Extract substrings with defined length from a string

Question

I have a string like this one

str = "this and that but also thit and that";

Now I want to extract, the first 3 letters before and after " and " so that the outcome is

{his and tha, hit and tha}

I tried it with

StringCases[str, 
 x__ ~~ " and " ~~ y__ :> {StringTake[x, -3], StringTake[y, 3]}]

but this extracts only the second substring {{"hit", "tha"}}. And using StringCases[str, _ ~~ " and " ~~ _] extracts only one letter {"s and t", "t and t"}. So is there a way to define a Blank with a particular length?

Answer 1

StringCases[str,   x__ ~~ " and " ~~ y__ /; StringLength[x] == StringLength[y] == 3]

or

StringCases[str, _ ~~ _ ~~ _ ~~ " and " ~~ _ ~~ _ ~~ _]

would work:

{"his and tha", "hit and tha"}

Answer 2

Using Repeated:

StringCases[
 str,
 Repeated[_, {3}] ~~ " and " ~~ Repeated[_, {3}],
 Overlaps -> True
 ]

{"his and tha", "hit and tha"}

Note the use of Overlaps which is necessary for strings such as

str = "this and that and thit and that";

With Overlaps -> True:

{"his and tha", "hat and thi", "hit and tha"}

with Overlaps -> False:

{"his and tha", "hit and tha"}

Answer 3

If the number of whitespaces before and after 'and' is unknown (but in each case there is at least one), a simple regex should do:

str = "this and that but also thit and that";
StringCases[str, RegularExpression[".{3}\s+and\s+.{3}"]]

{his and tha, hit and tha}

Alternatively, a positive lookahead may be used:

StringCases[str, RegularExpression["(?=(.{3}\s+and\s+.{3}))\\1"]]
StringCases[str, RegularExpression["(?=(.{3}\s+and\s+.{3}))"]:> "$1"]

{his and tha, hit and tha}

{his and tha, hit and tha}

---

For strings such as "this and that and thit and that" considered by C.E above:

str2 = "this and that and thit and that";
StringCases[str2, RegularExpression[".{3}\s+and\s+.{3}"], Overlaps:> True]
StringCases[str2, RegularExpression["(?=(.{3}\s+and\s+.{3}))\\1"],Overlaps->True]
StringCases[str2, RegularExpression["(?=(.{3}\s+and\s+.{3}))"]:> "$1"]

{his and tha, hat and thi, hit and tha}

{his and tha, hat and thi, hit and tha}

{his and tha, hat and thi, hit and tha}

In addition posix character classes are supported:

StringCases[str2, RegularExpression["[[:alnum:]]{3}[[:blank:]][Aa]nd[[:blank:]][[:alnum:]]{3}"], Overlaps:> True]

{his and tha, hat and thi, hit and tha}

---

For positive lookaheads, compare (see here):

s = "123456789123456789";
StringCases[s,RegularExpression["(?=(\d{10}))\\1"]]
StringCases[s,RegularExpression["(?=(\d{10}))\\1"],Overlaps:>True]
StringCases[s,RegularExpression["(?=(\d{10}))"]:> "$1"]

{1234567891}

{1234567891, 2345678912, 3456789123, 4567891234, 5678912345, 6789123456, 7891234567, 8912345678, 9123456789}

{1234567891, 2345678912, 3456789123, 4567891234, 5678912345, 6789123456, 7891234567, 8912345678, 9123456789}