Add conmorfeusz as synergy of morfeusz and concraft-pl

index.js

@@ -0,0 +1,157 @@
+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;
+  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: ${line}`);
+    }
+    
+    const { opening, ending, ...data } = match.groups;
+    
+    if (opening) {
+      inSegment = true;
+      segment = [];
+    }
+    
+    segment.push(data);
+    
+    if (ending) {
+      inSegment = false;
+      output.push(segment);
+      segment = undefined;      
+    }        
+  }
+
+  return output;
+}
+
+async function runConcraftLocally(input) {  
+  const { stdout, stderr } = spawnSync(CONCRAFT_BIN, ["tag", CONCRAFT_MODEL], { input });
+
+  return stdout.toString()
+}
+
+async function invokeConcraftRemotely(dag) {
+  const response = await fetch(`${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: a.tags.split(":").map(t => t.split(".")[0]).join(":") }))
+    .map(a => `${[...Object.values(a), "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);
+    
+    output.push(element);
+  }
+
+  return output;
+}
+
+async function danalyse(input) {
+  const analysis = await analyse(input);
+  return await desambiguate(analysis);
+}
+
+function generate(text, ...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: ${line}`);
+    }
+    
+    const { opening, ending, ...data } = match.groups;
+    
+    if (opening) {
+      inSegment = true;
+      segment = [];
+    }
+    
+    segment.push(data);
+    
+    if (ending) {
+      inSegment = false;
+      output.push(segment);
+      segment = undefined;      
+    }        
+  }
+  
+  return output.flatMap(a => a).filter(entry => tags.every(tag => entry.tags.includes(tag)));
+}
+
+async function main([ interpreter, script, action, ...args ]) {
+  const fn = {
+    analyse,
+    danalyse,
+    generate
+  }[action];
+
+  if (!fn) {
+    throw new Error(`Unknown action: '${action}'`);
+  }
+
+  const output = await fn(...args);
+
+  console.log(JSON.stringify(output, undefined, 2));
+}
+
+
+main(process.argv);