doc-view: better access to current page's text

1 files changed, 39 insertions, 22 deletions

diff --git a/lib/doc/jao-doc-view.el b/lib/doc/jao-doc-view.el
index cce158e..3631e98 100644
--- a/lib/doc/jao-doc-view.el
+++ b/lib/doc/jao-doc-view.el
@@ -47,7 +47,7 @@
               jao-doc-view--goer (or goto-page 'jao-doc-view-goto-page)))
 
 (defun jao-doc-view--imenu-create-index ()
-  (jao-pdf-imenu-index #'jao-doc-view--goer #jao-doc-view--imenu-file))
+  (jao-pdf-imenu-index 'jao-doc-view--goer 'jao-doc-view--imenu-file))
 
 (add-hook 'doc-view-mode-hook #'jao-pdf-set-up-imenu)
 
@@ -76,34 +76,51 @@
 
 (advice-add 'doc-view-goto-page :before #'jao-doc-view--trail-push)
 
+;;; Extract text
+(defun jao-doc-view-page-text (&optional re-render no-select)
+  (interactive "P")
+  (let* ((pno (doc-view-current-page))
+         (in buffer-file-name)
+         (cdir (or (doc-view--current-cache-dir) "/tmp"))
+         (out (format "%s/p%s.txt" cdir pno)))
+    (when (and (file-exists-p out) re-render)
+      (delete-file out))
+    (unless (file-exists-p out)
+      (shell-command-to-string (format "mutool convert -o %s %s %s" out in pno)))
+    (if no-select
+        out
+      (find-file out)
+      (view-mode))))
+
+(define-key doc-view-mode-map "t" #'jao-doc-view-page-text)
+
 ;;; Find URLs
+(defun jao-doc-view--full-txt ()
+  (expand-file-name "doc.txt" (doc-view--current-cache-dir)))
+
+(defun jao-doc-view--collect-urls (file)
+  (with-current-buffer (find-file-noselect file)
+    (goto-char (point-min))
+    (let ((urls nil))
+      (while (re-search-forward "https?://" nil t)
+        (push (thing-at-point-url-at-point) urls))
+      urls)))
+
 (defun jao-doc-view--page-urls (&optional all)
-  (if doc-view--current-converter-processes
-      (message "DocView: please wait till conversion finished.")
-    (let ((txt (expand-file-name "doc.txt" (doc-view--current-cache-dir)))
-          (page (doc-view-current-page))
-          (pd-rx "^")
-          (urls))
-      (if (file-readable-p txt)
-          (with-current-buffer (find-file-noselect txt)
-            (goto-char (point-min))
-            (unless all (re-search-forward pd-rx nil t (1- page)))
-            (let ((end (save-excursion
-                         (if (and (not all) (re-search-forward pd-rx nil t))
-                             (point)
-                           (point-max)))))
-              (while (re-search-forward "https?://" end t)
-                (push (thing-at-point-url-at-point) urls))
-              urls))
-        (doc-view-doc->txt txt (lambda () (jao-doc-view--page-urls all)))
-        'wait))))
+  (cond ((and all (not (file-exists-p (jao-doc-view--full-txt))))
+         (message "Full text not extracted yet: doing so!")
+         (doc-view-doc->txt txt (lambda () (message "Text extracted")))
+         'wait)
+        (all (jao-doc-view--collect-urls (jao-doc-view--full-txt)))
+        (t (jao-doc-view--collect-urls (jao-doc-view-page-text nil t)))))
 
 (defun jao-doc-view-visit-url (all)
   "Visit URL displayed in this page."
-  (interactive "P" doc-view-mode)
+  (interactive "P")
   (let ((urls (jao-doc-view--page-urls all)))
     (cond ((eq 'wait urls) (message "Extracting text, please wait and retry."))
-          ((zerop (length urls)) (message "No URLs in this page"))
+          ((zerop (length urls))
+           (message "No URLs in this %s" (if all "document" "page")))
           (t (when-let (url (completing-read "URL: " urls nil nil
                                              (when (null (cdr urls)) (car urls))))
                (browse-url url))))))