@iebh/reflib 2.2.3 → 2.3.1

package/modules/endnoteXml.js

@@ -1,13 +1,22 @@
 import camelCase from '../shared/camelCase.js';
 import Emitter from '../shared/emitter.js';
 
-// TODO: CF: Don't need to import both, it depends if we are on browser or node
-import * as htmlparser2 from "htmlparser2";
-// FIXME: CF: Browsers freak out without pollyfills if this is imported
-import {WritableStream as XMLParser} from 'htmlparser2/lib/WritableStream';
+// Only import WritableStream on node
+// This is needed even with the "browser" field of package.json
+let XMLParser;
+if (typeof window === 'undefined') {
+	// We are in node env
+	import('htmlparser2/lib/WritableStream').then(module => {
+		XMLParser = module.WritableStream;
+	});
+} else {
+	// Handle or ignore in browser environment
+	console.log("WritableStream not loaded in browser.");
+}
+
 
 /**
-* @see modules/interface.js
+* @see modules/inhterface.js
 */
 export function readStream(stream) {
 	let emitter = Emitter();
@@ -27,12 +36,65 @@ export function readStream(stream) {
 
 
 	/**
-	* Whether to append incomming text blocks to the previous block
+	* Whether to append incoming text blocks to the previous block
 	* This is necessary as XMLParser splits text into multiple calls so we need to know whether to append or treat this item as a continuation of the previous one
 	* @type {boolean}
 	*/
 	let textAppend = false;
 
+	class XMLParserBrowser {
+		constructor(passedParserOptions) {
+			// Stores XML in text form once read
+			this.text = "";
+			this.emitter = emitter;
+			// Add event listeners to mimic htmlparser2 behavior in the browser
+			this.emitter.on('opentag', passedParserOptions.onopentag);
+			this.emitter.on('closetag', passedParserOptions.onclosetag);
+			this.emitter.on('text', passedParserOptions.ontext);
+		}
+
+		write(data) {
+			// CF: TODO: Parse data as it comes in chunks for better memory efficiency
+			this.text += data;
+		}
+
+		end() {
+			this.parseXML(this.text);
+			// Free memory
+			this.text = ''
+			this.emitter.emit('end');
+		}
+
+		parseXML(xmlString) {
+			let parser = new DOMParser();
+			let doc = parser.parseFromString(xmlString, 'application/xml');
+			this.traverseNode(doc.documentElement);
+		}
+
+		traverseNode(node) {
+			if (node.nodeType === Node.ELEMENT_NODE) {
+				let name = camelCase(node.nodeName);
+				let attrs = Array.from(node.attributes).reduce((acc, attr) => {
+					acc[attr.name] = attr.value;
+					return acc;
+				}, {});
+
+				this.emitter.emit('opentag', name, attrs);
+
+				for (let child of node.childNodes) {
+					this.traverseNode(child);
+				}
+
+				this.emitter.emit('closetag', name);
+			} else if (node.nodeType === Node.TEXT_NODE) {
+				let text = node.nodeValue.trim();
+				if (text) {
+					this.emitter.emit('text', text);
+				}
+			}
+		}
+	}
+
 	/**
 	 * The options/callbacks for the parser
 	 * @type {Object}
@@ -49,7 +111,7 @@ export function readStream(stream) {
 		},
 		onclosetag(name) {
 			if (name == 'record') {
-				if (ref.title) ref.title = ref.title // htmlparser2 handles the '<title>' tag in a really bizare way so we have to pull apart the <style> bits when parsing
+				if (ref.title) ref.title = ref.title // htmlparser2 handles the '<title>' tag in a really bizarre way so we have to pull apart the <style> bits when parsing
 					.replace(/^.*<style.*>(.*)<\/style>.*$/m, '$1')
 					.replace(/^\s+/, '')
 					.replace(/\s+$/, '')
@@ -106,34 +168,26 @@ export function readStream(stream) {
 	// Queue up the parser in the next tick (so we can return the emitter first)
 	setTimeout(() => {
 
-		if (typeof stream.pipe === 'function') {
+		if (stream.isBrowser === true) {
 			// We are on the node.js client
-			let parser = new XMLParser(parserOptions);
+			console.log('Loading EndNote library as node.js')
+			let parser = new XMLParserBrowser(parserOptions);
 			stream.on('data', ()=> emitter.emit('progress', stream.bytesRead))
 			stream.pipe(parser)
 			return;
 		}
 
-		// TODO: CF: We may want to consider moving to a DIY parser for speed and memory efficiency
-		if (typeof stream.getReader === 'function') {
-			// We are on the browser
-			var reader = stream.getReader();
-			var parser = new htmlparser2.Parser(parserOptions);
-			parseXMLOnBrowser();
+		else if (typeof stream.pipe === 'function') {
+			// We are on the node.js client
+			console.log('Loading EndNote library as node.js')
+			let parser = new XMLParser(parserOptions);
+			stream.on('data', ()=> emitter.emit('progress', stream.bytesRead))
+			stream.pipe(parser)
 			return;
 		}
 
-		function parseXMLOnBrowser() {
-			reader.read().then(({done, value}) => {
-				if (done) {
-					parser.end();
-				} else {
-					var text = new TextDecoder().decode(value);
-					parser.write(text);
-					text = null; // Free up memory
-					parseXMLOnBrowser();
-				}
-			})
+		else {
+			console.error('Error determining if on browser or node')
 		}
 
 	})
@@ -146,8 +200,8 @@ export function readStream(stream) {
 * @see modules/interface.js
 * @param {Object} [options] Additional options to use when parsing
 * @param {string} [options.defaultType='journalArticle'] Default citation type to assume when no other type is specified
-* @param {string} [options.filePath="c:\\"] "Fake" internal source file path the citation lirary was exported from, must end with backslashes
-* @param {string} [options.fileName="EndNote.enl"] "Fake" internal source file name the citation lirary was exported from
+* @param {string} [options.filePath="c:\\"] "Fake" internal source file path the citation library was exported from, must end with backslashes
+* @param {string} [options.fileName="EndNote.enl"] "Fake" internal source file name the citation library was exported from
 * @param {function} [options.formatDate] Date formatter to translate between a JS Date object and the EndNote YYYY-MM-DD format
 */
 export function writeStream(stream, options) {
@@ -409,7 +463,7 @@ export let translations = {
 			{rl: 'grant', rawText: 'Grant', rawId: 54},
 			{rl: 'hearing', rawText: 'Hearing', rawId: 14},
 			{rl: 'journalArticle', rawText: 'Journal Article', rawId: 17},
-			{rl: 'legalRuleOrRegulation', rawText:', Legal Rule or Regulation', rawId: 50},
+			{rl: 'legalRuleOrRegulation', rawText: 'Legal Rule or Regulation', rawId: 50},
 			{rl: 'magazineArticle', rawText: 'Magazine Article', rawId: 19},
 			{rl: 'manuscript', rawText: 'Manuscript', rawId: 36},
 			{rl: 'map', rawText: 'Map', rawId: 20},

package/modules/json.js

@@ -1,6 +1,45 @@
 import Emitter from '../shared/emitter.js';
 import JSONStream from 'JSONStream';
 
+class BrowserJSONStream {
+	constructor() {
+			this.text = '';
+			this.emitter = Emitter();
+	}
+
+	write(data) {
+		// CF: TODO: Parse data as it comes in chunks for better memory efficiency
+		this.text += data
+	}
+
+	end() {
+		try {
+			// Parse this.text as JSON
+			const jsonArray = JSON.parse(this.text);
+			// Free memory
+			this.text = ''
+
+			// For each entry in the json array (as ref):
+			jsonArray.forEach(ref => {
+				this.emitter.emit('ref', {
+					recNumber: this.recNumber++,
+					...ref,
+				});
+			});
+
+			// Finished
+			this.emitter.emit('end');
+		} catch (e) {
+			console.error('Error parsing final JSON:', e);
+			this.emitter.emit('error', e);
+		}
+	}
+
+	on(event, listener) {
+		this.emitter.on(event, listener);
+	}
+}
+
 /**
 * @see modules/interface.js
 */
@@ -11,15 +50,36 @@ export function readStream(stream) {
 	// Queue up the parser in the next tick (so we can return the emitter first)
 	setTimeout(()=> {
 		stream.on('data', ()=> emitter.emit('progress', stream.bytesRead));
-		stream.pipe(
-			JSONStream.parse('*')
-				.on('data', ref => emitter.emit('ref', {
-					recNumber: recNumber++,
-					...ref,
-				}))
-				.on('end', ()=> emitter.emit('end'))
-				.on('error', emitter.emit.bind('error'))
-		)
+
+		if (stream.isBrowser === true) {
+			// On browser
+			console.log('Parsing JSON natively in browser');
+			const browserJSONStream = new BrowserJSONStream();
+			browserJSONStream.on('ref', (data) => {
+					emitter.emit('ref', data);
+			});
+			browserJSONStream.on('end', () => emitter.emit('end'));
+			browserJSONStream.on('error', (error) => emitter.emit('error', error));
+			stream.pipe(browserJSONStream);
+	}
+
+		else if (typeof stream.pipe === 'function') {
+			// On node.js
+			console.log('Parsing JSON with node.js library')
+			const nodeJSONStream = JSONStream.parse('*')
+			nodeJSONStream
+			.on('data', ref => emitter.emit('ref', {
+				recNumber: recNumber++,
+				...ref,
+			}))
+			.on('end', ()=> emitter.emit('end'))
+			.on('error', emitter.emit.bind('error'));
+			stream.pipe(nodeJSONStream)
+		}
+
+		else {
+			console.error('Error determining if on browser or node')
+		}
 	});
 
 	return emitter;
@@ -54,7 +114,7 @@ export function writeStream(stream, options) {
 			stream.write(']');
 			return new Promise((resolve, reject) =>
 				stream.end(err => err ? reject(err) : resolve())
-			);
-		},
-	};
+		);
+	},
+};
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@iebh/reflib",
-  "version": "2.2.3",
+  "version": "2.3.1",
   "description": "Reference / Citation reference library utilities",
   "scripts": {
     "lint": "eslint lib modules shared test",
@@ -17,8 +17,8 @@
     "citations",
     "library"
   ],
-  "author": [
-    "Matt Carter <m@ttcarter.com> (https://github.com/hash-bang)",
+  "author": "Matt Carter <m@ttcarter.com> (https://github.com/hash-bang)",
+  "contributors": [
     "Connor Forbes <cforbes.software@gmail.com> (https://github.com/connorf25)"
   ],
   "license": "MIT",
@@ -38,6 +38,11 @@
     },
     "./*": "./lib/*.js"
   },
+  "browser": {
+    "htmlparser2/lib/WritableStream": false,
+    "htmlparser2/lib/esm/WritableStream": false,
+    "JSONStream": false
+  },
   "devDependencies": {
     "@momsfriendlydevco/eslint-config": "^1.0.9",
     "chai": "^4.3.10",

package/shared/streamEmitter.js

@@ -10,7 +10,7 @@ import Emitter from '../shared/emitter.js';
 * @emits end Emitted as `()` when the input stream has closed
 * @emits error Emitted as `(Error)` on any read error
 */
-export default function streamEmitter(inStream) {
+export default function streamEmitter (inStream) {
 	if (inStream.getReader) { // Assume browser compatible ReadableStream
 		/**
 		* MC's tiny ReadableStream -> stream.Readable / Emitter pattern
@@ -21,6 +21,7 @@ export default function streamEmitter(inStream) {
 		*/
 		let reader = new Emitter();
 		Object.assign(reader, {
+			isBrowser: true, // Tells us we are in browser env
 			bytesRead: 0,
 			reader: inStream.getReader(),
 			textDecoder: new TextDecoder('utf-8'),