String matching: Difference between revisions

=={{header|Bracmat}}==

Bracmat does pattern matching in expressions <code><i>subject</i>:<i>pattern</i></code> and in strings <code>@(<i>subject</i>:<i>pattern</i>)</code>. The (sub)pattern <code>?</code> is a wild card.

<lang>( (sentence="I want a number such that that number will be even.")

& out$(@(!sentence:I ?) & "sentence starts with 'I'" | "sentence does not start with 'I'")

& out$(@(!sentence:? such ?) & "sentence contains 'such'" | "sentence does not contain 'such'")

& out$(@(!sentence:? "even.") & "sentence ends with 'even.'" | "sentence does not end with 'even.'")

& 0:?N

& ( @(!sentence:? be (? & !N+1:?N & ~))

| out$str$("sentence contains " !N " occurrences of 'be'")

)

)</lang>

In the last line, Bracmat is forced by the always failing node <code>~</code> to backtrack until all occurrences of 'be' are found. Thereafter the pattern match expression fails. The interesting part is the side effect: while backtracking, the accumulator <code>N</code> keeps track of how many are found.

Output:

<lang>sentence starts with 'I'

sentence contains 'such'

sentence ends with 'even.'

sentence contains 3 occurrences of 'be'</lang>