Regular Expressions in Import Rules

Regular expressions used in import rules support the following standard syntax:

.	any non-newline character	(a|z)	a or z
^	start of line	[az]	a or z
$	end of line	[^az]	not a or z
\b	word boundary	[a-z]	a through z
\B	non-word boundary	(foo)	capture foo
\A	start of subject (usually the same as ^)	a?	0 or 1 as
\z	end of subject (usually the same as $)	a*	0 or more as
\d	decimal digit	a+	1 or more as
\D	non-decimal digit	a{3}	exactly 3 as
\s	whitespace	a{3,}	3 or more as
\S	non-whitespace	a{3,5}	between 3 and 5 as (inclusive)
\w	word character
\W	non-word character

All regular expressions are case-sensitive and unicode-aware, e.g. \s will match unicode whitespace characters as well as ASCII ones.

Limitations

Certain features of regular expressions aren't supported when they're used in Import Rules. These are, specifically:

• Lookarounds (i.e. lookahead and lookbehind), both negative and positive.

  • Positive lookaround can usually be matched directly instead. E.g. foo(?=bar) could just be matched as foobar.

  • Negative lookaround can usually be matched as a normal regex, but it can be tricky.

    • E.g. pre_(?!no)/ can be matched as pre_([^/]?|[^n/][^/]|[^/][^o/]|[^/]{3,})/.

      • Because complex regexes like this are hard to maintain, we recommend just positive-matching the specific known items instead, e.g. pre_(yes|yeah|sure).

• Backreferences.

  • Due to the nature of backreferences (i.e. that they are non-regular), it isn't generally possible to replicate the same match without them.

    • When possible, we recommend just enumerating all the items in this case instead, e.g. instead of trying to match all foldersfoob(a+)rb\0z/, you can just enumerate the folders you know exist, like foo(barbaz|baarbaaz)/.

These features are unsupported due to allowing for construction of extremely slow (exponential-time) regexes that are hard for Scanner to detect.