Desktop, Cli: Fixed various bugs related to the import of ENEX files as HTML

CliClient/tests/EnexToHtml.js

@@ -89,20 +89,20 @@ describe('EnexToHtml', function() {
 		}],
 	});
 
-	it('fails when not given a matching resource', asyncTest(async () => {
-		// To test the promise-unexpectedly-resolved case, add `audioResource` to the array.
-		const resources = [];
-		const inputFile = fileWithPath('en-media--image.enex');
-		const enexInput = await shim.fsDriver().readFile(inputFile);
-		const promisedOutput = enexXmlToHtml(enexInput, resources);
+	// it('fails when not given a matching resource', asyncTest(async () => {
+	// 	// To test the promise-unexpectedly-resolved case, add `audioResource` to the array.
+	// 	const resources = [];
+	// 	const inputFile = fileWithPath('en-media--image.enex');
+	// 	const enexInput = await shim.fsDriver().readFile(inputFile);
+	// 	const promisedOutput = enexXmlToHtml(enexInput, resources);
 
-		promisedOutput.then(() => {
-			// Promise should not be resolved
-			expect(false).toEqual(true);
-		}, (reason) => {
-			expect(reason)
-				.toBe('Hash with no associated resource: 89ce7da62c6b2832929a6964237e98e9');
-		});
-	}));
+	// 	promisedOutput.then(() => {
+	// 		// Promise should not be resolved
+	// 		expect(false).toEqual(true);
+	// 	}, (reason) => {
+	// 		expect(reason)
+	// 			.toBe('Hash with no associated resource: 89ce7da62c6b2832929a6964237e98e9');
+	// 	});
+	// }));
 
 });

CliClient/tests/enex_to_html/attachment.html

@@ -2,6 +2,5 @@
   <div><a href="joplin://21ca2b948f222a38802940ec7e2e5de3" hash="21ca2b948f222a38802940ec7e2e5de3" type="application/pdf" style="cursor:pointer;" alt="attachment-1">attachment-1</a></div>
   <div>
     <br>
-    <br>
   </div>
 </en-note>

CliClient/tests/enex_to_html/en-media--audio.html

@@ -8,6 +8,5 @@
   </div>
   <div>
     <br>
-    <br>
   </div>
 </en-note>

CliClient/tests/enex_to_html/en-media--image.html

@@ -4,11 +4,9 @@
   <div>
     <input type="checkbox" onclick="return false;">A test for <i>italic</i>
     <br>
-    <br>
   </div>
   <div>
     <br>
-    <br>
   </div>
   <div><i><img src=":/89ce7da62c6b2832929a6964237e98e9" hash="89ce7da62c6b2832929a6964237e98e9" type="image/jpeg" alt=""></i></div>
 </en-note>

ReactNativeClient/lib/htmlUtils.js

@@ -7,6 +7,28 @@ const htmlentities = new Entities().encode;
 const imageRegex = /<img([\s\S]*?)src=["']([\s\S]*?)["']([\s\S]*?)>/gi;
 const anchorRegex = /<a([\s\S]*?)href=["']([\s\S]*?)["']([\s\S]*?)>/gi;
 
+const selfClosingElements = [
+	'area',
+	'base',
+	'basefont',
+	'br',
+	'col',
+	'command',
+	'embed',
+	'frame',
+	'hr',
+	'img',
+	'input',
+	'isindex',
+	'keygen',
+	'link',
+	'meta',
+	'param',
+	'source',
+	'track',
+	'wbr',
+];
+
 class HtmlUtils {
 	headAndBodyHtml(doc) {
 		const output = [];
@@ -15,6 +37,10 @@ class HtmlUtils {
 		return output.join('\n');
 	}
 
+	isSelfClosingTag(tagName) {
+		return selfClosingElements.includes(tagName.toLowerCase());
+	}
+
 	extractImageUrls(html) {
 		if (!html) return [];
 

ReactNativeClient/lib/import-enex-html-gen.js

@@ -1,6 +1,9 @@
 const stringToStream = require('string-to-stream');
 const cleanHtml = require('clean-html');
 const resourceUtils = require('lib/resourceUtils.js');
+const { isSelfClosingTag } = require('lib/htmlUtils');
+const Entities = require('html-entities').AllHtmlEntities;
+const htmlentities = new Entities().encode;
 
 function addResourceTag(lines, resource, attributes) {
 	// Note: refactor to use Resource.markdownTag
@@ -56,7 +59,7 @@ function enexXmlToHtml_(stream, resources) {
 		}
 	};
 
-	return new Promise((resolve, reject) => {
+	return new Promise((resolve) => {
 		const options = {};
 		const strict = false;
 		const saxStream = require('sax').createStream(strict, options);
@@ -69,12 +72,11 @@ function enexXmlToHtml_(stream, resources) {
 
 		saxStream.on('error', function(e) {
 			console.warn(e);
-			// reject(e);
 		});
 
 
 		saxStream.on('text', function(text) {
-			section.lines.push(text);
+			section.lines.push(htmlentities(text));
 		});
 
 		saxStream.on('opentag', function(node) {
@@ -110,7 +112,7 @@ function enexXmlToHtml_(stream, resources) {
 					}
 
 					if (!found) {
-						reject(`Hash with no associated resource: ${hash}`);
+						// console.warn(`Hash with no associated resource: ${hash}`);
 					}
 				}
 
@@ -122,16 +124,16 @@ function enexXmlToHtml_(stream, resources) {
 				}
 			} else if (tagName == 'en-todo') {
 				section.lines.push('<input type="checkbox" onclick="return false;" />');
-			} else if (node.isSelfClosing) {
-				section.lines.push(`<${tagName}${attributesStr}>`);
-			} else {
+			} else if (isSelfClosingTag(tagName)) {
 				section.lines.push(`<${tagName}${attributesStr}/>`);
+			} else {
+				section.lines.push(`<${tagName}${attributesStr}>`);
 			}
 		});
 
-		saxStream.on('closetag', function(n) {
-			const tagName = n ? n.toLowerCase() : n;
-			section.lines.push(`</${tagName}>`);
+		saxStream.on('closetag', function(node) {
+			const tagName = node ? node.toLowerCase() : node;
+			if (!isSelfClosingTag(tagName)) section.lines.push(`</${tagName}>`);
 		});
 
 		saxStream.on('attribute', function() {});
@@ -151,19 +153,19 @@ async function enexXmlToHtml(xmlString, resources, options = {}) {
 	const stream = stringToStream(xmlString);
 	const result = await enexXmlToHtml_(stream, resources, options);
 
-	try {
-		const preCleaning = result.content.lines.join(''); // xmlString
+	const preCleaning = result.content.lines.join('');
 	const final = await beautifyHtml(preCleaning);
 	return final.join('');
-	} catch (error) {
-		console.warn(error);
-	}
 }
 
 const beautifyHtml = (html) => {
 	return new Promise((resolve) => {
-		const options = { wrap: 0 };
-		cleanHtml.clean(html, options, (...cleanedHtml) => resolve(cleanedHtml));
+		try {
+			cleanHtml.clean(html, { wrap: 0 }, (...cleanedHtml) => resolve(cleanedHtml));
+		} catch (error) {
+			console.warn(`Could not clean HTML - the "unclean" version will be used: ${error.message}: ${html.trim().substr(0, 512).replace(/[\n\r]/g, ' ')}...`);
+			resolve([html]);
+		}
 	});
 };
 

ReactNativeClient/lib/import-enex-md-gen.js

@@ -688,7 +688,7 @@ function enexXmlToMdArray(stream, resources) {
 					}
 
 					if (!found) {
-						console.warn(`Hash with no associated resource: ${hash}`);
+						// console.warn(`Hash with no associated resource: ${hash}`);
 					}
 				}