* lisp/org-element.el (org-element--cache-root): New function.
(org-element--cache-sync, org-element-cache-get): Use new function.
This is a workaround for `avl-tree--root' bug before Emacs 24. Thanks
to Daniel Clemente for reporting it.
* lisp/org-element.el (org-element-at-point, org-element-cache-get,
org-element-cache-put, org-element--cache-sync): Complete rewrite to
use new data structure.
(org-element-context, org-element-cache-reset): Slight change in order
to use new tools to access cached data.
(org-element-cache-merge-changes-threshold): Renamed from
`org-element--cache-merge-changes-threshold'.
(org-element-cache-sync-idle-time): Renamed from
`org-element--cache-sync-idle-time'.
(org-element--cache-objects): New variable.
Now elements are stored in AVL tree and objects in a hash table. Also
moved functions relative to cache into a specific section of the file.
* lisp/org-element.el (org-element--cache-sync): Fix cache update when
a only contents of an element are modified.
(org-element--cache-shift-positions): Add optional argument allowing
partial shifting.
* lisp/org-element.el (org-element-at-point): Simplify opportunistic
shortcut by looking only for blank lines and headlines or
inlinetasks. Since regexp is simpler, we can afford to make more
searches.
* lisp/org-element.el (org-element--cache-stable-types): New variable.
(org-element--cache-sync): Do not store elements with missing parents
in cache.
The bug was introduced in 71c8474ae9.
Parsing doesn't always start from beginning of section, which means
that missing parents aren't always repaired while moving to the
current element. Therefore, we need to remove any element from cache
with an invalid parent property.
* lisp/org-element.el (org-element-secondary-p): New function.
* lisp/ox.el (org-export-get-previous-element,
org-export-get-next-element): Use new function.
* testing/lisp/test-org-element.el (test-org-element/secondary-p): New
test.
* lisp/org-element.el (org-element-at-point): Rewrite function. Remove
optional argument.
Include an opportunistic search at the beginning of the function. It
drastically improves speed on large sections for a small overhead on
small ones.
* lisp/org-element.el (org-element-timestamp-interpreter):
Ignore :raw-value and build timestamp from numeric properties
instead.
* testing/lisp/test-org-element.el: Update tests.
* testing/lisp/test-ox.el: Update tests.
This change allow to modify and update a timestamp easily without
requiring to reset :raw-value first, which was not intuitive.
* lisp/org-element.el (org-element--cache-sync): Properly remove from
cache last element in a plain list when blank lines at the end of
that list are modified.
* lisp/org-element.el (org-element-interpret-data): Change signature.
One can now define pseudo elements and objects types.
(org-element--interpret-data-1): New function.
* testing/lisp/test-org-element.el: Add tests.
* testing/lisp/test-ox.el: Update tests.
A pseudo element or object is a new element or object type that is
created and treated locally within an export back-end. The back-end
provides a translator for it and it is ignored when interpreted back
into Org syntax.
* lisp/org-element.el (org-element-comment-parser): Fix infloop when
a non-empty blank line follows a comment at the end of the buffer.
* testing/lisp/test-org-element.el: Add test.
* lisp/org-element.el (org-element-headline-interpreter): Take into
consideration `org-odd-levels-only' when building a headline.
* lisp/ox-org.el (org-org-headline): Correctly set transcoded headline
level during subtree export.
* lisp/org-element.el (org-element-latex-or-entity-successor,
org-element-latex-fragment-parser): Use `org-latex-regexps' instead
of `org-format-latex-options'.
* testing/lisp/test-org-element.el: Fix tests.
`org-latex-regexps' defines the correct syntax for LaTeX code within
an Org buffer. On the other hand, `org-format-latex-options'
determines which syntax can be previewed within the buffer.
* lisp/org-element.el (org-element-babel-call-parser,
org-element-inline-babel-call-parser): Do not call
`org-babel-lob-get-info' since it is out of parser's scope.
(org-element-babel-call-interpreter,
org-element-inline-babel-call-interpreter): Apply changes to parsers.
* lisp/org-element.el (org-element-center-block-parser,
org-element-drawer-parser, org-element-dynamic-block-parser,
org-element-item-parser, org-element-quote-block-parser,
org-element-comment-block-parser, org-element-export-block-parser,
org-element-verse-block-parser, org-element-special-block-parser,
org-element-example-block-parser, org-element-headline-parser,
org-element-inlinetask-parser): Remove :hiddenp property.
* lisp/org.el (org-end-of-line, org-down-element): Use an equivalent
of :hiddenp property.
* testing/lisp/test-org-element.el: Remove tests.
The property is removed because it is buggy (e.g., when there's a link
just after a block opening line) and not really useful.
* lisp/org-element.el (org-element-at-point): If point is at the end
of the buffer, and that buffer ends with a list, and there's no
final newline, return last element in last item instead of plain
list.
* testing/lisp/test-org-element.el: Add test.
This fix allows to fill correctly the following construct:
- Item with a very long line ending at the end of buffer.
* lisp/org-element.el (org-element-footnote-definition-parser): Fix
value for :contents-begin when first line of footnote definition is
empty besides the label.
* testing/lisp/test-org-element.el (test-org-element/footnote-definition-parser):
Add test.
* lisp/org-element.el (org-element-at-point): Return correct element
when point is on a blank line just below a headline.
* testing/lisp/test-org-element.el: Add test.
* lisp/org-element.el (org-element-text-markup-successor,
org-element-latex-or-entity-successor,
org-element-export-snippet-successor,
org-element-footnote-reference-successor,
org-element-inline-babel-call-successor,
org-element-inline-src-block-successor,
org-element-line-break-successor, org-element-link-successor,
org-element-plain-link-successor, org-element-macro-successor,
org-element-radio-target-successor,
org-element-statistics-cookie-successor,
org-element-sub/superscript-successor,
org-element-table-cell-successor, org-element-target-successor,
org-element-timestamp-successor): Remove LIMIT argument.
(org-element--parse-objects, org-element--get-next-object-candidates):
Apply signature change to successors.
(org-element-context): Narrow buffer around object containers so
parsing of objects when using this function is done under the same
restrictions as in buffer parsing.
* testing/lisp/test-org-element.el: Add test.
* lisp/org-element.el (org-element-plain-list-parser): Fix infloop
when parsing a list at the end of buffer, if buffer doesn't end at
a line beginning.
* testing/lisp/test-org-element.el: Add test.
* lisp/org-element.el (org-element-link-parser): Do not url-decode
parsed links.
The function now assumes link at point is correctly encoded if needed.
`org-insert-link' already automatically url-encode generated links,
and link types not handled by this function (e.g., custom-id links) do
not need such encoding.
* lisp/org-element.el (org-element-item-interpreter): Correctly
interpret back plain lists with "*" items. This fixes "This is not
a list" error returned in this case.
* testing/lisp/test-org-element.el: Add test.
* lisp/org-element.el (org-element--current-element): Fix
org-meta-return error at the end of buffer.
* testing/lisp/test-org-element.el: Add test.
Thanks to Nicolas Richard for reporting this.
* lisp/org-element.el (org-element-special-block-parser): Fix typo in
regexp matching block type. Also quote the type so it can contain
special characters.
* testing/lisp/test-org-element.el: Add test.
* lisp/org.el (org-property-re): Improve definition so that this regex
can be used in all situations. Extend docstring with explanation of
matching groups.
(org-at-property-p): Implement using `org-element-at-point'.
(org-entry-properties, org-buffer-property-keys, org-indent-line):
Use `org-property-re' and adjust match group numbers accordingly.
* lisp/org-element.el (org-element-node-property-parser): Use
`org-property-re' and adjust match group numbers accordingly. Move
`looking-at' out of the let clause to not rely on the unspecified
evaluation order inside the let.
* lisp/org-element.el (org-element--list-struct): New function.
(org-element-plain-list-parser, org-element--current-element): Use new
function.
This patch removes dependency on org-list.el to parsing lists. For
now, it leads to code duplication, but, ultimately (i.e., when parsing
will be faster), org-list.el will delegate the parsing job to
org-element. The new implementation is also faster than the previous one.
* lisp/org-element.el (org-element--remove-indentation): New function.
(org-element-example-block-parser, org-element-src-block-parser): Use
new function.
(org-element-src-block-interpreter): Update function according to
change.
* lisp/ox.el (org-export-unravel-code): Do not remove any indentation
since it now happens at the parser level.
* testing/lisp/test-org-element.el: Add tests.
* testing/lisp/test-ox.el: Refactor tests.
* lisp/org-element.el (org-element--current-element): Be stricter when
matching arguments in LaTeX environments. In particular, do not
allow anything else than options and arguments in the opening line.
* testing/lisp/test-org-element.el: Add tests.
* lisp/org-element.el (org-element--current-element): Allow the
opening string of a LaTeX environment to contain additional
arguments.
* testing/lisp/test-org-element.el: Add tests.
* lisp/org-element.el (org-element-fixed-width-interpreter): Fix
interpretation of fixed-width elements with a nil or empty string
value.
* testing/lisp/test-org-element.el: Add tests.
* lisp/org-element.el (org-element-macro-parser): Allow to escape
escaping character before a comma. Also do not trim spaces at
argument boundaries.
* doc/org.texi (Macro replacement): Update documentation about
possible locations and escaping mechanism.
* testing/lisp/test-org-element.el: Add tests.
With this patch, macro's arguments are read as the following:
"a,b" -> '("a" "b")
"a\,b" -> '("a,b")
"a\\,b" -> '("a\" "b")
"a\\\,b" -> '("a\,b")
"a\\\\,b" -> '(a"\\" "b")
Note that with the patch, you only need to escape backslashes before
a comma:
"a\\b\,c" -> '("a\\b,c")
* org-element.el (org-element-paragraph-separate): Use
new name `org-list-allow-alphabetical'.
* org-list.el (org-list-allow-alphabetical): Rename from
`org-alphabetical-lists'.
(org-list-empty-line-terminates-plain-lists): Rename from
`org-empty-line-terminates-plain-lists'.
(org-checkbox-hierarchical-statistics): Rename from
`org-hierarchical-checkbox-statistics'.
* org.texi (Plain lists, Checkboxes): Use non-obsolete
variable names.
* lisp/org-element.el (org-element-all-successors): Add `plain-link'
successor.
(org-element-object-restrictions): Remove `link' within `link'. Allow
`plain-link' instead.
(org-element-plain-link-successor): New function.
* testing/lisp/test-org-element.el: Add test.
Plain links within links are needed for the following kind of syntax:
[[http://orgmode.org][file:unicorn.png]]
No other link type is allowed within links.
Nicolas Goaziou
Bastien Guerry
Achim Gratz