compromise
Version:
modest natural language processing
17 lines (16 loc) • 421 B
JavaScript
import maybeMatch from './maybeMatch.js'
// tokenize first, then only tag sentences required
const lazyParse = function (input, reg) {
let net = reg
if (typeof reg === 'string') {
net = this.buildNet([{ match: reg }])
}
let doc = this.tokenize(input)
let m = maybeMatch(doc, net)
if (m.found) {
m.compute(['index', 'tagger'])
return m.match(reg)
}
return doc.none()
}
export default lazyParse