select() result can't be used as a string

[anonymous]

Hi,

I have encountered a problem using py:match that doesn't look right to me.

In my templates, I use customs elements like <tag ident="xhtml">XHTML</tag> and try to turn them into links with the following line:

<py:match path="tag">
  <a href="${url_for('tag', ident=select('@ident'))}">${select('text()')}</a>
</py:match>

Although the ident attribute's value is inserted with

title="${select('@ident')}"

and

title="${(select('@ident'))}"

it is not with

href="${url_for('tag', ident=select('@ident'))}"

which uses an URL generation function from ​Routes. Since the round braces don't seem to be the problem, is it the apostrophs or the equal sign of the keyword argument? I think the markup extract shown above definitely makes sense and did already work with Kid (with some syntax changes, of course).

What is the cause of this behaviour? I use the latest Markup version from SVN as of today.

Thanks, Jochen Kupperschmidt

[anonymous]

Oops! After some more tries, an exception informed me that I was dealing with a Stream instance. After a look into that class, I came up with this, doing what I expected:

<py:match path="tag"><a href="${url_for('tag', ident=str(select('@ident')))}">${select('text()')}</a></py:match>

So easy.

But would it be useful to have Markup directly convert it into a string for use in expressions like this?

[cmlenz]

While I agree that having to wrap select() in str() is a bit of a kludge, I don't currently see a good way around that.

select() always returns Stream objects, and if you pass it to a function in an expression, who says that function doesn't actually expect a stream?

I'm changing the summary to reflect the issue, but I currently don't see a solution.

[anonymous]

As long as the docs reflect that select() returns a stream (which they currently don't, as far as I can tell), that would be sufficient, I think.

[cmlenz]

(improved description formatting)

[jochen.kupperschmidt@…]

Since the initial point of this ticket is not a defect and a way to work with it is documented here, I'm closing this one.