@xuda.io/drive_module 1.1.951 → 1.1.953

package/index.js

@@ -25,7 +25,8 @@ const fse = require("fs-extra");
 const crypto = require("node:crypto");
 const unzipper = require("unzipper");
 const fs = require("fs");
-
+const tesseract = require("node-tesseract-ocr");
+const franc = require("franc");
 ////////////////////////////////////////////
 
 [
@@ -1722,16 +1723,11 @@ const ocr_drive_file = async function (app_id, doc) {
     );
 
     var ocr = "";
+    const generalConfig = { lang: "eng", oem: 1, psm: 3 };
 
     switch (doc.mime.split("/")[0]) {
       case "image": {
-        const tesseract = require("node-tesseract-ocr");
-        ocr = await tesseract.recognize(target_file, {
-          // lang: "eng",
-          lang: "eng+heb",
-          oem: 1,
-          psm: 3,
-        });
+        ocr = await tesseract.recognize(target_file, generalConfig);
 
         break;
       }
@@ -1740,15 +1736,10 @@ const ocr_drive_file = async function (app_id, doc) {
         switch (doc.mime.split("/")[1]) {
           case "pdf": {
             const { convertPDF } = require("pdf2image");
-            const tesseract = require("node-tesseract-ocr");
 
             const images = await convertPDF(target_file);
             for await (const imageData of images) {
-              ocr += await tesseract.recognize(imageData.path, {
-                // lang: "eng",
-                oem: 1,
-                psm: 3,
-              });
+              ocr += await tesseract.recognize(imageData.path, generalConfig);
             }
             break;
           }
@@ -1806,129 +1797,6 @@ const ocr_drive_file = async function (app_id, doc) {
   }
 };
 
-// const ocr_drive_file = async function (doc) {
-//   const { drive_type, target_file } = doc;
-//   var doc_ret;
-//   try {
-//     doc.ocr_stat = 2;
-//     doc.ocr_stat_date = Date.now();
-//     switch (drive_type) {
-//       case "workspace":
-//       case "studio":
-//         const app_id_reference = await _common.get_project_app_id(doc.app_id);
-//         await db_module.save_app_couch_doc(app_id_reference, doc);
-//         doc_ret = await db_module.get_app_couch_doc(app_id_reference, doc._id);
-//         break;
-
-//       case "user":
-//         await db_module.save_couch_doc("xuda_drive", doc);
-//         doc_ret = await db_module.get_couch_doc("xuda_drive", doc._id);
-//         break;
-
-//       default:
-//         throw new Error("drive_type error");
-//         break;
-//     }
-
-//     doc = await get_drive_doc("file",drive_type,)//doc_ret.data;
-
-//     var ocr = "";
-
-//     switch (doc.mime.split("/")[0]) {
-//       case "image": {
-//         const tesseract = require("node-tesseract-ocr");
-//         ocr = await tesseract.recognize(target_file, {
-//           lang: "eng",
-//           oem: 1,
-//           psm: 3,
-//         });
-
-//         break;
-//       }
-
-//       case "application": {
-//         switch (doc.mime.split("/")[1]) {
-//           case "pdf": {
-//             const { convertPDF } = require("pdf2image");
-//             const tesseract = require("node-tesseract-ocr");
-
-//             const images = await convertPDF(target_file);
-//             for await (const imageData of images) {
-//               ocr += await tesseract.recognize(imageData.path, {
-//                 lang: "eng",
-//                 oem: 1,
-//                 psm: 3,
-//               });
-//             }
-//             break;
-//           }
-
-//           case "vnd.openxmlformats-officedocument.wordprocessingml.document": {
-//             const mammoth = require("mammoth");
-//             ocr = (await mammoth.extractRawText({ path: target_file })).value;
-//             break;
-//           }
-
-//           case "nd.ms-excel":
-//           case "vnd.openxmlformats-officedocument.spreadsheetml.sheet": {
-//             const xlsx = require("node-xlsx").default;
-
-//             const sheets = xlsx.parse(target_file);
-//             sheets.forEach((sheet) => {
-//               console.log(`Sheet: ${sheet.name}`);
-//               sheet.data.forEach((row) => {
-//                 ocr += row.join(" "); // Join all cells in the row
-//               });
-//             });
-//             break;
-//           }
-
-//           case "vnd.ms-powerpoint":
-//           case "vnd.openxmlformats-officedocument.presentationml.presentation":
-//             // tbd
-//             break;
-
-//           default:
-//             ocr = (await fs_module.read_file(target_file)).data;
-//             break;
-//         }
-//         break;
-//       }
-
-//       case "text": {
-//         ocr = (await fs_module.read_file(target_file)).data;
-//         break;
-//       }
-
-//       default:
-//         break;
-//     }
-
-//     doc.ocr = ocr.replace(/<\/?[^>]+(>|$)/g, "").replace(/[\n\t]/g, "");
-//     doc.ocr_stat = 3;
-//   } catch (err) {
-//     doc.ocr_stat = 4;
-//   } finally {
-//     doc.ocr_stat_ts = Date.now();
-
-//     var save_ret;
-//     switch (drive_type) {
-//       case "workspace":
-//       case "studio":
-//         const app_id_reference = await _common.get_project_app_id(doc.app_id);
-//         await db_module.save_app_couch_doc(app_id_reference, doc);
-//         break;
-
-//       case "user":
-//         await db_module.save_couch_doc("xuda_drive", doc);
-//         break;
-
-//       default:
-//         break;
-//     }
-//   }
-// };
-
 const validate_drive_type = function (drive_type) {
   if (!["studio", "workspace", "user"].includes(drive_type)) {
     throw new Error(

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@xuda.io/drive_module",
-  "version": "1.1.951",
+  "version": "1.1.953",
   "description": "Xuda Drive Server Module",
   "main": "index.js",
   "dependencies": {