>>> URL(r=request, c='hello', f='world', args=['hi'], vars={'q':'greetings'},
>>> anchor='the_world')
'/welcome/hello/world/hi#the_world?q=greetings'
>>> URL(c='hello', f='world', args=['hi'], vars={'q':'greetings'},
>>> anchor='the_world')
'/welcome/hello/world/hi#the_world?q=greetings'
Attached is the diff
This should not effect any current usages of URL, but from this point
forward we shouldn't have to specify r=request!
--
Thadeus
On Wed, Jul 28, 2010 at 5:21 PM, mdipierro <[email protected]> wrote:
> ok
>
> On Jul 28, 4:52 pm, Thadeus Burgess <[email protected]> wrote:
>> For a little more advanced version of what Massimo just posted:
>>
>> This allows you to use both.
>>
>> URL(r=request, c=<controller>, f=.....)
>> and
>> URL(c=<controller>, f=<function>.....)
>>
>> http://packages.python.org/web2py_utils/init.html#gurlhttp://hg.thadeusb.com/Web/web2py_utils/file/697470f78d16/web2py_util...
>>
>> The reason we have to pass the request to URL each time is because it
>> is just imported into the context. Since it is just a python function,
>> it has no knowledge of request.
>>
>> Why can't we do something like gURL from within web2py? When building
>> the context, why can't we make URL a loaded function that will get
>> request passed to it, exactly how gURL works.
>>
>> Effectively, we won't have to pass request into URL function anymore... ever.
>>
>> If your interested, I can work on the patch.
>>
>> --
>> Thadeus
>>
>> On Wed, Jul 28, 2010 at 3:56 PM, mdipierro <[email protected]> wrote:
>> > 1) It is a bit more complex because of static, appadmin, and admin.
>> > You can map
>>
>> > /function/var1/var2
>>
>> > into
>>
>> > /app/controller/function/var1/var2
>>
>> > using a file routes.py
>> > ------- begin file
>> > routes_in=[
>> > ('/admin/$anything','/admin/$anything'),
>> > ('/static/$anything','/app/static/$anything'),
>> > ('/appadmin/$anything','/app/appadmin/$anything'),
>> > ('/$anything','/app/default/$anything'),
>> > ]
>> > routes_out=[(y,x) for (x,y) in routes_in]
>> > ----- end file
>>
>> > 2) in db.py define
>>
>> > def url(f,args={},vars={}):
>> > return URL(r=request,f=f,args=args,vars=vars)
>> > def go(f,args={},vars={},flash=''):
>> > session.flash=flash
>> > return go(url(f=f,args=args,vars=vars))
>>
>> > and use
>>
>> > url('index')
>>
>> > or
>>
>> > go('index',flash='you are being redirected')
>>
>> > I do it all the time
>>
>> > On Jul 28, 3:29 pm, VP <[email protected]> wrote:
>> >> I'm a newbie, so my questions probably have easy answers, but anyway.
>> >> Among a few things, there are two I don't understand and think can be
>> >> simplified.
>>
>> >> 1. this url: app/controller/function/var1/var2
>>
>> >> For me at least, most of the time I probably have only one
>> >> controller. If there is one controller, may be we should get rid of
>> >> "controller" in the url? This will be sufficient: app/f/a/b/c
>>
>> >> 2. Similarly, most of the time I have only one app (current app).
>> >> But I use URL() a lot and every time I have to pass in request like
>> >> this URL(r=request, f='foo').
>>
>> >> Why do I have to pass in request if I have only one app, one
>> >> controller?
>>
>> >> Furthermore, isn't request supposed to be global? If so, do we have
>> >> to pass it in?
>
diff -r a3a934f1f4f5 gluon/compileapp.py
--- a/gluon/compileapp.py Wed Jul 28 16:33:43 2010 -0500
+++ b/gluon/compileapp.py Wed Jul 28 17:51:22 2010 -0500
@@ -216,6 +216,11 @@
environment = {}
for key in html.__all__:
environment[key] = getattr(html, key)
+
+ # Overwrite the URL function with a proxy
+ # url function which contains this request.
+ environment['URL'] = html._gURL(request)
+
for key in validators.__all__:
environment[key] = getattr(validators, key)
if not request.env:
diff -r a3a934f1f4f5 gluon/html.py
--- a/gluon/html.py Wed Jul 28 16:33:43 2010 -0500
+++ b/gluon/html.py Wed Jul 28 17:51:22 2010 -0500
@@ -200,6 +200,25 @@
if regex_crlf.search(url):
raise SyntaxError, 'CRLF Injection Detected'
return rewrite.filter_out(url, env)
+
+def _gURL(request):
+ """
+ A proxy function for URL which contains knowledge
+ of a given request object.
+
+ Usage is exactly like URL except you do not have
+ to specify r=request!
+ """
+ def _URL(*args, **kwargs):
+ # If they use URL as just passing along
+ # args, we don't want to overwrite it and
+ # cause issues.
+ if not kwargs.has_key('r') and len(args) < 4:
+ kwargs['r'] = request
+
+ return URL(*args, **kwargs)
+ _URL.__doc__ = URL.__doc__
+ return _URL
ON = True
