- Zope - Zope lists

Tree Tag Options
by Martijn Pieters 17 Feb '99

17 Feb '99

At Tue, 02 Feb 1999 13:01:32 -0800 Amos Latteier wrote a nice explanation of the branches attribute of the tree tag. But when I played around with the source of the tree tag (TreeTag.py), I discovered more toys. This is has become an unofficial documentation of the #tree tag. Excuse the typos, for it has become very late. Have fun. Attributes of #tree assume_children=[0|1] default 0 If set to a nonzero value, all items are assumed to have subitems, and will therefore allways have a + sign in front of them when not yet expanded. Only when a user expands an item it is checked for subitems. This could be a good option when the retrieval of children is a slow process. branches="methodname" default="tpValues" If an object has defined this method, it should return a list of ids that are it's subobjects. The default, tpValues, is defined on the Items class (from which just about every objetc inherits), and Folders and Database Adapters define their own versions. If you use objectValues as method, you get the same effect on Folders, but not on Database Adapters. branches_expr="expression" branches_expr works like the expr attribute of tags like #if. You could therefore do:  to get a tree of only Folder and Document subobjects. This is easier than defining your own External Method for branches="", like Amos described. id="method_or_propertyname" default="tpId" Method or property name to determine the ID of the object for tree-state purposes. The Ids saved in the tree state cookie (and the tree-state local var) are realy just integers so the state var is not too big, but unusable for your average Zope developer. If you want to do anything with the state, you might want to define this as id="id", for a more readable form. Just do  and see for yourself. No default Zope object currently defines a tpId method or property. When no tpId method or property is defined, an internally generated integer is used (pickle id?). leaves="property" If set, this property is assumed to be a DTML Method within objects that have no sub-objects. It is displayed as part of the tree, no additional code is needed in your DTML. The leave DTML method is handed a special standard_html_header and standard_html_footer. nowrap=[0|1] default 0 If set to a nonzero value, the tree table will be rendered with the NOWRAP attribute, which means the browser won't wrap the item labels when spacially challenged. url="method_or_propertyname" default="tpURL" Method or property to be called on object to determine the relative URL of the object. The default, tpURL, is defined by the Item class, and the Gladfly Database Adapter defines it's own version, so it can show you it's tables and columns in the management screen. If you want to override the URL translation, define an External Method, like Amos described, and use that method with url="methodname". single=[0|1] default 0 If set to a nonzero value, only one branch on a level can be expanded. Any other expanded branches will collapse. Seriously cool. skip_unauthorized=[0|1] default 0 If set to a nonzero value, no ValidationErrors will be raised trying to read objects to which the user has no access. Variables in namespace set by tree tag (for every object being rendered): tree-root-url Relative URL of the DTML Document or Method this tag is in. tree-colspan The number of levels deep the tree is being rendered. tree-state The current state of the tree, a list of ids and sublists. tree-item-url The URL of the current object being rendered, relative to the document the #tree tag is in. This is generated using the url="methodname" method. tree-level Level of this object in the tree, starting at 0. tree-item-expanded Flag indicating wether this item is expanded or not. Variables in namespace that influence the tree tag: expand-all All branches expanded if this name is defined and nonzero (like in a cookie, or in the REQUEST object. Try http://URL/to/object/with/tree/tag?expand-all=1) collapse-all All branches collapsed if present and nonzero. tree-s This is the cookie and query name that contains the state of the tree. This is a zlib compressed, base64 encoded nested list of ID's. Have fun making one yourself. I think you better start with the TreeTag.py code. tree-e List of ID's to expand. See tree-s on the format. tree-c List of ID's to collapse. Again, see tree-s. Are we going to collect this kind of data somewhere? Like in a tag reference? Is there a protocol to define your own DTML tags?

1 0

RE: [Zope] ZOPE ODBC on Linux
by Paul Everitt 16 Feb '99

16 Feb '99

> > Randall Marbach wrote: > > > > > > Is it possible for a Zope 1.9 installation running on a > linux platform (redhat > > > 5.2) to access a MS ACCESS database located on a > different NT server via the > > > ODBC support built into Zope? > > > > As Access is not a client-server database, you are unable to use it > > even from another Windows (NT or 9x) machine. > > Uh, you could do something complicated involving e.g. RemoteCall from > Linux to a Win32 box, and a PythonCOM program on Win32 box. Hate to talk about something we aren't publicly releasing, but...we've been doing this for about two years now. We had a Principia Product that called a Network Client that allowed a local object to be a proxy for a remote "searchable object" (read: SQL Method). The two processes communicated via HTTP. The initial request was marshalled, rotor'd, then base64'd and packed in the body of the HTTP request to the second server which ran the query and packaged up the results in the same way. Tracebacks on the server were re-raised on the client process. We later added caching. Thus for a long time the digicool.com site, running on Linux, authenticated users against data stored in Lotus Notes on NT. The response time was pretty spiffy. This project came from the original customer project that created our relational database integration back at the end of 1996. We federated queries for the Department of Defense across a Sybase and Oracle databases running on different machines in different time zones. We continue to use it in our consulting gigs (in fact, a customer that is onsite here today is getting it). I imagine some enterprising soul could craft up a cool solution using XML-RPC or the like... --Paul Paul Everitt Digital Creations paul(a)digicool.com 540.371.6909

1 0

a question on external methods
by David Wagle 16 Feb '99

16 Feb '99

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Can someone tell me what I'm doing wrong? I have the VERY simple external methods test.xm and test2.xm defined. They are: from DateTime import DateTime def dwTime(self): t=DateTime() return DateTime.timeTime(t) def dwName(self,REQUEST=None): return "%s" % REQUEST['AUTHENTICATED_USER'] They both work fine from the Zope/manage screen using the 'try-it' tab. When I use them in a test file:     The first () works just fine. but the addition of the second, causes the following run-time error: Traceback (innermost last): File /usr/local/Zope-1.9.0-src/lib/python/ZPublisher/Publish.py, line 879, in publish_module File /usr/local/Zope-1.9.0-src/lib/python/ZPublisher/Publish.py, line 595, in publish (Info: /test_area/test1) File /usr/local/Zope-1.9.0-src/lib/python/OFS/Document.py, line 181, in __call__ (Object: test1) File /usr/local/Zope-1.9.0-src/lib/python/OFS/Document.py, line 177, in __call__ (Object: test1) File /usr/local/Zope-1.9.0-src/lib/python/DocumentTemplate/DT_String.py, line 514, in __call__ (Object: test1) File /usr/local/Zope-1.9.0-src/lib/python/Products/ExternalMethod/ExternalMethod.py, line 266, in __call__ (Object: Item) (Info: ((), {}, (None,))) TypeError: (see above) - --------------------------------------------- A computer without Microsoft products is like a dog without bricks chained to its head - --------------------------------------------- -----BEGIN PGP SIGNATURE----- Version: PGP for Personal Privacy 5.0 Comment: Processed by Mailcrypt 3.5.1, an Emacs/PGP interface Charset: noconv iQA/AwUBNsnKG9tGhckXZmDjEQJJpwCfeziCLyeaVw0/h3FfcAE2JJ3uQrQAoNbJ E+U2HC4yTS/aZQmJb0XO6YLk =ApHz -----END PGP SIGNATURE-----

3 4

ZOPE ODBC on Linux
by Randall Marbach 16 Feb '99

16 Feb '99

Is it possible for a Zope 1.9 installation running on a linux platform (redhat 5.2) to access a MS ACCESS database located on a different NT server via the ODBC support built into Zope? If not, is there some other way to do this? TIA Randy ____________________________________________________________________ More than just email--Get your FREE Netscape WebMail account today at http://home.netscape.com/netcenter/mail

5 5

"proxy roles", search engine for documentation.
by Anthony Baxter 16 Feb '99

16 Feb '99

Hi, Can someone give a 1-para description of proxy roles? The ZMG mentions them, but not what they are, or do. My guess is that it's like a Unix setuid program - anyone viewing the document gets the changed permissions, for the purpose of loading other objects into that object... On a related note - could a search engine be added to the Zope Documentation area of the site? Please? :) Anthony

1 0

RE: [Zope] a question on external methods
by Brian Lloyd 16 Feb '99

16 Feb '99

> I can't readily explain why this is the case, but it looks like either > a) REQUEST is not being passed in, or > b) REQUEST is None > > The following method: > def dwName(self,REQUEST=None): > > if REQUEST is not None: > return "%s" % REQUEST['AUTHENTICATED_USER'] > else: > return "" > > will run. It will produce the desired result if called directly > (by going to the URL of the method). It does not seem to produce > anything when called from a DTML Method. > > And that's as much as I know :) > > Kevin If you call it from the web, it will work because ZPublisher will automatically pass the REQUEST argument. If you call it from DTML, you have to pass REQUEST yourself:  Brian Lloyd brian(a)digicool.com Software Engineer 540.371.6909 Digital Creations http://www.digicool.com

1 0

RE: [Zope] Extend copy/move protocol to object instantiation?
by Brian Lloyd 16 Feb '99

16 Feb '99

> The copy/move protocol in 'CopySupport.py' is quite nice in that it > allows the object which is being moved/copied to participate in the > operation, including rejecting the operation or cleaning up at the > conclusion of the operation. I found a need for something > similar when > objects are first installed in a folder(ish) location: 'scope out > the neighborhood' and 'make yourself at home' hooks. > > ----------------------------------------------------- > > class anyClass: > > ... > > def addsubs(self): > '''add sub category field names''' > resobj=self.aq_parent.catlist1() > res =[] > for field in resobj._searchable_result_columns(): > res.append(field) > self.fields=res > > def makeyourselfathome(self): > """parent specific installation hook""" > self.addsubs() > > def addAnyClassObject(self,id,title,REQUEST=None): > """Create an Object and install it in its parent Folder > The argument 'self' will be bound to the parent Folder. > """ > ob=anyClass(id) > self._setObject(id, ob) > ob.__of__(self).makeyourselfathome() > if REQUEST is not None: > return self.manage_main(self,REQUEST) > > ------------------------------------------------------------- > > Would it make sense to add another 'op' type to > CopySource._NotifyOfCopyTo() and CopySource._postCopy() to indicate a > 'first home' operation? > > MAM I'd much rather see an eventual Zope event model handle things like this. In the absense of such a beast, we were forced to come up with a workable solution for copy and paste, but its probably not appropriate to try to make that general- purpose. I think it would be much better if you could eventually just define "event handler" methods that interact with a well-known event model to deal with such things, a la: def onObjectAdd(self, event): # Somehow use information in the event to # do some fixup on the added object, etc. ... def onCopyBegin(self, event): ... def onCopyFinish(self, event): ... Brian Lloyd brian(a)digicool.com Software Engineer 540.371.6909 Digital Creations http://www.digicool.com

1 0

Extend copy/move protocol to object instantiation?
by Mark Millikan 16 Feb '99

16 Feb '99

The copy/move protocol in 'CopySupport.py' is quite nice in that it allows the object which is being moved/copied to participate in the operation, including rejecting the operation or cleaning up at the conclusion of the operation. I found a need for something similar when objects are first installed in a folder(ish) location: 'scope out the neighborhood' and 'make yourself at home' hooks. ----------------------------------------------------- class anyClass: ... def addsubs(self): '''add sub category field names''' resobj=self.aq_parent.catlist1() res =[] for field in resobj._searchable_result_columns(): res.append(field) self.fields=res def makeyourselfathome(self): """parent specific installation hook""" self.addsubs() def addAnyClassObject(self,id,title,REQUEST=None): """Create an Object and install it in its parent Folder The argument 'self' will be bound to the parent Folder. """ ob=anyClass(id) self._setObject(id, ob) ob.__of__(self).makeyourselfathome() if REQUEST is not None: return self.manage_main(self,REQUEST) ------------------------------------------------------------- Would it make sense to add another 'op' type to CopySource._NotifyOfCopyTo() and CopySource._postCopy() to indicate a 'first home' operation? MAM

1 0

Re: [Zope] Prevent image-reloading...
by Michael Bernstein 16 Feb '99

16 Feb '99

Andy Smith wrote: > <untested> > >  > <A HREF="pages/> <IMG SRC="images/"></A> >  > > The urls that are generated are identical in each case. > > What's funny is that today I've come across the exact *opposite* problem. > I've got a gif that is being generated dynamically that should be being > updated as the result of use input, but the cached version keep popping up. > > If you know what makes one gif reload and another come back from the cache > then I'm all ears! Ok, It looks as though you are generating relative URL's for those images and links. it is usually much safer to generate fully qualified URL's. Lets say that you have this directory structure: / /images /a /b /c it is entirely possible to go to the folowing URL: www.yourdomain.com/a/b/c/b/c/a/b/a/c/b/images/someimage.gif and actually retrieve the image. Directories are subject to acquisition just like Methods are, so beware of referring to sibling directories as sub-directories when creating relative URL's. The browser (correctly) sees most of these images as having distinct URL's, and so does not retrieve them from the cache. I hope this solves your problem...I'm not sure what to do about your non-reloading image though. Michael.

2 1

New patch MailHost.py part deux
by Martijn Pieters 16 Feb '99

16 Feb '99

I updated the patch for MailHost.py (see my earlier mail), to fix to more problems: - Single dots on a line in the message body are now escaped (so they won't prematurely end the transmission). - The body is now quoted-printable. This _should_ solve any problems your SMTP host might have with the MailHost output. And you can find the patches on: http://www.xs4all.nl/~zmj/Zope/MailHost.1.9.0.patch or http://www.xs4all.nl/~zmj/Zope/MailHost.1.10.0pr1.patch Depending on your Zope version. -- M.J. Pieters, Web Developer | ATMM http://www.atmm.nl | Tel: +31-35-6254545 Fax: +31-35-6254555 | mailto:mj@atmm.nl http://www.atmm.nl/~mj | PGP: http://wwwkeys.nl.pgp.net:11371/pks/lookup?op=get&search=0xA8A32149 ------------------------------------------

1 0