Linter to check style guide compliance on all polysquare projects
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.

Polysquare Generic File Linter


Travis CI (Ubuntu) AppVeyor (Windows) Coverage PyPI Licence
Travis AppVeyor Coveralls PyPIVersionPyPIPythons License

Checks each file passed in for compliance with polysquare style guidelines.

  • headerblock/filename: Checks that the first line of the file has a line which matches /path/to/file from the source root
  • headerblock/desc_space: Checks that the second line of the headerblock is as empty comment
  • headerblock/space_copyright: Checks that the second last line of the headerblock is an empty comment
  • headerblock/copyright: Checks that the last line of the headerblock contains an appropriate short-form copyright notice
  • file/newline_last_char: Checks that the last line is just a \n
  • file/spelling_error: Checks that docstrings and comments do not contain spelling errors or technical-like terms that do not appear in the rest of the source file
  • file/trailing_whitespace: Checks that no line contains trailing whitespace

Main Linter Usage

usage: polysquare-generic-file-linter [-h] [--checks]
                                      [--whitelist [LIST [LIST ...]]]
                                      [--blacklist [LIST [LIST ...]]]
                                      [--spellcheck-cache SPELLCHECK_CACHE]
                                      [--log-technical-terms-to LOG]
                                      [--stamp-file-path STAMP_FILE_PATH]
                                      [--block-regexps [BLOCK [BLOCK ...]]]
                                      [FILE [FILE ...]]

Lint for Polysquare style guide

positional arguments:
  FILE                  read FILE

optional arguments:
  -h, --help            show this help message and exit
  --checks              list available checks
  --whitelist [WHITELIST [WHITELIST ...]]
                        list of checks that should only be run
  --blacklist [BLACKLIST [BLACKLIST ...]]
                        list of checks that should never be run
  --fix-what-you-can    fix errors automatically
  --spellcheck-cache SPELLCHECK_CACHE
                        path to spell-checking cache file
  --log-technical-terms-to LOG_TECHNICAL_TERMS_TO
                        path to file to log technical terms to
  --stamp-file-path STAMP_FILE_PATH
                        path to directory to store cached results
  --block-regexps [BLOCK_REGEXES [BLOCK_REGEXES ...]]
                        Regular expressions to exclude from all checks.


Of some interest to others may be the spell-checking functionality. The file/spelling_error check will scan any inline documentation (docstrings and comments) in your code for spelling errors and misused technical terms. If you want a string, (because, for example, it contains user-facing text) to be considered, just make it a python-style docstring by using three quotes.

Ordinary words

The spell-checker will check any ordinary word, those being words with roman alphabetical characters and an apostrophe (') against a list of words in the American English dictionary as generated by SCOWL at level 50 with abbreviations and hacker-terms on. You can also specify your own domain specific words by providing a file called DICTIONARY in the project root directory. Words are checked against both lists on a case-insensitive basis.

Technical words

Certain words, once separated by the check, will be treated as "technical" words as opposed to ordinary words. They will be checked against the list of valid symbols detected from the surrounding source code. For instance, the following code will trigger an error, because the term _CustomTerm wasn't defined in the source file (CustomTerm was, however):

class CustomTerm:

    """_CustomTerm is a certain type of class."""

Ignoring certain expressions

Sometimes it does not make sense to run spellcheck or check certain expressions against the list of detected technical words. This is often the case where comments might contain inline markup or metadata which looks and behaves like code. The check can be told to ignore anything matching a user-specified regex in order to handle this case. Just pass the regex to --block-regexps.

Removal of punctuation

The check will do its best to remove surrounding punctuation around words such that only those words are checked against the word lists. However, punctuation must follow standard English grammar rules in order for words around them to be considered as ordinary words instead of technical words. For instance, a space must appear before an opening parenthesis. But a nested opening parenthesis can appear directly after another opening The golden rule is that if it looks like something which could be code, the surrounding words will be treated as code and not as ordinary English words.

Speeding up execution

Whoosh, the spellchecking engine behind file/spelling_error needs to generate some data structures in order to quickly find corrections for words. Generating these data structures with the long word list that is shipped with this tool by default can take a few seconds. Obviously, this would be undesirable if this tool is to be used multiple times or as part of a script. You can pass --spellcheck-cache and a path to a directory to store cache files to cache the result of these data structures between invocations.

Stand-alone spellchecking

If you just want to run spellcheck on the code comments and inline documentation, then you can use the --whitelist option to only run that check. Just pass it with --whitelist file/spelling_error.

If you want to run spellcheck on an entire file, a special tool called spellcheck-linter is provided which also serves that purpose. It will check all ordinary looking words against the user-provided DICTIONARY and the built-in American English dictionary. If --technical-terms and a path to a filename containing technical terms is provided, it will also check that technical looking terms exist in this file.

Spellcheck Linter Usage

usage: spellcheck-linter [-h] [--spellcheck-cache SPELLCHECK_CACHE]
                         [--technical-terms TECHNICAL_TERMS]
                         [--technical-terms-dependencies [[DEPENDENCY ...]]]
                         [--stamp-file-path STAMP_FILE_PATH]
                         [FILE [FILE ...]]

Find spelling errors

positional arguments:
  FILE                  read FILE

optional arguments:
  -h, --help            show this help message and exit
  --spellcheck-cache SPELLCHECK_CACHE
                        path to spell-checking cache file
  --technical-terms TECHNICAL_TERMS
                        path to file to source technical terms from
  --stamp-file-path STAMP_FILE_PATH
                        path to directory to store cached results

Technical terms

A technical terms file is just a list of symbols in a text file. As a matter of convenience, this can be automatically generated for you by passing --log-technical-terms-to to polysquare-generic-file-linter when checking the inline documentation of those files. The second argument after this switch should be the path to a filename where technical terms are to be stored. On each invocation, the union of the current file contents and the technical terms detected will be written back to the file.

Embedding the checking API

The exported API in polysquarelinter.spelling isn't by any means stable right now, but it can be embedded into an application with some ease.

The Dictionary class encapsulates the Whoosh spellchecking API, word-lookup and caching functionality. The dictionary_sources keyword argument indicates a list of files from which the words in passed to the Dictionary were sourced from. If any of these files has a newer timestamp than the dictionary cache, then the dictionary cache will be regenerated.

The spellcheck_region function takes a list of lines and runs spellcheck and a check for invalid technical words on each word in those lines. It will handle punctuation and other syntactical markets appropriately in either case. Both the valid_words_dictionary and technical_words_dictionary can be either None or an instance of Dictionary. The user_words argument is simply a set of words that the user has indicated are always valid.

The spellcheckable_and_shadow_contents splits a file into spellcheckable chunks (made out of _ChunkInfo) and "shadow contents", which make up the rest of the file. The shadow contents are usually just the code around the inline documentation. The data member of _ChunkInfo is a list of lines, effectively representing the region which should be spellchecked. line_offset and col_offset indicate the line and column offset into the main contents. If you are reporting errors, any error in SpellcheckError as returned by spellcheck_region will be returned relative to the contents passed to it and not to the whole file. Use the offsets in _ChunkInfo to turn these into absolute offsets into the file being checked itself.


Internally, polysquare-generic-file-linter and spellcheck-linter cache their results using the jobstamps library. If you want to redirect where the cache files are written, you can pass --stamp-file-path to either tool.