Add more classes structures

4 files changed, 134 insertions, 0 deletions

diff --git a/lib/parsers/asciidoc.js b/lib/parsers/asciidoc.js
new file mode 100644
index 0000000..84e619d
--- /dev/null
+++ b/lib/parsers/asciidoc.js
@@ -0,0 +1,27 @@
+var Asciidoctor = require('asciidoctor.js');
+var htmlParser = require('./html');
+
+var asciidoctor = Asciidoctor();
+var opal = asciidoctor.Opal;
+
+var processor = null;
+var useExtensions = true;
+
+if (useExtensions) {
+    processor = asciidoctor.Asciidoctor(true);
+} else {
+    processor = asciidoctor.Asciidoctor();
+}
+
+
+// Convert asciidoc to HTML
+function asciidocToHTML(content) {
+    var options = opal.hash2(['attributes'], {'attributes': 'showtitle'});
+    return processor.$convert(content, options);
+}
+
+
+module.exports = htmlParser.inherits({
+    extensions: ['.adoc', '.asciidoc'],
+    toHTML: asciidocToHTML
+});
diff --git a/lib/parsers/html.js b/lib/parsers/html.js
new file mode 100644
index 0000000..8f4ed34
--- /dev/null
+++ b/lib/parsers/html.js
@@ -0,0 +1,90 @@
+var _ = require('lodash');
+var cheerio = require('cheerio');
+
+// Parse summary and returns a list of sections
+function parseSummary(html) {
+    var sections = [];
+    var $ = cheerio.load(html);
+
+    // Find main container
+    var $body = getContainer($);
+
+    // Extract sections, and parse
+    var $lists = $body.find('> ul, > ol');
+
+    $lists.each(function() {
+        sections.push({
+            articles: parseList($(this), $)
+        });
+    });
+
+    return sections;
+}
+
+// Parse readme and extract title, description
+function parseReadme(html) {
+    var $ = cheerio.load(html);
+
+    // Find main container
+    var $body = getContainer($);
+
+    return {
+        title: $body.find('h1:first-child').text().trim(),
+        description: $body.find('div.paragraph').first().text().trim()
+    };
+}
+
+// Return a page container (html, body tag or directly the root element)
+function getContainer($) {
+    var $body = $('body, html').first();
+    if (!$body) $body = $;
+
+    return $body;
+}
+
+// Parse a ul list and return list of chapters recursvely
+function parseList($ul, $) {
+    var articles = [];
+
+    $ul.children('li').each(function() {
+        var article = {};
+
+        var $li = $(this);
+
+        var $text = $li.find('> p, > span');
+        var $a = $li.find('> a, > p a, > span a');
+
+        article.title = $text.text();
+        if ($a.length > 0) {
+            article.title = $a.first().text();
+            article.ref = $a.attr('href');
+        }
+
+        // Inner list, with children article
+        var $sub = $li.find('> ol, > ul, > .olist > ol');
+        article.articles = parseList($sub, $);
+
+        articles.push(article);
+    });
+
+    return articles;
+}
+
+
+// Inherit from the html parser
+function inherits(opts) {
+    var parser = _.defaults(opts, {
+        toHTML: _.identity
+    });
+
+    parser.readme = _.compose(opts.toHTML, parseReadme);
+    parser.summary = _.compose(opts.toHTML, parseSummary);
+
+    return parser;
+}
+
+
+module.exports = inherits({
+    extensions: ['.html']
+});
+module.exports.inherits = inherits;
diff --git a/lib/parsers/index.js b/lib/parsers/index.js
new file mode 100644
index 0000000..d650386
--- /dev/null
+++ b/lib/parsers/index.js
@@ -0,0 +1,11 @@
+
+var PARSERS = {
+    html: require('./html'),
+    markdown: require('./markdown'),
+    asciidoc: require('./asciidoc')
+};
+
+
+module.exports = {
+
+};
diff --git a/lib/parsers/markdown.js b/lib/parsers/markdown.js
new file mode 100644
index 0000000..aac2858
--- /dev/null
+++ b/lib/parsers/markdown.js
@@ -0,0 +1,6 @@
+
+
+
+module.exports = {
+    extensions: ['.md', '.markdown', '.mdown']
+};