Skip lines that we don't care about.

The documents we're reading to memory can be huge, and in some cases
we only care about a small portion of it.  Rather than parsing all
lines, we can skip lines we don't care about and only parse the
remaining lines.