Table of contents

General information

Online demo

Parser of the tiny subset of German grammar

Parser of the tiny subset of English grammar

Parser of arithmetic expressions

vargrammar=newtinynlp.Grammar(['R -> N','S -> S add_sub M | M','M -> M mul_div T | T','N -> S lt_gt S | S','T -> num | ( S )',]);grammar.terminalSymbols=function(token){if('<'===token||'>'===token)return['lt_gt'];if('+'===token||'-'===token)return['add_sub'];if('*'===token||'/'===token)return['mul_div'];if('('===token)return['('];if(')'===token)return[')'];// Otherwise - token considered as a number:return['num'];}

Usage

Grammar with hardcoded terminal symbols

// Define grammarvargrammar=newtinynlp.Grammar([// Define grammar production rules'R -> S','S -> S add_sub M | M | num','M -> M mul_div T | T | num','T -> num',// Define terminal symbols'num -> 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 0','add_sub -> + | -','mul_div -> * | /']);// You have to tokenize input by yourself!// Creating array of tokensvartokens='2 + 3 * 4'.split(' ');// ParsingvarrootRule='R';varchart=tinynlp.parse(tokens,grammar,rootRule);// Get array with all parsed trees// In case of ambiguous grammar - there might be more than 1 parsing treevartrees=chart.getFinishedRoot(rootRule).traverse();// Iterate over all parsed trees and display them on HTML pagefor(variintrees){console.log(JSON.stringify(trees[i]))}

Customizing logic of tokens classification into terminal symbols

vargrammar=newtinynlp.Grammar(['R -> S','S -> S add_sub M | M | num','M -> M mul_div T | T | num','T -> num',]);// Define function, which will classify tokens into terminal typesgrammar.terminalSymbols=function(token){// Actually, this method might be customized// to use some more sophisticated classification mechanismsif('+'===token||'-'===token)return['add_sub'];if('*'===token||'/'===token)return['mul_div'];if(token.match(/^\d+$/))return['num'];// It is also possible that classifier returns// more than one terminal symbol, e.g.: ['term1', 'term2']// Otherwise:thrownewError("Can't recognize token: "+token);}// You have to tokenize input by yourself!// Creating array of tokensvartokens='2 + 3 * 4'.split(' ');// ParsingvarrootRule='R';varchart=tinynlp.parse(tokens,grammar,rootRule);// Get array with all parsed trees// In case of ambiguous grammar - there might be more than 1 parsing treevartrees=chart.getFinishedRoot(rootRule).traverse();// Iterate over all parsed trees and display them on HTML pagefor(variintrees){console.log(JSON.stringify(trees[i]))}

Traversing parsed trees

functiontoNestedList(tree){if(!tree.subtrees||tree.subtrees.length==0){return'<li>'+tree.root+'</li>';}varbuilder=[];builder.push('<li>');builder.push(tree.root);builder.push('<ul>')for(variintree.subtrees){builder.push(toNestedList(tree.subtrees[i]))}builder.push('</ul>')builder.push('</li>')returnbuilder.join('');}

// Iterate over all parsed trees and display them on HTML pagefor(variintrees){htmlRepresentstion='<ul>'+toNestedList(trees[i])+'</ul>'// embed htmlRepresentstion into HTML page}

Parsing tiny subset of English language grammar

vargrammar=newtinynlp.Grammar(['S -> NP VP','VP -> VP PP | V NP | V','PP -> P NP','NP -> Det N | N']);grammar.terminalSymbols=function(token){if(token=='eats')return['V'];if(token=='fish')return['N'];if(token=='fork')return['N'];if(token=='she')return['N'];if(token=='a')return['Det'];if(token=='with')return['P'];// otherwise:return[];}// Tokenizing sentencevartokens='she eats a fish with a fork'.split(' ');// ParsingvarrootRule='S';varchart=tinynlp.parse(tokens,grammar,rootRule);// Get array with all parsed trees// In case of ambiguous grammar - there might be more than 1 parsing treevartrees=chart.getFinishedRoot(rootRule).traverse();for(variintrees){// visit each parse tree}

Used Tools and Libraries

This library is used by...