Commit | Line | Data |
---|---|---|
367f7f81 LMI |
1 | ;;; shr.el --- Simple HTML Renderer |
2 | ||
3 | ;; Copyright (C) 2010 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 | ||
3d319c8f | 33 | (eval-when-compile (require 'cl)) |
71e691a5 | 34 | (require 'browse-url) |
2fe10446 KY |
35 | (unless (aref (char-category-set (make-char 'japanese-jisx0208 33 35)) ?>) |
36 | (load "kinsoku" nil t)) | |
71e691a5 | 37 | |
870409d4 G |
38 | (defgroup shr nil |
39 | "Simple HTML Renderer" | |
40 | :group 'mail) | |
41 | ||
42 | (defcustom shr-max-image-proportion 0.9 | |
43 | "How big pictures displayed are in relation to the window they're in. | |
44 | A value of 0.7 means that they are allowed to take up 70% of the | |
45 | width and height of the window. If they are larger than this, | |
46 | and Emacs supports it, then the images will be rescaled down to | |
47 | fit these criteria." | |
48 | :version "24.1" | |
49 | :group 'shr | |
50 | :type 'float) | |
51 | ||
52 | (defcustom shr-blocked-images nil | |
53 | "Images that have URLs matching this regexp will be blocked." | |
54 | :version "24.1" | |
55 | :group 'shr | |
56 | :type 'regexp) | |
57 | ||
afba0c4b JD |
58 | (defcustom shr-table-line ?- |
59 | "Character used to draw table line." | |
60 | :group 'shr | |
030158f3 | 61 | :type 'character) |
afba0c4b JD |
62 | |
63 | (defcustom shr-table-corner ?+ | |
6b7df8d3 G |
64 | "Character used to draw table corner." |
65 | :group 'shr | |
030158f3 | 66 | :type 'character) |
6b7df8d3 G |
67 | |
68 | (defcustom shr-hr-line ?- | |
69 | "Character used to draw hr line." | |
afba0c4b | 70 | :group 'shr |
030158f3 | 71 | :type 'character) |
afba0c4b | 72 | |
d0e0de31 JD |
73 | (defcustom shr-width fill-column |
74 | "Frame width to use for rendering." | |
75 | :type 'integer | |
76 | :group 'shr) | |
77 | ||
130e977f LMI |
78 | (defvar shr-content-function nil |
79 | "If bound, this should be a function that will return the content. | |
80 | This is used for cid: URLs, and the function is called with the | |
81 | cid: URL as the argument.") | |
82 | ||
66627fa9 G |
83 | ;;; Internal variables. |
84 | ||
870409d4 G |
85 | (defvar shr-folding-mode nil) |
86 | (defvar shr-state nil) | |
87 | (defvar shr-start nil) | |
a41c2e6d | 88 | (defvar shr-indentation 0) |
130e977f | 89 | (defvar shr-inhibit-images nil) |
66627fa9 | 90 | (defvar shr-list-mode nil) |
3d319c8f | 91 | (defvar shr-content-cache nil) |
83ffd571 | 92 | (defvar shr-kinsoku-shorten nil) |
99e65b2d | 93 | (defvar shr-table-depth 0) |
870409d4 | 94 | |
71e691a5 G |
95 | (defvar shr-map |
96 | (let ((map (make-sparse-keymap))) | |
97 | (define-key map "a" 'shr-show-alt-text) | |
98 | (define-key map "i" 'shr-browse-image) | |
99 | (define-key map "I" 'shr-insert-image) | |
100 | (define-key map "u" 'shr-copy-url) | |
101 | (define-key map "v" 'shr-browse-url) | |
cdf1fca4 | 102 | (define-key map "o" 'shr-save-contents) |
71e691a5 G |
103 | (define-key map "\r" 'shr-browse-url) |
104 | map)) | |
105 | ||
66627fa9 G |
106 | ;; Public functions and commands. |
107 | ||
108 | ;;;###autoload | |
109 | (defun shr-insert-document (dom) | |
3d319c8f | 110 | (setq shr-content-cache nil) |
66627fa9 G |
111 | (let ((shr-state nil) |
112 | (shr-start nil)) | |
113 | (shr-descend (shr-transform-dom dom)))) | |
114 | ||
115 | (defun shr-copy-url () | |
116 | "Copy the URL under point to the kill ring. | |
117 | If called twice, then try to fetch the URL and see whether it | |
118 | redirects somewhere else." | |
119 | (interactive) | |
120 | (let ((url (get-text-property (point) 'shr-url))) | |
121 | (cond | |
122 | ((not url) | |
123 | (message "No URL under point")) | |
124 | ;; Resolve redirected URLs. | |
125 | ((equal url (car kill-ring)) | |
126 | (url-retrieve | |
127 | url | |
128 | (lambda (a) | |
129 | (when (and (consp a) | |
130 | (eq (car a) :redirect)) | |
131 | (with-temp-buffer | |
132 | (insert (cadr a)) | |
133 | (goto-char (point-min)) | |
134 | ;; Remove common tracking junk from the URL. | |
135 | (when (re-search-forward ".utm_.*" nil t) | |
136 | (replace-match "" t t)) | |
137 | (message "Copied %s" (buffer-string)) | |
138 | (copy-region-as-kill (point-min) (point-max))))))) | |
139 | ;; Copy the URL to the kill ring. | |
140 | (t | |
141 | (with-temp-buffer | |
142 | (insert url) | |
143 | (copy-region-as-kill (point-min) (point-max)) | |
144 | (message "Copied %s" url)))))) | |
145 | ||
146 | (defun shr-show-alt-text () | |
147 | "Show the ALT text of the image under point." | |
148 | (interactive) | |
149 | (let ((text (get-text-property (point) 'shr-alt))) | |
150 | (if (not text) | |
151 | (message "No image under point") | |
152 | (message "%s" text)))) | |
153 | ||
154 | (defun shr-browse-image () | |
155 | "Browse the image under point." | |
156 | (interactive) | |
8b6f6573 | 157 | (let ((url (get-text-property (point) 'image-url))) |
66627fa9 G |
158 | (if (not url) |
159 | (message "No image under point") | |
160 | (message "Browsing %s..." url) | |
161 | (browse-url url)))) | |
162 | ||
3d319c8f LMI |
163 | (defun shr-insert-image () |
164 | "Insert the image under point into the buffer." | |
165 | (interactive) | |
8b6f6573 | 166 | (let ((url (get-text-property (point) 'image-url))) |
3d319c8f LMI |
167 | (if (not url) |
168 | (message "No image under point") | |
169 | (message "Inserting %s..." url) | |
170 | (url-retrieve url 'shr-image-fetched | |
171 | (list (current-buffer) (1- (point)) (point-marker)) | |
172 | t)))) | |
173 | ||
66627fa9 G |
174 | ;;; Utility functions. |
175 | ||
870409d4 G |
176 | (defun shr-transform-dom (dom) |
177 | (let ((result (list (pop dom)))) | |
178 | (dolist (arg (pop dom)) | |
179 | (push (cons (intern (concat ":" (symbol-name (car arg))) obarray) | |
180 | (cdr arg)) | |
181 | result)) | |
182 | (dolist (sub dom) | |
183 | (if (stringp sub) | |
953d41c4 | 184 | (push (cons 'text sub) result) |
870409d4 G |
185 | (push (shr-transform-dom sub) result))) |
186 | (nreverse result))) | |
187 | ||
870409d4 | 188 | (defun shr-descend (dom) |
a41c2e6d | 189 | (let ((function (intern (concat "shr-tag-" (symbol-name (car dom))) obarray))) |
870409d4 G |
190 | (if (fboundp function) |
191 | (funcall function (cdr dom)) | |
192 | (shr-generic (cdr dom))))) | |
193 | ||
194 | (defun shr-generic (cont) | |
195 | (dolist (sub cont) | |
196 | (cond | |
953d41c4 | 197 | ((eq (car sub) 'text) |
870409d4 | 198 | (shr-insert (cdr sub))) |
a41c2e6d | 199 | ((listp (cdr sub)) |
870409d4 G |
200 | (shr-descend sub))))) |
201 | ||
66627fa9 | 202 | (defun shr-insert (text) |
6b7df8d3 G |
203 | (when (and (eq shr-state 'image) |
204 | (not (string-match "\\`[ \t\n]+\\'" text))) | |
66627fa9 G |
205 | (insert "\n") |
206 | (setq shr-state nil)) | |
207 | (cond | |
208 | ((eq shr-folding-mode 'none) | |
209 | (insert text)) | |
210 | (t | |
73db8b08 KY |
211 | (when (and (string-match "\\`[ \t\n]" text) |
212 | (not (bolp)) | |
213 | (not (eq (char-after (1- (point))) ? ))) | |
214 | (insert " ")) | |
215 | (dolist (elem (split-string text)) | |
216 | (when (and (bolp) | |
217 | (> shr-indentation 0)) | |
218 | (shr-indent)) | |
219 | ;; The shr-start is a special variable that is used to pass | |
220 | ;; upwards the first point in the buffer where the text really | |
221 | ;; starts. | |
222 | (unless shr-start | |
223 | (setq shr-start (point))) | |
224 | ;; No space is needed behind a wide character categorized as | |
b41c2f65 KY |
225 | ;; kinsoku-bol, between characters both categorized as nospace, |
226 | ;; or at the beginning of a line. | |
73db8b08 | 227 | (let (prev) |
20438017 KY |
228 | (when (and (eq (preceding-char) ? ) |
229 | (or (= (line-beginning-position) (1- (point))) | |
73db8b08 KY |
230 | (and (aref fill-find-break-point-function-table |
231 | (setq prev (char-after (- (point) 2)))) | |
232 | (aref (char-category-set prev) ?>)) | |
233 | (and (aref fill-nospace-between-words-table prev) | |
234 | (aref fill-nospace-between-words-table | |
235 | (aref elem 0))))) | |
236 | (delete-char -1))) | |
237 | (insert elem) | |
e7102c0a KY |
238 | (let (found) |
239 | (while (and (> (current-column) shr-width) | |
240 | (progn | |
241 | (setq found (shr-find-fill-point)) | |
242 | (not (eolp)))) | |
fe98a42f KY |
243 | (when (eq (preceding-char) ? ) |
244 | (delete-char -1)) | |
245 | (insert "\n") | |
246 | (unless found | |
e7102c0a KY |
247 | (put-text-property (1- (point)) (point) 'shr-break t) |
248 | ;; No space is needed at the beginning of a line. | |
249 | (when (eq (following-char) ? ) | |
250 | (delete-char 1))) | |
251 | (when (> shr-indentation 0) | |
252 | (shr-indent)) | |
253 | (end-of-line)) | |
254 | (insert " "))) | |
73db8b08 KY |
255 | (unless (string-match "[ \t\n]\\'" text) |
256 | (delete-char -1))))) | |
66627fa9 | 257 | |
6b7df8d3 | 258 | (defun shr-find-fill-point () |
83ffd571 KY |
259 | (when (> (move-to-column shr-width) shr-width) |
260 | (backward-char 1)) | |
261 | (let (failed) | |
262 | (while (not | |
263 | (or (setq failed (= (current-column) shr-indentation)) | |
264 | (eq (preceding-char) ? ) | |
265 | (eq (following-char) ? ) | |
266 | (aref fill-find-break-point-function-table (preceding-char)))) | |
267 | (backward-char 1)) | |
268 | (if failed | |
20438017 | 269 | ;; There's no breakable point, so we give it up. |
83ffd571 KY |
270 | (progn |
271 | (end-of-line) | |
272 | (while (aref fill-find-break-point-function-table (preceding-char)) | |
273 | (backward-char 1)) | |
274 | nil) | |
275 | (or (eolp) | |
276 | ;; Don't put kinsoku-bol characters at the beginning of a line, | |
277 | ;; or kinsoku-eol characters at the end of a line, | |
278 | (let ((count 4)) | |
9f47ba2d KY |
279 | (if (or shr-kinsoku-shorten |
280 | (and (aref (char-category-set (preceding-char)) ?<) | |
281 | (progn | |
282 | (setq count (1- count)) | |
283 | (backward-char 1) | |
284 | t))) | |
83ffd571 | 285 | (while (and |
9f47ba2d KY |
286 | (>= (setq count (1- count)) 0) |
287 | (not (memq (preceding-char) (list ?\C-@ ?\n ? ))) | |
83ffd571 KY |
288 | (or (aref (char-category-set (preceding-char)) ?<) |
289 | (aref (char-category-set (following-char)) ?>))) | |
290 | (backward-char 1)) | |
9f47ba2d | 291 | (while (and (>= (setq count (1- count)) 0) |
389b76fa G |
292 | (aref (char-category-set (following-char)) ?>) |
293 | (aref fill-find-break-point-function-table | |
294 | (following-char))) | |
83ffd571 | 295 | (forward-char 1))) |
b643306f LMI |
296 | (when (eq (following-char) ? ) |
297 | (forward-char 1)) | |
83ffd571 | 298 | t))))) |
6b7df8d3 | 299 | |
66627fa9 G |
300 | (defun shr-ensure-newline () |
301 | (unless (zerop (current-column)) | |
302 | (insert "\n"))) | |
a41c2e6d G |
303 | |
304 | (defun shr-ensure-paragraph () | |
305 | (unless (bobp) | |
f7aa248a | 306 | (if (<= (current-column) shr-indentation) |
71e691a5 G |
307 | (unless (save-excursion |
308 | (forward-line -1) | |
309 | (looking-at " *$")) | |
a41c2e6d G |
310 | (insert "\n")) |
311 | (if (save-excursion | |
312 | (beginning-of-line) | |
5d2ef6db | 313 | (looking-at " *$")) |
a41c2e6d G |
314 | (insert "\n") |
315 | (insert "\n\n"))))) | |
316 | ||
66627fa9 | 317 | (defun shr-indent () |
f7aa248a G |
318 | (when (> shr-indentation 0) |
319 | (insert (make-string shr-indentation ? )))) | |
870409d4 | 320 | |
a41c2e6d | 321 | (defun shr-fontize-cont (cont &rest types) |
870409d4 G |
322 | (let (shr-start) |
323 | (shr-generic cont) | |
a41c2e6d G |
324 | (dolist (type types) |
325 | (shr-add-font (or shr-start (point)) (point) type)))) | |
870409d4 | 326 | |
6c85a14f LMI |
327 | ;; Add an overlay in the region, but avoid putting the font properties |
328 | ;; on blank text at the start of the line, and the newline at the end, | |
329 | ;; to avoid ugliness. | |
870409d4 | 330 | (defun shr-add-font (start end type) |
6c85a14f LMI |
331 | (save-excursion |
332 | (goto-char start) | |
333 | (while (< (point) end) | |
334 | (when (bolp) | |
335 | (skip-chars-forward " ")) | |
336 | (let ((overlay (make-overlay (point) (min (line-end-position) end)))) | |
337 | (overlay-put overlay 'face type)) | |
338 | (if (< (line-end-position) end) | |
339 | (forward-line 1) | |
340 | (goto-char end))))) | |
870409d4 | 341 | |
71e691a5 G |
342 | (defun shr-browse-url () |
343 | "Browse the URL under point." | |
344 | (interactive) | |
345 | (let ((url (get-text-property (point) 'shr-url))) | |
181cb5fb G |
346 | (cond |
347 | ((not url) | |
348 | (message "No link under point")) | |
349 | ((string-match "^mailto:" url) | |
be3c11b3 | 350 | (browse-url-mailto url)) |
181cb5fb G |
351 | (t |
352 | (browse-url url))))) | |
71e691a5 | 353 | |
cdf1fca4 LMI |
354 | (defun shr-save-contents (directory) |
355 | "Save the contents from URL in a file." | |
356 | (interactive "DSave contents of URL to directory: ") | |
357 | (let ((url (get-text-property (point) 'shr-url))) | |
358 | (if (not url) | |
359 | (message "No link under point") | |
360 | (url-retrieve (shr-encode-url url) | |
361 | 'shr-store-contents (list url directory))))) | |
362 | ||
363 | (defun shr-store-contents (status url directory) | |
364 | (unless (plist-get status :error) | |
365 | (when (or (search-forward "\n\n" nil t) | |
366 | (search-forward "\r\n\r\n" nil t)) | |
367 | (write-region (point) (point-max) | |
368 | (expand-file-name (file-name-nondirectory url) | |
369 | directory))))) | |
370 | ||
870409d4 G |
371 | (defun shr-image-fetched (status buffer start end) |
372 | (when (and (buffer-name buffer) | |
373 | (not (plist-get status :error))) | |
374 | (url-store-in-cache (current-buffer)) | |
375 | (when (or (search-forward "\n\n" nil t) | |
376 | (search-forward "\r\n\r\n" nil t)) | |
377 | (let ((data (buffer-substring (point) (point-max)))) | |
378 | (with-current-buffer buffer | |
379 | (let ((alt (buffer-substring start end)) | |
380 | (inhibit-read-only t)) | |
381 | (delete-region start end) | |
99e65b2d G |
382 | (goto-char start) |
383 | (shr-put-image data alt)))))) | |
870409d4 G |
384 | (kill-buffer (current-buffer))) |
385 | ||
99e65b2d | 386 | (defun shr-put-image (data alt) |
4abff904 JD |
387 | (if (display-graphic-p) |
388 | (let ((image (ignore-errors | |
389 | (shr-rescale-image data)))) | |
390 | (when image | |
99e65b2d G |
391 | (insert-image image (or alt "*")))) |
392 | (insert alt))) | |
870409d4 G |
393 | |
394 | (defun shr-rescale-image (data) | |
395 | (if (or (not (fboundp 'imagemagick-types)) | |
396 | (not (get-buffer-window (current-buffer)))) | |
397 | (create-image data nil t) | |
398 | (let* ((image (create-image data nil t)) | |
a41c2e6d | 399 | (size (image-size image t)) |
870409d4 G |
400 | (width (car size)) |
401 | (height (cdr size)) | |
402 | (edges (window-inside-pixel-edges | |
403 | (get-buffer-window (current-buffer)))) | |
404 | (window-width (truncate (* shr-max-image-proportion | |
405 | (- (nth 2 edges) (nth 0 edges))))) | |
406 | (window-height (truncate (* shr-max-image-proportion | |
407 | (- (nth 3 edges) (nth 1 edges))))) | |
408 | scaled-image) | |
409 | (when (> height window-height) | |
410 | (setq image (or (create-image data 'imagemagick t | |
411 | :height window-height) | |
412 | image)) | |
413 | (setq size (image-size image t))) | |
414 | (when (> (car size) window-width) | |
415 | (setq image (or | |
416 | (create-image data 'imagemagick t | |
417 | :width window-width) | |
418 | image))) | |
419 | image))) | |
420 | ||
85a45a69 GM |
421 | ;; url-cache-extract autoloads url-cache. |
422 | (declare-function url-cache-create-filename "url-cache" (url)) | |
423 | (autoload 'mm-disable-multibyte "mm-util") | |
be3c11b3 | 424 | (autoload 'browse-url-mailto "browse-url") |
85a45a69 | 425 | |
870409d4 G |
426 | (defun shr-get-image-data (url) |
427 | "Get image data for URL. | |
428 | Return a string with image data." | |
429 | (with-temp-buffer | |
430 | (mm-disable-multibyte) | |
71e691a5 | 431 | (when (ignore-errors |
ab67634f | 432 | (url-cache-extract (url-cache-create-filename (shr-encode-url url))) |
71e691a5 G |
433 | t) |
434 | (when (or (search-forward "\n\n" nil t) | |
435 | (search-forward "\r\n\r\n" nil t)) | |
436 | (buffer-substring (point) (point-max)))))) | |
870409d4 | 437 | |
40de2c6d KY |
438 | (defun shr-image-displayer (content-function) |
439 | "Return a function to display an image. | |
440 | CONTENT-FUNCTION is a function to retrieve an image for a cid url that | |
441 | is an argument. The function to be returned takes three arguments URL, | |
442 | START, and END." | |
443 | `(lambda (url start end) | |
f8d8a97b KY |
444 | (when url |
445 | (if (string-match "\\`cid:" url) | |
446 | ,(when content-function | |
447 | `(let ((image (funcall ,content-function | |
448 | (substring url (match-end 0))))) | |
449 | (when image | |
450 | (goto-char start) | |
451 | (shr-put-image image | |
452 | (prog1 | |
453 | (buffer-substring-no-properties start end) | |
454 | (delete-region start end)))))) | |
455 | (url-retrieve url 'shr-image-fetched | |
456 | (list (current-buffer) start end) | |
457 | t))))) | |
40de2c6d | 458 | |
66627fa9 G |
459 | (defun shr-heading (cont &rest types) |
460 | (shr-ensure-paragraph) | |
461 | (apply #'shr-fontize-cont cont types) | |
462 | (shr-ensure-paragraph)) | |
463 | ||
85a45a69 GM |
464 | (autoload 'widget-convert-button "wid-edit") |
465 | ||
de635afe G |
466 | (defun shr-urlify (start url) |
467 | (widget-convert-button | |
468 | 'url-link start (point) | |
469 | :help-echo url | |
470 | :keymap shr-map | |
471 | url) | |
472 | (put-text-property start (point) 'shr-url url)) | |
473 | ||
474 | (defun shr-encode-url (url) | |
475 | "Encode URL." | |
476 | (browse-url-url-encode-chars url "[)$ ]")) | |
477 | ||
66627fa9 G |
478 | ;;; Tag-specific rendering rules. |
479 | ||
480 | (defun shr-tag-p (cont) | |
481 | (shr-ensure-paragraph) | |
f7aa248a | 482 | (shr-indent) |
66627fa9 G |
483 | (shr-generic cont) |
484 | (shr-ensure-paragraph)) | |
485 | ||
036d93bc KY |
486 | (defun shr-tag-div (cont) |
487 | (shr-ensure-newline) | |
488 | (shr-indent) | |
489 | (shr-generic cont) | |
490 | (shr-ensure-newline)) | |
491 | ||
66627fa9 G |
492 | (defun shr-tag-b (cont) |
493 | (shr-fontize-cont cont 'bold)) | |
494 | ||
495 | (defun shr-tag-i (cont) | |
496 | (shr-fontize-cont cont 'italic)) | |
497 | ||
498 | (defun shr-tag-em (cont) | |
499 | (shr-fontize-cont cont 'bold)) | |
500 | ||
530f7b67 LMI |
501 | (defun shr-tag-strong (cont) |
502 | (shr-fontize-cont cont 'bold)) | |
503 | ||
66627fa9 G |
504 | (defun shr-tag-u (cont) |
505 | (shr-fontize-cont cont 'underline)) | |
506 | ||
507 | (defun shr-tag-s (cont) | |
508 | (shr-fontize-cont cont 'strike-through)) | |
509 | ||
2e76c12c LMI |
510 | (defun shr-parse-style (style) |
511 | (when style | |
512 | (let ((plist nil)) | |
513 | (dolist (elem (split-string style ";")) | |
514 | (when elem | |
515 | (setq elem (split-string elem ":")) | |
516 | (when (and (car elem) | |
517 | (cadr elem)) | |
518 | (let ((name (replace-regexp-in-string "^ +\\| +$" "" (car elem))) | |
519 | (value (replace-regexp-in-string "^ +\\| +$" "" (cadr elem)))) | |
520 | (push (cons (intern name obarray) | |
521 | value) | |
522 | plist))))) | |
523 | plist))) | |
524 | ||
66627fa9 G |
525 | (defun shr-tag-a (cont) |
526 | (let ((url (cdr (assq :href cont))) | |
527 | (start (point)) | |
528 | shr-start) | |
529 | (shr-generic cont) | |
de635afe G |
530 | (shr-urlify (or shr-start start) url))) |
531 | ||
532 | (defun shr-tag-object (cont) | |
99e65b2d G |
533 | (let ((start (point)) |
534 | url) | |
535 | (dolist (elem cont) | |
536 | (when (eq (car elem) 'embed) | |
537 | (setq url (or url (cdr (assq :src (cdr elem)))))) | |
538 | (when (and (eq (car elem) 'param) | |
539 | (equal (cdr (assq :name (cdr elem))) "movie")) | |
540 | (setq url (or url (cdr (assq :value (cdr elem))))))) | |
de635afe G |
541 | (when url |
542 | (shr-insert " [multimedia] ") | |
99e65b2d G |
543 | (shr-urlify start url)) |
544 | (shr-generic cont))) | |
545 | ||
546 | (defun shr-tag-video (cont) | |
547 | (let ((image (cdr (assq :poster cont))) | |
548 | (url (cdr (assq :src cont))) | |
549 | (start (point))) | |
550 | (shr-tag-img nil image) | |
551 | (shr-urlify start url))) | |
ab67634f | 552 | |
99e65b2d G |
553 | (defun shr-tag-img (cont &optional url) |
554 | (when (or url | |
555 | (and cont | |
556 | (cdr (assq :src cont)))) | |
68f6bd17 KY |
557 | (when (and (> (current-column) 0) |
558 | (not (eq shr-state 'image))) | |
559 | (insert "\n")) | |
560 | (let ((alt (cdr (assq :alt cont))) | |
99e65b2d | 561 | (url (or url (cdr (assq :src cont))))) |
68f6bd17 KY |
562 | (let ((start (point-marker))) |
563 | (when (zerop (length alt)) | |
953d41c4 | 564 | (setq alt "*")) |
68f6bd17 | 565 | (cond |
99e65b2d G |
566 | ((or (member (cdr (assq :height cont)) '("0" "1")) |
567 | (member (cdr (assq :width cont)) '("0" "1"))) | |
568 | ;; Ignore zero-sized or single-pixel images. | |
569 | ) | |
68f6bd17 KY |
570 | ((and (not shr-inhibit-images) |
571 | (string-match "\\`cid:" url)) | |
572 | (let ((url (substring url (match-end 0))) | |
573 | image) | |
574 | (if (or (not shr-content-function) | |
575 | (not (setq image (funcall shr-content-function url)))) | |
576 | (insert alt) | |
99e65b2d | 577 | (shr-put-image image alt)))) |
68f6bd17 KY |
578 | ((or shr-inhibit-images |
579 | (and shr-blocked-images | |
580 | (string-match shr-blocked-images url))) | |
581 | (setq shr-start (point)) | |
582 | (let ((shr-state 'space)) | |
b354bc53 KY |
583 | (if (> (string-width alt) 8) |
584 | (shr-insert (truncate-string-to-width alt 8)) | |
68f6bd17 KY |
585 | (shr-insert alt)))) |
586 | ((url-is-cached (shr-encode-url url)) | |
99e65b2d | 587 | (shr-put-image (shr-get-image-data url) alt)) |
68f6bd17 KY |
588 | (t |
589 | (insert alt) | |
590 | (ignore-errors | |
591 | (url-retrieve (shr-encode-url url) 'shr-image-fetched | |
592 | (list (current-buffer) start (point-marker)) | |
593 | t)))) | |
68f6bd17 KY |
594 | (put-text-property start (point) 'keymap shr-map) |
595 | (put-text-property start (point) 'shr-alt alt) | |
8b6f6573 LMI |
596 | (put-text-property start (point) 'image-url url) |
597 | (put-text-property start (point) 'image-displayer | |
40de2c6d | 598 | (shr-image-displayer shr-content-function)) |
99e65b2d | 599 | (put-text-property start (point) 'help-echo alt) |
68f6bd17 | 600 | (setq shr-state 'image))))) |
66627fa9 G |
601 | |
602 | (defun shr-tag-pre (cont) | |
603 | (let ((shr-folding-mode 'none)) | |
604 | (shr-ensure-newline) | |
f7aa248a | 605 | (shr-indent) |
66627fa9 G |
606 | (shr-generic cont) |
607 | (shr-ensure-newline))) | |
608 | ||
609 | (defun shr-tag-blockquote (cont) | |
610 | (shr-ensure-paragraph) | |
f7aa248a | 611 | (shr-indent) |
66627fa9 G |
612 | (let ((shr-indentation (+ shr-indentation 4))) |
613 | (shr-generic cont)) | |
614 | (shr-ensure-paragraph)) | |
a41c2e6d G |
615 | |
616 | (defun shr-tag-ul (cont) | |
617 | (shr-ensure-paragraph) | |
618 | (let ((shr-list-mode 'ul)) | |
3d319c8f LMI |
619 | (shr-generic cont)) |
620 | (shr-ensure-paragraph)) | |
a41c2e6d G |
621 | |
622 | (defun shr-tag-ol (cont) | |
3d319c8f | 623 | (shr-ensure-paragraph) |
a41c2e6d | 624 | (let ((shr-list-mode 1)) |
3d319c8f LMI |
625 | (shr-generic cont)) |
626 | (shr-ensure-paragraph)) | |
a41c2e6d G |
627 | |
628 | (defun shr-tag-li (cont) | |
f7aa248a G |
629 | (shr-ensure-paragraph) |
630 | (shr-indent) | |
8028ed5c LMI |
631 | (let* ((bullet |
632 | (if (numberp shr-list-mode) | |
633 | (prog1 | |
634 | (format "%d " shr-list-mode) | |
635 | (setq shr-list-mode (1+ shr-list-mode))) | |
636 | "* ")) | |
637 | (shr-indentation (+ shr-indentation (length bullet)))) | |
638 | (insert bullet) | |
639 | (shr-generic cont))) | |
a41c2e6d G |
640 | |
641 | (defun shr-tag-br (cont) | |
1e463294 | 642 | (unless (bobp) |
f7aa248a G |
643 | (insert "\n") |
644 | (shr-indent)) | |
a41c2e6d G |
645 | (shr-generic cont)) |
646 | ||
647 | (defun shr-tag-h1 (cont) | |
648 | (shr-heading cont 'bold 'underline)) | |
649 | ||
650 | (defun shr-tag-h2 (cont) | |
651 | (shr-heading cont 'bold)) | |
652 | ||
653 | (defun shr-tag-h3 (cont) | |
654 | (shr-heading cont 'italic)) | |
655 | ||
656 | (defun shr-tag-h4 (cont) | |
657 | (shr-heading cont)) | |
658 | ||
659 | (defun shr-tag-h5 (cont) | |
660 | (shr-heading cont)) | |
661 | ||
662 | (defun shr-tag-h6 (cont) | |
663 | (shr-heading cont)) | |
664 | ||
3d319c8f LMI |
665 | (defun shr-tag-hr (cont) |
666 | (shr-ensure-newline) | |
6b7df8d3 | 667 | (insert (make-string shr-width shr-hr-line) "\n")) |
3d319c8f | 668 | |
66627fa9 | 669 | ;;; Table rendering algorithm. |
a41c2e6d | 670 | |
a0ec382a LMI |
671 | ;; Table rendering is the only complicated thing here. We do this by |
672 | ;; first counting how many TDs there are in each TR, and registering | |
673 | ;; how wide they think they should be ("width=45%", etc). Then we | |
674 | ;; render each TD separately (this is done in temporary buffers, so | |
675 | ;; that we can use all the rendering machinery as if we were in the | |
676 | ;; main buffer). Now we know how much space each TD really takes, so | |
677 | ;; we then render everything again with the new widths, and finally | |
678 | ;; insert all these boxes into the main buffer. | |
6c769311 | 679 | (defun shr-tag-table-1 (cont) |
71e691a5 G |
680 | (setq cont (or (cdr (assq 'tbody cont)) |
681 | cont)) | |
130e977f | 682 | (let* ((shr-inhibit-images t) |
99e65b2d | 683 | (shr-table-depth (1+ shr-table-depth)) |
83ffd571 | 684 | (shr-kinsoku-shorten t) |
a0ec382a | 685 | ;; Find all suggested widths. |
130e977f | 686 | (columns (shr-column-specs cont)) |
a0ec382a | 687 | ;; Compute how many characters wide each TD should be. |
71e691a5 | 688 | (suggested-widths (shr-pro-rate-columns columns)) |
a0ec382a LMI |
689 | ;; Do a "test rendering" to see how big each TD is (this can |
690 | ;; be smaller (if there's little text) or bigger (if there's | |
691 | ;; unbreakable text). | |
71e691a5 | 692 | (sketch (shr-make-table cont suggested-widths)) |
a7dcc87b | 693 | (sketch-widths (shr-table-widths sketch suggested-widths))) |
030158f3 | 694 | ;; This probably won't work very well. |
83ffd571 KY |
695 | (when (> (+ (loop for width across sketch-widths |
696 | summing (1+ width)) | |
697 | shr-indentation 1) | |
030158f3 G |
698 | (frame-width)) |
699 | (setq truncate-lines t)) | |
a0ec382a | 700 | ;; Then render the table again with these new "hard" widths. |
130e977f | 701 | (shr-insert-table (shr-make-table cont sketch-widths t) sketch-widths)) |
a0ec382a LMI |
702 | ;; Finally, insert all the images after the table. The Emacs buffer |
703 | ;; model isn't strong enough to allow us to put the images actually | |
704 | ;; into the tables. | |
99e65b2d G |
705 | (when (zerop shr-table-depth) |
706 | (dolist (elem (shr-find-elements cont 'img)) | |
707 | (shr-tag-img (cdr elem))))) | |
130e977f | 708 | |
6c769311 KY |
709 | (defun shr-tag-table (cont) |
710 | (shr-ensure-paragraph) | |
711 | (let* ((caption (cdr (assq 'caption cont))) | |
712 | (header (cdr (assq 'thead cont))) | |
713 | (body (or (cdr (assq 'tbody cont)) cont)) | |
714 | (footer (cdr (assq 'tfoot cont))) | |
715 | (nheader (if header (shr-max-columns header))) | |
716 | (nbody (if body (shr-max-columns body))) | |
717 | (nfooter (if footer (shr-max-columns footer)))) | |
718 | (shr-tag-table-1 | |
3c066373 KY |
719 | (nconc |
720 | (if caption `((tr (td ,@caption)))) | |
721 | (if header | |
722 | (if footer | |
723 | ;; hader + body + footer | |
724 | (if (= nheader nbody) | |
725 | (if (= nbody nfooter) | |
726 | `((tr (td (table (tbody ,@header ,@body ,@footer))))) | |
727 | (nconc `((tr (td (table (tbody ,@header ,@body))))) | |
728 | (if (= nfooter 1) | |
729 | footer | |
730 | `((tr (td (table (tbody ,@footer)))))))) | |
731 | (nconc `((tr (td (table (tbody ,@header))))) | |
732 | (if (= nbody nfooter) | |
733 | `((tr (td (table (tbody ,@body ,@footer))))) | |
734 | (nconc `((tr (td (table (tbody ,@body))))) | |
735 | (if (= nfooter 1) | |
736 | footer | |
737 | `((tr (td (table (tbody ,@footer)))))))))) | |
738 | ;; header + body | |
739 | (if (= nheader nbody) | |
740 | `((tr (td (table (tbody ,@header ,@body))))) | |
741 | (if (= nheader 1) | |
742 | `(,@header (tr (td (table (tbody ,@body))))) | |
743 | `((tr (td (table (tbody ,@header)))) | |
744 | (tr (td (table (tbody ,@body)))))))) | |
745 | (if footer | |
746 | ;; body + footer | |
747 | (if (= nbody nfooter) | |
748 | `((tr (td (table (tbody ,@body ,@footer))))) | |
749 | (nconc `((tr (td (table (tbody ,@body))))) | |
2526f423 | 750 | (if (= nfooter 1) |
3c066373 KY |
751 | footer |
752 | `((tr (td (table (tbody ,@footer)))))))) | |
753 | (if caption | |
754 | `((tr (td (table (tbody ,@body))))) | |
755 | body))))))) | |
6c769311 | 756 | |
130e977f LMI |
757 | (defun shr-find-elements (cont type) |
758 | (let (result) | |
759 | (dolist (elem cont) | |
760 | (cond ((eq (car elem) type) | |
761 | (push elem result)) | |
762 | ((consp (cdr elem)) | |
763 | (setq result (nconc (shr-find-elements (cdr elem) type) result))))) | |
764 | (nreverse result))) | |
71e691a5 G |
765 | |
766 | (defun shr-insert-table (table widths) | |
767 | (shr-insert-table-ruler widths) | |
768 | (dolist (row table) | |
769 | (let ((start (point)) | |
770 | (height (let ((max 0)) | |
771 | (dolist (column row) | |
772 | (setq max (max max (cadr column)))) | |
773 | max))) | |
774 | (dotimes (i height) | |
775 | (shr-indent) | |
776 | (insert "|\n")) | |
777 | (dolist (column row) | |
778 | (goto-char start) | |
a7dcc87b | 779 | (let ((lines (nth 2 column)) |
130e977f LMI |
780 | (overlay-lines (nth 3 column)) |
781 | overlay overlay-line) | |
71e691a5 | 782 | (dolist (line lines) |
130e977f | 783 | (setq overlay-line (pop overlay-lines)) |
3d319c8f LMI |
784 | (end-of-line) |
785 | (insert line "|") | |
786 | (dolist (overlay overlay-line) | |
787 | (let ((o (make-overlay (- (point) (nth 0 overlay) 1) | |
788 | (- (point) (nth 1 overlay) 1))) | |
789 | (properties (nth 2 overlay))) | |
790 | (while properties | |
791 | (overlay-put o (pop properties) (pop properties))))) | |
792 | (forward-line 1)) | |
71e691a5 G |
793 | ;; Add blank lines at padding at the bottom of the TD, |
794 | ;; possibly. | |
795 | (dotimes (i (- height (length lines))) | |
796 | (end-of-line) | |
83ffd571 | 797 | (insert (make-string (string-width (car lines)) ? ) "|") |
71e691a5 G |
798 | (forward-line 1))))) |
799 | (shr-insert-table-ruler widths))) | |
800 | ||
801 | (defun shr-insert-table-ruler (widths) | |
83ffd571 KY |
802 | (when (and (bolp) |
803 | (> shr-indentation 0)) | |
804 | (shr-indent)) | |
afba0c4b | 805 | (insert shr-table-corner) |
71e691a5 | 806 | (dotimes (i (length widths)) |
afba0c4b | 807 | (insert (make-string (aref widths i) shr-table-line) shr-table-corner)) |
71e691a5 G |
808 | (insert "\n")) |
809 | ||
a7dcc87b G |
810 | (defun shr-table-widths (table suggested-widths) |
811 | (let* ((length (length suggested-widths)) | |
812 | (widths (make-vector length 0)) | |
813 | (natural-widths (make-vector length 0))) | |
71e691a5 G |
814 | (dolist (row table) |
815 | (let ((i 0)) | |
816 | (dolist (column row) | |
817 | (aset widths i (max (aref widths i) | |
818 | (car column))) | |
a7dcc87b G |
819 | (aset natural-widths i (max (aref natural-widths i) |
820 | (cadr column))) | |
821 | (setq i (1+ i))))) | |
863b61d6 KY |
822 | (let ((extra (- (apply '+ (append suggested-widths nil)) |
823 | (apply '+ (append widths nil)))) | |
a7dcc87b G |
824 | (expanded-columns 0)) |
825 | (when (> extra 0) | |
826 | (dotimes (i length) | |
827 | ;; If the natural width is wider than the rendered width, we | |
828 | ;; want to allow the column to expand. | |
829 | (when (> (aref natural-widths i) (aref widths i)) | |
830 | (setq expanded-columns (1+ expanded-columns)))) | |
831 | (dotimes (i length) | |
832 | (when (> (aref natural-widths i) (aref widths i)) | |
833 | (aset widths i (min | |
834 | (1+ (aref natural-widths i)) | |
835 | (+ (/ extra expanded-columns) | |
836 | (aref widths i)))))))) | |
71e691a5 G |
837 | widths)) |
838 | ||
839 | (defun shr-make-table (cont widths &optional fill) | |
840 | (let ((trs nil)) | |
841 | (dolist (row cont) | |
842 | (when (eq (car row) 'tr) | |
a0ec382a LMI |
843 | (let ((tds nil) |
844 | (columns (cdr row)) | |
845 | (i 0) | |
846 | column) | |
847 | (while (< i (length widths)) | |
848 | (setq column (pop columns)) | |
849 | (when (or (memq (car column) '(td th)) | |
850 | (null column)) | |
71e691a5 G |
851 | (push (shr-render-td (cdr column) (aref widths i) fill) |
852 | tds) | |
853 | (setq i (1+ i)))) | |
854 | (push (nreverse tds) trs)))) | |
855 | (nreverse trs))) | |
856 | ||
857 | (defun shr-render-td (cont width fill) | |
858 | (with-temp-buffer | |
3d319c8f LMI |
859 | (let ((cache (cdr (assoc (cons width cont) shr-content-cache)))) |
860 | (if cache | |
861 | (insert cache) | |
862 | (let ((shr-width width) | |
863 | (shr-indentation 0)) | |
864 | (shr-generic cont)) | |
865 | (delete-region | |
866 | (point) | |
867 | (+ (point) | |
868 | (skip-chars-backward " \t\n"))) | |
869 | (push (cons (cons width cont) (buffer-string)) | |
870 | shr-content-cache))) | |
71e691a5 G |
871 | (goto-char (point-min)) |
872 | (let ((max 0)) | |
873 | (while (not (eobp)) | |
874 | (end-of-line) | |
875 | (setq max (max max (current-column))) | |
876 | (forward-line 1)) | |
877 | (when fill | |
878 | (goto-char (point-min)) | |
a0ec382a LMI |
879 | ;; If the buffer is totally empty, then put a single blank |
880 | ;; line here. | |
881 | (if (zerop (buffer-size)) | |
882 | (insert (make-string width ? )) | |
883 | ;; Otherwise, fill the buffer. | |
884 | (while (not (eobp)) | |
885 | (end-of-line) | |
886 | (when (> (- width (current-column)) 0) | |
887 | (insert (make-string (- width (current-column)) ? ))) | |
888 | (forward-line 1)))) | |
a7dcc87b G |
889 | (if fill |
890 | (list max | |
891 | (count-lines (point-min) (point-max)) | |
892 | (split-string (buffer-string) "\n") | |
893 | (shr-collect-overlays)) | |
894 | (list max | |
895 | (shr-natural-width)))))) | |
896 | ||
897 | (defun shr-natural-width () | |
898 | (goto-char (point-min)) | |
899 | (let ((current 0) | |
900 | (max 0)) | |
901 | (while (not (eobp)) | |
902 | (end-of-line) | |
903 | (setq current (+ current (current-column))) | |
904 | (unless (get-text-property (point) 'shr-break) | |
905 | (setq max (max max current) | |
906 | current 0)) | |
907 | (forward-line 1)) | |
908 | max)) | |
130e977f LMI |
909 | |
910 | (defun shr-collect-overlays () | |
911 | (save-excursion | |
912 | (goto-char (point-min)) | |
913 | (let ((overlays nil)) | |
914 | (while (not (eobp)) | |
915 | (push (shr-overlays-in-region (point) (line-end-position)) | |
916 | overlays) | |
917 | (forward-line 1)) | |
918 | (nreverse overlays)))) | |
919 | ||
920 | (defun shr-overlays-in-region (start end) | |
921 | (let (result) | |
922 | (dolist (overlay (overlays-in start end)) | |
923 | (push (list (if (> start (overlay-start overlay)) | |
924 | (- end start) | |
925 | (- end (overlay-start overlay))) | |
926 | (if (< end (overlay-end overlay)) | |
927 | 0 | |
928 | (- end (overlay-end overlay))) | |
929 | (overlay-properties overlay)) | |
930 | result)) | |
931 | (nreverse result))) | |
71e691a5 G |
932 | |
933 | (defun shr-pro-rate-columns (columns) | |
934 | (let ((total-percentage 0) | |
935 | (widths (make-vector (length columns) 0))) | |
936 | (dotimes (i (length columns)) | |
a7dcc87b | 937 | (setq total-percentage (+ total-percentage (aref columns i)))) |
71e691a5 G |
938 | (setq total-percentage (/ 1.0 total-percentage)) |
939 | (dotimes (i (length columns)) | |
940 | (aset widths i (max (truncate (* (aref columns i) | |
941 | total-percentage | |
a7dcc87b | 942 | (- shr-width (1+ (length columns))))) |
71e691a5 G |
943 | 10))) |
944 | widths)) | |
945 | ||
946 | ;; Return a summary of the number and shape of the TDs in the table. | |
947 | (defun shr-column-specs (cont) | |
948 | (let ((columns (make-vector (shr-max-columns cont) 1))) | |
949 | (dolist (row cont) | |
950 | (when (eq (car row) 'tr) | |
951 | (let ((i 0)) | |
952 | (dolist (column (cdr row)) | |
953 | (when (memq (car column) '(td th)) | |
954 | (let ((width (cdr (assq :width (cdr column))))) | |
955 | (when (and width | |
956 | (string-match "\\([0-9]+\\)%" width)) | |
957 | (aset columns i | |
958 | (/ (string-to-number (match-string 1 width)) | |
130e977f LMI |
959 | 100.0)))) |
960 | (setq i (1+ i))))))) | |
71e691a5 G |
961 | columns)) |
962 | ||
963 | (defun shr-count (cont elem) | |
964 | (let ((i 0)) | |
965 | (dolist (sub cont) | |
966 | (when (eq (car sub) elem) | |
967 | (setq i (1+ i)))) | |
968 | i)) | |
969 | ||
970 | (defun shr-max-columns (cont) | |
971 | (let ((max 0)) | |
972 | (dolist (row cont) | |
973 | (when (eq (car row) 'tr) | |
130e977f LMI |
974 | (setq max (max max (+ (shr-count (cdr row) 'td) |
975 | (shr-count (cdr row) 'th)))))) | |
71e691a5 G |
976 | max)) |
977 | ||
f3fd95db | 978 | (provide 'shr) |
367f7f81 LMI |
979 | |
980 | ;;; shr.el ends here |