Only indent nested includes when indent is activated

Only indent content without section when "indent" is set

(Hopefully) fix the internal link and media handling

As proposed in issue #39 the link is now resolved using resolve_id and
then again transformed into an absolute link by prepending ":".

Detect more cases where additional sections aren't necessary

Move the section edit recognition into the helper/wrap syntax

Move existence check to the helper in order to avoid code duplication

Wrap content that is not in a section in a section

At the beginning of every page there might be content that is not in a
section, that is the content before the first header. When such a page
is in

Wrap content that is not in a section in a section

At the beginning of every page there might be content that is not in a
section, that is the content before the first header. When such a page
is included in another page at a higher level this content without a
section hasn't been indented properly. Now when needed a new section
around that content is created.

show more ...

Add compatibility code for the blog plugin

Add section edit button handling

This commit fixes section edit buttons. Now section edit buttons of
included pages don't interfere with buttons of the parent page anymore
and all section edits of t

Add section edit button handling

This commit fixes section edit buttons. Now section edit buttons of
included pages don't interfere with buttons of the parent page anymore
and all section edits of the included page are closed at the end of the
included page. Furthermore the id and revision of section edit buttons
of included pages are correct so they should actually work. A
redirect_id like in other edit buttons ensures that the user is
redirected to the correct page after the edit.

show more ...

Merge branch 'master' into render_instructions_rewrite

Conflicts:
syntax/meta.php

vim-modelines: Remove no longer allowed enc=utf-8

Remove superfluous and wrong cleanID call

This call to cleanID prevented things like :id, ..:id etc. from working
and is superfluous as resolve_pageid calls cleanID anyway.

Restructuring, first version of new cache handling

This commit replaces user dependent include keys by just one cache for
each page. This means the cache needs to be regenerated more often when
e.g.

Restructuring, first version of new cache handling

This commit replaces user dependent include keys by just one cache for
each page. This means the cache needs to be regenerated more often when
e.g. the edit permissions are different. As this is just a renderer
cache this shouldn't be a big problem. The main reason for this is that
the metadata won't be correctly updated when a different cache file is
used for every user.

In the metadata all include instructions are stored so when the cache is
checked the list of pages that actually should be included can be
reconstructed and then compared to the list of pages that has been
included when the page/metadata has been rendered.

The order of the includes and the respective parent pages are taken into
consideration so something like the inclusion of two user-dependent
pages that include each other won't break this cache handling. I'm not
sure if all scenarios are handled by this cache correctly, but I
can't think of a scenario where the page should be rendered but
neither the order nor the parent id nor one of the included pages has
been changed. When one included page is replaced by another one because
of these placeholders all sub-includes of that page will still be
checked but as then the cache will be purged anyway this doesn't matter.

I've decided against just storing one level of include instructions in
the metadata of every page and for storing everything in the root as
reading a lot of files should imho be avoided. The metadata that is read
might be outdated, but outdated metadata can only be a result of a
change in one of the included pages which is handled by the
file dependencies.

show more ...

New attempt of a rewrite using instructions during rendering

This patch makes the include plugin no longer do all the work in the
instructions but instead during rendering the instructions of the ch

New attempt of a rewrite using instructions during rendering

This patch makes the include plugin no longer do all the work in the
instructions but instead during rendering the instructions of the child
pages are converted and executed in the renderer of the page itself.
This means all advantages of working at instruction level are preserved,
but tricky caching problems are avoided. This is still not working
completely:

- The cache doesn't work at all, if your metadata is fresh the renderer
cache won't be used at all
- No tests have been done if ACL support is actually working (but it
should)
- Problems with the box plugin that have been fixed might reappear
- The support for different levels hasn't been tested that thoroughly
- The blog plugin will be broken with this version, perhaps
compatibility fixes will be added again later
- The metadata cache isn't refreshed when it should, at least in the
current git version of DokuWiki

show more ...

Resolve page id before the ACL check

removed deprecated pass-by-reference

fixed some instruction handling

added possiblity to include tagged pages closes #28

fixed macro substitution

cleanID() prior to acl checks

don't close open sections in some cases (box plugin)

fixed header calculation (hopefully)

switched to new plugin.info.txt layout

another attempt to fix the metadata issue

sort pages in namespace mode