
1085 (if caption
1086 `((tr (td (table (tbody ,@body)))))
1087 body)))))
1088 (when bgcolor
1089 (shr-colorize-region start (point) (cdr (assq 'color shr-stylesheet))
1090 bgcolor))))
1092 (defun shr-find-elements (cont type)
1093 (let (result)
1094 (dolist (elem cont)
1095 (cond ((eq (car elem) type)
1096 (push elem result))
1097 ((consp (cdr elem))
1098 (setq result (nconc (shr-find-elements (cdr elem) type) result)))))
1099 (nreverse result)))
1101 (defun shr-insert-table (table widths)
1102 (shr-insert-table-ruler widths)
1103 (dolist (row table)
1104 (let ((start (point))
1105 (height (let ((max 0))
1106 (dolist (column row)
1107 (setq max (max max (cadr column))))
1108 max)))
1109 (dotimes (i height)
1110 (shr-indent)
1111 (insert shr-table-vertical-line "\n"))
1112 (dolist (column row)
1113 (goto-char start)
1114 (let ((lines (nth 2 column))
1115 (overlay-lines (nth 3 column))
1116 overlay overlay-line)
1117 (dolist (line lines)
1118 (setq overlay-line (pop overlay-lines))
1119 (end-of-line)
1120 (insert line shr-table-vertical-line)
1121 (dolist (overlay overlay-line)
1122 (let ((o (make-overlay (- (point) (nth 0 overlay) 1)
1123 (- (point) (nth 1 overlay) 1)))
1124 (properties (nth 2 overlay)))
1125 (while properties
1126 (overlay-put o (pop properties) (pop properties)))))
1127 (forward-line 1))
1128 ;; Add blank lines at padding at the bottom of the TD,
1129 ;; possibly.
1130 (dotimes (i (- height (length lines)))
1131 (end-of-line)
1132 (let ((start (point)))
1133 (insert (make-string (string-width (car lines)) ? )
1134 shr-table-vertical-line)
1135 (when (nth 4 column)
1136 (shr-put-color start (1- (point)) :background (nth 4 column))))
1137 (forward-line 1)))))
1138 (shr-insert-table-ruler widths)))
1140 (defun shr-insert-table-ruler (widths)
1141 (when (and (bolp)
1142 (> shr-indentation 0))
1143 (shr-indent))
1144 (insert shr-table-corner)
1145 (dotimes (i (length widths))
1146 (insert (make-string (aref widths i) shr-table-horizontal-line)
1147 shr-table-corner))
1148 (insert "\n"))
1150 (defun shr-table-widths (table suggested-widths)
1151 (let* ((length (length suggested-widths))
1152 (widths (make-vector length 0))
1153 (natural-widths (make-vector length 0)))
1154 (dolist (row table)
1155 (let ((i 0))
1156 (dolist (column row)
1157 (aset widths i (max (aref widths i)
1158 (car column)))
1159 (aset natural-widths i (max (aref natural-widths i)
1160 (cadr column)))
1161 (setq i (1+ i)))))
1162 (let ((extra (- (apply '+ (append suggested-widths nil))
1163 (apply '+ (append widths nil))))
1164 (expanded-columns 0))
1165 (when (> extra 0)
1166 (dotimes (i length)
1167 ;; If the natural width is wider than the rendered width, we
1168 ;; want to allow the column to expand.
1169 (when (> (aref natural-widths i) (aref widths i))
1170 (setq expanded-columns (1+ expanded-columns))))
1171 (dotimes (i length)
1172 (when (> (aref natural-widths i) (aref widths i))
1173 (aset widths i (min
1174 (1+ (aref natural-widths i))
1175 (+ (/ extra expanded-columns)
1176 (aref widths i))))))))
1177 widths))
1179 (defun shr-make-table (cont widths &optional fill)
1180 (let ((trs nil))
1181 (dolist (row cont)
1182 (when (eq (car row) 'tr)
1183 (let ((tds nil)
1184 (columns (cdr row))
1185 (i 0)
1186 column)
1187 (while (< i (length widths))
1188 (setq column (pop columns))
1189 (when (or (memq (car column) '(td th))
1190 (null column))
1191 (push (shr-render-td (cdr column) (aref widths i) fill)
1192 tds)
1193 (setq i (1+ i))))
1194 (push (nreverse tds) trs))))
1195 (nreverse trs)))
1197 (defun shr-render-td (cont width fill)
1198 (with-temp-buffer
1199 (let ((bgcolor (cdr (assq :bgcolor cont)))
1200 (fgcolor (cdr (assq :fgcolor cont)))
1201 (style (cdr (assq :style cont)))
1202 (shr-stylesheet shr-stylesheet)
1203 overlays actual-colors)
1204 (when style
1205 (setq style (and (string-match "color" style)
1206 (shr-parse-style style))))
1207 (when bgcolor
1208 (setq style (nconc (list (cons 'background-color bgcolor)) style)))
1209 (when fgcolor
1210 (setq style (nconc (list (cons 'color fgcolor)) style)))
1211 (when style
1212 (setq shr-stylesheet (append style shr-stylesheet)))
1213 (let ((cache (cdr (assoc (cons width cont) shr-content-cache))))
1214 (if cache
1215 (progn
1216 (insert (car cache))
1217 (let ((end (length (car cache))))
1218 (dolist (overlay (cadr cache))
1219 (let ((new-overlay
1220 (make-overlay (1+ (- end (nth 0 overlay)))
1221 (1+ (- end (nth 1 overlay)))))
1222 (properties (nth 2 overlay)))
1223 (while properties
1224 (overlay-put new-overlay
1225 (pop properties) (pop properties)))))))
1226 (let ((shr-width width)
1227 (shr-indentation 0))
1228 (shr-descend (cons 'td cont)))
1229 (delete-region
1230 (point)
1231 (+ (point)
1232 (skip-chars-backward " \t\n")))
1233 (push (list (cons width cont) (buffer-string)
1234 (shr-overlays-in-region (point-min) (point-max)))
1235 shr-content-cache)))
1236 (goto-char (point-min))
1237 (let ((max 0))
1238 (while (not (eobp))
1239 (end-of-line)
1240 (setq max (max max (current-column)))
1241 (forward-line 1))
1242 (when fill
1243 (goto-char (point-min))
1244 ;; If the buffer is totally empty, then put a single blank
1245 ;; line here.
1246 (if (zerop (buffer-size))
1247 (insert (make-string width ? ))
1248 ;; Otherwise, fill the buffer.
1249 (while (not (eobp))
1250 (end-of-line)
1251 (when (> (- width (current-column)) 0)
1252 (insert (make-string (- width (current-column)) ? )))
1253 (forward-line 1)))
1254 (when style
1255 (setq actual-colors
1256 (shr-colorize-region
1257 (point-min) (point-max)
1258 (cdr (assq 'color shr-stylesheet))
1259 (cdr (assq 'background-color shr-stylesheet))))))
1260 (if fill
1261 (list max
1262 (count-lines (point-min) (point-max))
1263 (split-string (buffer-string) "\n")
1264 (shr-collect-overlays)
1265 (car actual-colors))
1266 (list max
1267 (shr-natural-width)))))))
1269 (defun shr-natural-width ()
1270 (goto-char (point-min))
1271 (let ((current 0)
1272 (max 0))
1273 (while (not (eobp))
1274 (end-of-line)
1275 (setq current (+ current (current-column)))
1276 (unless (get-text-property (point) 'shr-break)
1277 (setq max (max max current)
1278 current 0))
1279 (forward-line 1))
1280 max))
1282 (defun shr-collect-overlays ()
1283 (save-excursion
1284 (goto-char (point-min))
1285 (let ((overlays nil))
1286 (while (not (eobp))
1287 (push (shr-overlays-in-region (point) (line-end-position))
1288 overlays)
1289 (forward-line 1))
1290 (nreverse overlays))))
1292 (defun shr-overlays-in-region (start end)
1293 (let (result)
1294 (dolist (overlay (overlays-in start end))
1295 (push (list (if (> start (overlay-start overlay))
1296 (- end start)
1297 (- end (overlay-start overlay)))
1298 (if (< end (overlay-end overlay))
1300 (- end (overlay-end overlay)))
1301 (overlay-properties overlay))
1302 result))
1303 (nreverse result)))
1305 (defun shr-pro-rate-columns (columns)
1306 (let ((total-percentage 0)
1307 (widths (make-vector (length columns) 0)))
1308 (dotimes (i (length columns))
1309 (setq total-percentage (+ total-percentage (aref columns i))))
1310 (setq total-percentage (/ 1.0 total-percentage))
1311 (dotimes (i (length columns))
1312 (aset widths i (max (truncate (* (aref columns i)
1313 total-percentage
1314 (- shr-width (1+ (length columns)))))
1315 10)))
1316 widths))
1318 ;; Return a summary of the number and shape of the TDs in the table.
1319 (defun shr-column-specs (cont)
1320 (let ((columns (make-vector (shr-max-columns cont) 1)))
1321 (dolist (row cont)
1322 (when (eq (car row) 'tr)
1323 (let ((i 0))
1324 (dolist (column (cdr row))
1325 (when (memq (car column) '(td th))
1326 (let ((width (cdr (assq :width (cdr column)))))
1327 (when (and width
1328 (string-match "\\([0-9]+\\)%" width))
1329 (aset columns i
1330 (/ (string-to-number (match-string 1 width))
1331 100.0))))
1332 (setq i (1+ i)))))))
1333 columns))
1335 (defun shr-count (cont elem)
1336 (let ((i 0))
1337 (dolist (sub cont)
1338 (when (eq (car sub) elem)
1339 (setq i (1+ i))))
1342 (defun shr-max-columns (cont)
1343 (let ((max 0))
1344 (dolist (row cont)
1345 (when (eq (car row) 'tr)
1346 (setq max (max max (+ (shr-count (cdr row) 'td)
1347 (shr-count (cdr row) 'th))))))
1348 max))
1350 (provide 'shr)
1352 ;;; shr.el ends here