(This might be a dupes, but I could not find it.)

To translate forms, we currently go the following route:

• Mark strings as translatable (only mark, do nothing else)
• wtforms keeps them for us
• Our field rendering tools render the complete field nearly completely ba hand to be able to translate it there. Using standard translation tools there.

This works, mostly. It does not work to translate elements of a select box, for example a title like "- Select Collection -". Because we only mark it, but we're not using our translation tools in the rendering. We could go for rendering that part by hand also.

BUT: wtforms recommends something else:

• Use lazy_pass_to_ugettext() to mark and create a autotranslating object. This feels like a string, but really becomes a translated string, when it is forced to a real string.
• wtforms keeps this thing as a normal object around.
• When rendering it, wtforms will force it to a string as late as possible, so that translations can take effect.

So the first idea is to mark strings using lazy_pass_to_ugettext() and just disable the translation in our rendering tools (just remove the _()!) and see, how it works out.

Old in any forms.py:

from mediagoblin.tools.translate import fake_ugettext_passthrough as _

New:

from mediagoblin.tools.translate import lazy_pass_to_ugettext as _

I found a special case which the binary stl files will be parsed as a ascii stl file (which is normal) and fail in an unexpected way, as well as a case where ascii stl files (and plausibly obj files) will fail when there is indentation present in the file.

Will push a fix in a few minutes.

Supporting PDF files would let mediagoblin have many more use cases, such as a collection of datasheets, research papers, and books.

A working pdf media type is at:

​https://www.gitorious.org/~alon/mediagoblin/alons-mediagoblin/commits/pdf

The test is not in there because I wasn't sure how it should be added - I have it written but the problems are:

1. it needs to run only if pdftocairo is available

So I can add some code to check for the existence of that executable and only then enable the media type and test.