node-discount garbage

With a markdown document hello.md like so:

# hello, world

Repeated calls to markdown.parse produce different kinds of garbage at the end:

node> markdown.parse(fs.readFileSync('hello.md', 'utf8'));
'<h1>hello, world</h1>^?>\n '

node> markdown.parse(fs.readFileSync('hello.md', 'utf8'));
'<h1>hello, world</h1>'

node> markdown.parse(fs.readFileSync('hello.md', 'utf8'));
'<h1>hello, world</h1>{\u0007\ufffd\u000f\ufffd\u0015'

node> markdown.parse(fs.readFileSync('hello.md', 'utf8'));
'<h1>hello, world</h1>\u0002'

node> markdown.parse(fs.readFileSync('hello.md', 'utf8'));
'<h1>hello, world</h1>'

node> markdown.parse(fs.readFileSync('hello.md', 'utf8'));
'<h1>hello, world</h1>^?C\b\ufffd^?C\b\ufffd\u0001'