Re: [O] finding a parent node

[Thorsten Jolitz]

address@hidden (Łukasz Stelmach) writes:

Hello,

> What is the best way to iterate over ((great)*grand)?parent headlines of
> the current one (at point), that meets some criteria.

not an answer to your question, but rather a related question I wanted
to post anyway, so I do it in this thread:

simple-test.org:
#+begin_src org
* header 1
  :PROPERTIES:
  :CUSTOM_ID: XYZ22
  :END:
* header 2
  [2013-06-28 Fr 11:01]
** subheader 1
Some text
** subheader 2

More text and a table

| label  | col1 | col2 |
|--------+------+------|
| string | 3    | 4    |

Text and a src-block

#+begin_src emacs-lisp
 (+ 3 4)
#+end_src

#+end_src

parse buffer:
#+begin_src emacs-lisp
(progn
  (goto-char (point-max))
  (newline 2)
  (insert
   (format "%s"
           (with-current-buffer
               (find-file-noselect
                "/path/to/simple-test.org")
             (setq tree (org-element-parse-buffer))))) )
#+end_src

excerpts form parse-tree:
#+begin_src emacs-lisp
  (org-data
   nil
   ;; first headline
   (headline
    (:raw-value header 1 :begin 1 :end55 :pre-blank0 :hiddenp
  outline :contents-begin 12 :contents-end 55 :level 1 :priority
  nil :tags nil :todo-keyword nil :todo-type nil :post-blank
  1 :footnote-section-p nil :archivedp nil :commentedp nil :quotedp
  nil :CUSTOM_ID XYZ22 :CATEGORY simple-test :title (header
  1) :parent #0)

    (section (:begin 12 :end 55 :contents-begin 12 :contents-end
  55 :post-blank 0 :parent #1)

     (property-drawer (:begin 12 :end 55 :hiddenp
  outline :contents-begin 27 :contents-end 47 :post-blank
  0 :post-affiliated 12 :parent #2) (node-property (:key
  CUSTOM_ID :value XYZ22 :begin 27 :end 47 :post-blank 0 :parent
  #3)))))

   ;; second headline
   (headline (:raw-value header 2 :begin 55 :end 295 :pre-blank
  0 :hiddenp outline :contents-begin 66 :contents-end 295 :level
  1 :priority nil :tags nil :todo-keyword nil :todo-type
  nil :post-blank 0 :footnote-section-p nil :archivedp
  nil :commentedp nil :quotedp nil :CATEGORY
  simple-test :title (header 2) :parent #0)

    (section (:begin 66 :end 90 :contents-begin 66 :contents-end
  90 :post-blank 0 :parent #1) (paragraph (:begin 66 :end
  90 :contents-begin 66 :contents-end 90 :post-blank
  0 :post-affiliated 66 :parent #2)

     (timestamp (:type inactive :raw-value [2013-06-28 Fr
  11:01] :year-start 2013 :month-start 6 :day-start 28 :hour-start
  11 :minute-start 1 :year-end 2013 :month-end 6 :day-end
  28 :hour-end 11 :minute-end 1 :begin 68 :end 89 :post-blank
  0 :parent #3)) )) ...) )

#+end_src

now I can use `org-element-map' to access the elements and their attributes:

#+begin_src emacs-lisp
(insert
 (format "\n\n%s"
         (org-element-map
             tree
             'timestamp
           '(lambda (X) (org-element-property :raw-value X)))))

;; => ([2013-06-28 Fr 11:01])

#+end_src

this gives me the first timestamp element and its property list:
#+begin_src emacs-lisp
(setq stamp1
      (org-element-map
          tree
          'timestamp
        'identity nil t))
#+end_src

but for further processing I always struggle with the :parent attribute and
the circularities in the list. 

I often would like to use just a sublist like this, with :parent being a Link
or ID and not the parent object that refers to its parent object and so on ...

#+begin_src emacs-lisp
(timestamp (:type inactive :raw-value [2013-06-28 Fr
  11:01] :year-start 2013 :month-start 6 :day-start 28 :hour-start
  11 :minute-start 1 :year-end 2013 :month-end 6 :day-end
  28 :hour-end 11 :minute-end 1 :begin 68 :end 89 :post-blank
  0 :parent <<ID of or LINK to #3>>))
#+end_src

I tried to use the NO-RECURSION arg of `org-element-map' for this, but often
the parent is a headline or section, and then 'org-element-map' doesn't enter
headlines or sections and returns nil, i.e. the timestamp is never found. 

Is there a way - using the parser/export framework toolbox - to avoid these
circularities, i.e. to either ignore the :parent attribute or better replace
the list object in it by something else that nevertheless identifies the
parent unmistakenly? 

I know that with special bookkeeping while walking the tree one can deal with
these circularities, maybe I'm just new to this stuff, but I feel they make
the usual processing of such a nested list a bit complicated. 

--
cheers,
Thorsten