cetem-publico/script.js

A wrapper for CETEMPúblico, an European Portuguese corpus of news extracts from the newspaper Público, with 180 million words tagged automatically using PALAVRAS.

const CETEMPublico = require('./lib/cetem-publico');
const file = '/home/andrefs/Desktop/trash_buffer/CETEMPublico/CETEMPublicoAnotado2019.gz';
//const cp = new CETEMPublico(file, {gz: true});
const cp = new CETEMPublico();

async function printLines(){
  for await (const line of cp.lines()){
    console.log(line);
  }
}

async function printTokens(){
  for await (const token of cp.tokens()){
    console.log(token.toString());
  }
}

// async function printStuff(){
//   for await (const thing of cp.process('ext')){
//     console.log(thing.toString());
//   }
// }

async function printStuff(){
  let i = 0;
  for await (const thing of cp.sentences()){
    i++;
     console.log(thing);
    console.log(i);
  }
}


//printLines();
//printTokens();
printStuff();