Improve description extraction and adapt unit tests

2 files changed, 12 insertions, 6 deletions

diff --git a/lib/page/html.js b/lib/page/html.js
index 5a13dcf..e8d3a85 100644
--- a/lib/page/html.js
+++ b/lib/page/html.js
@@ -184,12 +184,18 @@ HTMLPipeline.prototype.applyAnnotations = function() {
 // Extract page description from html
 // This can totally be improved
 HTMLPipeline.prototype.extractDescription = function() {
-    var $p = this.$('p');
+    var $ = this.$;
+    var $p = $('p').first();
+    var $next = $p.nextUntil('h1,h2,h3,h4,h5,h6,pre,blockquote,ul,ol,div');
 
     var description = $p.text().trim();
-    if (description.length > 300) {
-        description = description.slice(0, 300).trim()+'...';
-    }
+
+    $next.each(function() {
+        description += ' ' + $(this).text().trim();
+    });
+
+    // Truncate description
+    description = _.trunc(description, 300);
 
     this.opts.onDescription(description);
 };
diff --git a/test/page.js b/test/page.js
index 0fc59e7..ed2b777 100644
--- a/test/page.js
+++ b/test/page.js
@@ -8,7 +8,7 @@ describe('Page', function() {
         return mock.setupDefaultBook({
             'README.md': ' # Hello World\n\nThis is a description',
             'heading.md': '# Hello\n\n## World',
-            'description.md': '# This is a title\n\nThis is the short description.\n\nNot this one.',
+            'description.md': '# This is a title\n\nThis is the short description.\n\nAnd the rest of the description.\n\n# Heading\n\nThis is not in the description',
             'frontmatter/description.md': '---\ndescription: Hello World\n---\n\n# This is a title\n\nThis is not the description',
             'frontmatter/var.md': '---\ntest: Hello World\n---\n\n{{ page.test }}',
 
@@ -141,7 +141,7 @@ describe('Page', function() {
 
             return page.toHTML(output)
             .then(function() {
-                page.attributes.description.should.equal('This is the short description.');
+                page.attributes.description.should.equal('This is the short description. And the rest of the description.');
             });
         });
     });