Decode URL entities to avoid broken links
[gnus] / lisp / gnus-html.el
index 1e6cdc5..4bac528 100644 (file)
 
 (eval-when-compile (require 'cl))
 (eval-when-compile (require 'mm-decode))
+
+(require 'gnus-art)
 (require 'mm-url)
+(require 'url)
+(require 'url-cache)
+(require 'xml)
+(require 'browse-url)
 
-(defcustom gnus-html-cache-directory (nnheader-concat gnus-directory "html-cache/")
-  "Where Gnus will cache images it downloads from the web."
+(defcustom gnus-html-image-cache-ttl (days-to-time 7)
+  "Time used to determine if we should use images from the cache."
   :version "24.1"
   :group 'gnus-art
-  :type 'directory)
+  :type 'integer)
 
-(defcustom gnus-html-cache-size 500000000
-  "The size of the Gnus image cache."
+(defcustom gnus-html-image-automatic-caching t
+  "Whether automatically cache retrieve images."
   :version "24.1"
   :group 'gnus-art
-  :type 'integer)
+  :type 'boolean)
 
 (defcustom gnus-html-frame-width 70
   "What width to use when rendering HTML."
@@ -56,7 +62,7 @@
   :group 'gnus-art
   :type 'regexp)
 
-(defcustom gnus-max-image-proportion 0.7
+(defcustom gnus-max-image-proportion 0.9
   "How big pictures displayed are in relation to the window they're in.
 A value of 0.7 means that they are allowed to take up 70% of the
 width and height of the window.  If they are larger than this,
@@ -70,11 +76,63 @@ fit these criteria."
   (let ((map (make-sparse-keymap)))
     (define-key map "u" 'gnus-article-copy-string)
     (define-key map "i" 'gnus-html-insert-image)
+    (define-key map "v" 'gnus-html-browse-url)
     map))
 
+(defvar gnus-html-displayed-image-map
+  (let ((map (make-sparse-keymap)))
+    (define-key map "a" 'gnus-html-show-alt-text)
+    (define-key map "i" 'gnus-html-browse-image)
+    (define-key map "\r" 'gnus-html-browse-url)
+    (define-key map "u" 'gnus-article-copy-string)
+    (define-key map [tab] 'widget-forward)
+    map))
+
+(eval-and-compile
+  (defalias 'gnus-html-encode-url-chars
+    (if (fboundp 'browse-url-url-encode-chars)
+       'browse-url-url-encode-chars
+      (lambda (text chars)
+       "URL-encode the chars in TEXT that match CHARS.
+CHARS is a regexp-like character alternative (e.g., \"[)$]\")."
+       (let ((encoded-text (copy-sequence text))
+             (s 0))
+         (while (setq s (string-match chars encoded-text s))
+           (setq encoded-text
+                 (replace-match (format "%%%x"
+                                        (string-to-char
+                                         (match-string 0 encoded-text)))
+                                t t encoded-text)
+                 s (1+ s)))
+         encoded-text))))
+  ;; XEmacs does not have window-inside-pixel-edges
+  (defalias 'gnus-window-inside-pixel-edges
+    (if (fboundp 'window-inside-pixel-edges)
+        'window-inside-pixel-edges
+      'window-pixel-edges)))
+
+(defun gnus-html-encode-url (url)
+  "Encode URL."
+  (gnus-html-encode-url-chars url "[)$ ]"))
+
+(defun gnus-html-cache-expired (url ttl)
+  "Check if URL is cached for more than TTL."
+  (cond (url-standalone-mode
+         (not (file-exists-p (url-cache-create-filename url))))
+        (t (let ((cache-time (url-is-cached url)))
+             (if cache-time
+                 (time-less-p
+                  (time-add
+                   cache-time
+                   ttl)
+                  (current-time))
+               t)))))
+
 ;;;###autoload
-(defun gnus-article-html (handle)
+(defun gnus-article-html (&optional handle)
   (let ((article-buffer (current-buffer)))
+    (unless handle
+      (setq handle (mm-dissect-buffer t)))
     (save-restriction
       (narrow-to-region (point) (point))
       (save-excursion
@@ -100,6 +158,7 @@ fit these criteria."
                                 "-I" "UTF-8"
                                 "-O" "UTF-8"
                                 "-o" "ext_halfdump=1"
+                                 "-o" "display_ins_del=2"
                                 "-o" "pre_conv=1"
                                 "-t" (format "%s" tab-width)
                                 "-cols" (format "%s" gnus-html-frame-width)
@@ -109,13 +168,104 @@ fit these criteria."
 
 (defvar gnus-article-mouse-face)
 
-(defun gnus-html-wash-tags ()
+(defun gnus-html-pre-wash ()
+  (goto-char (point-min))
+  (while (re-search-forward " *<pre_int> *</pre_int> *\n" nil t)
+    (replace-match "" t t))
+  (goto-char (point-min))
+  (while (re-search-forward "<a name[^\n>]+>" nil t)
+    (replace-match "" t t)))
+
+(defun gnus-html-wash-images ()
+  "Run through current buffer and replace img tags by images."
   (let (tag parameters string start end images url)
-    (mm-url-decode-entities)
     (goto-char (point-min))
-    (while (re-search-forward " *<pre_int> *</pre_int>\n" nil t)
-      (replace-match "" t t))
+    ;; Search for all the images first.
+    (while (re-search-forward "<img_alt \\([^>]*\\)>" nil t)
+      (setq parameters (match-string 1)
+           start (match-beginning 0))
+      (delete-region start (point))
+      (when (search-forward "</img_alt>" (line-end-position) t)
+       (delete-region (match-beginning 0) (match-end 0)))
+      (setq end (point))
+      (when (string-match "src=\"\\([^\"]+\\)" parameters)
+       (setq url (gnus-html-encode-url (match-string 1 parameters)))
+       (gnus-message 8 "gnus-html-wash-tags: fetching image URL %s" url)
+       (if (string-match "^cid:\\(.*\\)" url)
+           ;; URLs with cid: have their content stashed in other
+           ;; parts of the MIME structure, so just insert them
+           ;; immediately.
+           (let ((handle (mm-get-content-id
+                          (setq url (match-string 1 url))))
+                 image)
+             (when handle
+               (mm-with-part handle
+                 (setq image (gnus-create-image (buffer-string)
+                                                nil t))))
+             (when image
+                (let ((string (buffer-substring start end)))
+                  (delete-region start end)
+                  (gnus-put-image image (gnus-string-or string "*") 'cid)
+                  (gnus-add-image 'cid image))))
+         ;; Normal, external URL.
+          (let ((alt-text (when (string-match "\\(alt\\|title\\)=\"\\([^\"]+\\)"
+                                              parameters)
+                            (xml-substitute-special (match-string 2 parameters)))))
+            (gnus-put-text-property start end 'gnus-image-url url)
+            (if (gnus-html-image-url-blocked-p
+                 url
+                 (if (buffer-live-p gnus-summary-buffer)
+                     (with-current-buffer gnus-summary-buffer
+                       gnus-blocked-images)
+                   gnus-blocked-images))
+                (progn
+                  (widget-convert-button
+                   'link start end
+                   :action 'gnus-html-insert-image
+                   :help-echo url
+                   :keymap gnus-html-image-map
+                   :button-keymap gnus-html-image-map)
+                  (let ((overlay (gnus-make-overlay start end))
+                        (spec (list url start end alt-text)))
+                    (gnus-overlay-put overlay 'local-map gnus-html-image-map)
+                    (gnus-overlay-put overlay 'gnus-image spec)
+                    (gnus-put-text-property
+                     start end
+                     'gnus-image spec)))
+              ;; Non-blocked url
+              (let ((width
+                     (when (string-match "width=\"?\\([0-9]+\\)" parameters)
+                       (string-to-number (match-string 1 parameters))))
+                    (height
+                     (when (string-match "height=\"?\\([0-9]+\\)" parameters)
+                       (string-to-number (match-string 1 parameters)))))
+                ;; Don't fetch images that are really small.  They're
+                ;; probably tracking pictures.
+                (when (and (or (null height)
+                               (> height 4))
+                           (or (null width)
+                               (> width 4)))
+                  (gnus-html-display-image url start end alt-text))))))))))
+
+(defun gnus-html-display-image (url start end alt-text)
+  "Display image at URL on text from START to END.
+Use ALT-TEXT for the image string."
+  (if (gnus-html-cache-expired url gnus-html-image-cache-ttl)
+      ;; We don't have it, so schedule it for fetching
+      ;; asynchronously.
+      (gnus-html-schedule-image-fetching
+       (current-buffer)
+       (list url alt-text))
+    ;; It's already cached, so just insert it.
+    (gnus-html-put-image (gnus-html-get-image-data url) url alt-text)))
+
+(defun gnus-html-wash-tags ()
+  (let (tag parameters string start end images url)
+    (gnus-html-pre-wash)
+    (gnus-html-wash-images)
+
     (goto-char (point-min))
+    ;; Then do the other tags.
     (while (re-search-forward "<\\([^ />]+\\)\\([^>]*\\)>" nil t)
       (setq tag (match-string 1)
            parameters (match-string 2)
@@ -128,72 +278,7 @@ fit these criteria."
       (setq end (point))
       (cond
        ;; Fetch and insert a picture.
-       ((equal tag "img_alt")
-        (when (string-match "src=\"\\([^\"]+\\)" parameters)
-         (setq url (match-string 1 parameters))
-          (gnus-message 8 "gnus-html-wash-tags: fetching image URL %s" url)
-         (if (string-match "^cid:\\(.*\\)" url)
-             ;; URLs with cid: have their content stashed in other
-             ;; parts of the MIME structure, so just insert them
-             ;; immediately.
-             (let ((handle (mm-get-content-id
-                            (setq url (match-string 1 url))))
-                   image)
-               (when handle
-                 (mm-with-part handle
-                   (setq image (gnus-create-image (buffer-string)
-                                                  nil t))))
-               (when image
-                 (let ((string (buffer-substring start end)))
-                   (delete-region start end)
-                   (gnus-put-image image (gnus-string-or string "*")))))
-           ;; Normal, external URL.
-           (if (gnus-html-image-url-blocked-p
-                url
-                (if (buffer-live-p gnus-summary-buffer)
-                    (with-current-buffer gnus-summary-buffer
-                      gnus-blocked-images)
-                  gnus-blocked-images))
-               (progn
-                 (widget-convert-button
-                  'link start end
-                  :action 'gnus-html-insert-image
-                  :help-echo url
-                  :keymap gnus-html-image-map
-                  :button-keymap gnus-html-image-map)
-                 (let ((overlay (gnus-make-overlay start end))
-                       (spec (list url
-