Allow 4-character abbreviations of mnemonic words ...

[yorickdowne]

... when using the existing-mnemonic workflow.

This addresses #167

Choices made:

Because of the way the dict lookups are written, I decided to normalize to 'NFKC' and slice to 4 characters or less. The normalization is there so that combining Unicode characters work and the string doesn't get sliced to 3 visible characters.

I have no particular attachment to this way of handling it. If you'd rather not normalize and do a partial match on dict keys instead, I am game. If there's a third method you favor, ditto.

new-mnemonic workflow has not been touched and still does a straight compare. One might discuss whether that workflow should allow abbreviations as well, but that might be a different PR.

This code passes the test suite.

[lukas131313]

R

[yorickdowne]

Please give this another look for inclusion.

[yorickdowne]

This is now failing one specific test mnemonic. Checksum calculation fails. Let me see why that is.

test_mnemonic = 'abaco abaco abaco abaco abaco abaco abaco abaco abaco abaco abaco abaco abaco abaco abaco abaco abaco agitare'

[yorickdowne]

I know why this fails, the logic to create a word language map is wrong. I'm glad this got caught before it made it in! Let me think on how to improve

[yorickdowne]

This now passes all tests. Logic has been changed. Previously I normalized and sliced to 4 characters while constructing word_language_map. The way Python works, this turned out to be a very bad idea: There are multiple overlaps in abbreviations, and the map ended up with only one word to language mapping, not several. For example, "abac" matches words in Italian, Spanish, and Portuguese, but the way this was written previously, word_language_map would only have the Portuguese version in it.

Instead, I am now using a nested for list comprehension that compares the full word language map against abbreviations, then flatten the resulting 2D list again. If there's a more elegant way of doing this, please share!

Because abbreviations can match several languages, and then merely fail the checksum, I am now continuing the search through the languages when a checksum fails, rather than returning None on the first failure.

[yorickdowne]

Please give this another look for inclusion.

Also, please give feedback on whether changing the two pass statements to continue would be welcome.

[yorickdowne]

Please give this another look for inclusion. If there are concerns, please raise them.

[yorickdowne]

Ping to get this considered for inclusion, please.