markdown_extract only expects strings

[metaodi]

When using ckanext-fluent to define a multilingual markdown field, the output currently fails with the following error:

Error - <type 'exceptions.AttributeError'>: 'dict' object has no attribute 'strip'

I could trace this error back to the following line of code:

if (text is None) or (text.strip() == ''):

PR follows to make sure the site doesn't break because of this.

[rossjones]

I haven't checked that overriding helpers is possible, but perhaps it might be better to override markdown_extract in ckanext-fluent instead?

[metaodi]

@rossjones well monkey patching could work yes. I'll check if this is possible. But maybe it would be good to have a "proper" way of doing this.

[rossjones]

I don't think you need to monkey-patch, you could use ITemplateHelpers - https://github.com/ckan/ckan/blob/master/ckan/plugins/interfaces.py#L888-L910 but it appears that it does stop you overwriting helpers at

ckan/ckan/config/environment.py

Line 70 in 9bf61b4

raise Exception('overwritting extra helper %s' % helper)

:(

[wardi]

You can use an IPackageController before_show method to replace the any core fluent fields with a single-language version. I'm planning to add one of these to ckanext-fluent when I can make it work well enough to handle fluent with core dataset fields like title and notes.

@metaodi I've tried to pitch the idea of core ckan adopting the ckanext-fluent behaviour for dicts in place of text but haven't been able to sell it to the others yet :-) Maybe move this issue to ckanext-fluent for now.

[metaodi]

@wardi Okay I'll look into that to replace the dict with a single language version. For now, I'll try to implement that in my own extension, if I can make it work, I'll try to integrate that in fluent as well.

[wardi]

closing in favour of #2678