Added savePDF option to recognize per #488; cleaned up code for linter

examples/browser/download-pdf.html

@@ -8,26 +8,29 @@
       <button id="download-pdf" disabled="true">Download PDF</button>
     </div>
     <textarea id="board" readonly rows="8" cols="80">Upload an image file</textarea>
-    <script>
+    <script type="module">
       const { createWorker } = Tesseract;
       const worker = await createWorker({
         corePath: '/node_modules/tesseract.js-core/tesseract-core.wasm.js',
+        workerPath: "/dist/worker.dev.js",
         logger: m => console.log(m),
       });
       const uploader = document.getElementById('uploader');
       const dlBtn = document.getElementById('download-pdf');
+      let pdf;
       const recognize = async ({ target: { files }  }) => {
         await worker.loadLanguage('eng');
         await worker.initialize('eng');
-        const { data: { text } } = await worker.recognize(files[0]);
+        const res = await worker.recognize(files[0], {savePDF: true});
+        pdf = res.data.pdf;
+        const text = res.data.text;
         const board = document.getElementById('board');
         board.value = text;
         dlBtn.disabled = false;
       };
       const downloadPDF = async () => {
         const filename = 'tesseract-ocr-result.pdf';
-        const { data } = await worker.getPDF('Tesseract OCR Result');
+        const blob = new Blob([new Uint8Array(pdf)], { type: 'application/pdf' });
-        const blob = new Blob([new Uint8Array(data)], { type: 'application/pdf' });
         if (navigator.msSaveBlob) {
           // IE 10+
           navigator.msSaveBlob(blob, filename);

src/createWorker.js

@@ -28,14 +28,14 @@ module.exports = async (_options = {}) => {
   const resolves = {};
   const rejects = {};
 
-  let resReject;
+  let workerResReject;
-  let resResolve; 
+  let workerResResolve;
-  const res = new Promise((resolve, reject) => {
+  const workerRes = new Promise((resolve, reject) => {
-    resResolve = resolve;
+    workerResResolve = resolve;
-    resReject = reject;
+    workerResReject = reject;
   });
-  let workerError = (event) => {resReject(event.message)};
+  const workerError = (event) => { workerResReject(event.message); };
-  
+
   let worker = spawnWorker(options);
   worker.onerror = workerError;
 
@@ -63,8 +63,8 @@ module.exports = async (_options = {}) => {
     })
   );
 
-  const load = (jobId) => (
+  const load = () => (
-    console.warn("`load` is depreciated and should be removed from code (workers now come pre-loaded)")
+    console.warn('`load` is depreciated and should be removed from code (workers now come pre-loaded)')
   );
 
   const loadInternal = (jobId) => (
@@ -145,13 +145,14 @@ module.exports = async (_options = {}) => {
     }))
   );
 
-  const getPDF = (title = 'Tesseract OCR Result', textonly = false, jobId) => (
+  const getPDF = (title = 'Tesseract OCR Result', textonly = false, jobId) => {
-    startJob(createJob({
+    console.log('`getPDF` function is depreciated. `recognize` option `savePDF` should be used instead.');
+    return startJob(createJob({
       id: jobId,
       action: 'getPDF',
       payload: { title, textonly },
-    }))
+    }));
-  );
+  };
 
   const detect = async (image, jobId) => (
     startJob(createJob({
@@ -189,7 +190,7 @@ module.exports = async (_options = {}) => {
       resolves[action]({ jobId, data: d });
     } else if (status === 'reject') {
       rejects[action](data);
-      if (action === "load") resReject(data);
+      if (action === 'load') workerResReject(data);
       if (errorHandler) {
         errorHandler(data);
       } else {
@@ -220,8 +221,7 @@ module.exports = async (_options = {}) => {
     terminate,
   };
 
-  loadInternal().then(() => resResolve(resolveObj)).catch(() => {});
+  loadInternal().then(() => workerResResolve(resolveObj)).catch(() => {});
-
-  return res;
 
+  return workerRes;
 };

src/index.d.ts

@@ -59,6 +59,9 @@ declare namespace Tesseract {
     saveImageOriginal: boolean
     saveImageGrey: boolean
     saveImageBinary: boolean
+    savePDF: boolean
+    pdfTitle: string
+    pdfTextOnly: boolean
     rotateAuto: boolean
     rotateRadians: float
   }
@@ -231,10 +234,11 @@ declare namespace Tesseract {
     box: string | null;
     unlv: string | null;
     sd: string | null;
-    imageOriginal: string;
+    imageOriginal: string | null;
-    imageGrey: string;
+    imageGrey: string | null;
-    imageBinary: string;
+    imageBinary: string | null;
-    rotateRadians: number;
+    rotateRadians: number | null;
+    pdf: number[] | null;
   }
 }
 

src/worker-script/index.js

@@ -200,6 +200,20 @@ const initialize = async ({
   }
 };
 
+const getPDFInternal = (title, textonly) => {
+  const pdfRenderer = new TessModule.TessPDFRenderer('tesseract-ocr', '/', textonly);
+  pdfRenderer.BeginDocument(title);
+  pdfRenderer.AddImage(api);
+  pdfRenderer.EndDocument();
+  TessModule._free(pdfRenderer);
+
+  return TessModule.FS.readFile('/tesseract-ocr.pdf');
+};
+
+const getPDF = async ({ payload: { title, textonly } }, res) => {
+  res.resolve(getPDFInternal(title, textonly));
+};
+
 const getImage = (type) => {
   api.WriteImage(type, '/image.png');
   const pngBuffer = TessModule.FS.readFile('/image.png');
@@ -211,7 +225,8 @@ const getImage = (type) => {
 const recognize = async ({
   payload: {
     image, options: {
-      rectangle: rec, saveImageOriginal, saveImageGrey, saveImageBinary, rotateAuto, rotateRadians,
+      rectangle: rec, saveImageOriginal, saveImageGrey, saveImageBinary, savePDF, pdfTitle,
+      pdfTextOnly, rotateAuto, rotateRadians,
     },
   },
 }, res) => {
@@ -263,12 +278,23 @@ const recognize = async ({
     const result = dump(TessModule, api, params);
     if (saveImageOriginal) {
       result.imageOriginal = getImage(imageType.ORIGINAL);
+    } else {
+      result.imageOriginal = null;
     }
     if (saveImageGrey) {
       result.imageGrey = getImage(imageType.GREY);
+    } else {
+      result.imageGrey = null;
     }
     if (saveImageBinary) {
       result.imageBinary = getImage(imageType.BINARY);
+    } else {
+      result.imageBinary = null;
+    }
+    if (savePDF) {
+      result.pdf = getPDFInternal(pdfTitle ?? 'Tesseract OCR Result', pdfTextOnly ?? false);
+    } else {
+      result.pdf = null;
     }
     result.rotateRadians = rotateRadiansFinal;
     res.resolve(result);
@@ -325,12 +351,18 @@ const threshold = async ({
     const result = {};
     if (saveImageOriginal) {
       result.imageOriginal = getImage(imageType.ORIGINAL);
+    } else {
+      result.imageOriginal = null;
     }
     if (saveImageGrey) {
       result.imageGrey = getImage(imageType.GREY);
+    } else {
+      result.imageGrey = null;
     }
     if (saveImageBinary) {
       result.imageBinary = getImage(imageType.BINARY);
+    } else {
+      result.imageBinary = null;
     }
     result.rotateRadians = rotateRadiansFinal;
     res.resolve(result);
@@ -340,16 +372,6 @@ const threshold = async ({
   }
 };
 
-const getPDF = async ({ payload: { title, textonly } }, res) => {
-  const pdfRenderer = new TessModule.TessPDFRenderer('tesseract-ocr', '/', textonly);
-  pdfRenderer.BeginDocument(title);
-  pdfRenderer.AddImage(api);
-  pdfRenderer.EndDocument();
-  TessModule._free(pdfRenderer);
-
-  res.resolve(TessModule.FS.readFile('/tesseract-ocr.pdf'));
-};
-
 const detect = async ({ payload: { image } }, res) => {
   try {
     const ptr = setImage(TessModule, api, image);
@@ -357,7 +379,7 @@ const detect = async ({ payload: { image } }, res) => {
 
     if (!api.DetectOS(results)) {
       TessModule._free(ptr);
-      
+
       res.resolve({
         tesseract_script_id: null,
         script: null,
@@ -422,18 +444,18 @@ exports.dispatchHandlers = (packet, send) => {
 
   latestJob = res;
 
-    ({
+  ({
-      load,
+    load,
-      FS,
+    FS,
-      loadLanguage,
+    loadLanguage,
-      initialize,
+    initialize,
-      setParameters,
+    setParameters,
-      recognize,
+    recognize,
-      threshold,
+    threshold,
-      getPDF,
+    getPDF,
-      detect,
+    detect,
-      terminate,
+    terminate,
-    })[packet.action](packet, res)
+  })[packet.action](packet, res)
     .catch((err) => res.reject(err.toString()));
 };