Fix ValueError thrown on specific byte sequence

[adamsqi]

There is this issue that appears in the recent version of the package:

ValueError: b'\r\n' is not in list

The bug has been fixed in the recent version of https://github.com/delimitry/aztec_code_generator
More about the issue and the fix: delimitry#7

[dlenski]

This PR doesn't apply to the master branch of this repository at all. Did you try it?

[adamsqi]

@dlenski
It doesn't apply, my bad. I forked the repo and can see lots of conflicts.

[dlenski]

As far as I can tell, the only part of the changes here that are actually relevant to the stated purpose are the following lines:

diff --git a/aztec_code_generator.py b/aztec_code_generator.py
index 65bbb28..844b8b5 100644
--- a/aztec_code_generator.py
+++ b/aztec_code_generator.py
@@ -351,7 +353,8 @@ def find_optimal_sequence(data):
                         last_mode = abbr_modes.get(char.replace('/S', '').replace('/L', ''))
                         break
                 if last_mode == 'punct':
-                    if cur_seq[x][-1] + c in punct_2_chars:
+                    # do not use mixed mode for '\r\n' as in mixed mode '\r' and '\n' are separate
+                    if cur_seq[x][-1] + c in punct_2_chars and x != 'mixed':
                         if cur_len[x] < next_len[x]:
                             next_len[x] = cur_len[x]
                             next_seq[x] = cur_seq[x][:-1] + [cur_seq[x][-1] + c]

Do you agree with that?

[adamsqi]

@dlenski
Yes it looks like this is the significant part except of formatting.

[dlenski]

I'm not convinced that the solution from #4 is a very clarifying or complete one. For now, I've just added a (failing) test to demonstrate the problem in a94f699.

[adamsqi]

@dlenski
I think this PR can be closed, since the fix needs to be added in a new PR.