gnus-html.el: Add extra logging.
[gnus] / lisp / gnus-html.el
index f6cf00d..5ad90a8 100644 (file)
 
 ;;; Code:
 
+(eval-when-compile (require 'cl))
+(eval-when-compile (require 'mm-decode))
+(require 'mm-url)
+
 (defcustom gnus-html-cache-directory (nnheader-concat gnus-directory "html-cache/")
   "Where Gnus will cache images it downloads from the web."
+  :version "24.1"
   :group 'gnus-art
   :type 'directory)
 
 (defcustom gnus-html-cache-size 500000000
   "The size of the Gnus image cache."
+  :version "24.1"
   :group 'gnus-art
   :type 'integer)
 
 (defcustom gnus-html-frame-width 70
   "What width to use when rendering HTML."
+  :version "24.1"
   :group 'gnus-art
   :type 'integer)
 
+(defcustom gnus-blocked-images "."
+  "Images that have URLs matching this regexp will be blocked."
+  :version "24.1"
+  :group 'gnus-art
+  :type 'regexp)
+
 ;;;###autoload
 (defun gnus-article-html (handle)
   (let ((article-buffer (current-buffer)))
     (save-restriction
       (narrow-to-region (point) (point))
       (save-excursion
-       (set-buffer (car handle))
-       (call-process-region (point-min) (point-max)
-                            "w3m" 
-                            nil article-buffer nil
-                            "-halfdump"
-                            "-t" (format "%s" tab-width)
-                            "-cols" (format "%s" gnus-html-frame-width)
-                            "-o" "display_image=off"
-                            "-T" "text/html"))
+       (mm-with-part handle
+         (let* ((coding-system-for-read 'utf-8)
+                (coding-system-for-write 'utf-8)
+                (default-process-coding-system
+                  (cons coding-system-for-read coding-system-for-write))
+                (charset (mail-content-type-get (mm-handle-type handle)
+                                                'charset)))
+           (when (and charset
+                      (setq charset (mm-charset-to-coding-system charset))
+                      (not (eq charset 'ascii)))
+             (mm-decode-coding-region (point-min) (point-max) charset))
+           (call-process-region (point-min) (point-max)
+                                "w3m" 
+                                nil article-buffer nil
+                                "-halfdump"
+                                "-no-cookie"
+                                "-I" "UTF-8"
+                                "-O" "UTF-8"
+                                "-o" "ext_halfdump=1"
+                                "-o" "pre_conv=1"
+                                "-t" (format "%s" tab-width)
+                                "-cols" (format "%s" gnus-html-frame-width)
+                                "-o" "display_image=off"
+                                "-T" "text/html"))))
       (gnus-html-wash-tags))))
 
+(defvar gnus-article-mouse-face)
+
 (defun gnus-html-wash-tags ()
-  (let (tag parameters string start end images)
+  (let (tag parameters string start end images url)
     (mm-url-decode-entities)
     (goto-char (point-min))
     (while (re-search-forward "<\\([^ />]+\\)\\([^>]*\\)>" nil t)
       (cond
        ;; Fetch and insert a picture.
        ((equal tag "img_alt")
-       (when (string-match "src=\"\\([^\"]+\\)" parameters)
-         (setq parameters (match-string 1 parameters))
-         (when (or (null mm-w3m-safe-url-regexp)
-                   (string-match mm-w3m-safe-url-regexp parameters))
-           (let ((file (gnus-html-image-id parameters)))
-             (if (file-exists-p file)
-                 ;; It's already cached, so just insert it.
-                 (when (gnus-html-put-image file (point))
-                   ;; Delete the ALT text.
-                   (delete-region start end))
-               ;; We don't have it, so schedule it for fetching
-               ;; asynchronously.
-               (push (list parameters
-                           (set-marker (make-marker) start)
-                           (point-marker))
-                     images))))))
+        (when (string-match "src=\"\\([^\"]+\\)" parameters)
+         (setq url (match-string 1 parameters))
+          (gnus-message 8 "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
+                 (delete-region start end)
+                 (gnus-put-image image)))
+           ;; Normal, external URL.
+           (when (or (null gnus-blocked-images)
+                     (not (string-match gnus-blocked-images url)))
+             (let ((file (gnus-html-image-id url)))
+               (if (file-exists-p file)
+                   ;; It's already cached, so just insert it.
+                   (when (gnus-html-put-image file (point))
+                     ;; Delete the ALT text.
+                     (delete-region start end))
+                 ;; We don't have it, so schedule it for fetching
+                 ;; asynchronously.
+                 (push (list url
+                             (set-marker (make-marker) start)
+                             (point-marker))
+                       images)))))))
        ;; Add a link.
        ((equal tag "a")
        (when (string-match "href=\"\\([^\"]+\\)" parameters)
-         (setq parameters (match-string 1 parameters))
+         (setq url (match-string 1 parameters))
+          (gnus-message 8 "Fetching link URL %s" url)
          (gnus-article-add-button start end
-                                  'browse-url parameters)
+                                  'browse-url url
+                                  url)
          (let ((overlay (gnus-make-overlay start end)))
            (gnus-overlay-put overlay 'evaporate t)
-           (gnus-overlay-put overlay 'gnus-button-url parameters)
+           (gnus-overlay-put overlay 'gnus-button-url url)
            (when gnus-article-mouse-face
              (gnus-overlay-put overlay 'mouse-face gnus-article-mouse-face)))))
        ;; Whatever.  Just ignore the tag.
       (gnus-html-schedule-image-fetching (current-buffer) (nreverse images)))))
 
 (defun gnus-html-schedule-image-fetching (buffer images)
+  (gnus-message 8 "Scheduling image fetching in buffer %s, images %s" buffer images)
   (let* ((url (caar images))
         (process (start-process
                   "images" nil "curl"
                   "-s" "--create-dirs"
+                  "--location"
+                  "--max-time" "60"
                   "-o" (gnus-html-image-id url)
                   url)))
+    (process-kill-without-query process)
     (set-process-sentinel process 'gnus-html-curl-sentinel)
-    (set-process-plist process (list 'images images
-                                    'buffer buffer))))
+    (gnus-set-process-plist process (list 'images images
+                                         'buffer buffer))))
 
 (defun gnus-html-image-id (url)
   (expand-file-name (sha1 url) gnus-html-cache-directory))
 
 (defun gnus-html-curl-sentinel (process event)
   (when (string-match "finished" event)
-    (let* ((images (getf (process-plist process) 'images))
-          (buffer (getf (process-plist process) 'buffer))
+    (let* ((images (gnus-process-get process 'images))
+          (buffer (gnus-process-get process 'buffer))
           (spec (pop images))
           (file (gnus-html-image-id (car spec))))
-      (when (and (file-exists-p file)
-                (buffer-live-p buffer))
-       (save-excursion
-         (set-buffer buffer)
-         (let ((buffer-read-only nil))
+      (when (and (buffer-live-p buffer)
+                ;; If the position of the marker is 1, then that
+                ;; means that the text it was in has been deleted;
+                ;; i.e., that the user has selected a different
+                ;; article before the image arrived.
+                (not (= (marker-position (cadr spec)) (point-min))))
+       (with-current-buffer buffer
+         (let ((inhibit-read-only t))
            (when (gnus-html-put-image file (cadr spec))
-             (delete-region (cadr spec) (caddr spec))))))
+             (delete-region (1+ (cadr spec)) (caddr spec))))))
       (when images
        (gnus-html-schedule-image-fetching buffer images)))))
 
 (defun gnus-html-put-image (file point)
-  (let ((image (ignore-errors
-                (create-image file))))
-    (if image
-       (progn
-         (put-image image point)
-         t)
-      (put-image (find-image '((:type xpm :file "lock-broken.xpm")))
-                point)
-      nil)))
+  (when (display-graphic-p)
+    (let ((image (ignore-errors
+                  (gnus-create-image file))))
+      (save-excursion
+       (goto-char point)
+       (if (and image
+                ;; Kludge to avoid displaying 30x30 gif images, which
+                ;; seems to be a signal of a broken image.
+                (not (and (listp image)
+                          (eq (plist-get (cdr image) :type) 'gif)
+                          (= (car (image-size image t)) 30)
+                          (= (cdr (image-size image t)) 30))))
+           (progn
+             (gnus-put-image image)
+             t)
+         (when (fboundp 'find-image)
+           (gnus-put-image (find-image
+                            '((:type xpm :file "lock-broken.xpm")))))
+         nil)))))
 
 (defun gnus-html-prune-cache ()
   (let ((total-size 0)
          (decf total-size (cadr file))
          (delete-file (nth 2 file)))))))
 
-(defun gnus-html-prefetch-images ()
-  (save-match-data
-    (let (urls)
-      (while (re-search-forward "<img.*src=[\"']\\([^\"']+\\)" nil t)
-       (let ((url (match-string 1)))
-         (when (or (null mm-w3m-safe-url-regexp)
-                   (string-match mm-w3m-safe-url-regexp url))
-           (unless (file-exists-p (gnus-html-image-id url))
-             (push url urls)
-             (push (gnus-html-image-id url) urls)
-             (push "-o" urls)))))
-      (apply 'start-process 
-            "images" nil "curl"
-            "-s" "--create-dirs"
-            urls))))
+;;;###autoload
+(defun gnus-html-prefetch-images (summary)
+  (let (blocked-images urls)
+    (when (buffer-live-p summary)
+      (with-current-buffer summary
+       (setq blocked-images gnus-blocked-images))
+      (save-match-data
+       (while (re-search-forward "<img.*src=[\"']\\([^\"']+\\)" nil t)
+         (let ((url (match-string 1)))
+           (if (or (null blocked-images)
+                    (not (string-match blocked-images url)))
+                (unless (file-exists-p (gnus-html-image-id url))
+                  (push url urls)
+                  (push (gnus-html-image-id url) urls)
+                  (push "-o" urls))
+              (gnus-message 8 "Image URL %s is blocked" url))))
+       (let ((process
+              (apply 'start-process 
+                     "images" nil "curl"
+                     "-s" "--create-dirs"
+                     "--location"
+                     "--max-time" "60"
+                     urls)))
+         (process-kill-without-query process))))))
+
+(provide 'gnus-html)
 
 ;;; gnus-html.el ends here