Split off from bug 867514 .
Even a conservative "dictionary of possible misspellings" approach may create too much noise when run on source code; particularly when highly abbrivated names are used.
Example case:
https://fuchsia-review.googlesource.com/c/tools/+/175949/2/symbolize/pipeline.go#109
Implementing this may require going through source files of different types and deciding what's "in a comment" and what's "in a string". This will probably involve tokenizing different types of source files in order to see where multi-line comments (or strings?) start in different languages such as Python, Go, etc.
(Side note: If we take the approach of tokenizing, and iterating through all tokens keeping track of whether we're in a comment etc., then at that point we might as well just load the codespell dictionary and use the words there directly without invoking the codespell Pythons script).
Comment 1 by diegomtzg@google.com
, Jul 25