Open CJ-Wright opened 6 years ago
Basic idea If https://github.com/conda-forge/conda-smithy/blob/master/conda_smithy/lint_recipe.py#L264 calculate the distance between the bad subsection and the expected subsections maybe using https://github.com/conda-forge/python-levenshtein-feedstock and if it is less than a threshold add a bit to the error asking if they meant one of the expected subsections.
For some of the lints it would be helpful if the linter suggested the valid name. Eg. for unexpected subsection name we could use a spell check algorithm to find the most likely spelling candidate (eg patch -> patches).