This commit fixes two bugs that occur when using the // operator:
1) The first is that using any of the pluralized scrapers with a
selector created from one or more // operators will potentially
return duplicate tags. This occurs if there are multiple ways in
which a given tag may satisfy the selector.
2) The // operator does not force a decent from the current
context. For example, if one attempts to match a nested div
<div id=a><div id=b></div></div> with the selector
"div" // "div" the outer div will be matched.
In order to fix this bug without regressing on performance the entire
selection engine had to be re-written. The engine now features linear
time matching for tags as opposed to the probably exponential prior
implementation.
This commit fixes two bugs that occur when using the
//operator:1) The first is that using any of the pluralized scrapers with a selector created from one or more
//operators will potentially return duplicate tags. This occurs if there are multiple ways in which a given tag may satisfy the selector.2) The
//operator does not force a decent from the current context. For example, if one attempts to match a nested div<div id=a><div id=b></div></div>with the selector"div" // "div"the outer div will be matched.In order to fix this bug without regressing on performance the entire selection engine had to be re-written. The engine now features linear time matching for tags as opposed to the probably exponential prior implementation.