Parameter parse_until doesn't compare to tag name.

[klein4]

django.template.Parser.parse() takes an argument parse_until, which the docs say is a tuple of tag names which indicate that parsing should halt. However, when Parser.parse() finds a new token of token_type=TOKEN_BLOCK, it tests token.contents in parse_until rather than token.contents.split_token()[0] in parse_until. The latter more closely matches the documented functionality.

This distinction is relevant when tokens after the initial tag contain arguments.

To see why this is useful, consider a custom block tag random which randomly chooses a nodelist according to its weight. It might be invoked like this:

{% random "2" %}
This node has a 2/7 chance of being chosen.
{% or "2" %}
This node has a 2/7 chance of being chosen.
{% or "3" %}
This node has a 3/7 chance of being chosen.
{% endrandom %}

As it currently stands, it is impossible to give non-initial tokens arguments as in this example. This may be considered a non-trivial patch, but it seems to align with the documented functionality of the parse_until parameter.

[klein4]

This is my first Django patch. Created with diff -crB from my library root. Apologies if the format is inappropriate.

[klein4]

Woops. I mistyped the line token.contents.split_token()[0] in parse_until in the description above. That should read token.split_contents()[0] in parse_until. I typed it correctly in my patch, however.

[Russell Keith-Magee]

I'm pretty sure this is a duplicate of #3100.

Regarding format - preferred format is the output of "svn diff".