On Mar 29, 2011, at 11:04 AM, Jonathan Lundell wrote:
> On Mar 29, 2011, at 10:34 AM, Anthony wrote:
>> When web2py writes A('my link', _href=URL(r=request, vars=request.vars)) to
>> HTML, it results in:
>>
>> <a href="/page?input_one=aaa®_no=bbb">my link</a>
>>
>> The problem is, the browser converts the "®" in the href to the
>> registered trademark symbol, so when you click the link, the requested URL
>> sent to web2py ends up including the trademark symbol in place of the "®"
>> portion of the URL (presumably, this would happen with any HTML entity, such
>> as ©, etc.).
>>
>> To avoid this kind of problem, I think href attributes are supposed to
>> include "&" in place of "&", so the link should have been written as:
>>
>> <a href="/page?input_one=aaa&reg_no=bbb">my link</a>
>>
>> So, why doesn't web2py write the URL properly? When a helper (such as the
>> A() helper) is converted to HTML, web2py passes its attributes to the
>> xmlescape() function, which calls cgi.escape(), which automatically converts
>> "&" to "&". However, the xmlescape() function does not escape objects
>> that have an 'xml' attribute, and because the URL() function returns an
>> object with an 'xml' attribute, it is not escaped.
>
> That's a good explanation. ISTR that not too long ago URL was changed to
> return XML(whatever), which is why the xml attribute is there. I don't
> remember why the change was made, though. Maybe Massimo does.
The change happened last August, around the time that the new template system
went in. The problem, according to Massimo at the time, was that without the
XML() call, something (I don't know what) was being double escaped.
I'm thinking that maybe URL should call cgi.escape on its own string before
returning it with XML().
>
>>
>> This should probably be fixed somehow, but for now, a workaround is to
>> explicitly call the 'xml' method of URL, which will convert it to a string
>> before it is passed to xmlescape(). So, change your code to:
>>
>> link = A('my link', _href=URL(r=request, vars=request.vars).xml())
>>
>> Note, this issue is not a problem if you manually enter an href that
>> includes an HTML entity. For example:
>>
>> link = A('my link', _href='/page?input_one=aaa®_no=bbb')
>>
>> will properly be converted to:
>>
>> <a href="/page?input_one=aaa&reg_no=bbb">my link</a>
>>
>> Best,
>> Anthony
>>
>>
>> On Tuesday, March 29, 2011 9:53:50 AM UTC-4, Jim Karsten wrote:
>> I have a variable named 'reg_no' that when used as a link paramater is
>> getting converted into a registered trademark symbol. Here is a simple
>> example that illustrates it.
>>
>> The controller:
>>
>> def page():
>> link = A('my link', _href=URL(r=request, vars=request.vars))
>> form = SQLFORM.factory(
>> Field('input_one', default=request.vars.input_one),
>> Field('reg_no', default=request.vars.reg_no),
>> )
>> return dict(form=form, link=link)
>>
>> The view:
>>
>> {{=form}}
>> {{=link}}
>>
>>
>> Enter values in both fields, then submit. Then click the link. The
>> link should refresh the page, with the form populated with the entered
>> values. If the second field is named 'input_two' it works. But when
>> it's named 'reg_no' the first input has 'aaaR_no=bbb' where 'aaa' and
>> 'bbb' are the entered values and R is the registered trademark symbol,
>> and the second input is blank.
>>
>> The link url is very different depending on the field name:
>>
>> 'input_two': http://www...com/demo/default/page?input_one=aaa&input_two=bbb
>> 'reg_no' : http://www...com/demo/default/page?input_one=aaa%C2%AE_no%3Dbbb
>>
>> I renamed 'reg_no' to something not starting with 'reg_' and I have it
>> working. I'm not sure if this is a bug or not.
>>
>> Any insight would be welcome.
>>
>> Jim Karsten
>
>
