cycle template tag should accept a single argument

[Gary Wilson]

When passed a single argument, the cycle tag should treat it as an iterable to cycle through. For example, with the context:

{'colors': ['red', 'blue', 'green']}

you should be able to have a template like:

{% for row in mydata %}
<tr class="{% cycle colors %}">...</tr>
{% endfor %}

that would output:

<tr class="red">...</tr>
<tr class="blue">...</tr>
<tr class="green">...</tr>
<tr class="red">...</tr>
...

[Simon G <dev@…>]

This sounds like a sensible improvement

[Malcolm Tredinnick]

I'm a little concerned that this is ambiguous when you're reading it: you can't easily tell if it's referring to a previously named "cycle" tag or creating a new one using a variable as an iterator.

Is there a better way to write this that won't cause confusion when reading templates?

[munhitsu]

So far I've managed to perform the request but maintaining compatibility with older approach makes my code not so... clean.

What do you think about a new keyword i.e.: cycletable?

Maybe its a time to refer to named cycle tag by variable reference? To be honest referring by name is not to clear for me.

[Gary Wilson]

Replying to mtredinnick:

I'm a little concerned that this is ambiguous when you're reading it: you can't easily tell if it's referring to a previously named "cycle" tag or creating a new one using a variable as an iterator.

Is it really that different from using the "as varname" syntax of other tags or the {% with %} tag though? The way I see it, when you encounter any variable name in the templates, you have to check where it came from. Was it passed in the context? Was it defined as the target of "as varname" syntax of another variable? Was it defined with a with tag? Is it a loop variable? etc. This would be no different.

[munhitsu]

Guys take a look at following unit test suite. Does it reflect your expectations?

{
'cycle20': ("{% cycle colors %}", {'colors': 'red'}, template.TemplateSyntaxError),
'cycle21': ("{% cycle colors %}{% cycle colors %}", {'colors': ['red', 'blue', 'green']}, 'redblue'),
'cycle22': ("{% cycle colors %}{% cycle colors %}{% cycle colors %}", {'colors': ['red', 'blue', 'green']}, 'redbluegreen'),
'cycle23': ("{% cycle colors %}{% cycle colors %}{% cycle colors %}{% cycle colors %}", {'colors': ['red', 'blue', 'green']}, 'redbluegreenred'),
'cycle24': ("{% cycle colors as color %}{% cycle color %}", {'colors': ['red', 'blue', 'green']}, 'redblue'),
'cycle25': ("{% cycle colors as color %}{% cycle color %}{% cycle color %}", {'colors': ['red', 'blue', 'green']}, 'redbluegreen'),
'cycle26': ("{% cycle colors as color %}{% cycle color %}{% cycle color %}{% cycle color %}", {'colors': ['red', 'blue', 'green']}, 'redbluegreenred'),
'cycle27': ("{% for i in test %}{% cycle colors %}{{ i }},{% endfor %}", {'test': range(5),'colors': ['a', 'b']}, 'a0,b1,a2,b3,a4,'),
}

[munhitsu]

Seems to be working. Old regression is passing. New has been added.
I've added a small enhancement - a cycle pushed to context. Thanks to it cycle internals can be used inside template.

        context['cycle'] = {
            # cycle internals
            'collection': self.collection,
            'name':self.variable_name,
            'value':value,
        }

One note. Last part of syntax parsing has to be done in render method. We need context to say if {% cycle colors %} refers to table or is a mistake.

[Gary Wilson]

Your cycle20 test case should not raise a TemplateSyntaxError, but rather should output "r" since a string is iterable. A TemplateSyntaxError should only be raised if the syntax for the tag in not correct, i.e. {% cycle colors as %}. If the object passed is not iterable, the template tag should fail silently and return an empty string. Also, your patch should allow any iterable instead of just lists. I also don't see a reason that we should be exposing the internal state of the tag. Could you explain your reasoning for wanting to do that?

[munhitsu]

v2

[munhitsu]

Replying to gwilson:

Thanks for the comments.

1. Functionality for test cycle20 updated. Do you have an idea how to prepare context with non iterable variable? I have no clue how to make unit test for it.

   'cycle20': ("{% cycle colors %}", {'colors': 'r'}, 'r')
   

2. Regarding silent fail... Actual implementation will throw TemplateSyntaxError for test cycle01 as original one. But i.e if 'a' would be defined in context than it will fall back to Variable('a').resolve(context).

   'cycle01': ('{% cycle a %}', {}, template.TemplateSyntaxError)
   

3. I've just added NameError exception catch as a backup to make sure we will fail silently
4. Cycle patch now iterates over everything that is iterable - see test cycle22. It's hashed as Python does not guarantee iteration sequence on hash.

   'cycle22': ("{% cycle colors %}{% cycle colors %}", {'colors': {'r': 1, 'g': 2}}, 'rg')
   

5. Regarding cycle internal state in tag. I must say that I was inspired by ForNode and cycle is a somehow close to it. Take a look at tests cycle28..29. That the only reasonable case that I see now to use cycle internals. It's your choice.

   'cycle28': ("{% for i in test %}{% cycle colors %}{{ cycle.value }}{{ i }},{% endfor %}", {'test': range(5),'colors': ['r', 'g', 'b']}, 'rr0,gg1,bb2,rr3,gg4,'),
   'cycle29': ("{% for i in test %}{% cycle colors %}{% cycle colors %}{{ i }},{% endfor %}", {'test': range(5),'colors': ['r', 'g', 'b']}, 'rg0,br1,gb2,rg3,br4,'),
   

[Julien Phalip]

A design decision hasn't been reached yet.

[Alex Gaynor]

Discussion with Carl: this new syntax would be ambiguous with the current syntax, however this is a good feature, and the current API is kind of silly. This is accepted, provisional on creating a new, more sane and consistent syntax, to be placed in the future module.

[Alexander Lazarević]

Is this still considered a desirable feature?

[Mariusz Felisiak]

Replying to Alexander Lazarević:

Is this still considered a desirable feature?

Yes, IMO nothing has changed.

[Alexander Lazarević]

I created a PR ​https://github.com/django/django/pull/17716

[Alexander Lazarević]

For the new PR I'm going to remove "Needs documentation", "Needs tests" and "Patch needs improvement". Please set these flags again if required (or I should have left them set in the first place).

[Alexander Lazarević]

I just put something together at ​https://sandbox.e11bits.com/ to showcase the feature more broadly. Maybe people can judge by that, if the feature is desirable and help with the code review? It's just a quick hack, so don't expect too much. The examples are more or less the ones found as testcases in the PR. (crossposted to discord)

[Natalia Bidart]

Replying to Alexander Lazarević:

I just put something together at ​https://sandbox.e11bits.com/ to showcase the feature more broadly. Maybe people can judge by that, if the feature is desirable and help with the code review? It's just a quick hack, so don't expect too much. The examples are more or less the ones found as testcases in the PR. (crossposted to discord)

Thank you Alexander for your work on this. In order to unify the conversation in a single place, I will reply to the forum thread with my view, hopefully we can agree on an implementation design there and then come back to the ticket.

As a reference, ​this is the forum post.

[Alexander Lazarević]

After the discussion in ​https://forum.djangoproject.com/t/cycle-template-tag-should-accept-a-single-argument/27010/3 I'm withdrawing the PR.

My implementation would have introduced an ambiguity that would have made it harder for users to use the cycle tag.

In the discussion @carlton mentioned that it could be worth "... to investigate [is] whether we could add * and *'* list and dictionary expansion to the DTL. cycle could benefit from list expansion.".

So I'm trying to come up with something in that direction instead.

[Mariusz Felisiak]

There is nothing to review, at least for now.