From 3ca7c004504990e753d90115dba37f617230f417 Mon Sep 17 00:00:00 2001
From: Pierre Leroux <p.leroux@gmx.com>
Date: Wed, 20 Nov 2024 16:45:34 +0100
Subject: [PATCH] feat: mupdfjs with wasm build to extract metadata and cover
 img

---
 package-lock.json            |  6 +++++
 package.json                 |  1 +
 src/main/pdf/extract.ts      | 50 +++++++++++++++++++++++++++++++++++-
 src/main/pdf/extract.type.ts |  8 +++---
 tsconfig.json                |  2 +-
 webpack.config.main.js       |  2 +-
 6 files changed, 62 insertions(+), 7 deletions(-)

diff --git a/package-lock.json b/package-lock.json
index 7b5b40f0b5..5d5d5bc242 100644
--- a/package-lock.json
+++ b/package-lock.json
@@ -37,6 +37,7 @@
         "match-sorter": "^8.0.0",
         "mathjax": "^3.2.2",
         "mime-types": "^2.1.35",
+        "mupdf": "github:edrlab/mupdf.js",
         "nanoid": "^5.0.8",
         "node-fetch": "^3.3.2",
         "proxy-agent": "^6.4.0",
@@ -19736,6 +19737,11 @@
         "multicast-dns": "cli.js"
       }
     },
+    "node_modules/mupdf": {
+      "version": "1.0.0",
+      "resolved": "git+ssh://git@github.com/edrlab/mupdf.js.git#928cad2fcf7db5e3bd46070f140a2713f873bb86",
+      "license": "AGPL-3.0-or-later"
+    },
     "node_modules/nanoid": {
       "version": "5.0.8",
       "resolved": "https://registry.npmjs.org/nanoid/-/nanoid-5.0.8.tgz",
diff --git a/package.json b/package.json
index 3cf7c51d3c..39b2171ca2 100644
--- a/package.json
+++ b/package.json
@@ -291,6 +291,7 @@
     "match-sorter": "^8.0.0",
     "mathjax": "^3.2.2",
     "mime-types": "^2.1.35",
+    "mupdf": "github:edrlab/mupdf.js",
     "nanoid": "^5.0.8",
     "node-fetch": "^3.3.2",
     "proxy-agent": "^6.4.0",
diff --git a/src/main/pdf/extract.ts b/src/main/pdf/extract.ts
index 2fc18805d2..24715e9c97 100644
--- a/src/main/pdf/extract.ts
+++ b/src/main/pdf/extract.ts
@@ -13,11 +13,59 @@ import { encodeURIComponent_RFC3986 } from "@r2-utils-js/_utils/http/UrlUtils";
 
 import { IInfo } from "./extract.type";
 
+import { readFile } from "node:fs/promises";
+
+import * as mupdfjs from "mupdf";
+
 const debug = debug_("readium-desktop:main/pdf/extract/index.ts");
 debug("_");
 
 type TExtractPdfData = [data: IInfo | undefined, coverPNG: Buffer | undefined];
-export const extractPDFData =
+
+export const extractPDFData = async (pdfPath: string): Promise<TExtractPdfData> => {
+
+    try {
+        const pdfBuffer = await readFile(pdfPath);
+
+        const doc = mupdfjs.PDFDocument.openDocument(pdfBuffer, "application/pdf");
+
+        const info: IInfo = {
+            Title: doc.getMetaData("info:Title"),
+            Subject: doc.getMetaData("info:Subject"),
+            Keywords: doc.getMetaData("info:Keywords"),
+            Author: doc.getMetaData("info:Author"),
+            Creator: doc.getMetaData("info:Creator"),
+            Producer: doc.getMetaData("info:Producer"),
+            CreationDate: doc.getMetaData("info:CreationDate"),
+            ModDate: doc.getMetaData("info:ModDate"),
+            numberOfPages: doc.countPages(),
+        };
+
+        const page = new mupdfjs.PDFPage(doc, 0);
+
+        const pixmap = page.toPixmap(mupdfjs.Matrix.identity, mupdfjs.ColorSpace.DeviceRGB, false, true);
+        const pngImage = pixmap.asPNG();
+        const img = Buffer.alloc(pngImage.byteLength);
+        for (let i = 0; i < img.length; ++i) {
+            img[i] = pngImage[i];
+        }
+
+        return [info, img];
+
+
+    } catch (e) {
+
+        debug("####");
+        debug("####");
+        debug(e);
+        debug("####");
+        debug("####");
+    }
+
+    return [undefined, undefined];
+};
+
+export const extractPDFDataPdfjs =
     async (pdfPath: string)
         : Promise<TExtractPdfData> => {
 
diff --git a/src/main/pdf/extract.type.ts b/src/main/pdf/extract.type.ts
index f5cc73f683..cdf856a6ed 100644
--- a/src/main/pdf/extract.type.ts
+++ b/src/main/pdf/extract.type.ts
@@ -7,10 +7,10 @@
 
 export interface IInfo {
     PDFFormatVersion?: string;
-    IsAcroFormPresent?: boolean;
-    IsCollectionPresent?: boolean;
-    IsLinearized?: boolean;
-    IsXFAPresent?: boolean;
+    // IsAcroFormPresent?: boolean;
+    // IsCollectionPresent?: boolean;
+    // IsLinearized?: boolean;
+    // IsXFAPresent?: boolean;
     Title?: string;
     Subject?: string;
     Keywords?: string;
diff --git a/tsconfig.json b/tsconfig.json
index 90206474a7..1f75d587cc 100644
--- a/tsconfig.json
+++ b/tsconfig.json
@@ -44,7 +44,7 @@
     "removeComments": true,
     "skipLibCheck": false,
     "module": "ES2020",
-    "moduleResolution": "Node",
+    "moduleResolution": "node10",
     "lib": [
       "es2020",
       "dom",
diff --git a/webpack.config.main.js b/webpack.config.main.js
index 4402d982ca..ad221403d1 100644
--- a/webpack.config.main.js
+++ b/webpack.config.main.js
@@ -45,7 +45,7 @@ const _externalsCache = new Set();
 if (nodeEnv !== "production") {
     const nodeExternals = require("webpack-node-externals");
     const neFunc = nodeExternals({
-        allowlist: ["timeout-signal", "nanoid", "normalize-url", "node-fetch", "data-uri-to-buffer", /^fetch-blob/, /^formdata-polyfill/],
+        allowlist: ["timeout-signal", "nanoid", "normalize-url", "node-fetch", "mupdf", "data-uri-to-buffer", /^fetch-blob/, /^formdata-polyfill/],
         importType: function (moduleName) {
             if (!_externalsCache.has(moduleName)) {
                 console.log(`WEBPACK EXTERNAL (MAIN): [${moduleName}]`);