Convert video tags to links

README.md

@@ -60,6 +60,7 @@ You can configure the behaviour of html-to-text with the following options:
  * `hideLinkHrefIfSameAsText` by default links are translated the following `<a href='link'>text</a>` => becomes => `text [link]`. If this option is set to true and `link` and `text` are the same, `[link]` will be hidden and only `text` visible.
  * `ignoreHref` ignore all document links if `true`.
  * `ignoreImage` ignore all document images if `true`.
+ * `ignoreVideo` ignore all document videos if `true`.
  * `preserveNewlines` by default, any newlines `\n` in a block of text will be removed. If `true`, these newlines will not be removed.
  * `decodeOptions` defines the text decoding options given to `he.decode`. For more informations see the [he](https://github.com/mathiasbynens/he) module.
  * `uppercaseHeadings` by default, headings (`<h1>`, `<h2>`, etc) are uppercased. Set to `false` to leave headings as they are.

bin/cli.js

@@ -8,6 +8,7 @@ var argv = optimist
 	.default('wordwrap', 80)
 	.default('ignore-href', false)
 	.default('ignore-image', false)
+	.default('ignore-video', false)
 	.argv;
 
 var text = '';
@@ -25,7 +26,8 @@ process.stdin.on('end', function end() {
 		tables: interpretTables(argv.tables),
 		wordwrap: argv.wordwrap,
 		ignoreHref: argv['ignore-href'],
-		ignoreImage: argv['ignore-image']
+		ignoreImage: argv['ignore-image'],
+		ignoreVideo: argv['ignore-video']
 	});
 	process.stdout.write(text + '\n', 'utf-8');
 });

example/test.html

@@ -142,5 +142,19 @@ <h2>Pretty printed Source Code</h2>
 	console.log(text);
 });
 		</pre>
+
+		<hr />
+		<h2>Img tag</h2>
+		<img src="http://example.com/image.png" />
+
+		<hr />
+		<h2>Video tag (src attribute)</h2>
+		<video src="http://example.com/video-with-src-attribute.mp4"></video>
+
+		<hr />
+		<h2>Video tag (source tag)</h2>
+		<video>
+			<source src="http://example.com/video-with-source-tag.mp4">
+		</video>
 	</body>
-</html>
+</html>

lib/formatter.js

@@ -33,6 +33,27 @@ function formatImage(elem, options) {
 	return (result);
 }
 
+function formatVideo(elem, options) {
+	if (options.ignoreVideo) {
+	  return '';
+	}
+
+	var result = '', attribs = elem.attribs || {};
+
+	if (attribs.src) {
+		result += '[' + attribs.src + ']';
+	} else if (elem.children.length > 0) {
+		_.each(elem.children, function(el) {
+			var elAttribs = el.attribs || {};
+			if (el.type === 'tag' && el.name.toLowerCase() === 'source') {
+				result += '[' + el.attribs.src + ']';
+			}
+		});
+	}
+
+	return (result);
+}
+
 function formatLineBreak(elem, fn, options) {
 	return '\n' + fn(elem.children, options);
 }
@@ -219,6 +240,7 @@ function formatTable(elem, fn, options) {
 
 exports.text = formatText;
 exports.image = formatImage;
+exports.video = formatVideo;
 exports.lineBreak = formatLineBreak;
 exports.paragraph = formatParagraph;
 exports.anchor = formatAnchor;

lib/html-to-text.js

@@ -112,6 +112,9 @@ function walk(dom, options, result) {
 					case 'img':
 						result += format.image(elem, options);
 						break;
+					case 'video':
+						result += format.video(elem, options);
+						break;
 					case 'a':
 						// Inline element needs its leading space to be trimmed if `result`
 						// currently ends with whitespace