Unbundle all dependencies.

[kennethreitz]

It's time.

[iurisilvio]

I agree! But why you closed the issue?

[kennethreitz]

@MorleyB might take a crack at this!

[hermanocabral]

please!

[kennethreitz]

@MorleyB need any help getting started?

[jayvdb]

De-vendoring dbfpy is going to be a bit more effort than the others.

The dbfpy package on PyPi is 2.3.1 , and the bundled version here is from an older release, 2.2.5, and it is slightly patched, and the tests here fail when using the latest PyPI package - the struct's are different.

The errors can be encountered with jayvdb@95cfef3, and viewed at https://travis-ci.org/jayvdb/tablib/jobs/505156956

The adaptions are not visible in git, but there is a bit of history at https://github.com/kennethreitz/tablib/pull/153 , and a bit more at https://github.com/phargogh/dbfpy3/commits/master

@phargogh might be able to help.

The diff which makes the tablib tests pass is

--- dbfpy-2.3.1/dbfpy/fields.py.orig    2019-03-12 16:16:12.434154549 +0700
+++ dbfpy-2.3.1/dbfpy/fields.py 2019-03-12 16:33:19.785761986 +0700
@@ -150,7 +150,7 @@
         return (
             _name +
             self.typeCode +
-            struct.pack("<L", self.start) +
+            chr(0) * 4 +
             chr(self.length) +
             chr(self.decimalCount) +
             chr(0) * 14

But I suspect that correct answer will be a more complicated mixture of the two codebases, as both have had bug fixes, and dbf comes in several flavours.

A few days ago I found another version of the codebase on GitHub, and asked if they are interested in releasing a new package at antmoth/dbfpy#1 , and @estelendur is interested.

[jayvdb]

Another approach is to switch to using a different library, such as pypi packages dbfread or dbf.