> 2. Nevod matches MULTIPLE patterns against document in ONE PASS. Patters/expressions are indexed by state machine and filtered effectively during matching.
Regexps do that too. Specifically, there is a whole class of "lexers"/"tokenizers" which are regexp-based, and which specialize in matching 1000's of patterns against the text, in one pass.
(You can kinda approximate it in regular regexp by doing multiple rules at once, each in it's own group, and seeing which groups are non-empty, like in "(rule1)|(rule2)|(rule3)" and so on. If your regexp library is good, it will match this in time independent from number of rules. But real lexers will have nicer syntax)
Regexps do that too. Specifically, there is a whole class of "lexers"/"tokenizers" which are regexp-based, and which specialize in matching 1000's of patterns against the text, in one pass.
(You can kinda approximate it in regular regexp by doing multiple rules at once, each in it's own group, and seeing which groups are non-empty, like in "(rule1)|(rule2)|(rule3)" and so on. If your regexp library is good, it will match this in time independent from number of rules. But real lexers will have nicer syntax)