On 2015-01-12 23:56, Nicolas Goaziou writes: > Alan Schmitt writes: > >> Should we assume the cookie is at the end of the headline or not? > > The cookie can be anywhere within the headline text. > >> I'd gladly do this, but to use the `org-element-map' function it seem >> that I need a parse tree. How can I get it? Or is it possible to use >> directly `org-element-map' on a buffer? > > You don't need to use `org-element-map'[fn:1]. However, you cannot just use > `re-search-forward' either. You probably need to map over entries (e.g., > with `org-map-entries'), apply some filter to current headline, and > compare it with link's path. This is slower than the current > implementation. > > Another option would be to ignore only contents of statistics cookies, > not the whole cookie. This way we still can turn path into a proper > regexp. Would the following work as a regexp builder that allows arbitrary space and cookies between each word (making sure there is at least one)? #+begin_src emacs-lisp (defun org-heading-regexp-build (s) (let* ((sp (reverse (org-split-string s))) (wspace "[ \t]") (wspaceopt (concat wspace "*")) (cookie (concat "\\(?:" wspaceopt "\\(?:\\[[0-9]+%\\]\\|\\[[0-9]+/[0-9]+\\]\\)" wspaceopt "\\)")) (sep (concat "\\(?:" wspace "+\\|" cookie "+\\)")) res) (dolist (w sp) (setq res (concat w sep res))) (concat sep res))) #+end_src This of course needs to be extended with the other headline features (todo keyword, tags, …), this is just to know if I'm on the right track. Best, Alan -- OpenPGP Key ID : 040D0A3B4ED2E5C7