diff2typo

[RainRat]

I'm experimenting with a tool that can turn diffs into typos data files.
https://github.com/RainRat/diff2typo

1. Find a change that fixed lots of typos.
2. Create a diff for that change, ie. git diff [parameters] > diff.txt
3. You can continue with different changes, even different repositories, all into one file ie. git diff [parameters] >> diff.txt
4. Run the tool. ie. python diff2typo.py --input_file diff.txt --output_file typos.txt --typos_tool_path typos/typos
5. It will do all the work: extract the corrections, sort, deduplicate, and format it ready for the "Non-critical corrections" thread.

[RainRat]

I did a major update:
-splitting compound words
-use dictionary file (to exclude candidates which are more of grammar fixes)
-output format (arrow, csv, table, list)
-improved docs everywhere (code comments, user messages, readme)

[NathanBaulch]

FWIW, I usually use the following command to get a rough summary but something more refined sounds good.

git diff HEAD^! --word-diff-regex='\w+' -U0 \
  | grep -E '\[\-.*\-\]\{\+.*\+\}' \
  | sed -r 's/.*\[\-(.*)\-\]\{\+(.*)\+\}.*/\1 -> \2/' \
  | sort | uniq -c | sort -n