scribe.js-ocr 0.6.0 → 0.6.1

package/js/fontEval.js

@@ -52,41 +52,25 @@ export async function evaluateFonts(pageArr, opt) {
   const evalNimbusRomNo9L = !!(opt ? FontCont.opt?.NimbusRomNo9L : FontCont.raw?.NimbusRomNo9L);
   const evalNimbusMono = !!(opt ? FontCont.opt?.NimbusMono : FontCont.raw?.NimbusMono);
 
-  // The browser version runs in parallel using workers, however the Node.js version runs sequentially,
-  // as the canvas package does not support workers, and trying to run in parallel causes problems.
-  // The logic is the same in both versions.
-  let fontMetricsTmp;
-  if (typeof process === 'undefined') {
-    const fontMetricsPromises = {
-      carlito: evalCarlito ? evalPagesFont('Carlito', pageArr, opt) : null,
-      nimbusSans: evalNimbusSans ? evalPagesFont('NimbusSans', pageArr, opt) : null,
-      century: evalCentury ? evalPagesFont('Century', pageArr, opt) : null,
-      palatino: evalPalatino ? evalPagesFont('Palatino', pageArr, opt) : null,
-      garamond: evalGaramond ? evalPagesFont('Garamond', pageArr, opt) : null,
-      nimbusRomNo9L: evalNimbusRomNo9L ? evalPagesFont('NimbusRomNo9L', pageArr, opt) : null,
-      nimbusMono: evalNimbusMono ? evalPagesFont('NimbusMono', pageArr, opt) : null,
-    };
-
-    fontMetricsTmp = {
-      carlito: await fontMetricsPromises.carlito,
-      nimbusSans: await fontMetricsPromises.nimbusSans,
-      century: await fontMetricsPromises.century,
-      palatino: await fontMetricsPromises.palatino,
-      garamond: await fontMetricsPromises.garamond,
-      nimbusRomNo9L: await fontMetricsPromises.nimbusRomNo9L,
-      nimbusMono: await fontMetricsPromises.nimbusMono,
-    };
-  } else {
-    fontMetricsTmp = {
-      carlito: evalCarlito ? await evalPagesFont('Carlito', pageArr, opt) : null,
-      nimbusSans: evalNimbusSans ? await evalPagesFont('NimbusSans', pageArr, opt) : null,
-      century: evalCentury ? await evalPagesFont('Century', pageArr, opt) : null,
-      palatino: evalPalatino ? await evalPagesFont('Palatino', pageArr, opt) : null,
-      garamond: evalGaramond ? await evalPagesFont('Garamond', pageArr, opt) : null,
-      nimbusRomNo9L: evalNimbusRomNo9L ? await evalPagesFont('NimbusRomNo9L', pageArr, opt) : null,
-      nimbusMono: evalNimbusMono ? await evalPagesFont('NimbusMono', pageArr, opt) : null,
-    };
-  }
+  const fontMetricsPromises = {
+    carlito: evalCarlito ? evalPagesFont('Carlito', pageArr, opt) : null,
+    nimbusSans: evalNimbusSans ? evalPagesFont('NimbusSans', pageArr, opt) : null,
+    century: evalCentury ? evalPagesFont('Century', pageArr, opt) : null,
+    palatino: evalPalatino ? evalPagesFont('Palatino', pageArr, opt) : null,
+    garamond: evalGaramond ? evalPagesFont('Garamond', pageArr, opt) : null,
+    nimbusRomNo9L: evalNimbusRomNo9L ? evalPagesFont('NimbusRomNo9L', pageArr, opt) : null,
+    nimbusMono: evalNimbusMono ? evalPagesFont('NimbusMono', pageArr, opt) : null,
+  };
+
+  const fontMetricsTmp = {
+    carlito: await fontMetricsPromises.carlito,
+    nimbusSans: await fontMetricsPromises.nimbusSans,
+    century: await fontMetricsPromises.century,
+    palatino: await fontMetricsPromises.palatino,
+    garamond: await fontMetricsPromises.garamond,
+    nimbusRomNo9L: await fontMetricsPromises.nimbusRomNo9L,
+    nimbusMono: await fontMetricsPromises.nimbusMono,
+  };
 
   const fontMetrics = {
     Carlito: fontMetricsTmp.carlito ? fontMetricsTmp.carlito.metricTotal / fontMetricsTmp.carlito.wordsTotal : null,

package/js/generalWorkerMain.js

@@ -147,9 +147,6 @@ export class gs {
   /** @type {?import('../tess/tesseract.esm.min.js').default} */
   static schedulerInner = null;
 
-  /** @type {?Function} */
-  static #resReady = null;
-
   /** @type {?Promise<void>} */
   static schedulerReady = null;
 
@@ -237,10 +234,6 @@ export class gs {
   static renderPageStaticImp = async (args) => (await gs.schedulerInner.addJob('renderPageStaticImp', args));
 
   static init = async () => {
-    gs.schedulerReady = new Promise((resolve, reject) => {
-      gs.#resReady = resolve;
-    });
-
     let workerN;
     if (opt.workerN) {
       workerN = opt.workerN;
@@ -272,8 +265,7 @@ export class gs {
 
     await Promise.all(resArr);
 
-    // @ts-ignore
-    gs.#resReady(true);
+    return;
   };
 
   /**
@@ -318,15 +310,14 @@ export class gs {
   /**
    * Gets the general scheduler if it exists, otherwise creates a new one.
    */
-  static getGeneralScheduler = async () => {
+  static getGeneralScheduler = () => {
     if (gs.schedulerReady) {
-      await gs.schedulerReady;
-      return;
+      return gs.schedulerReady;
     }
 
-    await gs.init();
+    gs.schedulerReady = gs.init();
 
-    return;
+    return gs.schedulerReady;
   };
 
   static clear = () => {
@@ -337,7 +328,6 @@ export class gs {
     gs.clear();
     await gs.schedulerInner.terminate();
     gs.schedulerInner = null;
-    gs.#resReady = null;
     gs.schedulerReady = null;
     gs.#resReadyTesseract = null;
     gs.schedulerReadyTesseract = null;

package/js/recognizeConvert.js

@@ -94,19 +94,9 @@ export const compareOCR = async (ocrA, ocrB, options) => {
     if (res.debugImg) debugImageArr[i] = res.debugImg;
   };
 
-  // This function is run in the main thread in Node.js, with no mechanism for queuing jobs.
-  // Therefore, this needs to be run one at a time in Node.js.
-  if (typeof process === 'undefined') {
-    const indices = [...Array(ocrA.length).keys()];
-    const compPromises = indices.map(async (i) => comparePageI(i));
-    await Promise.allSettled(compPromises);
-  } else {
-    // This needs to be run one at a time in Node.js, as this is run in the main thread,
-    // and there is no mechanism for queuing jobs, so side effects will interfere with each other.
-    for (let i = 0; i < ocrA.length; i++) {
-      await comparePageI(i);
-    }
-  }
+  const indices = [...Array(ocrA.length).keys()];
+  const compPromises = indices.map(async (i) => comparePageI(i));
+  await Promise.allSettled(compPromises);
 
   return { ocr: ocrArr, metrics: metricsArr, debug: debugImageArr };
 };
@@ -193,8 +183,6 @@ export const recognizePageImp = async (n, legacy, lstm, areaMode, tessOptions =
   // is to get debugging images for layout analysis rather than get text.
   const runRecognition = legacy || lstm;
 
-  await gs.getGeneralScheduler();
-
   const resArr = await gs.recognizeAndConvert2({
     image: nativeN.src,
     options: config,