@xuda.io/drive_module 1.1.908 → 1.1.910

package/index.js

@@ -1694,6 +1694,7 @@ exports.check_drive_file = async function (req, job_id, headers, file_obj) {
 };
 
 const ocr_drive_file = async function (doc) {
+  return;
   const { drive_type, target_file } = doc;
   var doc_ret;
   try {
@@ -1717,7 +1718,7 @@ const ocr_drive_file = async function (doc) {
         break;
     }
 
-    doc = doc_ret.data;
+    doc = await get_drive_doc("file", drive_type); //doc_ret.data;
 
     var ocr = "";
 
@@ -1816,6 +1817,129 @@ const ocr_drive_file = async function (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(
@@ -1860,7 +1984,7 @@ exports.run_drive_pending_ocr = async function (req) {
 
     var app_res = await db_module.find_couch_query("xuda_master", app_opt);
 
-    opt.selector.docType = "workspace__drive";
+    opt.selector.docType = "workspace_drive";
     for (let app of app_res.docs) {
       const app_drive_res = await db_module.find_app_couch_query(app._id, opt);
       for await (let doc of app_drive_res.docs) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@xuda.io/drive_module",
-  "version": "1.1.908",
+  "version": "1.1.910",
   "description": "Xuda Drive Server Module",
   "main": "index.js",
   "dependencies": {