@geographika I thought it is implemented by:

        for excluded_path in config.options.excluded_paths:
            if fnmatch.fnmatch(hit.path, excluded_path):
return False

in https://github.com/piwik/piwik-log-analytics/blob/master/import_logs.py#L1990-L1992 or was i maybe mistaken?

Thanks for the quick reply! Apologies for the first incorrect pull request (I left a print statement in).

That only seems to be checking the --excluded_paths option, whereas the docs seem to imply you can do the following to import all .log files in a folder (which didn't work for me prior to the changes in this pull request):

python import_logs.py *.log

The files to process are gathered by:

self.options, self.filenames = option_parser.parse_args(sys.argv[1:])

And then looped through later on with:

        for filename in config.filenames:
            parser.parse(filename)

The check_path function (that uses excluded_path) seems to be called when parsing the lines in a file.

I see - it was probably working already here because linux bash will expand the *log and write the actual file names.

Would it be possible to use the fnmatch instead of glob for consistency with the other wildcards?

That would explain it. Yes I'm running this on a Windows server.

I'd prefer glob as it has the advantage of being able to get files in subdirectories, whereas this would all need to be coded with fnmatch and os.lisdir and some recursion (glob uses fnmatch internally).

The check for a * wildcard is a bit hacky though. Could any log filenames on linux contain a *? This isn't possible on Windows, but is technically possible on Unix - https://en.wikipedia.org/wiki/Filename#Reserved_characters_and_words