This is lovely work btw – very overdue to have good django wiki, and this looks very promising.

One small, low-priority issue here, and I'm not sure what the correct behaviour would actually be: When searching a site which is using markdown formatting, the search previews display the markup tags (## heading, [^footnote] etc)

Actually formatting the text might not be desirable, but I think stripping tags and displaying plaintext might be nicer.

I would like to grant a user access to edit a page without giving that user the ability to edit the page's permissions. Perhaps there should be another permission flag for editing permissions?

Add a charfield for title or short description for images and allow images to be added independently from articles.

Why?

I was thinking of something like google-images so you could search or browse categorised images containing short description and optionally if the image is related with articles it would list links to articles within wiki.

because when I upload pictures, and place them on the wiki, it shows me to save the changes.

I added django comments framework and comment form into wiki/article.html

after posting comment, i had this error:

AttributeError at /comments/post/

'function' object has no attribute 'startswith'

Request Method: POST
Request URL: http://127.0.0.1:8000/comments/post/
Django Version: 1.4
Exception Type: AttributeError
Exception Value:

'function' object has no attribute 'startswith'

Exception Location: /home/vladka/PycharmProjects/wiki/wikidharma/wiki/models/init.py in reverse, line 66
Python Executable: /home/vladka/PycharmProjects/wiki/env/bin/python
Python Version: 2.7.3
Server time: Срд, 22 Авг 2012 01:13:31 -0500

...
/home/vladka/PycharmProjects/wiki/wikidharma/wiki/models/init.py in reverse

if args[0].startswith('wiki:'): ...

Variable Value
args = ( function confirmed at 0x2a47c80 ,)
kwargs = {}

i guess you need to change init.py to something like this:

if type(args[0]) == str and args[0].startswith('wiki:'):

Right now, deleting an article marks all child articles as deleted. To restore the article all children must be undeleted, whether or not they were deleted before their parent was deleted then restored.

For example:

foo/
   bar-normal
   bar-deleted

bar-normal and bar-deleted are children of foo. bar-deleted starts out as deleted. If we delete foo, it will mark bar-normal as deleted. If we restore foo, it will mark both bar-normal and bar-deleted as undeleted! This means a deletion is not safely reversible.

Another bad side effect case is when one article has two urlpaths - pets/dog and animals/dog. If pets/ is deleted, dog will be too which makes it disappear from animals/.

I would like to propose an alternate deleting scheme. Instead of recursively deleting children, only the target article is deleted. When a urlpath is resolved, it checks that all parent articles have not been deleted. In the previous example where pets/ is deleted, accessing pets/dog would show up as deleted but accessing animals/dog would work fine.

This suggests that asking if an article is deleted should work at the urlpath level. A method is_deleted() could return a tuple (result_bool, first_deleted_ancestor), where first_deleted_ancestor is useful because it is the article a user must restore to view their target article.

Idea here:
http://dabapps.com/blog/higher-level-query-api-django-orm/

Snippet here:
http://djangosnippets.org/snippets/2117/

Related objects often accessed:

• article.select_related() <- since all articles include a call to current_revision
• urlpath.get_ancestors().select_related() <- since most views also include a breadcrumb list
• On the History page: revisions.select_related <- since all revisions point to previous_revision

More ideas!

This is a big performance boost, since it will avoid many on-the-fly SQL queries.

wiki/markitup/sets/frontend/images could not be found

Steps to reproduce:

1. Create a plugin with a wiki_plugin.py and an urlpattern in the BasePlugin.
2. Point to a view that has "from wiki import models"

It won't work because of circularity.

Theres an error on wiki test page:

TypeError at /Awesomeness/HowDoesThisWorktheSubarticleFeature/_dir/

active() takes exactly 2 arguments (1 given)

http://wiki.overtag.dk/Awesomeness/HowDoesThisWorktheSubarticleFeature/_dir/

One of my favorite features from MediaWiki is Templates. I'd love to see something similar in django-wiki. Templates are so useful for standardizing the way you create certain types of documents.

The basic usage model as I used it was to create a template page for each type of document that I want people to create. Then when people create new pages, there's a very clear set of choices for how they might lay out their content. It's also fantastic as an aid when learning the wiki syntax because it can give you a lot of examples using whatever conventions are common on the site.

In MediaWiki, selecting a template on a new or existing page will overwrite whatever is in the page currently. I don't know if that's good or bad -- once you know how it behaves, you adapt accordingly. That features isn't important to me, but the ability to lay out standard document styles and then to let people see them on page creation is.

Don't know if this code is used some where, but i think here is error:
https://github.com/benjaoming/django-wiki/blob/master/wiki/plugins/images/forms.py#L50

return super(SidebarForm, self).save(_args, *_kwargs)

should be super(RevisionForm, self).save(_args, *_kwargs), no ?

if you press the "Add article +" - it saves the current article, if you press the "Settings" - nothing happens

Right now, the changes pages allows a user to merge the current article with a deleted revision. This works fine because the deleted revision has a copy of the full text before it was deleted. It is kind of odd though. Should it be allowed?

I get an error in template located in wiki/plugins/images/render.html

Invalid blocktag empty.

I suppose that this block tag may be used only in this context:
https://docs.djangoproject.com/en/dev/ref/templates/builtins/?from=olddocs#for-empty

Second bug is in
wiki/plugins/images/render.html
wiki/plugins/images/sidebar.html

where I get invalidblocktag {% endthumbnail %} which isn't needed at all.

In views/article SearchView - line 467 there's

articles = models.Article.objects.filter(Q(current_revision__title__contains=self.query) |
Q(current_revision__content__contains=self.query))

in my opinion there should be
articles = models.Article.objects.filter(Q(current_revision__title__iexact=self.query) |
Q(current_revision__content__contains=self.query))

So articles with slugs containing capitalized first letter will be findable :)

Hi,
Just wanted to know if multilingual is planned or not? Sure I could do something like /mywiki/fr/XXX but that would be better to support a language namespace out of the box I think.

Thanks,

Guillaume

I got this error when trying to access the wiki for the first time.

TemplateSyntaxError at /wiki/_accounts/login/
You must enable the 'sekizai.context_processors.sekizai' template context processor or use 'sekizai.context.SekizaiContext' to render your templates.

This is fixed adding sekizai.context_processors.sekizai to TEMPLATE_CONTEXT_PROCESSORS:

TEMPLATE_CONTEXT_PROCESSORS = (
    (...)
    "sekizai.context_processors.sekizai",
)

Maybe this should be added to the README?

IntegrityError at /admin/wiki/article/add/
ORA-01400: cannot insert NULL into ("ISV"."WIKI_ARTICLEREVISION"."REVISION_NUMBER")

It would be really cool to have a different start page which would contain newest articles, best articles etc.

Hi,

First of all, thanks for this effort, this is just awesome to see such a good wiki being developed :)

I'm trying to embed an article into a page of my website, but I'd like the article not to be binded to an URLPath, since it's more linked to my already exiistng page. Is this use case possible at the moment ?

As a temporary fix, I've created a namespace for these pages (say tags/XXX) where I store them, but that should be optional in my case.

Thanks,

Guillaume

Are there any chances that django-wiki will get a search included out-of-the-box? Would be awesome. Love the Project so far.

It would be amazing if django-wiki had integrated support for academic citations. One of the biggest frustrations for me with mediawiki is the mess that citations get in, and the fact that there's no way of adding citations in a structured format (for example to facilitate browsing by author in separate views from the wiki).

My current project is a research wiki which aims to collate the latest findings in a particular field for both students and researchers, and I think it would be great if citations were dealt with in a similar way to attachments, but with an extension to the markdown syntax (based on pandoc's?) which allows you to create in-text citations from the citekey. What I want ultimately is to be able to add citations in wiki pages, but pull them out for a 'latest publications' listing elsewhere.

CLS is definitely seems the way to go, and citeproc.py does a lot of what would be needed in terms of processing bibtex. Django-publications may also be interesting as a starting point for a storage model.

• http://citationstyles.org
• https://github.com/brechtm/citeproc-py
• https://github.com/lucastheis/django-publications
• http://johnmacfarlane.net/pandoc/README.html

First of all, I have to say that this is a fantastic much-needed project! You guys are doing a great job and I'm looking forward to watching this project keep blooming.

Steps to reproduce:

1. Click edit article
2. Make some changes
3. Realise you want to add image
4. Click on images, Select a file, select add image
   Result:
   The image is uploaded, but it reloads the pages and all of your other changes are lost.

Expectation:
Ideally when the page reloads it will still show what you had written up to that point. Alternatively having at least a warning before submitting the form would be useful (as is already on the Todo list)

And now for something completely different:
(I hope it isn't too bad ettiquete to add other things in this same posting).
Have you considered using google's diff-match-patch library? At a cursory glance, I prefer its diffs and patches
I'd like to experiment with using TinyMCE for markup/editing, when do you think the code will be stable enough to play around with this?

At the moment it's impossible to put javascript dependencies (e.g. jquery, bootstrap) after the page content because the article doesn't display if you do. I haven't checked why this is, but page loads are much faster if the js links come after the content (css should come first though).

There is a pretty big lag when loading a new preview in a modal box. Can we blank out the box until then?

after today update i had this error.

AttributeError at /
'EmptyQuerySet' object has no attribute 'select_related_common'

Here is traceback:

/home/vladka/PycharmProjects/wiki/wikidharma/wiki/decorators.py in wrapper
if urlpath.is_deleted(): # This also checks all ancestors ...
▶ Local vars

/home/vladka/PycharmProjects/wiki/wikidharma/wiki/models/urlpath.py in is_deleted
return self.first_deleted_ancestor() is not None ...
▼ Local vars
self URLPath: (root)

/home/vladka/PycharmProjects/wiki/wikidharma/wiki/models/urlpath.py in first_deleted_ancestor
def is_deleted(self):
"""
Returns True if this article or any of its ancestors have been deleted
"""
return self.first_deleted_ancestor() is not None

def first_deleted_ancestor(self):
    for ancestor in self.cached_ancestors + [self]: ...
        if ancestor.article.current_revision.deleted == True:
            return ancestor
    return None

@transaction.commit_manually
def delete_subtree(self):

▼ Local vars
self URLPath: (root)

/home/vladka/PycharmProjects/wiki/wikidharma/wiki/models/urlpath.py in cached_ancestors
add_selecte_related on one of these ancestors will not occur an additional
sql query, as they were retrieved with a select_related.

    If the cached ancestors were not set explicitly, they will be retrieved from
    the database.
    """
    if not hasattr(self, "_cached_ancestors"):
        self._cached_ancestors = list(self.get_ancestors().select_related_common() ) ...
    #self._cached_ancestors = []
    return self._cached_ancestors

@cached_ancestors.setter
def cached_ancestors(self, ancestors):

▼ Local vars
self URLPath: (root)

What are you thinking about writing unittest ?
maybe i will make it ...

With the "groups" and "others" permissions, it is difficult to limit access to anonymous users. For example, I have a page that I want administrators to be able to edit, regular users can only view, and anonymous users cannot have any access.

I see three different approaches to fixing this problem. In increasing complexity:

1. Configure site-wide permissions for anonymous users. (The same for every article.)
2. Add another group of permissions to every article "anonymous_read", "anonymous_write".
3. Allow more than set of (group, read_policy, write_policy) to articles.

I think 3 would add a lot of complexity. 1 would suite my needs personally, but I don't know if it would be flexible enough for everyone. 2 seems like a winner.

It doesn't work with safe_mode as a preprocessor. Hopefully it's easy to turn into a blockprocessor.

I am not sure, but I believe that for a user to delete an article, the user should have write access to the parent of that article.

In the edX setup we will have the root article and several course pages.

/
  6.002x/
    FAQ/
  3.091x/
    crystals/

The root article is not editable by students. 6.002x and 3.091x are editable, but it would be nice if a student couldn't delete those pages.

Also, in general deleting an article is useful when someone accidentally creates an article and needs to get rid of it. Creating an article relies on the edit permissions of the parent. Deleting should probably do the same.

It might be tricky (or impossible) to find a parent for a given article though. Can't some articles have two parents?

Comments?

UnicodeEncodeError at /teksty-na-sajt/tehnicheskie/vneshnie-i-vnutrennie-faktory-obespecheniya-kaches/_plugin/attachments/download/1/
'ascii' codec can't encode characters in position 21-26: ordinal not in range(128), HTTP response headers must be in US-ASCII format
Request Method: GET
Request URL: http://wiki.loc/teksty-na-sajt/tehnicheskie/vneshnie-i-vnutrennie-faktory-obespecheniya-kaches/_plugin/attachments/download/1/
Django Version: 1.4.1
Exception Type: UnicodeEncodeError
Exception Value:
'ascii' codec can't encode characters in position 21-26: ordinal not in range(128), HTTP response headers must be in US-ASCII format
Exception Location: /usr/home/vladka/wiki.loc/env/lib/python2.7/site-packages/django/http/init.py in _convert_to_ascii, line 594
Python Executable: /usr/home/vladka/wiki.loc/env/bin/python
Python Version: 2.7.2
Python Path:
['/usr/home/vladka/wiki.loc/wikidharma',
'/usr/home/vladka/wiki.loc/env/lib/python2.7/site-packages/setuptools-0.6c11-py2.7.egg',
'/usr/home/vladka/wiki.loc/env/lib/python2.7/site-packages/pip-1.0.2-py2.7.egg',
'/usr/home/vladka/wiki.loc/env/lib/python27.zip',
'/usr/home/vladka/wiki.loc/env/lib/python2.7',
'/usr/home/vladka/wiki.loc/env/lib/python2.7/plat-freebsd8',
'/usr/home/vladka/wiki.loc/env/lib/python2.7/lib-tk',
'/usr/home/vladka/wiki.loc/env/lib/python2.7/lib-old',
'/usr/home/vladka/wiki.loc/env/lib/python2.7/lib-dynload',
'/usr/local/lib/python2.7',
'/usr/local/lib/python2.7/plat-freebsd8',
'/usr/local/lib/python2.7/lib-tk',
'/usr/home/vladka/wiki.loc/env/lib/python2.7/site-packages',
'/usr/home/vladka/wiki.loc/env/lib/python2.7/site-packages/PIL']
Server time: Sat, 15 Sep 2012 08:26:42 -0500

/usr/home/vladka/wiki.loc/wikidharma/wiki/core/http.py in send_file
response["Content-Disposition"] = "attachment; filename=%s" % filename ...
▶ Local vars
/usr/home/vladka/wiki.loc/env/lib/python2.7/site-packages/django/http/init.py in setitem
header, value = self._convert_to_ascii(header, value) ...
▶ Local vars
/usr/home/vladka/wiki.loc/env/lib/python2.7/site-packages/django/http/init.py in _convert_to_ascii
value = value.encode('us-ascii')

When creating root I get:

AttributeError at /wiki

'EmptyQuerySet' object has no attribute 'can_read'

Exception Value:

'EmptyQuerySet' object has no attribute 'can_read'

Exception Location: /home/easy2/domains/lekarzeweb.pl/lekarzeweb_dev/doctorsweb/../lib/wiki/models/article.py in get_children, line 101

../lib/wiki/models/article.py in get_children

                yield descendant



    def get_children(self, max_num=None, user_can_read=None, **kwargs):

        """NB! This generator is expensive, so use it with care!!"""

        cnt = 0

        for obj in self.articleforobject_set.filter(is_mptt=True):

            if user_can_read:

                objects = obj.content_object.get_children().filter(**kwargs).can_read(user_can_read)

...

            else:

                objects = obj.content_object.get_children().filter(**kwargs)

            for child in objects.order_by('articles__article__current_revision__title'):

                cnt += 1

                if max_num and cnt > max_num: return

                yield child

It would be neat if there were key shortcuts for editing/saving. Mousetrap (https://github.com/ccampbell/mousetrap) is a really nice (and light) library for capturing key events cross platform. Adding it would be as simple as:

<script src="{{ STATIC_URL }}js/mousetrap.min.js"></script>

And then in a script block in the article template:

$("#id_content").addClass('mousetrap'); // this enables capture of key events in a text area (not the default)
Mousetrap.bind(['ctrl+s'], function(e) { 
          document.getElementById('article_edit_form').target=''; document.getElementById('article_edit_form').action='/_edit/'; 
      });


  Mousetrap.bind(['ctrl+e'], function(e) {
      window.location.href = "{% url 'wiki:edit' article_id=article.id path=urlpath.path %}";
  });

At the same time, it might be worth cleaning up the javascript link generation to use {% url %} and also query, given it's already a dependency.

How would you guys feel about changing the pattern for logging in/editing pages. At present it feels weird to be on an article but not logged in, then login and be redirected to some other page, then navigate back to start editing.

A more normal flow would be to click 'edit', then be redirected to a login page, and then if successful back to the edit page. In fact, this is what happens if you manually edit the url (adding '_edit') when logged out.

So my suggestion is to always show the 'edit' button on articles (perhaps alongside view source if you think it's important) and use LOGIN_REDIRECT_URL as a fallback for when the user clicks a login link.

./manage.py syncdb
/usr/local/lib/python2.6/site-packages/South-0.7.5-py2.6.egg/south/db/oracle.py:25: UserWarning: ! WARNING: South's Oracle support is still alpha. Be wary of possible bugs.
warnings.warn("! WARNING: South's Oracle support is still alpha. "
Syncing...
Creating tables ...
Creating table wiki_article
Creating table wiki_articleforobject
Creating table wiki_articlerevision
Traceback (most recent call last):
File "./manage.py", line 10, in
...
django.db.utils.DatabaseError: ORA-00910: specified length too long for its datatype

Database Oracle 11g, NLS_NCHAR_CHARACTERSET AL16UTF16
Problem in column UserMessage: "USER_MESSAGE" NVARCHAR2(2056).
Oracle Docs: The maximum column size allowed is 4000 characters when the national character set is UTF8 and 2000 when it is AL16UTF16.

I got the following error then tried to run syncdb:

django.db.utils.DatabaseError: tablespace "notify" does not exist

I think the notifications should use the default tablespace, since the setup of a extra tablespace adds unnecesary complexity

manage.py migrate wiki.plugins.images

Error on SQL DML:

CREATE TABLE "IMAGES_IMAGE" ("REVISIONPLUGIN_PTR_ID" NUMBER(11) NOT NULL PRIMARY KEY, "IMAGE" NVARCHAR2(100) DEFAULT '' NULL , "CAPTION" NVARCHAR2(2056) DEFAULT '' NULL );

Caption NVARCHAR2 must be equal or less 2000 byte length for Oracle NLS_CHARACTERSET = AL32UTF8

I got this error when trying to access the wiki for the first time

if add article with very big slug (for example 200 symbols), and add this article - everything seems looks fine, but you can't see this article in breadcrumbs list and some other places.

Problem that if you added article with huge slug, new record in articleforobject table wasn't created for this article

When installing django-wiki on an existing application, you must:

• Add at least mptt, sekizai, and django.contrib.humanize to settings.INSTALLED_APPS
• run syncdb
• setup urls.

To setup urls, I've added:

url(r'^wiki/', include('wiki.urls', namespace='wiki')),

Maybe this should be reflected on the README

In breadcrumbs.html:

{% for ancestor in urlpath.get_ancestors.all %}

This code means that we cannot use the grandness of prefetch_related. Instead, start putting the breadcrumbs in the context. This can be done from the ArticleMixing view class.

When objects are deleted so should their files.

I'm getting this error when trying to add an attachment to a page:

Transaction managed block ended with pending COMMIT/ROLLBACK

The traceback is below. I can see you are doing something off with manual commits of transactions... I've never actually used django's transactions before though.

Environment:

Request Method: POST
Request URL: http://127.0.0.1:8000/Practice/_plugin/attachments/

Django Version: 1.4
Python Version: 2.7.2
Installed Applications:
('django.contrib.admin',
'django.contrib.auth',
'django.contrib.contenttypes',
'django.contrib.redirects',
'django.contrib.sessions',
'django.contrib.sites',
'django.contrib.sitemaps',
'django.contrib.staticfiles',
'django.contrib.markup',
'django.contrib.humanize',
'storages',
'cms',
'mptt',
'menus',
'sekizai',
'cms.plugins.picture',
'cms.plugins.link',
'cms.plugins.file',
'cms.plugins.snippet',
'cms.plugins.teaser',
'cmsplugin_markup',
'registration',
'south',
'publications',
'research',
'categories',
'categories.editor',
'compressor',
'wiki',
'wiki.plugins.attachments',
'wiki.plugins.links',
'wiki.plugins.help',
'wiki.plugins.notifications',
'wikicitations',
'django_notify',
'overextends')
Installed Middleware:
('django.contrib.sessions.middleware.SessionMiddleware',
'django.contrib.auth.middleware.AuthenticationMiddleware',
'django.contrib.redirects.middleware.RedirectFallbackMiddleware',
'django.middleware.common.CommonMiddleware',
'django.middleware.csrf.CsrfViewMiddleware',
'django.contrib.messages.middleware.MessageMiddleware',
'cms.middleware.page.CurrentPageMiddleware',
'cms.middleware.user.CurrentUserMiddleware',
'cms.middleware.toolbar.ToolbarMiddleware')

Traceback:
File "/Users/ben/django/placebo/v/lib/python2.7/site-packages/django/core/handlers/base.py" in get_response

1.                     response = callback(request, _callback_args, *_callback_kwargs)
   

   File "/Users/ben/django/placebo/v/lib/python2.7/site-packages/django/views/generic/base.py" in view

2.         return self.dispatch(request, _args, *_kwargs)
   

   File "/Users/ben/django/placebo/v/lib/python2.7/site-packages/django/utils/decorators.py" in _wrapper

3.         return bound_func(_args, *_kwargs)
   

   File "/Users/ben/django/placebo/v/lib/python2.7/site-packages/wiki/decorators.py" in wrapper

4.     return func(request, article, _args, *_kwargs)
   

   File "/Users/ben/django/placebo/v/lib/python2.7/site-packages/django/utils/decorators.py" in bound_func

5.             return func(self, _args2, *_kwargs2)
   

   File "/Users/ben/django/placebo/v/lib/python2.7/site-packages/wiki/plugins/attachments/views.py" in dispatch

6.     return super(AttachmentView, self).dispatch(request, article, _args, *_kwargs)
   

   File "/Users/ben/django/placebo/v/lib/python2.7/site-packages/wiki/views/mixins.py" in dispatch

7.     return super(ArticleMixin, self).dispatch(request, _args, *_kwargs)
   

   File "/Users/ben/django/placebo/v/lib/python2.7/site-packages/django/views/generic/base.py" in dispatch

8.     return handler(request, _args, *_kwargs)
   

   File "/Users/ben/django/placebo/v/lib/python2.7/site-packages/django/views/generic/edit.py" in post

9.         return self.form_valid(form)
   

   File "/Users/ben/django/placebo/v/lib/python2.7/site-packages/django/db/transaction.py" in inner

10.             return func(_args, *_kwargs)
    

    File "/Users/ben/django/placebo/v/lib/python2.7/site-packages/django/db/transaction.py" in exit

11.     self.exiting(exc_value, self.using)
    

    File "/Users/ben/django/placebo/v/lib/python2.7/site-packages/django/db/transaction.py" in exiting

12.     leave_transaction_management(using=using)
    

    File "/Users/ben/django/placebo/v/lib/python2.7/site-packages/django/db/transaction.py" in leave_transaction_management

13. connection.leave_transaction_management()
    

    File "/Users/ben/django/placebo/v/lib/python2.7/site-packages/django/db/backends/init.py" in leave_transaction_management

14.         raise TransactionManagementError("Transaction managed block ended with "
    

Exception Type: TransactionManagementError at /Practice/_plugin/attachments/
Exception Value: Transaction managed block ended with pending COMMIT/ROLLBACK

Perhaps with the ability to specify a starting point in the tree, and a depth, and the ability to drill down the hierarchy as well as navigate to individual pages.

I was trying to turn off signups for a wiki which is going to be public view by invitation only to edit.

I tried setting WIKI_ACCOUNT_HANDLING = False, and then including
(r'^accounts/login/$', 'django.contrib.auth.views.login'),

in my urls.py, but I still get a reverse error because of this line in base.html

{% url 'wiki:login' %}

Can I suggest a setting to allow moderation of user signups? Or using writing a backend for django-registration to allow for pluggable registration backends?

Maybe i'm wrong but this line:

https://github.com/benjaoming/django-wiki/blob/master/wiki/plugins/images/models.py#L42

will create error: Unknown column 'wiki_revisionplugin.current_revision_id' in 'field list'

In my opinion these kind of links might be way too long.

http://wiki.overtag.dk/TestTitle1/Ete/Tettet/Tetetete/Tetetete/Dfbdbdfb/Dfbdfbdfbdfb/Fbdfbdbdfbfd/Dfbdfbdfbdfb/

For such article it should look like this:

http://wiki.overtag.dk/TestTitle1/1,Dfbdfbdfbdfb/

First link is a core category, last link is last node for particular tree.