Ben Welsh

I agree this is the major weakness of the library. The issue is that the CPI database from the BLS is quite large. It is packaged here as a SQLite...

If this can be done, we can move the pandas imports of models inside the to_dataframe methods as well.

We now have a tidy page scan of T&C thanks to the Notre Dame library. https://archive.org/details/tulips-and-chimneys/page/n5/mode/2up So if anyone wants to follow @JoeGermuska's lead on the XLI Poem files, I...

I think that linking to the Hathi images is good enough, especially if they're prefer us not to download. If we get a scan of our own at a later...

Adding the hyperlink to the poem's YAML file seems right. I just encourage you to structure it as an array so it can support cases where the poems runs over...

I think I'd vote for the second one, unless you object.

I think your proposed change to the `page_scans` data makes a lot of sense. Will give us more flexibility.

If you'd like to template the source into page, one place it could go is the book metadata, which should be available in that context. I think it would only...

I'm okay with hammering out a short term fix to get it working for you. I am open to deprecating python-fs in favor of pathlib.

It's been a good while since I worked through these details, but I'm pretty sure Travis pulls it off with a mock. https://github.com/palewire/django-bakery/blob/master/.travis.yml#L13-L49