conmorfeusz/index.js

const { spawnSync } = require("child_process");

async function analyse(input) {    
  const { stdout } = spawnSync(MORFEUSZ_ANALYSER_BIN, [], { input });

  const linePattern = /^\s?(?<opening>\[)?(?<start>\d+),(?<end>\d+),(?<segment>,|(.+?)),(?<lemma>,|(.+?)),(?<tags>[\w:.]+),(?<frequency>_|(.+?)),(?<qualifiers>_|(.+?))(?<ending>\])?$/;
  
  const output = [];
  let segment = undefined;  
  
  for (const line of stdout.toString().split(/\r?\n|\r/)) {
    if (line.trim().length === 0) {
      continue;
    }
    
    const match = line.match(linePattern);
    
    if (!match) {
      throw new Error(`Following line does not match the pattern: ${line}`);
    }
    
    const { opening, ending, ...data } = match.groups;
    
    if (opening) {
      inSegment = true;
      segment = [];
    }
    
    segment.push({ ...data, tags: parseTags(data.tags), rtags: data.tags });
    
    if (ending) {
      inSegment = false;
      output.push(segment);
      segment = undefined;      
    }        
  }

  return output;
}

async function runConcraftLocally(input) {  
  const { stdout, stderr } = spawnSync(CONCRAFT_BIN, ["tag", CONCRAFT_MODEL], { input });

  const error = stderr.toString();

  if (error?.trim()?.length > 0) {
    throw new Error(error);
  }

  return stdout.toString()
}

async function invokeConcraftRemotely(dag) {
  const response = await fetch(`${CONCRAFT_SERVER_URL}/parse`, {
    method: 'POST',
    headers: {
      'Content-Type': 'application/json'
    },
    body: JSON.stringify({
      dag
    })
  });

  const data = await response.json();  
  return data.dag;
}

async function desambiguate(analysis) {
  const input = analysis
    .flatMap(a => a)
    .map(a => ({ ...a, tags: Object.values(a.tags).filter(t => t).map(tag => tag.split(".")[0]).join(":") }))
    .map(a => `${[...Object.keys(a).filter(k => !['rtags'].includes(k)).map(k => a[k]), "0.0", "_", "_", "_"].join("\t")}`)           
    .join("\n");


  const response = await (CONCRAFT_MODE === 'local' ? runConcraftLocally : invokeConcraftRemotely)(input);

  const keys = ["start", "end", "segment", "lemma", "tags", "frequency", "qualifiers", "prob", "interp_meta", "eos", "seg_meta", "disamb"];
  const output = [];
  for (const line of response.split(/\r?\n|\r/)) {
    if (line.trim().length === 0) {
      continue;
    }

    const properties = line.split(/\t/).map((prop, index) => [keys[index], prop === '_' ? '' : prop]);
    const element = Object.fromEntries(properties);

    element.start = Number.parseInt(element.start);
    element.end = Number.parseInt(element.end);
    element.disamb = element.disamb === 'disamb';    
    element.eos = element.eos === 'eos';
    element.prob = Number.parseFloat(element.prob);    
    element.rtags = element.tags;
    element.tags = parseTags(element.rtags);
    
    output.push(element);
  }

  return output;
}

async function danalyse(input) {
  const analysis = await analyse(input);
  return await desambiguate(analysis);
}

function generate(input, ...tags) {  
  const { stdout } = spawnSync(MORFEUSZ_GENERATOR_BIN, [], { input });
  
  const linePattern = /^\s?(?<opening>\[)?(?<segment>,|(.+?)),(?<lemma>,|(.+?)),(?<tags>[\w:.]+),(?<frequency>_|(.+?)),(?<qualifiers>_|(.+?))(?<ending>\])?$/;
  
  const output = [];
        
  let segment = undefined;
  let inSegment = false;
  
  for (const line of stdout.toString().split(/\r?\n|\r/)) {
    if (line.trim().length === 0) {
      continue;
    }
    
    const match = line.match(linePattern);
    
    if (!match) {
      throw new Error(`Following line does not match the pattern: ${line}`);
    }
    
    const { opening, ending, ...data } = match.groups;
    
    if (opening) {
      inSegment = true;
      segment = [];
    }
    
    segment.push(data);
    
    if (ending) {
      inSegment = false;
      output.push(segment);
      segment = undefined;      
    }        
  }
  
  return output.flatMap(a => a).filter(entry => tags.every(tag => entry.tags.includes(tag)));
}

function parseTags(tags) {
  const [type, ...rest] = tags.split(":");

  const parsers = {
    adv: (degree) => ({ degree }),
    imps: (aspect) => ({ aspect }),
    inf: (aspect) => ({ aspect }),
    pant: (aspect) => ({ aspect }),
    pcon: (aspect) => ({ aspect }),
    qub: (vocalicity) => ({ vocalicity }),
    prep: (c, vocalicity) => ({ case: c, vocalicity }),
    siebie: (c) => ({ case: c }),
    subst: (number, c, gender,) => ({ number, case: c, gender }),
    depr: (number, c, gender,) => ({ number, case: c, gender }),
    ger: (number, c, gender, aspect, negation) => ({ number, case: c, gender, aspect, negation }),
    ppron12: (number, c, gender, person, accentability) => ({ number, case: c, gender, person, accentability }),
    ppron3: (number, c, gender, person, accentability, postprepositionality) => ({ number, case: c, gender, person, accentability, postprepositionality }),
    num: (number, c, gender, accommodability) => ({ number, case: c, gender, accommodability }),
    numcol: (number, c, gender, accommodability) => ({ number, case: c, gender, accommodability }),
    adj: (number, c, gender, degree) => ({ number, case: c, gender, degree }),
    pact: (number, c, gender, aspect, negation) => ({ number, case: c, gender, aspect, negation }),
    ppas: (number, c, gender, aspect, negation) => ({ number, case: c, gender, aspect, negation }),
    winien: (number, gender, aspect,) => ({ number, gender, aspect }),
    praet: (number, gender, aspect, agglutination) => ({ number, gender, aspect, agglutination }),
    bedzie: (number, person, aspect,) => ({ number, person, aspect }),
    fin: (number, person, aspect,) => ({ number, person, aspect }),
    impt: (number, person, aspect,) => ({ number, person, aspect }),
    aglt: (number, person, aspect, vocalicity) => ({ number, person, aspect, vocalicity }),
  };

  return { 
    type,
    ... (parsers[type]?.(...rest) ?? {})
  }
}

async function main([ interpreter, script, action, ...args ]) {
  const fn = {
    analyse,
    danalyse,
    generate
  }[action];

  if (!fn) {
    throw new Error(`Unknown action: '${action}'`);
  }

  const output = await fn(...args);

  console.log(JSON.stringify(output, undefined, 2));
}


main(process.argv);