This addresses <http://ikiwiki.info/todo/aggregate_to_internal_pages/>
in a simple way. With this approach, a flag day is required, on which all
users of aggregated pages start to inline them using the internal() pagespec;
after that, the aggregateinternal option can safely be switched on in the
setup file (and the old aggregated pages can be deleted by hand).
This ensures that the same link is reached as is used on pages,
so browsers will know that the link on pages has been visited, and color it
appropriately.
The constructor can fail with a useless error message if module fail to
load. Work around this by evaling it, and checking for failures, and
printing CGI::Session->errstr to get a more useful message.
This is truely horribly disgusting. CGI::tmpFileName, in current perls, is
an undocumented function (which should be a clue..) that takes the original
filename of an uploaded attachment, and returns the name of the tempfile
that CGI has stored it in.
In old perls, though, CGI::tmpFileName does not take a filename. It takes
a key from the object's {'.tmpfiles'} hash. This key is something
crazy like '*Fh::fh00001group' -- apparently the stringification of a
filehandle object.
Just to add to the fun, tmpFileName doesn't take the key, it expects a
refernce to the key. Argh?!
But the fun doesn't stop there, because in perl 5.8, CGI.pm is also broken
in two other ways. The upload() method is supposed to return a filehandle
to the temp file. It doesn't. The param() method is supposed to return
a filehandle to the temp file, that stringifies to the original filename.
It returns just the original filename, no filehandle.
Combine all these bugs, and you end up with this disgusting commit. Since
I have no way to get the filehandle, I *need* to get the tempfile name.
If I had the filehandle, I could probably pass it into tmpFileName, and
it might strigify to the right key name. But I don't, so the only way to
determine the key is to grub through the .tmpfiles hash ourselves.
And finally, one the temp file name is discovered, a filehandle can finally
be obtained by (re)opening it.
I recommend that this commit be reverted when perl 5.8 is a mercifully
faded memory.
I'm really, really, really glad I'm actually being paid for working on
this right now!
* The editpage form now uses the raw page name, not the page title, in its
'page' cgi parameter. Using the title was ambiguous and made it
impossible to tell between some pages, like "foo/bar" and "foo__47__bar",
sometimes causing the wrong page to be edited.
* This change means that some edit links need to be updated.
Force a rebuild on upgrade to this version.
* Above change also allowed really fixing escaped slashes from the blogpost
form.
Currently includes UI, and a few tests of the attachment, as well as the
framework to extend pagespecs to test attachments. Does not actually save
the file yet.
* toc: Revert change in 2.45 that made it run at sanitize time. This breaks
use of toc in a sidebar.
* Call format hooks when generating page previews, thus fixing toc display
there, as well as fixing inlins to again display in page previews, since
it's started using format hooks. This also allows several other things,
like embed, that use format hooks, to work during page preview time.
* Format hooks should not rely on getting an entire html document, as they
will only get the body during page preview.
* toggle: Deal with preview mode when adding javascript.
This change needs libtext-wikicreole-perl (>= 0.05-2).
Also removing custom link function, there's no need for it -
if it is not defined, the unmodified markup will be returned.
Can't sort by titles; the tree building logic requires that the list be
sorted by page name.
Setting linktext => $page is not the same as omitting it entirely. So some
contortions to only set linktext when the page name is not being shown.
This occurred when a plugin, loaded earlier, filled out a template in its
checkconfig, before recentchanges's checkconfig had run. Since such a
template won't be a recentchanges template, just test for the value being
uninitialized and skip processing.
Using the title obscured path info, and made search results look
inconsistent. Since nothing else uses the title like that, it didn't make
sense for search to.
Because the search plugin needed it, also because it's one of the few
plugins that didn't already have it.
I also considered adding it to htmlize, but I really cannot imagine caring
what the destpage is when htmlizing. (I'll probably be poven wrong later.)
Something has changed in CGI.pm in perl 5.10. It used to not care
if STDIN was opened using :utf8, but now it'll mis-encode utf-8 values
when used that way by ikiwiki. Now I have to binmode(STDIN) before
instantiating the CGI object.
In 57bba4dac1, I changed from decoding
CGI::Formbuilder fields to utf-8, to decoding cgi parameters before setting
up the form object. As of perl 5.10, that approach no longer has any effect
(reason unknown). To get correctly encoded values in FormBuilder forms,
they must once again be decoded after the form is set up.
As noted in 57bba4da, this can cause one set of problems for
formbuilder_setup hooks if decode_form_utf8 is called before the hooks, and
a different set if it's called after. To avoid both sets of problems, call
it both before and after. (Only remaining problem is the sheer ugliness and
inefficiency of that..)
I think that these changes will also work with older perl versions, but I
haven't checked.
Also, in the case of the poll plugin, the cgi parameter needs to be
explcitly decoded before it is used to handle utf-8 values. (This may have
always been broken, not sure if it's related to perl 5.10 or not.)
Turns out duplicate index files do not need to be stored when usedirs is in
use, just when it's not. Ikiwiki is quite consistent about using page/ when
usedirs is in use. (The only exception is the search plugin, which needs
fixing.)
This also includes significant code cleanup, removal of a incorrect special
case for empty files, and addition of a workaround for a bug in the amazon
perl module.
The fix involved embedding the session id in the forms, and not allowing the
forms to be submitted if the embedded id does not match the session id.
In the case of the preferences form, if the session id is not embedded,
then the CGI parameters are cleared. This avoids a secondary attack where the
link to the preferences form prefills password or other fields, and
the user hits "submit" without noticing these prefilled values.
In the case of the editpage form, the anonok plugin can allow anyone to edit,
and so I chose not to guard against CSRF attacks against users who are not
logged in. Otherwise, it also embeds the session id and checks it.
For page editing, I assume that the user will notice if content or commit
message is changed because of CGI parameters, and won't blndly hit save page.
So I didn't block those CGI paramters. (It's even possible to use those CGI
parameters, for good, not for evil, I guess..)
The only other CSRF attack I can think of in ikiwiki involves the poll plugin.
It's certianly possible to set up a link that causes the user to unknowingly
vote in a poll. However, the poll plugin is not intended to be used for things
that people would want to attack, since anyone can after all edit the poll page
and fill in any values they like. So this "attack" is ignorable.
tag 473987 +patch
thanks
Hi,
The issue is that we need to convert relative links to absolute
ones for atom and rss feeds -- but there are two types of
relative links. The first kind, relative to the current
document ( href="some/path") is handled correctly. The second
kind of relative url is is relative to the http server
base (href="/semi-abs/path"), and that broke.
It broke because we just prepended the url of the current
document to the href (http://host/path/to/this-doc/ + link),
which gave us, in the first place:
http://host/path/to/this-doc/some/path [correct], and
http://host/path/to/this-doc//semi-abs/path [wrong]
The fix is to calculate the base for the http server (the base of
the wiki does not help, since the base of the wiki can be
different from the base of the http server -- I have, for example,
"url => http://host.name.mine/blog/manoj/"), and prepend that to
the relative references that start with a /.
This has been tested.
Signed-off-by: Manoj Srivastava <srivasta@debian.org>
destpage does not normally need to be worried about when creating other files
as part of the process of rendering a page. Using destpage results in
inlined pages creating two copies of such files. It works to not use destpage
in this case because the inlining page depends on the source page, so if the
source page is modified or deleted the inlining page will be updated.
Instead of using the XML-RPC v2 extension <nil/>, which Perl's
XML::RPC::Parser does not (yet) support (Joey's patch is pending), we
agreed on a sentinel: {'null':''}, that is, a hash with a single key
"null" pointing to the empty string.
The Python proxy automatically converts None appropriately and raises an
exception if a hook function should, by weird coincidence, attempt to
return {'null':''}.
Signed-off-by: martin f. krafft <madduck@madduck.net>
Markdown is slow. Especially if it has to process an enormous page. The
most common enormous page is currently the recentchanges page, which gets
processed a lot, and contains very little actual markdown. Most of it is a
big <div>, which markdown skips ... slowly.
This is a rather sick optimisation to work around markdown's speed issues.
Now inline inserts a small, dummy div, allows markdown to quickly render
the actual page content, then replaces the dummy with the actual inlined
pages later.
Results: Rendering just a recentchanges page, with diffs included, dropped
from 4.5 seconds to 2.7 seconds on my laptop. Building the entire wiki
dropped from 46.6 seconds to 39.5 seconds.
(It would be better if inline were a *post*-processor directive.)
xml rpc only allows functions to return a single value, no lists. So getargv
needs to return a list reference, which means that the caller will see an xml
rpc array.
This works around a perl crasher bug, and also avoids bloating pages
with enormous diffs.
rcs_recentchanges modified to return a list in an array context.
If a diff of the firsst commit in a git repo was requested, it would fail and
print to stderr since first^ isn't valid. Using git show will always work.
set to the destination page. This avoids need for hacks to munge the urls
in preview mode, which fixes several bugs.
* Several destpage fixes in plugins.
Adds an optional xrds-location parameter to the openid meta handler,
which allows for XRDS delegation.
A good document on XRDS is
http://www.windley.com/archives/2007/05/using_xrds.shtml
Signed-off-by: martin f. krafft <madduck@madduck.net>
correct, and it's certianly not correct now, since the wiki is locked
before rcs_commit is ever called, and should not be unlocked by
rcs_commit either.
Markdown is such a splintered mess.. The current debian package provides
only Text::Markdown::Markdown, while all versions of Text::Markdown support
Text::Markdown::markdown, and old versions also support the capitalised version,
while new ones don't.
It's getting to the point where `grep /markdown/i %symbol_table` is the only
sane way to figure out what function to call..
* rcs_diff is a new function that rcs modules should implement.
* Implemented rcs_diff for git, svn, and tla (tla version untested).
Mercurial and monotone still todo.
Add special handling for <meta name="robots" ...> which needs not be
scrubbed as it's harmless.
Signed-off-by: martin f. krafft <madduck@madduck.net>
(cherry picked from commit b15d0299a7f7b147e89d8a202d6cca1c21491af2)
A new regexp fixes this bug:
http://ikiwiki.info/bugs/No_link_for_blog_items_when_filename_contains_a_colon/
I traced this down to htmlscrubber. If disabled,
it works. If enabled, then $safe_url_regexp
determines the URL unsafe because of the colon and
hence removes the src attribute.
Digging into this, I find that RFC 3986 pretty
much discourages colons in filenames:
"""
A path segment that contains a colon character
(e.g., "this:that") cannot be used as the first
segment of a relative-path reference, as it would
be mistaken for a scheme name. Such a segment must
be preceded by a dot-segment (e.g., "./this:that")
to make a relative- path reference.
"""
on the other hand, with usedirs, any link to
another page will be prepended by ../ anyway, so
that makes them okay again.
The solution still seems not to use colons.
In any case, htmlscrubber should get a new regexp,
courtesy of dato.
I have tested and verified this.
Signed-off-by: martin f. krafft <madduck@madduck.net>
As was already done for linkfication, links generated in a prevew page
are relative to the top of the wiki, so it has to be told that the destpage
is there.
I was using "" to indicate this, but that may confuse some preprocessor
plugins, which treat parameters with an empry value specially (sparkline is one
such). Instead, use "/", which is more accurate anyway and works just as well.
which forced a scan of the page to make available metadata that
appeared after the inline directive. Problem is that scan made it forget
about any other files rendered due to the page. The scan also turns out
to be unnecessary now, since meta persistently stores state and it's
always available. So it was just removed.
(as preserving the full list across preview would be tricky). Userdirs
were still being offered as an option there, remove them.
* Fix a bug where user A created a page concurrently with user B, and
when B previewed it would redirect B to A's new page, losing B's work.
Instead, don't redirect and let conflict handling resolve it.
containing ikiwiki.cgi, but this should not change the urls to the style
sheets etc. Add a new forcebareurl parameter to misctemplate to allow
it to do that.
of XML::RPC's default of us-ascii. Allows interoperation with
python's xmlrpc library, which threw invalid encoding exceptions and
caused the rst plugin to hang.
Some browsers interpret about: URIs like a limited version of data:
URIs. In particular, some versions of Internet Explorer interpret
arbitrary HTML content in about: URIs.
just avoid actually writing the files. This is necessary because ikiwiki
saves state after a preview (in case it actually *did* write files),
and if will_render isn't called its security checks will get upset
when the page is saved. Thanks to Edward Betts for his help tracking this
tricky bug down.
- On commits, replace "mtn sync" bidirectional with "mtn push" single
direction. No need to pull changes when doing a commit. mtn sync
is still called in rcs_update.
- Support for viewing differences via patches using viewmtn.
Now aggregation will not lock the wiki. Any changes made during aggregaton are
merged in with the changed state accumulated while aggregating. A separate
lock file prevents multiple concurrent aggregators. Garbage collection
of orphaned guids is much improved. loadstate() is only called once
per process, so tricky support for reloading wiki state is not needed.
(Tested fairly thuroughly.)
since this leads to too many problems with web caching, especially with
inlined pages. Properly solving this would involve tracking every page
that contributes to a page's content and using the youngest of them all,
as well as special cases for things like the version plugin, and it's just
too complex to do.
license, and copyright. This can be used to create custom RecentChanges.
* meta: To support the pagespec functions, metadata about pages has to be
retained as pagestate.
* Fix encoding bug when pagestate values contained spaces.
I kept it to a simple global configuration, rather than using the
preprocessor directive for recentchanges, because that had chicken and egg
problems and seemed overcomplicated. This should work reasonably well,
though it would be good to add some more metadata so that more customised
recentchanges pages can be made.
This makes it a lot quicker to deal with lots of recentchanges pages
appearing and disappearing. It avoids needing to clutter up pagespecs with
exclusions for those pages, by making normal pagespecs not match them.
This is important to do because until will_render is called, ikiwiki doesn't
know that the page exists. This avoids recentchanges re-writing every change
page every run.
If a page type starts with an underscore, hide it from the list of page types
in the edit form, and don't allow editing pages of that type. This allows
for plugins to add page types for internal use.
gettext choked on a Unicode apostrophe in the aggregate plugin, which
appeared in a new error message in commit
4f872b5633. Replace it with an ASCII
apostrophe.
the code, since that process can change internal state as needed, and
it will automatically be cleaned up for the parent process, which proceeds
to render the changes.
The -c option to git log/diff-tree produces "merged" diffs with a
different format from normal ones. However, the existing diff-tree
parser only accepted non-merged diff lines. Merged diff lines caused
the parser to get out of sync. This patch adds a full parser for diffs
with any number of parents. See the "DIFF FORMAT FOR MERGES" section in
the git-diff-tree man page for more information.
Signed-off-by: Brian Downing <bdowning@lavos.net>
remove the enclosing paragraph and newline markdown wraps it in.
This allows removing several hacks around this markdown behavior from
other plugins that htmlize fragements of pages.
returned (and not run in some cases) rather than the plugins directly
forcing a user to log in.
* opendiscussion: allow editing of the toplevel discussion page,
and, indirectly, allow creating new discussion pages.
* decode_form_utf8 only fixed the utf-8 encoding for fields that were
registered at the time it was called, which was before the
formbuilder_setup hook. Fields added by the hook didn't get decoded.
But it can't be put after the hook either, since plugins using the hook
need to be able to use form values. To fix this dilemma, it's been changed
to a decode_cgi_utf8, which is called on the cgi query object, before the
form is set up, and decodes *all* cgi parameters.
the needsbuild hook. This resulted in feeds not being removed when pages
were updated, and probably other bugs.
* aggregate: Avoid uninitialised value warning when removing a feed that
has an expired guid.
links required meta to be run during scan, which complicated its data
storage, since it had to clear data stored during the scan pass to avoid
duplicating it during the normal preprocessing pass.
* If you used "meta link", you should switch to either "meta openid" (for
openid delegations), or tags (for internal, invisible links). I assume
that nobody really used "meta link" for external, non-openid links, since
the htmlscrubber ate those. (Tell me differently and I'll consider bringing
back that support.)
* meta: Improved data storage.
* meta: Drop the hackish filter hook that was used to clear
stored data before preprocessing, this hack was ugly, and broken (cf:
liw's disappearing openids).
* aggregate: Convert filter hook to a needsbuild hook.
inserting them into the html template. This ensures that markdown
acts on them, even if the value is expanded inside a block-level html
element in the html template. Closes: #454058
* Use a div in the note template rather than a span.
so that more than one plugin can use this hook.
I believe this is a safe change, since only passwordauth uses this hook.
(If some other plugin already used it, it would have broken passwordauth!)
It would be better if it were a formbuilder hook. But the formbuilder hook
is wacked.. I may need to change how that hook works, which would mean
changing the only current user of it, passwordauth).
and forces rebuilds of the pages that contain calendars. So
running ikiwiki --refresh at midnight is now enough, no need for a full
wiki rebuild each midnight.
* calendar: Work around block html parsing bug in markdown 1.0.1 by
enclosing the calendar in an extra div.
Simon McVittie
Joey Hess
Gabriel McManus
Bernd Zeimetz
Thomas Schwinge
Manoj Srivastava
martin f. krafft
Adeodato Simó
Josh Triplett
Jelmer Vernooij
Brian Downing
Patrick Winnertz
Joey Hess
Nis Martensen