Merge branch 'feature/blueprint' into develop

.gitignore

@@ -64,3 +64,4 @@ typings/
 .idea
 .project
 .settings
+*.pdf

cli.js

@@ -17,13 +17,15 @@ configure();
 
 function with_common_options(cmd) {
 	return cmd
+		.option('-b, --blueprint [blueprint]', 'Path to the blueprint')
 		.option('-o, --output [output]', 'Path for the generated bundle')
 		.option('--template [template]', 'Path to custom HTML template')
 		.option('--style [stylesheet]', 'Path to custom CSS')
 		.option('--css [style]', 'Additional CSS style')
 		.option('--individual', 'Export each web page as an individual file')
-		.option('--no-amp', "Don't prefer the AMP version of the web page")
-		.option('--toc', 'Generate a Table of Contents');
+		.option('--cover', 'Generate Cover')
+		.option('--toc', 'Generate TOC')
+		.option('--no-amp', "Don't prefer the AMP version of the web page");
 }
 
 program.version(pkg.version);

index.js

@@ -12,7 +12,7 @@ const Readability = require('./vendor/readability');
 const pkg = require('./package.json');
 
 const spinner = ora();
-
+const blueprints = require('./src/blueprints');
 const {
 	ampToHtml,
 	fixLazyLoadedImages,
@@ -65,23 +65,26 @@ function configure() {
 	Fetch a web page and clean the HTML
 	-----------------------------------
  */
-async function cleanup(url, options) {
+async function cleanup(item, blueprint) {
 	try {
-		spinner.start(`Fetching: ${url}`);
-		const content = (await got(url, {
+		spinner.start(`Fetching: ${item.url}`);
+		const content = (await got(item.url, {
 			headers: {
 				'user-agent': `percollate/${pkg.version}`
 			}
 		})).body;
 		spinner.succeed();
 
 		spinner.start('Enhancing web page');
-		const dom = createDom({ url, content });
+		const dom = createDom({ url: item.url, content });
 
 		const amp = dom.window.document.querySelector('link[rel=amphtml]');
-		if (amp && options.amp) {
+		if (amp && blueprint.options.amp) {
 			spinner.succeed('Found AMP version');
-			return cleanup(amp.href, options);
+			return cleanup(
+				Object.assign({}, item, { url: amp.href }),
+				blueprint
+			);
 		}
 
 		/* 
@@ -104,11 +107,12 @@ async function cleanup(url, options) {
 		}).parse();
 
 		spinner.succeed();
+
 		const _id = Math.random()
 			.toString(36)
 			.replace(/[^a-z]+/g, '')
 			.substr(2, 10);
-		return { ...parsed, _id, url };
+		return Object.assign({}, parsed, item, { _id: _id });
 	} catch (error) {
 		spinner.fail(error.message);
 		throw error;
@@ -119,24 +123,25 @@ async function cleanup(url, options) {
 	Bundle the HTML files into a PDF
 	--------------------------------
  */
-async function bundle(items, options) {
+async function bundle(blueprint) {
 	spinner.start('Generating temporary HTML file');
 	const temp_file = tmp.tmpNameSync({ postfix: '.html' });
 
-	const stylesheet = resolve(options.style || './templates/default.css');
-	const style = fs.readFileSync(stylesheet, 'utf8') + (options.css || '');
-	const generateToc = options.toc;
+	const stylesheet = resolve(blueprint.document.css);
+	let style = fs.readFileSync(stylesheet, 'utf8');
+	if (blueprint.cover.generate) {
+		style += fs.readFileSync(resolve(blueprint.cover.css), 'utf8');
+	}
+	if (blueprint.toc.generate) {
+		style += fs.readFileSync(resolve(blueprint.toc.css), 'utf8');
+	}
 
 	const html = nunjucks.renderString(
-		fs.readFileSync(
-			resolve(options.template || './templates/default.html'),
-			'utf8'
-		),
+		fs.readFileSync(resolve(blueprint.document.template), 'utf8'),
 		{
-			items,
+			items: blueprint.document.items,
 			style,
-			stylesheet, // deprecated
-			generateToc
+			blueprint
 		}
 	);
 
@@ -180,7 +185,11 @@ async function bundle(items, options) {
 
 	fs.writeFileSync(temp_file, html);
 
-	spinner.succeed(`Temporary HTML file: file://${temp_file}`);
+	spinner.succeed(
+		`Processed ${
+			blueprint.document.items.length
+		} items, temporary HTML file: file://${temp_file}`
+	);
 
 	spinner.start('Saving PDF');
 
@@ -190,7 +199,7 @@ async function bundle(items, options) {
 			Allow running with no sandbox
 			See: https://github.com/danburzo/percollate/issues/26
 		 */
-		args: options.sandbox
+		args: blueprint.options.sandbox
 			? undefined
 			: ['--no-sandbox', '--disable-setuid-sandbox'],
 		defaultViewport: {
@@ -213,9 +222,11 @@ async function bundle(items, options) {
 		in case we're bundling many web pages.
 	 */
 	const output_path =
-		options.output ||
-		(items.length === 1
-			? `${slugify(items[0].title || 'Untitled page')}.pdf`
+		blueprint.options.output ||
+		(blueprint.document.items.length === 1
+			? `${slugify(
+					blueprint.document.items[0].title || 'Untitled page'
+			  )}.pdf`
 			: `percollate-${Date.now()}.pdf`);
 
 	await page.pdf({
@@ -236,18 +247,25 @@ async function bundle(items, options) {
 	Generate PDF
  */
 async function pdf(urls, options) {
-	if (!urls.length) return;
-	let items = [];
-	for (let url of urls) {
-		let item = await cleanup(url, options);
-		if (options.individual) {
-			await bundle([item], options);
-		} else {
-			items.push(item);
-		}
+	const blueprint = blueprints.fromCommandLineOptions(urls, options);
+	if (!blueprint.document.items || !blueprint.document.items.length) {
+		return;
 	}
-	if (!options.individual) {
-		await bundle(items, options);
+	blueprint.document.items = await Promise.all(
+		blueprint.document.items.map(async function(item) {
+			return await cleanup(item, blueprint);
+		})
+	);
+	if (blueprint.options.individual) {
+		await Promise.all(
+			blueprint.document.items.map(async function(item) {
+				let itemBlueprint = Object.assign({}, blueprint);
+				itemBlueprint.document.items = [item];
+				await bundle(blueprint);
+			})
+		);
+	} else {
+		await bundle(blueprint);
 	}
 }
 

src/blueprints.js

@@ -0,0 +1,84 @@
+const fs = require('fs');
+
+function defaultBlueprint() {
+	return {
+		cover: _defaultCover(),
+		toc: _defaultToc(),
+		document: _defaultDocument(),
+		options: _defaultOptions()
+	};
+}
+
+function fromCommandLineOptions(urls, options) {
+	let blueprint = null;
+	if (options.blueprint) {
+		blueprint = JSON.parse(fs.readFileSync(options.blueprint, 'utf8'));
+	} else {
+		blueprint = _parseCommandLineOptions(options, defaultBlueprint());
+	}
+	if (urls && urls.length > 0) {
+		blueprint.document.items = urls.map(function(url) {
+			return { url: url };
+		});
+	}
+	return blueprint;
+}
+
+function _parseCommandLineOptions(options, blueprint) {
+	if (options.cover) {
+		blueprint.cover['generate'] = options.cover;
+	}
+	if (options.toc) {
+		blueprint.toc['generate'] = options.toc;
+	}
+	if (options.template) {
+		blueprint.document['template'] = options.template;
+	}
+	if (options.style) {
+		blueprint.document['css'] = options.style;
+	}
+	blueprint.options['sandbox'] = options.sandbox;
+	blueprint.options['output'] = options.output;
+	blueprint.options['individual'] = options.individual;
+	blueprint.options['no-amp'] = options.amp;
+	return blueprint;
+}
+
+function _defaultOptions() {
+	return {
+		output: 'percollate.pdf',
+		individual: false,
+		amp: true
+	};
+}
+
+function _defaultCover() {
+	return {
+		generate: false,
+		template: './templates/default_cover.html',
+		css: './templates/default_cover.css',
+		assets: {}
+	};
+}
+
+function _defaultToc() {
+	return {
+		generate: false,
+		template: './templates/default_toc.html',
+		css: './templates/default_toc.css',
+		assets: {}
+	};
+}
+
+function _defaultDocument() {
+	return {
+		template: './templates/default.html',
+		css: './templates/default.css',
+		assets: {},
+		items: []
+	};
+}
+
+module.exports = {
+	fromCommandLineOptions
+};

templates/default.css

@@ -84,10 +84,6 @@ pre code {
 	display: none;
 }
 
-nav.toc {
-	page-break-after: always;
-}
-
 /*
 	Article formatting
 	----------------------------------------------------

templates/default.html

@@ -10,15 +10,8 @@
 </head>
 <body>
 
-	{% if generateToc %}
-	<nav class="toc">
-		<h1>Items</h1>
-		<ol>
-			{% for item in items %}
-			<li class="toc__line"><a href="#{{ item._id }}">{{ item.title }}</a></li>
-			{% endfor %}
-		</ol>
-	</nav>
+	{% if blueprint.toc.generate %}
+	{% include blueprint.toc.template %}
 	{% endif %}
 
 	{% for item in items %}

templates/default_toc.css

@@ -0,0 +1,3 @@
+nav.toc {
+	page-break-after: always;
+}

templates/default_toc.html

@@ -0,0 +1,8 @@
+<nav class="toc">
+	<h1>Items</h1>
+	<ol>
+		{% for item in items %}
+		<li class="toc__line"><a href="#{{ item._id }}">{{ item.title }}</a></li>
+		{% endfor %}
+	</ol>
+</nav>