Add support for SDT

chrome/content/zotero/xpcom/pdfWorker/manager.js

@@ -24,6 +24,7 @@
 */
 
 const WORKER_URL = 'chrome://zotero/content/xpcom/pdfWorker/worker.js';
+const ASSETS_URL = 'chrome://zotero/content/xpcom/pdfWorker/';
 const CMAPS_URL = 'resource://zotero/reader/pdf/web/cmaps/';
 const STANDARD_FONTS_URL = 'resource://zotero/reader/pdf/web/standard_fonts/';
 const WASM_URL = 'resource://zotero/reader/pdf/web/wasm/';
@@ -143,6 +144,20 @@ class PDFWorker {
 					Zotero.debug('Failed to fetch wasm data:');
 					Zotero.debug(e);
 				}
+				try {
+					if (message.action === 'FetchData') {
+						let response = await Zotero.HTTP.request(
+							'GET',
+							ASSETS_URL + message.data,
+							{ responseType: 'arraybuffer' }
+						);
+						respData = new Uint8Array(response.response);
+					}
+				}
+				catch (e) {
+					Zotero.debug(`Failed to fetch data (${message.data}):`);
+					Zotero.debug(e);
+				}
 				try {
 					if (message.action === 'SaveRenderedAnnotation') {
 						let { libraryID, annotationKey, buf } = message.data;
@@ -679,6 +694,58 @@ class PDFWorker {
 		}, isPriority);
 	}
 
+	/**
+	 * Get Zotero Structured Text JSON for a PDF, EPUB, or snapshot attachment
+	 *
+	 * @param {Integer} itemID Attachment item id
+	 * @param {Boolean} [isPriority]
+	 * @param {String} [password]
+	 * @returns {Promise<Object|null>}
+	 */
+	async getStructuredData(itemID, isPriority, password) {
+		return this._enqueue(async () => {
+			let attachment = await Zotero.Items.getAsync(itemID);
+
+			Zotero.debug(`Getting structured data from item ${attachment.libraryKey}`);
+			let t = new Date();
+
+			let contentType = attachment.attachmentContentType;
+			if (!(attachment.isPDFAttachment()
+					|| attachment.isEPUBAttachment()
+					|| attachment.isSnapshotAttachment())) {
+				throw new Error('Item must be a PDF, EPUB, or snapshot attachment');
+			}
+
+			let path = await attachment.getFilePathAsync();
+			if (!path) {
+				return null;
+			}
+			let buf = await IOUtils.read(path);
+			buf = new Uint8Array(buf).buffer;
+
+			try {
+				var result = await this._query('getStructuredData', {
+					buf, contentType, password
+				}, [buf]);
+			}
+			catch (e) {
+				let error = new Error(`Worker 'getStructuredData' failed: ${JSON.stringify({ error: e.message })}`);
+				try {
+					error.name = JSON.parse(e.message).name;
+				}
+				catch (e) {
+					Zotero.logError(e);
+				}
+				Zotero.logError(error);
+				throw error;
+			}
+
+			Zotero.debug(`Extracted structured data for item ${attachment.libraryKey} in ${new Date() - t} ms`);
+
+			return result;
+		}, isPriority);
+	}
+
 	/**
 	 * Get data for recognizer-server
 	 *

chrome/content/zotero/xpcom/reader.js

@@ -631,6 +631,18 @@ class ReaderInstance {
 				// As above
 				setTimeout(() => this._openReadAloudVoicesDialog({ lang, tier, ftl }));
 			},
+			getSDT: (password) => {
+				return new this._iframeWindow.Promise(async (resolve) => {
+					try {
+						let sdt = await Zotero.PDFWorker.getStructuredData(this._item.id, true, password);
+						resolve(sdt ? Cu.cloneInto(sdt, this._iframeWindow) : null);
+					}
+					catch (e) {
+						Zotero.logError(e);
+						resolve(null);
+					}
+				});
+			},
 		}, this._iframeWindow, { cloneFunctions: true }));
 
 		this._resolveInitPromise();

document-worker

@@ -1 +1 @@
-Subproject commit fd642b38287f1e59aaf8e02c3132da6d3daa39c1
+Subproject commit 0631cda250d1068544ee1b86e0072be1108b175c

js-build/document-worker.js

@@ -29,14 +29,14 @@ async function getDocumentWorker(signatures) {
 			await exec(
 				`cd ${tmpDir}`
 				+ ` && (test -f ${filename} || curl -f ${url} -o ${filename})`
-				+ ` && unzip -o ${filename} worker.js -d ${targetDir}`
+				+ ` && unzip -o ${filename} -d ${targetDir}`
 			);
 		}
 		catch (e) {
 			console.error(e);
 			await exec('npm ci', { cwd: modulePath });
 			await exec('npm run build', { cwd: modulePath });
-			await fs.copy(path.join(modulePath, 'build', 'worker.js'), path.join(targetDir, 'worker.js'));
+			await fs.copy(path.join(modulePath, 'build'), targetDir);
 		}
 		signatures['document-worker'] = { hash };
 	}

reader

@@ -1 +1 @@
-Subproject commit 46d009be885cf7cf2543d4bfae50483d85a9c01b
+Subproject commit 55c09dd963fe9baaceb75d4fdc584cae8a97f228