Show commit status icon in commits table (#1688)

* Show commit status icon in commits table

* Add comments

* Fix icons

* Few more places where commit table is displayed

* Change integration test to use goquery for parsing html

* Add integration tests for commit table and status icons

* Fix status to return lates status correctly on all databases

* Rewrote lates commit status selects

18 files changed, 4007 insertions, 0 deletions

diff --git a/vendor/github.com/PuerkitoBio/goquery/LICENSE b/vendor/github.com/PuerkitoBio/goquery/LICENSE
new file mode 100644
index 0000000000..f743d3728a
--- /dev/null
+++ b/vendor/github.com/PuerkitoBio/goquery/LICENSE
@@ -0,0 +1,12 @@
+Copyright (c) 2012-2016, Martin Angers & Contributors
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
+
+* Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
+
+* Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
+
+* Neither the name of the author nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
diff --git a/vendor/github.com/PuerkitoBio/goquery/README.md b/vendor/github.com/PuerkitoBio/goquery/README.md
new file mode 100644
index 0000000000..7b01a27178
--- /dev/null
+++ b/vendor/github.com/PuerkitoBio/goquery/README.md
@@ -0,0 +1,126 @@
+# goquery - a little like that j-thing, only in Go 
+[![build status](https://secure.travis-ci.org/PuerkitoBio/goquery.png)](http://travis-ci.org/PuerkitoBio/goquery) [![GoDoc](https://godoc.org/github.com/PuerkitoBio/goquery?status.png)](http://godoc.org/github.com/PuerkitoBio/goquery) [![Sourcegraph Badge](https://sourcegraph.com/github.com/PuerkitoBio/goquery/-/badge.svg)](https://sourcegraph.com/github.com/PuerkitoBio/goquery?badge)
+
+
+goquery brings a syntax and a set of features similar to [jQuery][] to the [Go language][go]. It is based on Go's [net/html package][html] and the CSS Selector library [cascadia][]. Since the net/html parser returns nodes, and not a full-featured DOM tree, jQuery's stateful manipulation functions (like height(), css(), detach()) have been left off.
+
+Also, because the net/html parser requires UTF-8 encoding, so does goquery: it is the caller's responsibility to ensure that the source document provides UTF-8 encoded HTML. See the [wiki][] for various options to do this.
+
+Syntax-wise, it is as close as possible to jQuery, with the same function names when possible, and that warm and fuzzy chainable interface. jQuery being the ultra-popular library that it is, I felt that writing a similar HTML-manipulating library was better to follow its API than to start anew (in the same spirit as Go's `fmt` package), even though some of its methods are less than intuitive (looking at you, [index()][index]...).
+
+## Installation
+
+Please note that because of the net/html dependency, goquery requires Go1.1+.
+
+    $ go get github.com/PuerkitoBio/goquery
+
+(optional) To run unit tests:
+
+    $ cd $GOPATH/src/github.com/PuerkitoBio/goquery
+    $ go test
+
+(optional) To run benchmarks (warning: it runs for a few minutes):
+
+    $ cd $GOPATH/src/github.com/PuerkitoBio/goquery
+    $ go test -bench=".*"
+
+## Changelog
+
+**Note that goquery's API is now stable, and will not break.**
+
+*    **2017-02-12 (v1.1.0)** : Add `SetHtml` and `SetText` (thanks to @glebtv).
+*    **2016-12-29 (v1.0.2)** : Optimize allocations for `Selection.Text` (thanks to @radovskyb).
+*    **2016-08-28 (v1.0.1)** : Optimize performance for large documents.
+*    **2016-07-27 (v1.0.0)** : Tag version 1.0.0.
+*    **2016-06-15** : Invalid selector strings internally compile to a `Matcher` implementation that never matches any node (instead of a panic). So for example, `doc.Find("~")` returns an empty `*Selection` object.
+*    **2016-02-02** : Add `NodeName` utility function similar to the DOM's `nodeName` property. It returns the tag name of the first element in a selection, and other relevant values of non-element nodes (see godoc for details). Add `OuterHtml` utility function similar to the DOM's `outerHTML` property (named `OuterHtml` in small caps for consistency with the existing `Html` method on the `Selection`).
+*    **2015-04-20** : Add `AttrOr` helper method to return the attribute's value or a default value if absent. Thanks to [piotrkowalczuk][piotr].
+*    **2015-02-04** : Add more manipulation functions - Prepend* - thanks again to [Andrew Stone][thatguystone].
+*    **2014-11-28** : Add more manipulation functions - ReplaceWith*, Wrap* and Unwrap - thanks again to [Andrew Stone][thatguystone].
+*    **2014-11-07** : Add manipulation functions (thanks to [Andrew Stone][thatguystone]) and `*Matcher` functions, that receive compiled cascadia selectors instead of selector strings, thus avoiding potential panics thrown by goquery via `cascadia.MustCompile` calls. This results in better performance (selectors can be compiled once and reused) and more idiomatic error handling (you can handle cascadia's compilation errors, instead of recovering from panics, which had been bugging me for a long time). Note that the actual type expected is a `Matcher` interface, that `cascadia.Selector` implements. Other matcher implementations could be used.
+*    **2014-11-06** : Change import paths of net/html to golang.org/x/net/html (see https://groups.google.com/forum/#!topic/golang-nuts/eD8dh3T9yyA). Make sure to update your code to use the new import path too when you call goquery with `html.Node`s.
+*    **v0.3.2** : Add `NewDocumentFromReader()` (thanks jweir) which allows creating a goquery document from an io.Reader.
+*    **v0.3.1** : Add `NewDocumentFromResponse()` (thanks assassingj) which allows creating a goquery document from an http response.
+*    **v0.3.0** : Add `EachWithBreak()` which allows to break out of an `Each()` loop by returning false. This function was added instead of changing the existing `Each()` to avoid breaking compatibility.
+*    **v0.2.1** : Make go-getable, now that [go.net/html is Go1.0-compatible][gonet] (thanks to @matrixik for pointing this out).
+*    **v0.2.0** : Add support for negative indices in Slice(). **BREAKING CHANGE** `Document.Root` is removed, `Document` is now a `Selection` itself (a selection of one, the root element, just like `Document.Root` was before). Add jQuery's Closest() method.
+*    **v0.1.1** : Add benchmarks to use as baseline for refactorings, refactor Next...() and Prev...() methods to use the new html package's linked list features (Next/PrevSibling, FirstChild). Good performance boost (40+% in some cases).
+*    **v0.1.0** : Initial release.
+
+## API
+
+goquery exposes two structs, `Document` and `Selection`, and the `Matcher` interface. Unlike jQuery, which is loaded as part of a DOM document, and thus acts on its containing document, goquery doesn't know which HTML document to act upon. So it needs to be told, and that's what the `Document` type is for. It holds the root document node as the initial Selection value to manipulate.
+
+jQuery often has many variants for the same function (no argument, a selector string argument, a jQuery object argument, a DOM element argument, ...). Instead of exposing the same features in goquery as a single method with variadic empty interface arguments, statically-typed signatures are used following this naming convention:
+
+*    When the jQuery equivalent can be called with no argument, it has the same name as jQuery for the no argument signature (e.g.: `Prev()`), and the version with a selector string argument is called `XxxFiltered()` (e.g.: `PrevFiltered()`)
+*    When the jQuery equivalent **requires** one argument, the same name as jQuery is used for the selector string version (e.g.: `Is()`)
+*    The signatures accepting a jQuery object as argument are defined in goquery as `XxxSelection()` and take a `*Selection` object as argument (e.g.: `FilterSelection()`)
+*    The signatures accepting a DOM element as argument in jQuery are defined in goquery as `XxxNodes()` and take a variadic argument of type `*html.Node` (e.g.: `FilterNodes()`)
+*    The signatures accepting a function as argument in jQuery are defined in goquery as `XxxFunction()` and take a function as argument (e.g.: `FilterFunction()`)
+*    The goquery methods that can be called with a selector string have a corresponding version that take a `Matcher` interface and are defined as `XxxMatcher()` (e.g.: `IsMatcher()`)
+
+Utility functions that are not in jQuery but are useful in Go are implemented as functions (that take a `*Selection` as parameter), to avoid a potential naming clash on the `*Selection`'s methods (reserved for jQuery-equivalent behaviour).
+
+The complete [godoc reference documentation can be found here][doc].
+
+Please note that Cascadia's selectors do not necessarily match all supported selectors of jQuery (Sizzle). See the [cascadia project][cascadia] for details. Invalid selector strings compile to a `Matcher` that fails to match any node. Behaviour of the various functions that take a selector string as argument follows from that fact, e.g. (where `~` is an invalid selector string):
+
+* `Find("~")` returns an empty selection because the selector string doesn't match anything.
+* `Add("~")` returns a new selection that holds the same nodes as the original selection, because it didn't add any node (selector string didn't match anything).
+* `ParentsFiltered("~")` returns an empty selection because the selector string doesn't match anything.
+* `ParentsUntil("~")` returns all parents of the selection because the selector string didn't match any element to stop before the top element.
+
+## Examples
+
+See some tips and tricks in the [wiki][].
+
+Adapted from example_test.go:
+
+```Go
+package main
+
+import (
+  "fmt"
+  "log"
+
+  "github.com/PuerkitoBio/goquery"
+)
+
+func ExampleScrape() {
+  doc, err := goquery.NewDocument("http://metalsucks.net")
+  if err != nil {
+    log.Fatal(err)
+  }
+
+  // Find the review items
+  doc.Find(".sidebar-reviews article .content-block").Each(func(i int, s *goquery.Selection) {
+    // For each item found, get the band and title
+    band := s.Find("a").Text()
+    title := s.Find("i").Text()
+    fmt.Printf("Review %d: %s - %s\n", i, band, title)
+  })
+}
+
+func main() {
+  ExampleScrape()
+}
+```
+
+## License
+
+The [BSD 3-Clause license][bsd], the same as the [Go language][golic]. Cascadia's license is [here][caslic].
+
+[jquery]: http://jquery.com/
+[go]: http://golang.org/
+[cascadia]: https://github.com/andybalholm/cascadia
+[bsd]: http://opensource.org/licenses/BSD-3-Clause
+[golic]: http://golang.org/LICENSE
+[caslic]: https://github.com/andybalholm/cascadia/blob/master/LICENSE
+[doc]: http://godoc.org/github.com/PuerkitoBio/goquery
+[index]: http://api.jquery.com/index/
+[gonet]: https://github.com/golang/net/
+[html]: http://godoc.org/golang.org/x/net/html
+[wiki]: https://github.com/PuerkitoBio/goquery/wiki/Tips-and-tricks
+[thatguystone]: https://github.com/thatguystone
+[piotr]: https://github.com/piotrkowalczuk
diff --git a/vendor/github.com/PuerkitoBio/goquery/array.go b/vendor/github.com/PuerkitoBio/goquery/array.go
new file mode 100644
index 0000000000..d7af5eee1c
--- /dev/null
+++ b/vendor/github.com/PuerkitoBio/goquery/array.go
@@ -0,0 +1,103 @@
+package goquery
+
+import (
+	"golang.org/x/net/html"
+)
+
+// First reduces the set of matched elements to the first in the set.
+// It returns a new Selection object, and an empty Selection object if the
+// the selection is empty.
+func (s *Selection) First() *Selection {
+	return s.Eq(0)
+}
+
+// Last reduces the set of matched elements to the last in the set.
+// It returns a new Selection object, and an empty Selection object if
+// the selection is empty.
+func (s *Selection) Last() *Selection {
+	return s.Eq(-1)
+}
+
+// Eq reduces the set of matched elements to the one at the specified index.
+// If a negative index is given, it counts backwards starting at the end of the
+// set. It returns a new Selection object, and an empty Selection object if the
+// index is invalid.
+func (s *Selection) Eq(index int) *Selection {
+	if index < 0 {
+		index += len(s.Nodes)
+	}
+
+	if index >= len(s.Nodes) || index < 0 {
+		return newEmptySelection(s.document)
+	}
+
+	return s.Slice(index, index+1)
+}
+
+// Slice reduces the set of matched elements to a subset specified by a range
+// of indices.
+func (s *Selection) Slice(start, end int) *Selection {
+	if start < 0 {
+		start += len(s.Nodes)
+	}
+	if end < 0 {
+		end += len(s.Nodes)
+	}
+	return pushStack(s, s.Nodes[start:end])
+}
+
+// Get retrieves the underlying node at the specified index.
+// Get without parameter is not implemented, since the node array is available
+// on the Selection object.
+func (s *Selection) Get(index int) *html.Node {
+	if index < 0 {
+		index += len(s.Nodes) // Negative index gets from the end
+	}
+	return s.Nodes[index]
+}
+
+// Index returns the position of the first element within the Selection object
+// relative to its sibling elements.
+func (s *Selection) Index() int {
+	if len(s.Nodes) > 0 {
+		return newSingleSelection(s.Nodes[0], s.document).PrevAll().Length()
+	}
+	return -1
+}
+
+// IndexSelector returns the position of the first element within the
+// Selection object relative to the elements matched by the selector, or -1 if
+// not found.
+func (s *Selection) IndexSelector(selector string) int {
+	if len(s.Nodes) > 0 {
+		sel := s.document.Find(selector)
+		return indexInSlice(sel.Nodes, s.Nodes[0])
+	}
+	return -1
+}
+
+// IndexMatcher returns the position of the first element within the
+// Selection object relative to the elements matched by the matcher, or -1 if
+// not found.
+func (s *Selection) IndexMatcher(m Matcher) int {
+	if len(s.Nodes) > 0 {
+		sel := s.document.FindMatcher(m)
+		return indexInSlice(sel.Nodes, s.Nodes[0])
+	}
+	return -1
+}
+
+// IndexOfNode returns the position of the specified node within the Selection
+// object, or -1 if not found.
+func (s *Selection) IndexOfNode(node *html.Node) int {
+	return indexInSlice(s.Nodes, node)
+}
+
+// IndexOfSelection returns the position of the first node in the specified
+// Selection object within this Selection object, or -1 if not found.
+func (s *Selection) IndexOfSelection(sel *Selection) int {
+	if sel != nil && len(sel.Nodes) > 0 {
+		return indexInSlice(s.Nodes, sel.Nodes[0])
+	}
+	return -1
+}
diff --git a/vendor/github.com/PuerkitoBio/goquery/doc.go b/vendor/github.com/PuerkitoBio/goquery/doc.go
new file mode 100644
index 0000000000..71146a7802
--- /dev/null
+++ b/vendor/github.com/PuerkitoBio/goquery/doc.go
@@ -0,0 +1,123 @@
+// Copyright (c) 2012-2016, Martin Angers & Contributors
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+// * Redistributions of source code must retain the above copyright notice,
+// this list of conditions and the following disclaimer.
+// * Redistributions in binary form must reproduce the above copyright notice,
+// this list of conditions and the following disclaimer in the documentation and/or
+// other materials provided with the distribution.
+// * Neither the name of the author nor the names of its contributors may be used to
+// endorse or promote products derived from this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS
+// OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY
+// AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+// DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
+// WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY
+// WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+/*
+Package goquery implements features similar to jQuery, including the chainable
+syntax, to manipulate and query an HTML document.
+
+It brings a syntax and a set of features similar to jQuery to the Go language.
+It is based on Go's net/html package and the CSS Selector library cascadia.
+Since the net/html parser returns nodes, and not a full-featured DOM
+tree, jQuery's stateful manipulation functions (like height(), css(), detach())
+have been left off.
+
+Also, because the net/html parser requires UTF-8 encoding, so does goquery: it is
+the caller's responsibility to ensure that the source document provides UTF-8 encoded HTML.
+See the repository's wiki for various options on how to do this.
+
+Syntax-wise, it is as close as possible to jQuery, with the same method names when
+possible, and that warm and fuzzy chainable interface. jQuery being the
+ultra-popular library that it is, writing a similar HTML-manipulating
+library was better to follow its API than to start anew (in the same spirit as
+Go's fmt package), even though some of its methods are less than intuitive (looking
+at you, index()...).
+
+It is hosted on GitHub, along with additional documentation in the README.md
+file: https://github.com/puerkitobio/goquery
+
+Please note that because of the net/html dependency, goquery requires Go1.1+.
+
+The various methods are split into files based on the category of behavior.
+The three dots (...) indicate that various "overloads" are available.
+
+* array.go : array-like positional manipulation of the selection.
+    - Eq()
+    - First()
+    - Get()
+    - Index...()
+    - Last()
+    - Slice()
+
+* expand.go : methods that expand or augment the selection's set.
+    - Add...()
+    - AndSelf()
+    - Union(), which is an alias for AddSelection()
+
+* filter.go : filtering methods, that reduce the selection's set.
+    - End()
+    - Filter...()
+    - Has...()
+    - Intersection(), which is an alias of FilterSelection()
+    - Not...()
+
+* iteration.go : methods to loop over the selection's nodes.
+    - Each()
+    - EachWithBreak()
+    - Map()
+
+* manipulation.go : methods for modifying the document
+    - After...()
+    - Append...()
+    - Before...()
+    - Clone()
+    - Empty()
+    - Prepend...()
+    - Remove...()
+    - ReplaceWith...()
+    - Unwrap()
+    - Wrap...()
+    - WrapAll...()
+    - WrapInner...()
+
+* property.go : methods that inspect and get the node's properties values.
+    - Attr*(), RemoveAttr(), SetAttr()
+    - AddClass(), HasClass(), RemoveClass(), ToggleClass()
+    - Html()
+    - Length()
+    - Size(), which is an alias for Length()
+    - Text()
+
+* query.go : methods that query, or reflect, a node's identity.
+    - Contains()
+    - Is...()
+
+* traversal.go : methods to traverse the HTML document tree.
+    - Children...()
+    - Contents()
+    - Find...()
+    - Next...()
+    - Parent[s]...()
+    - Prev...()
+    - Siblings...()
+
+* type.go : definition of the types exposed by goquery.
+    - Document
+    - Selection
+    - Matcher
+
+* utilities.go : definition of helper functions (and not methods on a *Selection)
+that are not part of jQuery, but are useful to goquery.
+    - NodeName
+    - OuterHtml
+*/
+package goquery
diff --git a/vendor/github.com/PuerkitoBio/goquery/expand.go b/vendor/github.com/PuerkitoBio/goquery/expand.go
new file mode 100644
index 0000000000..f0c6c86d8c
--- /dev/null
+++ b/vendor/github.com/PuerkitoBio/goquery/expand.go
@@ -0,0 +1,46 @@
+package goquery
+
+import "golang.org/x/net/html"
+
+// Add adds the selector string's matching nodes to those in the current
+// selection and returns a new Selection object.
+// The selector string is run in the context of the document of the current
+// Selection object.
+func (s *Selection) Add(selector string) *Selection {
+	return s.AddNodes(findWithMatcher([]*html.Node{s.document.rootNode}, compileMatcher(selector))...)
+}
+
+// AddMatcher adds the matcher's matching nodes to those in the current
+// selection and returns a new Selection object.
+// The matcher is run in the context of the document of the current
+// Selection object.
+func (s *Selection) AddMatcher(m Matcher) *Selection {
+	return s.AddNodes(findWithMatcher([]*html.Node{s.document.rootNode}, m)...)
+}
+
+// AddSelection adds the specified Selection object's nodes to those in the
+// current selection and returns a new Selection object.
+func (s *Selection) AddSelection(sel *Selection) *Selection {
+	if sel == nil {
+		return s.AddNodes()
+	}
+	return s.AddNodes(sel.Nodes...)
+}
+
+// Union is an alias for AddSelection.
+func (s *Selection) Union(sel *Selection) *Selection {
+	return s.AddSelection(sel)
+}
+
+// AddNodes adds the specified nodes to those in the
+// current selection and returns a new Selection object.
+func (s *Selection) AddNodes(nodes ...*html.Node) *Selection {
+	return pushStack(s, appendWithoutDuplicates(s.Nodes, nodes, nil))
+}
+
+// AndSelf adds the previous set of elements on the stack to the current set.
+// It returns a new Selection object containing the current Selection combined
+// with the previous one.
+func (s *Selection) AndSelf() *Selection {
+	return s.AddSelection(s.prevSel)
+}
diff --git a/vendor/github.com/PuerkitoBio/goquery/filter.go b/vendor/github.com/PuerkitoBio/goquery/filter.go
new file mode 100644
index 0000000000..9138ffb33e
--- /dev/null
+++ b/vendor/github.com/PuerkitoBio/goquery/filter.go
@@ -0,0 +1,163 @@
+package goquery
+
+import "golang.org/x/net/html"
+
+// Filter reduces the set of matched elements to those that match the selector string.
+// It returns a new Selection object for this subset of matching elements.
+func (s *Selection) Filter(selector string) *Selection {
+	return s.FilterMatcher(compileMatcher(selector))
+}
+
+// FilterMatcher reduces the set of matched elements to those that match
+// the given matcher. It returns a new Selection object for this subset
+// of matching elements.
+func (s *Selection) FilterMatcher(m Matcher) *Selection {
+	return pushStack(s, winnow(s, m, true))
+}
+
+// Not removes elements from the Selection that match the selector string.
+// It returns a new Selection object with the matching elements removed.
+func (s *Selection) Not(selector string) *Selection {
+	return s.NotMatcher(compileMatcher(selector))
+}
+
+// NotMatcher removes elements from the Selection that match the given matcher.
+// It returns a new Selection object with the matching elements removed.
+func (s *Selection) NotMatcher(m Matcher) *Selection {
+	return pushStack(s, winnow(s, m, false))
+}
+
+// FilterFunction reduces the set of matched elements to those that pass the function's test.
+// It returns a new Selection object for this subset of elements.
+func (s *Selection) FilterFunction(f func(int, *Selection) bool) *Selection {
+	return pushStack(s, winnowFunction(s, f, true))
+}
+
+// NotFunction removes elements from the Selection that pass the function's test.
+// It returns a new Selection object with the matching elements removed.
+func (s *Selection) NotFunction(f func(int, *Selection) bool) *Selection {
+	return pushStack(s, winnowFunction(s, f, false))
+}
+
+// FilterNodes reduces the set of matched elements to those that match the specified nodes.
+// It returns a new Selection object for this subset of elements.
+func (s *Selection) FilterNodes(nodes ...*html.Node) *Selection {
+	return pushStack(s, winnowNodes(s, nodes, true))
+}
+
+// NotNodes removes elements from the Selection that match the specified nodes.
+// It returns a new Selection object with the matching elements removed.
+func (s *Selection) NotNodes(nodes ...*html.Node) *Selection {
+	return pushStack(s, winnowNodes(s, nodes, false))
+}
+
+// FilterSelection reduces the set of matched elements to those that match a
+// node in the specified Selection object.
+// It returns a new Selection object for this subset of elements.
+func (s *Selection) FilterSelection(sel *Selection) *Selection {
+	if sel == nil {
+		return pushStack(s, winnowNodes(s, nil, true))
+	}
+	return pushStack(s, winnowNodes(s, sel.Nodes, true))
+}
+
+// NotSelection removes elements from the Selection that match a node in the specified
+// Selection object. It returns a new Selection object with the matching elements removed.
+func (s *Selection) NotSelection(sel *Selection) *Selection {
+	if sel == nil {
+		return pushStack(s, winnowNodes(s, nil, false))
+	}
+	return pushStack(s, winnowNodes(s, sel.Nodes, false))
+}
+
+// Intersection is an alias for FilterSelection.
+func (s *Selection) Intersection(sel *Selection) *Selection {
+	return s.FilterSelection(sel)
+}
+
+// Has reduces the set of matched elements to those that have a descendant
+// that matches the selector.
+// It returns a new Selection object with the matching elements.
+func (s *Selection) Has(selector string) *Selection {
+	return s.HasSelection(s.document.Find(selector))
+}
+
+// HasMatcher reduces the set of matched elements to those that have a descendant
+// that matches the matcher.
+// It returns a new Selection object with the matching elements.
+func (s *Selection) HasMatcher(m Matcher) *Selection {
+	return s.HasSelection(s.document.FindMatcher(m))
+}
+
+// HasNodes reduces the set of matched elements to those that have a
+// descendant that matches one of the nodes.
+// It returns a new Selection object with the matching elements.
+func (s *Selection) HasNodes(nodes ...*html.Node) *Selection {
+	return s.FilterFunction(func(_ int, sel *Selection) bool {
+		// Add all nodes that contain one of the specified nodes
+		for _, n := range nodes {
+			if sel.Contains(n) {
+				return true
+			}
+		}
+		return false
+	})
+}
+
+// HasSelection reduces the set of matched elements to those that have a
+// descendant that matches one of the nodes of the specified Selection object.
+// It returns a new Selection object with the matching elements.
+func (s *Selection) HasSelection(sel *Selection) *Selection {
+	if sel == nil {
+		return s.HasNodes()
+	}
+	return s.HasNodes(sel.Nodes...)
+}
+
+// End ends the most recent filtering operation in the current chain and
+// returns the set of matched elements to its previous state.
+func (s *Selection) End() *Selection {
+	if s.prevSel != nil {
+		return s.prevSel
+	}
+	return newEmptySelection(s.document)
+}
+
+// Filter based on the matcher, and the indicator to keep (Filter) or
+// to get rid of (Not) the matching elements.
+func winnow(sel *Selection, m Matcher, keep bool) []*html.Node {
+	// Optimize if keep is requested
+	if keep {
+		return m.Filter(sel.Nodes)
+	}
+	// Use grep
+	return grep(sel, func(i int, s *Selection) bool {
+		return !m.Match(s.Get(0))
+	})
+}
+
+// Filter based on an array of nodes, and the indicator to keep (Filter) or
+// to get rid of (Not) the matching elements.
+func winnowNodes(sel *Selection, nodes []*html.Node, keep bool) []*html.Node {
+	if len(nodes)+len(sel.Nodes) < minNodesForSet {
+		return grep(sel, func(i int, s *Selection) bool {
+			return isInSlice(nodes, s.Get(0)) == keep
+		})
+	}
+
+	set := make(map[*html.Node]bool)
+	for _, n := range nodes {
+		set[n] = true
+	}
+	return grep(sel, func(i int, s *Selection) bool {
+		return set[s.Get(0)] == keep
+	})
+}
+
+// Filter based on a function test, and the indicator to keep (Filter) or
+// to get rid of (Not) the matching elements.
+func winnowFunction(sel *Selection, f func(int, *Selection) bool, keep bool) []*html.Node {
+	return grep(sel, func(i int, s *Selection) bool {
+		return f(i, s) == keep
+	})
+}
diff --git a/vendor/github.com/PuerkitoBio/goquery/iteration.go b/vendor/github.com/PuerkitoBio/goquery/iteration.go
new file mode 100644
index 0000000000..e246f2e0e6
--- /dev/null
+++ b/vendor/github.com/PuerkitoBio/goquery/iteration.go
@@ -0,0 +1,39 @@
+package goquery
+
+// Each iterates over a Selection object, executing a function for each
+// matched element. It returns the current Selection object. The function
+// f is called for each element in the selection with the index of the
+// element in that selection starting at 0, and a *Selection that contains
+// only that element.
+func (s *Selection) Each(f func(int, *Selection)) *Selection {
+	for i, n := range s.Nodes {
+		f(i, newSingleSelection(n, s.document))
+	}
+	return s
+}
+
+// EachWithBreak iterates over a Selection object, executing a function for each
+// matched element. It is identical to Each except that it is possible to break
+// out of the loop by returning false in the callback function. It returns the
+// current Selection object.
+func (s *Selection) EachWithBreak(f func(int, *Selection) bool) *Selection {
+	for i, n := range s.Nodes {
+		if !f(i, newSingleSelection(n, s.document)) {
+			return s
+		}
+	}
+	return s
+}
+
+// Map passes each element in the current matched set through a function,
+// producing a slice of string holding the returned values. The function
+// f is called for each element in the selection with the index of the
+// element in that selection starting at 0, and a *Selection that contains
+// only that element.
+func (s *Selection) Map(f func(int, *Selection) string) (result []string) {
+	for i, n := range s.Nodes {
+		result = append(result, f(i, newSingleSelection(n, s.document)))
+	}
+
+	return result
+}
diff --git a/vendor/github.com/PuerkitoBio/goquery/manipulation.go b/vendor/github.com/PuerkitoBio/goquery/manipulation.go
new file mode 100644
index 0000000000..ebb4ffe128
--- /dev/null
+++ b/vendor/github.com/PuerkitoBio/goquery/manipulation.go
@@ -0,0 +1,573 @@
+package goquery
+
+import (
+	"strings"
+
+	"golang.org/x/net/html"
+)
+
+// After applies the selector from the root document and inserts the matched elements
+// after the elements in the set of matched elements.
+//
+// If one of the matched elements in the selection is not currently in the
+// document, it's impossible to insert nodes after it, so it will be ignored.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) After(selector string) *Selection {
+	return s.AfterMatcher(compileMatcher(selector))
+}
+
+// AfterMatcher applies the matcher from the root document and inserts the matched elements
+// after the elements in the set of matched elements.
+//
+// If one of the matched elements in the selection is not currently in the
+// document, it's impossible to insert nodes after it, so it will be ignored.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) AfterMatcher(m Matcher) *Selection {
+	return s.AfterNodes(m.MatchAll(s.document.rootNode)...)
+}
+
+// AfterSelection inserts the elements in the selection after each element in the set of matched
+// elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) AfterSelection(sel *Selection) *Selection {
+	return s.AfterNodes(sel.Nodes...)
+}
+
+// AfterHtml parses the html and inserts it after the set of matched elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) AfterHtml(html string) *Selection {
+	return s.AfterNodes(parseHtml(html)...)
+}
+
+// AfterNodes inserts the nodes after each element in the set of matched elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) AfterNodes(ns ...*html.Node) *Selection {
+	return s.manipulateNodes(ns, true, func(sn *html.Node, n *html.Node) {
+		if sn.Parent != nil {
+			sn.Parent.InsertBefore(n, sn.NextSibling)
+		}
+	})
+}
+
+// Append appends the elements specified by the selector to the end of each element
+// in the set of matched elements, following those rules:
+//
+// 1) The selector is applied to the root document.
+//
+// 2) Elements that are part of the document will be moved to the new location.
+//
+// 3) If there are multiple locations to append to, cloned nodes will be
+// appended to all target locations except the last one, which will be moved
+// as noted in (2).
+func (s *Selection) Append(selector string) *Selection {
+	return s.AppendMatcher(compileMatcher(selector))
+}
+
+// AppendMatcher appends the elements specified by the matcher to the end of each element
+// in the set of matched elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) AppendMatcher(m Matcher) *Selection {
+	return s.AppendNodes(m.MatchAll(s.document.rootNode)...)
+}
+
+// AppendSelection appends the elements in the selection to the end of each element
+// in the set of matched elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) AppendSelection(sel *Selection) *Selection {
+	return s.AppendNodes(sel.Nodes...)
+}
+
+// AppendHtml parses the html and appends it to the set of matched elements.
+func (s *Selection) AppendHtml(html string) *Selection {
+	return s.AppendNodes(parseHtml(html)...)
+}
+
+// AppendNodes appends the specified nodes to each node in the set of matched elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) AppendNodes(ns ...*html.Node) *Selection {
+	return s.manipulateNodes(ns, false, func(sn *html.Node, n *html.Node) {
+		sn.AppendChild(n)
+	})
+}
+
+// Before inserts the matched elements before each element in the set of matched elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) Before(selector string) *Selection {
+	return s.BeforeMatcher(compileMatcher(selector))
+}
+
+// BeforeMatcher inserts the matched elements before each element in the set of matched elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) BeforeMatcher(m Matcher) *Selection {
+	return s.BeforeNodes(m.MatchAll(s.document.rootNode)...)
+}
+
+// BeforeSelection inserts the elements in the selection before each element in the set of matched
+// elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) BeforeSelection(sel *Selection) *Selection {
+	return s.BeforeNodes(sel.Nodes...)
+}
+
+// BeforeHtml parses the html and inserts it before the set of matched elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) BeforeHtml(html string) *Selection {
+	return s.BeforeNodes(parseHtml(html)...)
+}
+
+// BeforeNodes inserts the nodes before each element in the set of matched elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) BeforeNodes(ns ...*html.Node) *Selection {
+	return s.manipulateNodes(ns, false, func(sn *html.Node, n *html.Node) {
+		if sn.Parent != nil {
+			sn.Parent.InsertBefore(n, sn)
+		}
+	})
+}
+
+// Clone creates a deep copy of the set of matched nodes. The new nodes will not be
+// attached to the document.
+func (s *Selection) Clone() *Selection {
+	ns := newEmptySelection(s.document)
+	ns.Nodes = cloneNodes(s.Nodes)
+	return ns
+}
+
+// Empty removes all children nodes from the set of matched elements.
+// It returns the children nodes in a new Selection.
+func (s *Selection) Empty() *Selection {
+	var nodes []*html.Node
+
+	for _, n := range s.Nodes {
+		for c := n.FirstChild; c != nil; c = n.FirstChild {
+			n.RemoveChild(c)
+			nodes = append(nodes, c)
+		}
+	}
+
+	return pushStack(s, nodes)
+}
+
+// Prepend prepends the elements specified by the selector to each element in
+// the set of matched elements, following the same rules as Append.
+func (s *Selection) Prepend(selector string) *Selection {
+	return s.PrependMatcher(compileMatcher(selector))
+}
+
+// PrependMatcher prepends the elements specified by the matcher to each
+// element in the set of matched elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) PrependMatcher(m Matcher) *Selection {
+	return s.PrependNodes(m.MatchAll(s.document.rootNode)...)
+}
+
+// PrependSelection prepends the elements in the selection to each element in
+// the set of matched elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) PrependSelection(sel *Selection) *Selection {
+	return s.PrependNodes(sel.Nodes...)
+}
+
+// PrependHtml parses the html and prepends it to the set of matched elements.
+func (s *Selection) PrependHtml(html string) *Selection {
+	return s.PrependNodes(parseHtml(html)...)
+}
+
+// PrependNodes prepends the specified nodes to each node in the set of
+// matched elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) PrependNodes(ns ...*html.Node) *Selection {
+	return s.manipulateNodes(ns, true, func(sn *html.Node, n *html.Node) {
+		// sn.FirstChild may be nil, in which case this functions like
+		// sn.AppendChild()
+		sn.InsertBefore(n, sn.FirstChild)
+	})
+}
+
+// Remove removes the set of matched elements from the document.
+// It returns the same selection, now consisting of nodes not in the document.
+func (s *Selection) Remove() *Selection {
+	for _, n := range s.Nodes {
+		if n.Parent != nil {
+			n.Parent.RemoveChild(n)
+		}
+	}
+
+	return s
+}
+
+// RemoveFiltered removes the set of matched elements by selector.
+// It returns the Selection of removed nodes.
+func (s *Selection) RemoveFiltered(selector string) *Selection {
+	return s.RemoveMatcher(compileMatcher(selector))
+}
+
+// RemoveMatcher removes the set of matched elements.
+// It returns the Selection of removed nodes.
+func (s *Selection) RemoveMatcher(m Matcher) *Selection {
+	return s.FilterMatcher(m).Remove()
+}
+
+// ReplaceWith replaces each element in the set of matched elements with the
+// nodes matched by the given selector.
+// It returns the removed elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) ReplaceWith(selector string) *Selection {
+	return s.ReplaceWithMatcher(compileMatcher(selector))
+}
+
+// ReplaceWithMatcher replaces each element in the set of matched elements with
+// the nodes matched by the given Matcher.
+// It returns the removed elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) ReplaceWithMatcher(m Matcher) *Selection {
+	return s.ReplaceWithNodes(m.MatchAll(s.document.rootNode)...)
+}
+
+// ReplaceWithSelection replaces each element in the set of matched elements with
+// the nodes from the given Selection.
+// It returns the removed elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) ReplaceWithSelection(sel *Selection) *Selection {
+	return s.ReplaceWithNodes(sel.Nodes...)
+}
+
+// ReplaceWithHtml replaces each element in the set of matched elements with
+// the parsed HTML.
+// It returns the removed elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) ReplaceWithHtml(html string) *Selection {
+	return s.ReplaceWithNodes(parseHtml(html)...)
+}
+
+// ReplaceWithNodes replaces each element in the set of matched elements with
+// the given nodes.
+// It returns the removed elements.
+//
+// This follows the same rules as Selection.Append.
+func (s *Selection) ReplaceWithNodes(ns ...*html.Node) *Selection {
+	s.AfterNodes(ns...)
+	return s.Remove()
+}
+
+// Set the html content of each element in the selection to specified html string.
+func (s *Selection) SetHtml(html string) *Selection {
+	return setHtmlNodes(s, parseHtml(html)...)
+}
+
+// Set the content of each element in the selection to specified content. The
+// provided text string is escaped.
+func (s *Selection) SetText(text string) *Selection {
+	return s.SetHtml(html.EscapeString(text))
+}
+
+// Unwrap removes the parents of the set of matched elements, leaving the matched
+// elements (and their siblings, if any) in their place.
+// It returns the original selection.
+func (s *Selection) Unwrap() *Selection {
+	s.Parent().Each(func(i int, ss *Selection) {
+		// For some reason, jquery allows unwrap to remove the <head> element, so
+		// allowing it here too. Same for <html>. Why it allows those elements to
+		// be unwrapped while not allowing body is a mystery to me.
+		if ss.Nodes[0].Data != "body" {
+			ss.ReplaceWithSelection(ss.Contents())
+		}
+	})
+
+	return s
+}
+
+// Wrap wraps each element in the set of matched elements inside the first
+// element matched by the given selector. The matched child is cloned before
+// being inserted into the document.
+//
+// It returns the original set of elements.
+func (s *Selection) Wrap(selector string) *Selection {
+	return s.WrapMatcher(compileMatcher(selector))
+}
+
+// WrapMatcher wraps each element in the set of matched elements inside the
+// first element matched by the given matcher. The matched child is cloned
+// before being inserted into the document.
+//
+// It returns the original set of elements.
+func (s *Selection) WrapMatcher(m Matcher) *Selection {
+	return s.wrapNodes(m.MatchAll(s.document.rootNode)...)
+}
+
+// WrapSelection wraps each element in the set of matched elements inside the
+// first element in the given Selection. The element is cloned before being
+// inserted into the document.
+//
+// It returns the original set of elements.
+func (s *Selection) WrapSelection(sel *Selection) *Selection {
+	return s.wrapNodes(sel.Nodes...)
+}
+
+// WrapHtml wraps each element in the set of matched elements inside the inner-
+// most child of the given HTML.
+//
+// It returns the original set of elements.
+func (s *Selection) WrapHtml(html string) *Selection {
+	return s.wrapNodes(parseHtml(html)...)
+}
+
+// WrapNode wraps each element in the set of matched elements inside the inner-
+// most child of the given node. The given node is copied before being inserted
+// into the document.
+//
+// It returns the original set of elements.
+func (s *Selection) WrapNode(n *html.Node) *Selection {
+	return s.wrapNodes(n)
+}
+
+func (s *Selection) wrapNodes(ns ...*html.Node) *Selection {
+	s.Each(func(i int, ss *Selection) {
+		ss.wrapAllNodes(ns...)
+	})
+
+	return s
+}
+
+// WrapAll wraps a single HTML structure, matched by the given selector, around
+// all elements in the set of matched elements. The matched child is cloned
+// before being inserted into the document.
+//
+// It returns the original set of elements.
+func (s *Selection) WrapAll(selector string) *Selection {
+	return s.WrapAllMatcher(compileMatcher(selector))
+}
+
+// WrapAllMatcher wraps a single HTML structure, matched by the given Matcher,
+// around all elements in the set of matched elements. The matched child is
+// cloned before being inserted into the document.
+//
+// It returns the original set of elements.
+func (s *Selection) WrapAllMatcher(m Matcher) *Selection {
+	return s.wrapAllNodes(m.MatchAll(s.document.rootNode)...)
+}
+
+// WrapAllSelection wraps a single HTML structure, the first node of the given
+// Selection, around all elements in the set of matched elements. The matched
+// child is cloned before being inserted into the document.
+//
+// It returns the original set of elements.
+func (s *Selection) WrapAllSelection(sel *Selection) *Selection {
+	return s.wrapAllNodes(sel.Nodes...)
+}
+
+// WrapAllHtml wraps the given HTML structure around all elements in the set of
+// matched elements. The matched child is cloned before being inserted into the
+// document.
+//
+// It returns the original set of elements.
+func (s *Selection) WrapAllHtml(html string) *Selection {
+	return s.wrapAllNodes(parseHtml(html)...)
+}
+
+func (s *Selection) wrapAllNodes(ns ...*html.Node) *Selection {
+	if len(ns) > 0 {
+		return s.WrapAllNode(ns[0])
+	}
+	return s
+}
+
+// WrapAllNode wraps the given node around the first element in the Selection,
+// making all other nodes in the Selection children of the given node. The node
+// is cloned before being inserted into the document.
+//
+// It returns the original set of elements.
+func (s *Selection) WrapAllNode(n *html.Node) *Selection {
+	if s.Size() == 0 {
+		return s
+	}
+
+	wrap := cloneNode(n)
+
+	first := s.Nodes[0]
+	if first.Parent != nil {
+		first.Parent.InsertBefore(wrap, first)
+		first.Parent.RemoveChild(first)
+	}
+
+	for c := getFirstChildEl(wrap); c != nil; c = getFirstChildEl(wrap) {
+		wrap = c
+	}
+
+	newSingleSelection(wrap, s.document).AppendSelection(s)
+
+	return s
+}
+
+// WrapInner wraps an HTML structure, matched by the given selector, around the
+// content of element in the set of matched elements. The matched child is
+// cloned before being inserted into the document.
+//
+// It returns the original set of elements.
+func (s *Selection) WrapInner(selector string) *Selection {
+	return s.WrapInnerMatcher(compileMatcher(selector))
+}
+
+// WrapInnerMatcher wraps an HTML structure, matched by the given selector,
+// around the content of element in the set of matched elements. The matched
+// child is cloned before being inserted into the document.
+//
+// It returns the original set of elements.
+func (s *Selection) WrapInnerMatcher(m Matcher) *Selection {
+	return s.wrapInnerNodes(m.MatchAll(s.document.rootNode)...)
+}
+
+// WrapInnerSelection wraps an HTML structure, matched by the given selector,
+// around the content of element in the set of matched elements. The matched
+// child is cloned before being inserted into the document.
+//
+// It returns the original set of elements.
+func (s *Selection) WrapInnerSelection(sel *Selection) *Selection {
+	return s.wrapInnerNodes(sel.Nodes...)
+}
+
+// WrapInnerHtml wraps an HTML structure, matched by the given selector, around
+// the content of element in the set of matched elements. The matched child is
+// cloned before being inserted into the document.
+//
+// It returns the original set of elements.
+func (s *Selection) WrapInnerHtml(html string) *Selection {
+	return s.wrapInnerNodes(parseHtml(html)...)
+}
+
+// WrapInnerNode wraps an HTML structure, matched by the given selector, around
+// the content of element in the set of matched elements. The matched child is
+// cloned before being inserted into the document.
+//
+// It returns the original set of elements.
+func (s *Selection) WrapInnerNode(n *html.Node) *Selection {
+	return s.wrapInnerNodes(n)
+}
+
+func (s *Selection) wrapInnerNodes(ns ...*html.Node) *Selection {
+	if len(ns) == 0 {
+		return s
+	}
+
+	s.Each(func(i int, s *Selection) {
+		contents := s.Contents()
+
+		if contents.Size() > 0 {
+			contents.wrapAllNodes(ns...)
+		} else {
+			s.AppendNodes(cloneNode(ns[0]))
+		}
+	})
+
+	return s
+}
+
+func parseHtml(h string) []*html.Node {
+	// Errors are only returned when the io.Reader returns any error besides
+	// EOF, but strings.Reader never will
+	nodes, err := html.ParseFragment(strings.NewReader(h), &html.Node{Type: html.ElementNode})
+	if err != nil {
+		panic("goquery: failed to parse HTML: " + err.Error())
+	}
+	return nodes
+}
+
+func setHtmlNodes(s *Selection, ns ...*html.Node) *Selection {
+	for _, n := range s.Nodes {
+		for c := n.FirstChild; c != nil; c = n.FirstChild {
+			n.RemoveChild(c)
+		}
+		for _, c := range ns {
+			n.AppendChild(cloneNode(c))
+		}
+	}
+	return s
+}
+
+// Get the first child that is an ElementNode
+func getFirstChildEl(n *html.Node) *html.Node {
+	c := n.FirstChild
+	for c != nil && c.Type != html.ElementNode {
+		c = c.NextSibling
+	}
+	return c
+}
+
+// Deep copy a slice of nodes.
+func cloneNodes(ns []*html.Node) []*html.Node {
+	cns := make([]*html.Node, 0, len(ns))
+
+	for _, n := range ns {
+		cns = append(cns, cloneNode(n))
+	}
+
+	return cns
+}
+
+// Deep copy a node. The new node has clones of all the original node's
+// children but none of its parents or siblings.
+func cloneNode(n *html.Node) *html.Node {
+	nn := &html.Node{
+		Type:     n.Type,
+		DataAtom: n.DataAtom,
+		Data:     n.Data,
+		Attr:     make([]html.Attribute, len(n.Attr)),
+	}
+
+	copy(nn.Attr, n.Attr)
+	for c := n.FirstChild; c != nil; c = c.NextSibling {
+		nn.AppendChild(cloneNode(c))
+	}
+
+	return nn
+}
+
+func (s *Selection) manipulateNodes(ns []*html.Node, reverse bool,
+	f func(sn *html.Node, n *html.Node)) *Selection {
+
+	lasti := s.Size() - 1
+
+	// net.Html doesn't provide document fragments for insertion, so to get
+	// things in the correct order with After() and Prepend(), the callback
+	// needs to be called on the reverse of the nodes.
+	if reverse {
+		for i, j := 0, len(ns)-1; i < j; i, j = i+1, j-1 {
+			ns[i], ns[j] = ns[j], ns[i]
+		}
+	}
+
+	for i, sn := range s.Nodes {
+		for _, n := range ns {
+			if i != lasti {
+				f(sn, cloneNode(n))
+			} else {
+				if n.Parent != nil {
+					n.Parent.RemoveChild(n)
+				}
+				f(sn, n)
+			}
+		}
+	}
+
+	return s
+}
diff --git a/vendor/github.com/PuerkitoBio/goquery/property.go b/vendor/github.com/PuerkitoBio/goquery/property.go
new file mode 100644
index 0000000000..411126db2c
--- /dev/null
+++ b/vendor/github.com/PuerkitoBio/goquery/property.go
@@ -0,0 +1,275 @@
+package goquery
+
+import (
+	"bytes"
+	"regexp"
+	"strings"
+
+	"golang.org/x/net/html"
+)
+
+var rxClassTrim = regexp.MustCompile("[\t\r\n]")
+
+// Attr gets the specified attribute's value for the first element in the
+// Selection. To get the value for each element individually, use a looping
+// construct such as Each or Map method.
+func (s *Selection) Attr(attrName string) (val string, exists bool) {
+	if len(s.Nodes) == 0 {
+		return
+	}
+	return getAttributeValue(attrName, s.Nodes[0])
+}
+
+// AttrOr works like Attr but returns default value if attribute is not present.
+func (s *Selection) AttrOr(attrName, defaultValue string) string {
+	if len(s.Nodes) == 0 {
+		return defaultValue
+	}
+
+	val, exists := getAttributeValue(attrName, s.Nodes[0])
+	if !exists {
+		return defaultValue
+	}
+
+	return val
+}
+
+// RemoveAttr removes the named attribute from each element in the set of matched elements.
+func (s *Selection) RemoveAttr(attrName string) *Selection {
+	for _, n := range s.Nodes {
+		removeAttr(n, attrName)
+	}
+
+	return s
+}
+
+// SetAttr sets the given attribute on each element in the set of matched elements.
+func (s *Selection) SetAttr(attrName, val string) *Selection {
+	for _, n := range s.Nodes {
+		attr := getAttributePtr(attrName, n)
+		if attr == nil {
+			n.Attr = append(n.Attr, html.Attribute{Key: attrName, Val: val})
+		} else {
+			attr.Val = val
+		}
+	}
+
+	return s
+}
+
+// Text gets the combined text contents of each element in the set of matched
+// elements, including their descendants.
+func (s *Selection) Text() string {
+	var buf bytes.Buffer
+
+	// Slightly optimized vs calling Each: no single selection object created
+	var f func(*html.Node)
+	f = func(n *html.Node) {
+		if n.Type == html.TextNode {
+			// Keep newlines and spaces, like jQuery
+			buf.WriteString(n.Data)
+		}
+		if n.FirstChild != nil {
+			for c := n.FirstChild; c != nil; c = c.NextSibling {
+				f(c)
+			}
+		}
+	}
+	for _, n := range s.Nodes {
+		f(n)
+	}
+
+	return buf.String()
+}
+
+// Size is an alias for Length.
+func (s *Selection) Size() int {
+	return s.Length()
+}
+
+// Length returns the number of elements in the Selection object.
+func (s *Selection) Length() int {
+	return len(s.Nodes)
+}
+
+// Html gets the HTML contents of the first element in the set of matched
+// elements. It includes text and comment nodes.
+func (s *Selection) Html() (ret string, e error) {
+	// Since there is no .innerHtml, the HTML content must be re-created from
+	// the nodes using html.Render.
+	var buf bytes.Buffer
+
+	if len(s.Nodes) > 0 {
+		for c := s.Nodes[0].FirstChild; c != nil; c = c.NextSibling {
+			e = html.Render(&buf, c)
+			if e != nil {
+				return
+			}
+		}
+		ret = buf.String()
+	}
+
+	return
+}
+
+// AddClass adds the given class(es) to each element in the set of matched elements.
+// Multiple class names can be specified, separated by a space or via multiple arguments.
+func (s *Selection) AddClass(class ...string) *Selection {
+	classStr := strings.TrimSpace(strings.Join(class, " "))
+
+	if classStr == "" {
+		return s
+	}
+
+	tcls := getClassesSlice(classStr)
+	for _, n := range s.Nodes {
+		curClasses, attr := getClassesAndAttr(n, true)
+		for _, newClass := range tcls {
+			if !strings.Contains(curClasses, " "+newClass+" ") {
+				curClasses += newClass + " "
+			}
+		}
+
+		setClasses(n, attr, curClasses)
+	}
+
+	return s
+}
+
+// HasClass determines whether any of the matched elements are assigned the
+// given class.
+func (s *Selection) HasClass(class string) bool {
+	class = " " + class + " "
+	for _, n := range s.Nodes {
+		classes, _ := getClassesAndAttr(n, false)
+		if strings.Contains(classes, class) {
+			return true
+		}
+	}
+	return false
+}
+
+// RemoveClass removes the given class(es) from each element in the set of matched elements.
+// Multiple class names can be specified, separated by a space or via multiple arguments.
+// If no class name is provided, all classes are removed.
+func (s *Selection) RemoveClass(class ...string) *Selection {
+	var rclasses []string
+
+	classStr := strings.TrimSpace(strings.Join(class, " "))
+	remove := classStr == ""
+
+	if !remove {
+		rclasses = getClassesSlice(classStr)
+	}
+
+	for _, n := range s.Nodes {
+		if remove {
+			removeAttr(n, "class")
+		} else {
+			classes, attr := getClassesAndAttr(n, true)
+			for _, rcl := range rclasses {
+				classes = strings.Replace(classes, " "+rcl+" ", " ", -1)
+			}
+
+			setClasses(n, attr, classes)
+		}
+	}
+
+	return s
+}
+
+// ToggleClass adds or removes the given class(es) for each element in the set of matched elements.
+// Multiple class names can be specified, separated by a space or via multiple arguments.
+func (s *Selection) ToggleClass(class ...string) *Selection {
+	classStr := strings.TrimSpace(strings.Join(class, " "))
+
+	if classStr == "" {
+		return s
+	}
+
+	tcls := getClassesSlice(classStr)
+
+	for _, n := range s.Nodes {
+		classes, attr := getClassesAndAttr(n, true)
+		for _, tcl := range tcls {
+			if strings.Contains(classes, " "+tcl+" ") {
+				classes = strings.Replace(classes, " "+tcl+" ", " ", -1)
+			} else {
+				classes += tcl + " "
+			}
+		}
+
+		setClasses(n, attr, classes)
+	}
+
+	return s
+}
+
+func getAttributePtr(attrName string, n *html.Node) *html.Attribute {
+	if n == nil {
+		return nil
+	}
+
+	for i, a := range n.Attr {
+		if a.Key == attrName {
+			return &n.Attr[i]
+		}
+	}
+	return nil
+}
+
+// Private function to get the specified attribute's value from a node.
+func getAttributeValue(attrName string, n *html.Node) (val string, exists bool) {
+	if a := getAttributePtr(attrName, n); a != nil {
+		val = a.Val
+		exists = true
+	}
+	return
+}
+
+// Get and normalize the "class" attribute from the node.
+func getClassesAndAttr(n *html.Node, create bool) (classes string, attr *html.Attribute) {
+	// Applies only to element nodes
+	if n.Type == html.ElementNode {
+		attr = getAttributePtr("class", n)
+		if attr == nil && create {
+			n.Attr = append(n.Attr, html.Attribute{
+				Key: "class",
+				Val: "",
+			})
+			attr = &n.Attr[len(n.Attr)-1]
+		}
+	}
+
+	if attr == nil {
+		classes = " "
+	} else {
+		classes = rxClassTrim.ReplaceAllString(" "+attr.Val+" ", " ")
+	}
+
+	return
+}
+
+func getClassesSlice(classes string) []string {
+	return strings.Split(rxClassTrim.ReplaceAllString(" "+classes+" ", " "), " ")
+}
+
+func removeAttr(n *html.Node, attrName string) {
+	for i, a := range n.Attr {
+		if a.Key == attrName {
+			n.Attr[i], n.Attr[len(n.Attr)-1], n.Attr =
+				n.Attr[len(n.Attr)-1], html.Attribute{}, n.Attr[:len(n.Attr)-1]
+			return
+		}
+	}
+}
+
+func setClasses(n *html.Node, attr *html.Attribute, classes string) {
+	classes = strings.TrimSpace(classes)
+	if classes == "" {
+		removeAttr(n, "class")
+		return
+	}
+
+	attr.Val = classes
+}
diff --git a/vendor/github.com/PuerkitoBio/goquery/query.go b/vendor/github.com/PuerkitoBio/goquery/query.go
new file mode 100644
index 0000000000..1a7f8732e9
--- /dev/null
+++ b/vendor/github.com/PuerkitoBio/goquery/query.go
@@ -0,0 +1,53 @@
+package goquery
+
+import "golang.org/x/net/html"
+
+// Is checks the current matched set of elements against a selector and
+// returns true if at least one of these elements matches.
+func (s *Selection) Is(selector string) bool {
+	if len(s.Nodes) > 0 {
+		return s.IsMatcher(compileMatcher(selector))
+	}
+
+	return false
+}
+
+// IsMatcher checks the current matched set of elements against a matcher and
+// returns true if at least one of these elements matches.
+func (s *Selection) IsMatcher(m Matcher) bool {
+	if len(s.Nodes) > 0 {
+		if len(s.Nodes) == 1 {
+			return m.Match(s.Nodes[0])
+		}
+		return len(m.Filter(s.Nodes)) > 0
+	}
+
+	return false
+}
+
+// IsFunction checks the current matched set of elements against a predicate and
+// returns true if at least one of these elements matches.
+func (s *Selection) IsFunction(f func(int, *Selection) bool) bool {
+	return s.FilterFunction(f).Length() > 0
+}
+
+// IsSelection checks the current matched set of elements against a Selection object
+// and returns true if at least one of these elements matches.
+func (s *Selection) IsSelection(sel *Selection) bool {
+	return s.FilterSelection(sel).Length() > 0
+}
+
+// IsNodes checks the current matched set of elements against the specified nodes
+// and returns true if at least one of these elements matches.
+func (s *Selection) IsNodes(nodes ...*html.Node) bool {
+	return s.FilterNodes(nodes...).Length() > 0
+}
+
+// Contains returns true if the specified Node is within,
+// at any depth, one of the nodes in the Selection object.
+// It is NOT inclusive, to behave like jQuery's implementation, and
+// unlike Javascript's .contains, so if the contained
+// node is itself in the selection, it returns false.
+func (s *Selection) Contains(n *html.Node) bool {
+	return sliceContains(s.Nodes, n)
+}
diff --git a/vendor/github.com/PuerkitoBio/goquery/traversal.go b/vendor/github.com/PuerkitoBio/goquery/traversal.go
new file mode 100644
index 0000000000..5fa5315ac5
--- /dev/null
+++ b/vendor/github.com/PuerkitoBio/goquery/traversal.go
@@ -0,0 +1,698 @@
+package goquery
+
+import "golang.org/x/net/html"
+
+type siblingType int
+
+// Sibling type, used internally when iterating over children at the same
+// level (siblings) to specify which nodes are requested.
+const (
+	siblingPrevUntil siblingType = iota - 3
+	siblingPrevAll
+	siblingPrev
+	siblingAll
+	siblingNext
+	siblingNextAll
+	siblingNextUntil
+	siblingAllIncludingNonElements
+)
+
+// Find gets the descendants of each element in the current set of matched
+// elements, filtered by a selector. It returns a new Selection object
+// containing these matched elements.
+func (s *Selection) Find(selector string) *Selection {
+	return pushStack(s, findWithMatcher(s.Nodes, compileMatcher(selector)))
+}
+
+// FindMatcher gets the descendants of each element in the current set of matched
+// elements, filtered by the matcher. It returns a new Selection object
+// containing these matched elements.
+func (s *Selection) FindMatcher(m Matcher) *Selection {
+	return pushStack(s, findWithMatcher(s.Nodes, m))
+}
+
+// FindSelection gets the descendants of each element in the current
+// Selection, filtered by a Selection. It returns a new Selection object
+// containing these matched elements.
+func (s *Selection) FindSelection(sel *Selection) *Selection {
+	if sel == nil {
+		return pushStack(s, nil)
+	}
+	return s.FindNodes(sel.Nodes...)
+}
+
+// FindNodes gets the descendants of each element in the current
+// Selection, filtered by some nodes. It returns a new Selection object
+// containing these matched elements.
+func (s *Selection) FindNodes(nodes ...*html.Node) *Selection {
+	return pushStack(s, mapNodes(nodes, func(i int, n *html.Node) []*html.Node {
+		if sliceContains(s.Nodes, n) {
+			return []*html.Node{n}
+		}
+		return nil
+	}))
+}
+
+// Contents gets the children of each element in the Selection,
+// including text and comment nodes. It returns a new Selection object
+// containing these elements.
+func (s *Selection) Contents() *Selection {
+	return pushStack(s, getChildrenNodes(s.Nodes, siblingAllIncludingNonElements))
+}
+
+// ContentsFiltered gets the children of each element in the Selection,
+// filtered by the specified selector. It returns a new Selection
+// object containing these elements. Since selectors only act on Element nodes,
+// this function is an alias to ChildrenFiltered unless the selector is empty,
+// in which case it is an alias to Contents.
+func (s *Selection) ContentsFiltered(selector string) *Selection {
+	if selector != "" {
+		return s.ChildrenFiltered(selector)
+	}
+	return s.Contents()
+}
+
+// ContentsMatcher gets the children of each element in the Selection,
+// filtered by the specified matcher. It returns a new Selection
+// object containing these elements. Since matchers only act on Element nodes,
+// this function is an alias to ChildrenMatcher.
+func (s *Selection) ContentsMatcher(m Matcher) *Selection {
+	return s.ChildrenMatcher(m)
+}
+
+// Children gets the child elements of each element in the Selection.
+// It returns a new Selection object containing these elements.
+func (s *Selection) Children() *Selection {
+	return pushStack(s, getChildrenNodes(s.Nodes, siblingAll))
+}
+
+// ChildrenFiltered gets the child elements of each element in the Selection,
+// filtered by the specified selector. It returns a new
+// Selection object containing these elements.
+func (s *Selection) ChildrenFiltered(selector string) *Selection {
+	return filterAndPush(s, getChildrenNodes(s.Nodes, siblingAll), compileMatcher(selector))
+}
+
+// ChildrenMatcher gets the child elements of each element in the Selection,
+// filtered by the specified matcher. It returns a new
+// Selection object containing these elements.
+func (s *Selection) ChildrenMatcher(m Matcher) *Selection {
+	return filterAndPush(s, getChildrenNodes(s.Nodes, siblingAll), m)
+}
+
+// Parent gets the parent of each element in the Selection. It returns a
+// new Selection object containing the matched elements.
+func (s *Selection) Parent() *Selection {
+	return pushStack(s, getParentNodes(s.Nodes))
+}
+
+// ParentFiltered gets the parent of each element in the Selection filtered by a
+// selector. It returns a new Selection object containing the matched elements.
+func (s *Selection) ParentFiltered(selector string) *Selection {
+	return filterAndPush(s, getParentNodes(s.Nodes), compileMatcher(selector))
+}
+
+// ParentMatcher gets the parent of each element in the Selection filtered by a
+// matcher. It returns a new Selection object containing the matched elements.
+func (s *Selection) ParentMatcher(m Matcher) *Selection {
+	return filterAndPush(s, getParentNodes(s.Nodes), m)
+}
+
+// Closest gets the first element that matches the selector by testing the
+// element itself and traversing up through its ancestors in the DOM tree.
+func (s *Selection) Closest(selector string) *Selection {
+	cs := compileMatcher(selector)
+	return s.ClosestMatcher(cs)
+}
+
+// ClosestMatcher gets the first element that matches the matcher by testing the
+// element itself and traversing up through its ancestors in the DOM tree.
+func (s *Selection) ClosestMatcher(m Matcher) *Selection {
+	return pushStack(s, mapNodes(s.Nodes, func(i int, n *html.Node) []*html.Node {
+		// For each node in the selection, test the node itself, then each parent
+		// until a match is found.
+		for ; n != nil; n = n.Parent {
+			if m.Match(n) {
+				return []*html.Node{n}
+			}
+		}
+		return nil
+	}))
+}
+
+// ClosestNodes gets the first element that matches one of the nodes by testing the
+// element itself and traversing up through its ancestors in the DOM tree.
+func (s *Selection) ClosestNodes(nodes ...*html.Node) *Selection {
+	set := make(map[*html.Node]bool)
+	for _, n := range nodes {
+		set[n] = true
+	}
+	return pushStack(s, mapNodes(s.Nodes, func(i int, n *html.Node) []*html.Node {
+		// For each node in the selection, test the node itself, then each parent
+		// until a match is found.
+		for ; n != nil; n = n.Parent {
+			if set[n] {
+				return []*html.Node{n}
+			}
+		}
+		return nil
+	}))
+}
+
+// ClosestSelection gets the first element that matches one of the nodes in the
+// Selection by testing the element itself and traversing up through its ancestors
+// in the DOM tree.
+func (s *Selection) ClosestSelection(sel *Selection) *Selection {
+	if sel == nil {
+		return pushStack(s, nil)
+	}
+	return s.ClosestNodes(sel.Nodes...)
+}
+
+// Parents gets the ancestors of each element in the current Selection. It
+// returns a new Selection object with the matched elements.
+func (s *Selection) Parents() *Selection {
+	return pushStack(s, getParentsNodes(s.Nodes, nil, nil))
+}
+
+// ParentsFiltered gets the ancestors of each element in the current
+// Selection. It returns a new Selection object with the matched elements.
+func (s *Selection) ParentsFiltered(selector string) *Selection {
+	return filterAndPush(s, getParentsNodes(s.Nodes, nil, nil), compileMatcher(selector))
+}
+
+// ParentsMatcher gets the ancestors of each element in the current
+// Selection. It returns a new Selection object with the matched elements.
+func (s *Selection) ParentsMatcher(m Matcher) *Selection {
+	return filterAndPush(s, getParentsNodes(s.Nodes, nil, nil), m)
+}
+
+// ParentsUntil gets the ancestors of each element in the Selection, up to but
+// not including the element matched by the selector. It returns a new Selection
+// object containing the matched elements.
+func (s *Selection) ParentsUntil(selector string) *Selection {
+	return pushStack(s, getParentsNodes(s.Nodes, compileMatcher(selector), nil))
+}
+
+// ParentsUntilMatcher gets the ancestors of each element in the Selection, up to but
+// not including the element matched by the matcher. It returns a new Selection
+// object containing the matched elements.
+func (s *Selection) ParentsUntilMatcher(m Matcher) *Selection {
+	return pushStack(s, getParentsNodes(s.Nodes, m, nil))
+}
+
+// ParentsUntilSelection gets the ancestors of each element in the Selection,
+// up to but not including the elements in the specified Selection. It returns a
+// new Selection object containing the matched elements.
+func (s *Selection) ParentsUntilSelection(sel *Selection) *Selection {
+	if sel == nil {
+		return s.Parents()
+	}
+	return s.ParentsUntilNodes(sel.Nodes...)
+}
+
+// ParentsUntilNodes gets the ancestors of each element in the Selection,
+// up to but not including the specified nodes. It returns a
+// new Selection object containing the matched elements.
+func (s *Selection) ParentsUntilNodes(nodes ...*html.Node) *Selection {
+	return pushStack(s, getParentsNodes(s.Nodes, nil, nodes))
+}
+
+// ParentsFilteredUntil is like ParentsUntil, with the option to filter the
+// results based on a selector string. It returns a new Selection
+// object containing the matched elements.
+func (s *Selection) ParentsFilteredUntil(filterSelector, untilSelector string) *Selection {
+	return filterAndPush(s, getParentsNodes(s.Nodes, compileMatcher(untilSelector), nil), compileMatcher(filterSelector))
+}
+
+// ParentsFilteredUntilMatcher is like ParentsUntilMatcher, with the option to filter the
+// results based on a matcher. It returns a new Selection object containing the matched elements.
+func (s *Selection) ParentsFilteredUntilMatcher(filter, until Matcher) *Selection {
+	return filterAndPush(s, getParentsNodes(s.Nodes, until, nil), filter)
+}
+
+// ParentsFilteredUntilSelection is like ParentsUntilSelection, with the
+// option to filter the results based on a selector string. It returns a new
+// Selection object containing the matched elements.
+func (s *Selection) ParentsFilteredUntilSelection(filterSelector string, sel *Selection) *Selection {
+	return s.ParentsMatcherUntilSelection(compileMatcher(filterSelector), sel)
+}
+
+// ParentsMatcherUntilSelection is like ParentsUntilSelection, with the
+// option to filter the results based on a matcher. It returns a new
+// Selection object containing the matched elements.
+func (s *Selection) ParentsMatcherUntilSelection(filter Matcher, sel *Selection) *Selection {
+	if sel == nil {
+		return s.ParentsMatcher(filter)
+	}
+	return s.ParentsMatcherUntilNodes(filter, sel.Nodes...)
+}
+
+// ParentsFilteredUntilNodes is like ParentsUntilNodes, with the
+// option to filter the results based on a selector string. It returns a new
+// Selection object containing the matched elements.
+func (s *Selection) ParentsFilteredUntilNodes(filterSelector string, nodes ...*html.Node) *Selection {
+	return filterAndPush(s, getParentsNodes(s.Nodes, nil, nodes), compileMatcher(filterSelector))
+}
+
+// ParentsMatcherUntilNodes is like ParentsUntilNodes, with the
+// option to filter the results based on a matcher. It returns a new
+// Selection object containing the matched elements.
+func (s *Selection) ParentsMatcherUntilNodes(filter Matcher, nodes ...*html.Node) *Selection {
+	return filterAndPush(s, getParentsNodes(s.Nodes, nil, nodes), filter)
+}
+
+// Siblings gets the siblings of each element in the Selection. It returns
+// a new Selection object containing the matched elements.
+func (s *Selection) Siblings() *Selection {
+	return pushStack(s, getSiblingNodes(s.Nodes, siblingAll, nil, nil))
+}
+
+// SiblingsFiltered gets the siblings of each element in the Selection
+// filtered by a selector. It returns a new Selection object containing the
+// matched elements.
+func (s *Selection) SiblingsFiltered(selector string) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingAll, nil, nil), compileMatcher(selector))
+}
+
+// SiblingsMatcher gets the siblings of each element in the Selection
+// filtered by a matcher. It returns a new Selection object containing the
+// matched elements.
+func (s *Selection) SiblingsMatcher(m Matcher) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingAll, nil, nil), m)
+}
+
+// Next gets the immediately following sibling of each element in the
+// Selection. It returns a new Selection object containing the matched elements.
+func (s *Selection) Next() *Selection {
+	return pushStack(s, getSiblingNodes(s.Nodes, siblingNext, nil, nil))
+}
+
+// NextFiltered gets the immediately following sibling of each element in the
+// Selection filtered by a selector. It returns a new Selection object
+// containing the matched elements.
+func (s *Selection) NextFiltered(selector string) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingNext, nil, nil), compileMatcher(selector))
+}
+
+// NextMatcher gets the immediately following sibling of each element in the
+// Selection filtered by a matcher. It returns a new Selection object
+// containing the matched elements.
+func (s *Selection) NextMatcher(m Matcher) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingNext, nil, nil), m)
+}
+
+// NextAll gets all the following siblings of each element in the
+// Selection. It returns a new Selection object containing the matched elements.
+func (s *Selection) NextAll() *Selection {
+	return pushStack(s, getSiblingNodes(s.Nodes, siblingNextAll, nil, nil))
+}
+
+// NextAllFiltered gets all the following siblings of each element in the
+// Selection filtered by a selector. It returns a new Selection object
+// containing the matched elements.
+func (s *Selection) NextAllFiltered(selector string) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingNextAll, nil, nil), compileMatcher(selector))
+}
+
+// NextAllMatcher gets all the following siblings of each element in the
+// Selection filtered by a matcher. It returns a new Selection object
+// containing the matched elements.
+func (s *Selection) NextAllMatcher(m Matcher) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingNextAll, nil, nil), m)
+}
+
+// Prev gets the immediately preceding sibling of each element in the
+// Selection. It returns a new Selection object containing the matched elements.
+func (s *Selection) Prev() *Selection {
+	return pushStack(s, getSiblingNodes(s.Nodes, siblingPrev, nil, nil))
+}
+
+// PrevFiltered gets the immediately preceding sibling of each element in the
+// Selection filtered by a selector. It returns a new Selection object
+// containing the matched elements.
+func (s *Selection) PrevFiltered(selector string) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingPrev, nil, nil), compileMatcher(selector))
+}
+
+// PrevMatcher gets the immediately preceding sibling of each element in the
+// Selection filtered by a matcher. It returns a new Selection object
+// containing the matched elements.
+func (s *Selection) PrevMatcher(m Matcher) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingPrev, nil, nil), m)
+}
+
+// PrevAll gets all the preceding siblings of each element in the
+// Selection. It returns a new Selection object containing the matched elements.
+func (s *Selection) PrevAll() *Selection {
+	return pushStack(s, getSiblingNodes(s.Nodes, siblingPrevAll, nil, nil))
+}
+
+// PrevAllFiltered gets all the preceding siblings of each element in the
+// Selection filtered by a selector. It returns a new Selection object
+// containing the matched elements.
+func (s *Selection) PrevAllFiltered(selector string) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingPrevAll, nil, nil), compileMatcher(selector))
+}
+
+// PrevAllMatcher gets all the preceding siblings of each element in the
+// Selection filtered by a matcher. It returns a new Selection object
+// containing the matched elements.
+func (s *Selection) PrevAllMatcher(m Matcher) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingPrevAll, nil, nil), m)
+}
+
+// NextUntil gets all following siblings of each element up to but not
+// including the element matched by the selector. It returns a new Selection
+// object containing the matched elements.
+func (s *Selection) NextUntil(selector string) *Selection {
+	return pushStack(s, getSiblingNodes(s.Nodes, siblingNextUntil,
+		compileMatcher(selector), nil))
+}
+
+// NextUntilMatcher gets all following siblings of each element up to but not
+// including the element matched by the matcher. It returns a new Selection
+// object containing the matched elements.
+func (s *Selection) NextUntilMatcher(m Matcher) *Selection {
+	return pushStack(s, getSiblingNodes(s.Nodes, siblingNextUntil,
+		m, nil))
+}
+
+// NextUntilSelection gets all following siblings of each element up to but not
+// including the element matched by the Selection. It returns a new Selection
+// object containing the matched elements.
+func (s *Selection) NextUntilSelection(sel *Selection) *Selection {
+	if sel == nil {
+		return s.NextAll()
+	}
+	return s.NextUntilNodes(sel.Nodes...)
+}
+
+// NextUntilNodes gets all following siblings of each element up to but not
+// including the element matched by the nodes. It returns a new Selection
+// object containing the matched elements.
+func (s *Selection) NextUntilNodes(nodes ...*html.Node) *Selection {
+	return pushStack(s, getSiblingNodes(s.Nodes, siblingNextUntil,
+		nil, nodes))
+}
+
+// PrevUntil gets all preceding siblings of each element up to but not
+// including the element matched by the selector. It returns a new Selection
+// object containing the matched elements.
+func (s *Selection) PrevUntil(selector string) *Selection {
+	return pushStack(s, getSiblingNodes(s.Nodes, siblingPrevUntil,
+		compileMatcher(selector), nil))
+}
+
+// PrevUntilMatcher gets all preceding siblings of each element up to but not
+// including the element matched by the matcher. It returns a new Selection
+// object containing the matched elements.
+func (s *Selection) PrevUntilMatcher(m Matcher) *Selection {
+	return pushStack(s, getSiblingNodes(s.Nodes, siblingPrevUntil,
+		m, nil))
+}
+
+// PrevUntilSelection gets all preceding siblings of each element up to but not
+// including the element matched by the Selection. It returns a new Selection
+// object containing the matched elements.
+func (s *Selection) PrevUntilSelection(sel *Selection) *Selection {
+	if sel == nil {
+		return s.PrevAll()
+	}
+	return s.PrevUntilNodes(sel.Nodes...)
+}
+
+// PrevUntilNodes gets all preceding siblings of each element up to but not
+// including the element matched by the nodes. It returns a new Selection
+// object containing the matched elements.
+func (s *Selection) PrevUntilNodes(nodes ...*html.Node) *Selection {
+	return pushStack(s, getSiblingNodes(s.Nodes, siblingPrevUntil,
+		nil, nodes))
+}
+
+// NextFilteredUntil is like NextUntil, with the option to filter
+// the results based on a selector string.
+// It returns a new Selection object containing the matched elements.
+func (s *Selection) NextFilteredUntil(filterSelector, untilSelector string) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingNextUntil,
+		compileMatcher(untilSelector), nil), compileMatcher(filterSelector))
+}
+
+// NextFilteredUntilMatcher is like NextUntilMatcher, with the option to filter
+// the results based on a matcher.
+// It returns a new Selection object containing the matched elements.
+func (s *Selection) NextFilteredUntilMatcher(filter, until Matcher) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingNextUntil,
+		until, nil), filter)
+}
+
+// NextFilteredUntilSelection is like NextUntilSelection, with the
+// option to filter the results based on a selector string. It returns a new
+// Selection object containing the matched elements.
+func (s *Selection) NextFilteredUntilSelection(filterSelector string, sel *Selection) *Selection {
+	return s.NextMatcherUntilSelection(compileMatcher(filterSelector), sel)
+}
+
+// NextMatcherUntilSelection is like NextUntilSelection, with the
+// option to filter the results based on a matcher. It returns a new
+// Selection object containing the matched elements.
+func (s *Selection) NextMatcherUntilSelection(filter Matcher, sel *Selection) *Selection {
+	if sel == nil {
+		return s.NextMatcher(filter)
+	}
+	return s.NextMatcherUntilNodes(filter, sel.Nodes...)
+}
+
+// NextFilteredUntilNodes is like NextUntilNodes, with the
+// option to filter the results based on a selector string. It returns a new
+// Selection object containing the matched elements.
+func (s *Selection) NextFilteredUntilNodes(filterSelector string, nodes ...*html.Node) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingNextUntil,
+		nil, nodes), compileMatcher(filterSelector))
+}
+
+// NextMatcherUntilNodes is like NextUntilNodes, with the
+// option to filter the results based on a matcher. It returns a new
+// Selection object containing the matched elements.
+func (s *Selection) NextMatcherUntilNodes(filter Matcher, nodes ...*html.Node) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingNextUntil,
+		nil, nodes), filter)
+}
+
+// PrevFilteredUntil is like PrevUntil, with the option to filter
+// the results based on a selector string.
+// It returns a new Selection object containing the matched elements.
+func (s *Selection) PrevFilteredUntil(filterSelector, untilSelector string) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingPrevUntil,
+		compileMatcher(untilSelector), nil), compileMatcher(filterSelector))
+}
+
+// PrevFilteredUntilMatcher is like PrevUntilMatcher, with the option to filter
+// the results based on a matcher.
+// It returns a new Selection object containing the matched elements.
+func (s *Selection) PrevFilteredUntilMatcher(filter, until Matcher) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingPrevUntil,
+		until, nil), filter)
+}
+
+// PrevFilteredUntilSelection is like PrevUntilSelection, with the
+// option to filter the results based on a selector string. It returns a new
+// Selection object containing the matched elements.
+func (s *Selection) PrevFilteredUntilSelection(filterSelector string, sel *Selection) *Selection {
+	return s.PrevMatcherUntilSelection(compileMatcher(filterSelector), sel)
+}
+
+// PrevMatcherUntilSelection is like PrevUntilSelection, with the
+// option to filter the results based on a matcher. It returns a new
+// Selection object containing the matched elements.
+func (s *Selection) PrevMatcherUntilSelection(filter Matcher, sel *Selection) *Selection {
+	if sel == nil {
+		return s.PrevMatcher(filter)
+	}
+	return s.PrevMatcherUntilNodes(filter, sel.Nodes...)
+}
+
+// PrevFilteredUntilNodes is like PrevUntilNodes, with the
+// option to filter the results based on a selector string. It returns a new
+// Selection object containing the matched elements.
+func (s *Selection) PrevFilteredUntilNodes(filterSelector string, nodes ...*html.Node) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingPrevUntil,
+		nil, nodes), compileMatcher(filterSelector))
+}
+
+// PrevMatcherUntilNodes is like PrevUntilNodes, with the
+// option to filter the results based on a matcher. It returns a new
+// Selection object containing the matched elements.
+func (s *Selection) PrevMatcherUntilNodes(filter Matcher, nodes ...*html.Node) *Selection {
+	return filterAndPush(s, getSiblingNodes(s.Nodes, siblingPrevUntil,
+		nil, nodes), filter)
+}
+
+// Filter and push filters the nodes based on a matcher, and pushes the results
+// on the stack, with the srcSel as previous selection.
+func filterAndPush(srcSel *Selection, nodes []*html.Node, m Matcher) *Selection {
+	// Create a temporary Selection with the specified nodes to filter using winnow
+	sel := &Selection{nodes, srcSel.document, nil}
+	// Filter based on matcher and push on stack
+	return pushStack(srcSel, winnow(sel, m, true))
+}
+
+// Internal implementation of Find that return raw nodes.
+func findWithMatcher(nodes []*html.Node, m Matcher) []*html.Node {
+	// Map nodes to find the matches within the children of each node
+	return mapNodes(nodes, func(i int, n *html.Node) (result []*html.Node) {
+		// Go down one level, becausejQuery's Find selects only within descendants
+		for c := n.FirstChild; c != nil; c = c.NextSibling {
+			if c.Type == html.ElementNode {
+				result = append(result, m.MatchAll(c)...)
+			}
+		}
+		return
+	})
+}
+
+// Internal implementation to get all parent nodes, stopping at the specified
+// node (or nil if no stop).
+func getParentsNodes(nodes []*html.Node, stopm Matcher, stopNodes []*html.Node) []*html.Node {
+	return mapNodes(nodes, func(i int, n *html.Node) (result []*html.Node) {
+		for p := n.Parent; p != nil; p = p.Parent {
+			sel := newSingleSelection(p, nil)
+			if stopm != nil {
+				if sel.IsMatcher(stopm) {
+					break
+				}
+			} else if len(stopNodes) > 0 {
+				if sel.IsNodes(stopNodes...) {
+					break
+				}
+			}
+			if p.Type == html.ElementNode {
+				result = append(result, p)
+			}
+		}
+		return
+	})
+}
+
+// Internal implementation of sibling nodes that return a raw slice of matches.
+func getSiblingNodes(nodes []*html.Node, st siblingType, untilm Matcher, untilNodes []*html.Node) []*html.Node {
+	var f func(*html.Node) bool
+
+	// If the requested siblings are ...Until, create the test function to
+	// determine if the until condition is reached (returns true if it is)
+	if st == siblingNextUntil || st == siblingPrevUntil {
+		f = func(n *html.Node) bool {
+			if untilm != nil {
+				// Matcher-based condition
+				sel := newSingleSelection(n, nil)
+				return sel.IsMatcher(untilm)
+			} else if len(untilNodes) > 0 {
+				// Nodes-based condition
+				sel := newSingleSelection(n, nil)
+				return sel.IsNodes(untilNodes...)
+			}
+			return false
+		}
+	}
+
+	return mapNodes(nodes, func(i int, n *html.Node) []*html.Node {
+		return getChildrenWithSiblingType(n.Parent, st, n, f)
+	})
+}
+
+// Gets the children nodes of each node in the specified slice of nodes,
+// based on the sibling type request.
+func getChildrenNodes(nodes []*html.Node, st siblingType) []*html.Node {
+	return mapNodes(nodes, func(i int, n *html.Node) []*html.Node {
+		return getChildrenWithSiblingType(n, st, nil, nil)
+	})
+}
+
+// Gets the children of the specified parent, based on the requested sibling
+// type, skipping a specified node if required.
+func getChildrenWithSiblingType(parent *html.Node, st siblingType, skipNode *html.Node,
+	untilFunc func(*html.Node) bool) (result []*html.Node) {
+
+	// Create the iterator function
+	var iter = func(cur *html.Node) (ret *html.Node) {
+		// Based on the sibling type requested, iterate the right way
+		for {
+			switch st {
+			case siblingAll, siblingAllIncludingNonElements:
+				if cur == nil {
+					// First iteration, start with first child of parent
+					// Skip node if required
+					if ret = parent.FirstChild; ret == skipNode && skipNode != nil {
+						ret = skipNode.NextSibling
+					}
+				} else {
+					// Skip node if required
+					if ret = cur.NextSibling; ret == skipNode && skipNode != nil {
+						ret = skipNode.NextSibling
+					}
+				}
+			case siblingPrev, siblingPrevAll, siblingPrevUntil:
+				if cur == nil {
+					// Start with previous sibling of the skip node
+					ret = skipNode.PrevSibling
+				} else {
+					ret = cur.PrevSibling
+				}
+			case siblingNext, siblingNextAll, siblingNextUntil:
+				if cur == nil {
+					// Start with next sibling of the skip node
+					ret = skipNode.NextSibling
+				} else {
+					ret = cur.NextSibling
+				}
+			default:
+				panic("Invalid sibling type.")
+			}
+			if ret == nil || ret.Type == html.ElementNode || st == siblingAllIncludingNonElements {
+				return
+			}
+			// Not a valid node, try again from this one
+			cur = ret
+		}
+	}
+
+	for c := iter(nil); c != nil; c = iter(c) {
+		// If this is an ...Until case, test before append (returns true
+		// if the until condition is reached)
+		if st == siblingNextUntil || st == siblingPrevUntil {
+			if untilFunc(c) {
+				return
+			}
+		}
+		result = append(result, c)
+		if st == siblingNext || st == siblingPrev {
+			// Only one node was requested (immediate next or previous), so exit
+			return
+		}
+	}
+	return
+}
+
+// Internal implementation of parent nodes that return a raw slice of Nodes.
+func getParentNodes(nodes []*html.Node) []*html.Node {
+	return mapNodes(nodes, func(i int, n *html.Node) []*html.Node {
+		if n.Parent != nil && n.Parent.Type == html.ElementNode {
+			return []*html.Node{n.Parent}
+		}
+		return nil
+	})
+}
+
+// Internal map function used by many traversing methods. Takes the source nodes
+// to iterate on and the mapping function that returns an array of nodes.
+// Returns an array of nodes mapped by calling the callback function once for
+// each node in the source nodes.
+func mapNodes(nodes []*html.Node, f func(int, *html.Node) []*html.Node) (result []*html.Node) {
+	set := make(map[*html.Node]bool)
+	for i, n := range nodes {
+		if vals := f(i, n); len(vals) > 0 {
+			result = appendWithoutDuplicates(result, vals, set)
+		}
+	}
+	return result
+}
diff --git a/vendor/github.com/PuerkitoBio/goquery/type.go b/vendor/github.com/PuerkitoBio/goquery/type.go
new file mode 100644
index 0000000000..e2169fa30d
--- /dev/null
+++ b/vendor/github.com/PuerkitoBio/goquery/type.go
@@ -0,0 +1,135 @@
+package goquery
+
+import (
+	"errors"
+	"io"
+	"net/http"
+	"net/url"
+
+	"github.com/andybalholm/cascadia"
+
+	"golang.org/x/net/html"
+)
+
+// Document represents an HTML document to be manipulated. Unlike jQuery, which
+// is loaded as part of a DOM document, and thus acts upon its containing
+// document, GoQuery doesn't know which HTML document to act upon. So it needs
+// to be told, and that's what the Document class is for. It holds the root
+// document node to manipulate, and can make selections on this document.
+type Document struct {
+	*Selection
+	Url      *url.URL
+	rootNode *html.Node
+}
+
+// NewDocumentFromNode is a Document constructor that takes a root html Node
+// as argument.
+func NewDocumentFromNode(root *html.Node) *Document {
+	return newDocument(root, nil)
+}
+
+// NewDocument is a Document constructor that takes a string URL as argument.
+// It loads the specified document, parses it, and stores the root Document
+// node, ready to be manipulated.
+func NewDocument(url string) (*Document, error) {
+	// Load the URL
+	res, e := http.Get(url)
+	if e != nil {
+		return nil, e
+	}
+	return NewDocumentFromResponse(res)
+}
+
+// NewDocumentFromReader returns a Document from a generic reader.
+// It returns an error as second value if the reader's data cannot be parsed
+// as html. It does *not* check if the reader is also an io.Closer, so the
+// provided reader is never closed by this call, it is the responsibility
+// of the caller to close it if required.
+func NewDocumentFromReader(r io.Reader) (*Document, error) {
+	root, e := html.Parse(r)
+	if e != nil {
+		return nil, e
+	}
+	return newDocument(root, nil), nil
+}
+
+// NewDocumentFromResponse is another Document constructor that takes an http response as argument.
+// It loads the specified response's document, parses it, and stores the root Document
+// node, ready to be manipulated. The response's body is closed on return.
+func NewDocumentFromResponse(res *http.Response) (*Document, error) {
+	if res == nil {
+		return nil, errors.New("Response is nil")
+	}
+	defer res.Body.Close()
+	if res.Request == nil {
+		return nil, errors.New("Response.Request is nil")
+	}
+
+	// Parse the HTML into nodes
+	root, e := html.Parse(res.Body)
+	if e != nil {
+		return nil, e
+	}
+
+	// Create and fill the document
+	return newDocument(root, res.Request.URL), nil
+}
+
+// CloneDocument creates a deep-clone of a document.
+func CloneDocument(doc *Document) *Document {
+	return newDocument(cloneNode(doc.rootNode), doc.Url)
+}
+
+// Private constructor, make sure all fields are correctly filled.
+func newDocument(root *html.Node, url *url.URL) *Document {
+	// Create and fill the document
+	d := &Document{nil, url, root}
+	d.Selection = newSingleSelection(root, d)
+	return d
+}
+
+// Selection represents a collection of nodes matching some criteria. The
+// initial Selection can be created by using Document.Find, and then
+// manipulated using the jQuery-like chainable syntax and methods.
+type Selection struct {
+	Nodes    []*html.Node
+	document *Document
+	prevSel  *Selection
+}
+
+// Helper constructor to create an empty selection
+func newEmptySelection(doc *Document) *Selection {
+	return &Selection{nil, doc, nil}
+}
+
+// Helper constructor to create a selection of only one node
+func newSingleSelection(node *html.Node, doc *Document) *Selection {
+	return &Selection{[]*html.Node{node}, doc, nil}
+}
+
+// Matcher is an interface that defines the methods to match
+// HTML nodes against a compiled selector string. Cascadia's
+// Selector implements this interface.
+type Matcher interface {
+	Match(*html.Node) bool
+	MatchAll(*html.Node) []*html.Node
+	Filter([]*html.Node) []*html.Node
+}
+
+// compileMatcher compiles the selector string s and returns
+// the corresponding Matcher. If s is an invalid selector string,
+// it returns a Matcher that fails all matches.
+func compileMatcher(s string) Matcher {
+	cs, err := cascadia.Compile(s)
+	if err != nil {
+		return invalidMatcher{}
+	}
+	return cs
+}
+
+// invalidMatcher is a Matcher that always fails to match.
+type invalidMatcher struct{}
+
+func (invalidMatcher) Match(n *html.Node) bool             { return false }
+func (invalidMatcher) MatchAll(n *html.Node) []*html.Node  { return nil }
+func (invalidMatcher) Filter(ns []*html.Node) []*html.Node { return nil }
diff --git a/vendor/github.com/PuerkitoBio/goquery/utilities.go b/vendor/github.com/PuerkitoBio/goquery/utilities.go
new file mode 100644
index 0000000000..b4c061a4d4
--- /dev/null
+++ b/vendor/github.com/PuerkitoBio/goquery/utilities.go
@@ -0,0 +1,161 @@
+package goquery
+
+import (
+	"bytes"
+
+	"golang.org/x/net/html"
+)
+
+// used to determine if a set (map[*html.Node]bool) should be used
+// instead of iterating over a slice. The set uses more memory and
+// is slower than slice iteration for small N.
+const minNodesForSet = 1000
+
+var nodeNames = []string{
+	html.ErrorNode:    "#error",
+	html.TextNode:     "#text",
+	html.DocumentNode: "#document",
+	html.CommentNode:  "#comment",
+}
+
+// NodeName returns the node name of the first element in the selection.
+// It tries to behave in a similar way as the DOM's nodeName property
+// (https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeName).
+//
+// Go's net/html package defines the following node types, listed with
+// the corresponding returned value from this function:
+//
+//     ErrorNode : #error
+//     TextNode : #text
+//     DocumentNode : #document
+//     ElementNode : the element's tag name
+//     CommentNode : #comment
+//     DoctypeNode : the name of the document type
+//
+func NodeName(s *Selection) string {
+	if s.Length() == 0 {
+		return ""
+	}
+	switch n := s.Get(0); n.Type {
+	case html.ElementNode, html.DoctypeNode:
+		return n.Data
+	default:
+		if n.Type >= 0 && int(n.Type) < len(nodeNames) {
+			return nodeNames[n.Type]
+		}
+		return ""
+	}
+}
+
+// OuterHtml returns the outer HTML rendering of the first item in
+// the selection - that is, the HTML including the first element's
+// tag and attributes.
+//
+// Unlike InnerHtml, this is a function and not a method on the Selection,
+// because this is not a jQuery method (in javascript-land, this is
+// a property provided by the DOM).
+func OuterHtml(s *Selection) (string, error) {
+	var buf bytes.Buffer
+
+	if s.Length() == 0 {
+		return "", nil
+	}
+	n := s.Get(0)
+	if err := html.Render(&buf, n); err != nil {
+		return "", err
+	}
+	return buf.String(), nil
+}
+
+// Loop through all container nodes to search for the target node.
+func sliceContains(container []*html.Node, contained *html.Node) bool {
+	for _, n := range container {
+		if nodeContains(n, contained) {
+			return true
+		}
+	}
+
+	return false
+}
+
+// Checks if the contained node is within the container node.
+func nodeContains(container *html.Node, contained *html.Node) bool {
+	// Check if the parent of the contained node is the container node, traversing
+	// upward until the top is reached, or the container is found.
+	for contained = contained.Parent; contained != nil; contained = contained.Parent {
+		if container == contained {
+			return true
+		}
+	}
+	return false
+}
+
+// Checks if the target node is in the slice of nodes.
+func isInSlice(slice []*html.Node, node *html.Node) bool {
+	return indexInSlice(slice, node) > -1
+}
+
+// Returns the index of the target node in the slice, or -1.
+func indexInSlice(slice []*html.Node, node *html.Node) int {
+	if node != nil {
+		for i, n := range slice {
+			if n == node {
+				return i
+			}
+		}
+	}
+	return -1
+}
+
+// Appends the new nodes to the target slice, making sure no duplicate is added.
+// There is no check to the original state of the target slice, so it may still
+// contain duplicates. The target slice is returned because append() may create
+// a new underlying array. If targetSet is nil, a local set is created with the
+// target if len(target) + len(nodes) is greater than minNodesForSet.
+func appendWithoutDuplicates(target []*html.Node, nodes []*html.Node, targetSet map[*html.Node]bool) []*html.Node {
+	// if there are not that many nodes, don't use the map, faster to just use nested loops
+	// (unless a non-nil targetSet is passed, in which case the caller knows better).
+	if targetSet == nil && len(target)+len(nodes) < minNodesForSet {
+		for _, n := range nodes {
+			if !isInSlice(target, n) {
+				target = append(target, n)
+			}
+		}
+		return target
+	}
+
+	// if a targetSet is passed, then assume it is reliable, otherwise create one
+	// and initialize it with the current target contents.
+	if targetSet == nil {
+		targetSet = make(map[*html.Node]bool, len(target))
+		for _, n := range target {
+			targetSet[n] = true
+		}
+	}
+	for _, n := range nodes {
+		if !targetSet[n] {
+			target = append(target, n)
+			targetSet[n] = true
+		}
+	}
+
+	return target
+}
+
+// Loop through a selection, returning only those nodes that pass the predicate
+// function.
+func grep(sel *Selection, predicate func(i int, s *Selection) bool) (result []*html.Node) {
+	for i, n := range sel.Nodes {
+		if predicate(i, newSingleSelection(n, sel.document)) {
+			result = append(result, n)
+		}
+	}
+	return result
+}
+
+// Creates a new Selection object based on the specified nodes, and keeps the
+// source Selection object on the stack (linked list).
+func pushStack(fromSel *Selection, nodes []*html.Node) *Selection {
+	result := &Selection{nodes, fromSel.document, fromSel}
+	return result
+}
diff --git a/vendor/github.com/andybalholm/cascadia/LICENSE b/vendor/github.com/andybalholm/cascadia/LICENSE
new file mode 100755
index 0000000000..ee5ad35acc
--- /dev/null
+++ b/vendor/github.com/andybalholm/cascadia/LICENSE
@@ -0,0 +1,24 @@
+Copyright (c) 2011 Andy Balholm. All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are
+met:
+
+   * Redistributions of source code must retain the above copyright
+notice, this list of conditions and the following disclaimer.
+   * Redistributions in binary form must reproduce the above
+copyright notice, this list of conditions and the following disclaimer
+in the documentation and/or other materials provided with the
+distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
diff --git a/vendor/github.com/andybalholm/cascadia/README.md b/vendor/github.com/andybalholm/cascadia/README.md
new file mode 100644
index 0000000000..9021cb92aa
--- /dev/null
+++ b/vendor/github.com/andybalholm/cascadia/README.md
@@ -0,0 +1,7 @@
+# cascadia
+
+[![](https://travis-ci.org/andybalholm/cascadia.svg)](https://travis-ci.org/andybalholm/cascadia)
+
+The Cascadia package implements CSS selectors for use with the parse trees produced by the html package.
+
+To test CSS selectors without writing Go code, check out [cascadia](https://github.com/suntong/cascadia) the command line tool, a thin wrapper around this package.
diff --git a/vendor/github.com/andybalholm/cascadia/parser.go b/vendor/github.com/andybalholm/cascadia/parser.go
new file mode 100644
index 0000000000..495db9ccfc
--- /dev/null
+++ b/vendor/github.com/andybalholm/cascadia/parser.go
@@ -0,0 +1,835 @@
+// Package cascadia is an implementation of CSS selectors.
+package cascadia
+
+import (
+	"errors"
+	"fmt"
+	"regexp"
+	"strconv"
+	"strings"
+
+	"golang.org/x/net/html"
+)
+
+// a parser for CSS selectors
+type parser struct {
+	s string // the source text
+	i int    // the current position
+}
+
+// parseEscape parses a backslash escape.
+func (p *parser) parseEscape() (result string, err error) {
+	if len(p.s) < p.i+2 || p.s[p.i] != '\\' {
+		return "", errors.New("invalid escape sequence")
+	}
+
+	start := p.i + 1
+	c := p.s[start]
+	switch {
+	case c == '\r' || c == '\n' || c == '\f':
+		return "", errors.New("escaped line ending outside string")
+	case hexDigit(c):
+		// unicode escape (hex)
+		var i int
+		for i = start; i < p.i+6 && i < len(p.s) && hexDigit(p.s[i]); i++ {
+			// empty
+		}
+		v, _ := strconv.ParseUint(p.s[start:i], 16, 21)
+		if len(p.s) > i {
+			switch p.s[i] {
+			case '\r':
+				i++
+				if len(p.s) > i && p.s[i] == '\n' {
+					i++
+				}
+			case ' ', '\t', '\n', '\f':
+				i++
+			}
+		}
+		p.i = i
+		return string(rune(v)), nil
+	}
+
+	// Return the literal character after the backslash.
+	result = p.s[start : start+1]
+	p.i += 2
+	return result, nil
+}
+
+func hexDigit(c byte) bool {
+	return '0' <= c && c <= '9' || 'a' <= c && c <= 'f' || 'A' <= c && c <= 'F'
+}
+
+// nameStart returns whether c can be the first character of an identifier
+// (not counting an initial hyphen, or an escape sequence).
+func nameStart(c byte) bool {
+	return 'a' <= c && c <= 'z' || 'A' <= c && c <= 'Z' || c == '_' || c > 127
+}
+
+// nameChar returns whether c can be a character within an identifier
+// (not counting an escape sequence).
+func nameChar(c byte) bool {
+	return 'a' <= c && c <= 'z' || 'A' <= c && c <= 'Z' || c == '_' || c > 127 ||
+		c == '-' || '0' <= c && c <= '9'
+}
+
+// parseIdentifier parses an identifier.
+func (p *parser) parseIdentifier() (result string, err error) {
+	startingDash := false
+	if len(p.s) > p.i && p.s[p.i] == '-' {
+		startingDash = true
+		p.i++
+	}
+
+	if len(p.s) <= p.i {
+		return "", errors.New("expected identifier, found EOF instead")
+	}
+
+	if c := p.s[p.i]; !(nameStart(c) || c == '\\') {
+		return "", fmt.Errorf("expected identifier, found %c instead", c)
+	}
+
+	result, err = p.parseName()
+	if startingDash && err == nil {
+		result = "-" + result
+	}
+	return
+}
+
+// parseName parses a name (which is like an identifier, but doesn't have
+// extra restrictions on the first character).
+func (p *parser) parseName() (result string, err error) {
+	i := p.i
+loop:
+	for i < len(p.s) {
+		c := p.s[i]
+		switch {
+		case nameChar(c):
+			start := i
+			for i < len(p.s) && nameChar(p.s[i]) {
+				i++
+			}
+			result += p.s[start:i]
+		case c == '\\':
+			p.i = i
+			val, err := p.parseEscape()
+			if err != nil {
+				return "", err
+			}
+			i = p.i
+			result += val
+		default:
+			break loop
+		}
+	}
+
+	if result == "" {
+		return "", errors.New("expected name, found EOF instead")
+	}
+
+	p.i = i
+	return result, nil
+}
+
+// parseString parses a single- or double-quoted string.
+func (p *parser) parseString() (result string, err error) {
+	i := p.i
+	if len(p.s) < i+2 {
+		return "", errors.New("expected string, found EOF instead")
+	}
+
+	quote := p.s[i]
+	i++
+
+loop:
+	for i < len(p.s) {
+		switch p.s[i] {
+		case '\\':
+			if len(p.s) > i+1 {
+				switch c := p.s[i+1]; c {
+				case '\r':
+					if len(p.s) > i+2 && p.s[i+2] == '\n' {
+						i += 3
+						continue loop
+					}
+					fallthrough
+				case '\n', '\f':
+					i += 2
+					continue loop
+				}
+			}
+			p.i = i
+			val, err := p.parseEscape()
+			if err != nil {
+				return "", err
+			}
+			i = p.i
+			result += val
+		case quote:
+			break loop
+		case '\r', '\n', '\f':
+			return "", errors.New("unexpected end of line in string")
+		default:
+			start := i
+			for i < len(p.s) {
+				if c := p.s[i]; c == quote || c == '\\' || c == '\r' || c == '\n' || c == '\f' {
+					break
+				}
+				i++
+			}
+			result += p.s[start:i]
+		}
+	}
+
+	if i >= len(p.s) {
+		return "", errors.New("EOF in string")
+	}
+
+	// Consume the final quote.
+	i++
+
+	p.i = i
+	return result, nil
+}
+
+// parseRegex parses a regular expression; the end is defined by encountering an
+// unmatched closing ')' or ']' which is not consumed
+func (p *parser) parseRegex() (rx *regexp.Regexp, err error) {
+	i := p.i
+	if len(p.s) < i+2 {
+		return nil, errors.New("expected regular expression, found EOF instead")
+	}
+
+	// number of open parens or brackets;
+	// when it becomes negative, finished parsing regex
+	open := 0
+
+loop:
+	for i < len(p.s) {
+		switch p.s[i] {
+		case '(', '[':
+			open++
+		case ')', ']':
+			open--
+			if open < 0 {
+				break loop
+			}
+		}
+		i++
+	}
+
+	if i >= len(p.s) {
+		return nil, errors.New("EOF in regular expression")
+	}
+	rx, err = regexp.Compile(p.s[p.i:i])
+	p.i = i
+	return rx, err
+}
+
+// skipWhitespace consumes whitespace characters and comments.
+// It returns true if there was actually anything to skip.
+func (p *parser) skipWhitespace() bool {
+	i := p.i
+	for i < len(p.s) {
+		switch p.s[i] {
+		case ' ', '\t', '\r', '\n', '\f':
+			i++
+			continue
+		case '/':
+			if strings.HasPrefix(p.s[i:], "/*") {
+				end := strings.Index(p.s[i+len("/*"):], "*/")
+				if end != -1 {
+					i += end + len("/**/")
+					continue
+				}
+			}
+		}
+		break
+	}
+
+	if i > p.i {
+		p.i = i
+		return true
+	}
+
+	return false
+}
+
+// consumeParenthesis consumes an opening parenthesis and any following
+// whitespace. It returns true if there was actually a parenthesis to skip.
+func (p *parser) consumeParenthesis() bool {
+	if p.i < len(p.s) && p.s[p.i] == '(' {
+		p.i++
+		p.skipWhitespace()
+		return true
+	}
+	return false
+}
+
+// consumeClosingParenthesis consumes a closing parenthesis and any preceding
+// whitespace. It returns true if there was actually a parenthesis to skip.
+func (p *parser) consumeClosingParenthesis() bool {
+	i := p.i
+	p.skipWhitespace()
+	if p.i < len(p.s) && p.s[p.i] == ')' {
+		p.i++
+		return true
+	}
+	p.i = i
+	return false
+}
+
+// parseTypeSelector parses a type selector (one that matches by tag name).
+func (p *parser) parseTypeSelector() (result Selector, err error) {
+	tag, err := p.parseIdentifier()
+	if err != nil {
+		return nil, err
+	}
+
+	return typeSelector(tag), nil
+}
+
+// parseIDSelector parses a selector that matches by id attribute.
+func (p *parser) parseIDSelector() (Selector, error) {
+	if p.i >= len(p.s) {
+		return nil, fmt.Errorf("expected id selector (#id), found EOF instead")
+	}
+	if p.s[p.i] != '#' {
+		return nil, fmt.Errorf("expected id selector (#id), found '%c' instead", p.s[p.i])
+	}
+
+	p.i++
+	id, err := p.parseName()
+	if err != nil {
+		return nil, err
+	}
+
+	return attributeEqualsSelector("id", id), nil
+}
+
+// parseClassSelector parses a selector that matches by class attribute.
+func (p *parser) parseClassSelector() (Selector, error) {
+	if p.i >= len(p.s) {
+		return nil, fmt.Errorf("expected class selector (.class), found EOF instead")
+	}
+	if p.s[p.i] != '.' {
+		return nil, fmt.Errorf("expected class selector (.class), found '%c' instead", p.s[p.i])
+	}
+
+	p.i++
+	class, err := p.parseIdentifier()
+	if err != nil {
+		return nil, err
+	}
+
+	return attributeIncludesSelector("class", class), nil
+}
+
+// parseAttributeSelector parses a selector that matches by attribute value.
+func (p *parser) parseAttributeSelector() (Selector, error) {
+	if p.i >= len(p.s) {
+		return nil, fmt.Errorf("expected attribute selector ([attribute]), found EOF instead")
+	}
+	if p.s[p.i] != '[' {
+		return nil, fmt.Errorf("expected attribute selector ([attribute]), found '%c' instead", p.s[p.i])
+	}
+
+	p.i++
+	p.skipWhitespace()
+	key, err := p.parseIdentifier()
+	if err != nil {
+		return nil, err
+	}
+
+	p.skipWhitespace()
+	if p.i >= len(p.s) {
+		return nil, errors.New("unexpected EOF in attribute selector")
+	}
+
+	if p.s[p.i] == ']' {
+		p.i++
+		return attributeExistsSelector(key), nil
+	}
+
+	if p.i+2 >= len(p.s) {
+		return nil, errors.New("unexpected EOF in attribute selector")
+	}
+
+	op := p.s[p.i : p.i+2]
+	if op[0] == '=' {
+		op = "="
+	} else if op[1] != '=' {
+		return nil, fmt.Errorf(`expected equality operator, found "%s" instead`, op)
+	}
+	p.i += len(op)
+
+	p.skipWhitespace()
+	if p.i >= len(p.s) {
+		return nil, errors.New("unexpected EOF in attribute selector")
+	}
+	var val string
+	var rx *regexp.Regexp
+	if op == "#=" {
+		rx, err = p.parseRegex()
+	} else {
+		switch p.s[p.i] {
+		case '\'', '"':
+			val, err = p.parseString()
+		default:
+			val, err = p.parseIdentifier()
+		}
+	}
+	if err != nil {
+		return nil, err
+	}
+
+	p.skipWhitespace()
+	if p.i >= len(p.s) {
+		return nil, errors.New("unexpected EOF in attribute selector")
+	}
+	if p.s[p.i] != ']' {
+		return nil, fmt.Errorf("expected ']', found '%c' instead", p.s[p.i])
+	}
+	p.i++
+
+	switch op {
+	case "=":
+		return attributeEqualsSelector(key, val), nil
+	case "!=":
+		return attributeNotEqualSelector(key, val), nil
+	case "~=":
+		return attributeIncludesSelector(key, val), nil
+	case "|=":
+		return attributeDashmatchSelector(key, val), nil
+	case "^=":
+		return attributePrefixSelector(key, val), nil
+	case "$=":
+		return attributeSuffixSelector(key, val), nil
+	case "*=":
+		return attributeSubstringSelector(key, val), nil
+	case "#=":
+		return attributeRegexSelector(key, rx), nil
+	}
+
+	return nil, fmt.Errorf("attribute operator %q is not supported", op)
+}
+
+var errExpectedParenthesis = errors.New("expected '(' but didn't find it")
+var errExpectedClosingParenthesis = errors.New("expected ')' but didn't find it")
+var errUnmatchedParenthesis = errors.New("unmatched '('")
+
+// parsePseudoclassSelector parses a pseudoclass selector like :not(p).
+func (p *parser) parsePseudoclassSelector() (Selector, error) {
+	if p.i >= len(p.s) {
+		return nil, fmt.Errorf("expected pseudoclass selector (:pseudoclass), found EOF instead")
+	}
+	if p.s[p.i] != ':' {
+		return nil, fmt.Errorf("expected attribute selector (:pseudoclass), found '%c' instead", p.s[p.i])
+	}
+
+	p.i++
+	name, err := p.parseIdentifier()
+	if err != nil {
+		return nil, err
+	}
+	name = toLowerASCII(name)
+
+	switch name {
+	case "not", "has", "haschild":
+		if !p.consumeParenthesis() {
+			return nil, errExpectedParenthesis
+		}
+		sel, parseErr := p.parseSelectorGroup()
+		if parseErr != nil {
+			return nil, parseErr
+		}
+		if !p.consumeClosingParenthesis() {
+			return nil, errExpectedClosingParenthesis
+		}
+
+		switch name {
+		case "not":
+			return negatedSelector(sel), nil
+		case "has":
+			return hasDescendantSelector(sel), nil
+		case "haschild":
+			return hasChildSelector(sel), nil
+		}
+
+	case "contains", "containsown":
+		if !p.consumeParenthesis() {
+			return nil, errExpectedParenthesis
+		}
+		if p.i == len(p.s) {
+			return nil, errUnmatchedParenthesis
+		}
+		var val string
+		switch p.s[p.i] {
+		case '\'', '"':
+			val, err = p.parseString()
+		default:
+			val, err = p.parseIdentifier()
+		}
+		if err != nil {
+			return nil, err
+		}
+		val = strings.ToLower(val)
+		p.skipWhitespace()
+		if p.i >= len(p.s) {
+			return nil, errors.New("unexpected EOF in pseudo selector")
+		}
+		if !p.consumeClosingParenthesis() {
+			return nil, errExpectedClosingParenthesis
+		}
+
+		switch name {
+		case "contains":
+			return textSubstrSelector(val), nil
+		case "containsown":
+			return ownTextSubstrSelector(val), nil
+		}
+
+	case "matches", "matchesown":
+		if !p.consumeParenthesis() {
+			return nil, errExpectedParenthesis
+		}
+		rx, err := p.parseRegex()
+		if err != nil {
+			return nil, err
+		}
+		if p.i >= len(p.s) {
+			return nil, errors.New("unexpected EOF in pseudo selector")
+		}
+		if !p.consumeClosingParenthesis() {
+			return nil, errExpectedClosingParenthesis
+		}
+
+		switch name {
+		case "matches":
+			return textRegexSelector(rx), nil
+		case "matchesown":
+			return ownTextRegexSelector(rx), nil
+		}
+
+	case "nth-child", "nth-last-child", "nth-of-type", "nth-last-of-type":
+		if !p.consumeParenthesis() {
+			return nil, errExpectedParenthesis
+		}
+		a, b, err := p.parseNth()
+		if err != nil {
+			return nil, err
+		}
+		if !p.consumeClosingParenthesis() {
+			return nil, errExpectedClosingParenthesis
+		}
+		if a == 0 {
+			switch name {
+			case "nth-child":
+				return simpleNthChildSelector(b, false), nil
+			case "nth-of-type":
+				return simpleNthChildSelector(b, true), nil
+			case "nth-last-child":
+				return simpleNthLastChildSelector(b, false), nil
+			case "nth-last-of-type":
+				return simpleNthLastChildSelector(b, true), nil
+			}
+		}
+		return nthChildSelector(a, b,
+				name == "nth-last-child" || name == "nth-last-of-type",
+				name == "nth-of-type" || name == "nth-last-of-type"),
+			nil
+
+	case "first-child":
+		return simpleNthChildSelector(1, false), nil
+	case "last-child":
+		return simpleNthLastChildSelector(1, false), nil
+	case "first-of-type":
+		return simpleNthChildSelector(1, true), nil
+	case "last-of-type":
+		return simpleNthLastChildSelector(1, true), nil
+	case "only-child":
+		return onlyChildSelector(false), nil
+	case "only-of-type":
+		return onlyChildSelector(true), nil
+	case "input":
+		return inputSelector, nil
+	case "empty":
+		return emptyElementSelector, nil
+	case "root":
+		return rootSelector, nil
+	}
+
+	return nil, fmt.Errorf("unknown pseudoclass :%s", name)
+}
+
+// parseInteger parses a  decimal integer.
+func (p *parser) parseInteger() (int, error) {
+	i := p.i
+	start := i
+	for i < len(p.s) && '0' <= p.s[i] && p.s[i] <= '9' {
+		i++
+	}
+	if i == start {
+		return 0, errors.New("expected integer, but didn't find it")
+	}
+	p.i = i
+
+	val, err := strconv.Atoi(p.s[start:i])
+	if err != nil {
+		return 0, err
+	}
+
+	return val, nil
+}
+
+// parseNth parses the argument for :nth-child (normally of the form an+b).
+func (p *parser) parseNth() (a, b int, err error) {
+	// initial state
+	if p.i >= len(p.s) {
+		goto eof
+	}
+	switch p.s[p.i] {
+	case '-':
+		p.i++
+		goto negativeA
+	case '+':
+		p.i++
+		goto positiveA
+	case '0', '1', '2', '3', '4', '5', '6', '7', '8', '9':
+		goto positiveA
+	case 'n', 'N':
+		a = 1
+		p.i++
+		goto readN
+	case 'o', 'O', 'e', 'E':
+		id, nameErr := p.parseName()
+		if nameErr != nil {
+			return 0, 0, nameErr
+		}
+		id = toLowerASCII(id)
+		if id == "odd" {
+			return 2, 1, nil
+		}
+		if id == "even" {
+			return 2, 0, nil
+		}
+		return 0, 0, fmt.Errorf("expected 'odd' or 'even', but found '%s' instead", id)
+	default:
+		goto invalid
+	}
+
+positiveA:
+	if p.i >= len(p.s) {
+		goto eof
+	}
+	switch p.s[p.i] {
+	case '0', '1', '2', '3', '4', '5', '6', '7', '8', '9':
+		a, err = p.parseInteger()
+		if err != nil {
+			return 0, 0, err
+		}
+		goto readA
+	case 'n', 'N':
+		a = 1
+		p.i++
+		goto readN
+	default:
+		goto invalid
+	}
+
+negativeA:
+	if p.i >= len(p.s) {
+		goto eof
+	}
+	switch p.s[p.i] {
+	case '0', '1', '2', '3', '4', '5', '6', '7', '8', '9':
+		a, err = p.parseInteger()
+		if err != nil {
+			return 0, 0, err
+		}
+		a = -a
+		goto readA
+	case 'n', 'N':
+		a = -1
+		p.i++
+		goto readN
+	default:
+		goto invalid
+	}
+
+readA:
+	if p.i >= len(p.s) {
+		goto eof
+	}
+	switch p.s[p.i] {
+	case 'n', 'N':
+		p.i++
+		goto readN
+	default:
+		// The number we read as a is actually b.
+		return 0, a, nil
+	}
+
+readN:
+	p.skipWhitespace()
+	if p.i >= len(p.s) {
+		goto eof
+	}
+	switch p.s[p.i] {
+	case '+':
+		p.i++
+		p.skipWhitespace()
+		b, err = p.parseInteger()
+		if err != nil {
+			return 0, 0, err
+		}
+		return a, b, nil
+	case '-':
+		p.i++
+		p.skipWhitespace()
+		b, err = p.parseInteger()
+		if err != nil {
+			return 0, 0, err
+		}
+		return a, -b, nil
+	default:
+		return a, 0, nil
+	}
+
+eof:
+	return 0, 0, errors.New("unexpected EOF while attempting to parse expression of form an+b")
+
+invalid:
+	return 0, 0, errors.New("unexpected character while attempting to parse expression of form an+b")
+}
+
+// parseSimpleSelectorSequence parses a selector sequence that applies to
+// a single element.
+func (p *parser) parseSimpleSelectorSequence() (Selector, error) {
+	var result Selector
+
+	if p.i >= len(p.s) {
+		return nil, errors.New("expected selector, found EOF instead")
+	}
+
+	switch p.s[p.i] {
+	case '*':
+		// It's the universal selector. Just skip over it, since it doesn't affect the meaning.
+		p.i++
+	case '#', '.', '[', ':':
+		// There's no type selector. Wait to process the other till the main loop.
+	default:
+		r, err := p.parseTypeSelector()
+		if err != nil {
+			return nil, err
+		}
+		result = r
+	}
+
+loop:
+	for p.i < len(p.s) {
+		var ns Selector
+		var err error
+		switch p.s[p.i] {
+		case '#':
+			ns, err = p.parseIDSelector()
+		case '.':
+			ns, err = p.parseClassSelector()
+		case '[':
+			ns, err = p.parseAttributeSelector()
+		case ':':
+			ns, err = p.parsePseudoclassSelector()
+		default:
+			break loop
+		}
+		if err != nil {
+			return nil, err
+		}
+		if result == nil {
+			result = ns
+		} else {
+			result = intersectionSelector(result, ns)
+		}
+	}
+
+	if result == nil {
+		result = func(n *html.Node) bool {
+			return n.Type == html.ElementNode
+		}
+	}
+
+	return result, nil
+}
+
+// parseSelector parses a selector that may include combinators.
+func (p *parser) parseSelector() (result Selector, err error) {
+	p.skipWhitespace()
+	result, err = p.parseSimpleSelectorSequence()
+	if err != nil {
+		return
+	}
+
+	for {
+		var combinator byte
+		if p.skipWhitespace() {
+			combinator = ' '
+		}
+		if p.i >= len(p.s) {
+			return
+		}
+
+		switch p.s[p.i] {
+		case '+', '>', '~':
+			combinator = p.s[p.i]
+			p.i++
+			p.skipWhitespace()
+		case ',', ')':
+			// These characters can't begin a selector, but they can legally occur after one.
+			return
+		}
+
+		if combinator == 0 {
+			return
+		}
+
+		c, err := p.parseSimpleSelectorSequence()
+		if err != nil {
+			return nil, err
+		}
+
+		switch combinator {
+		case ' ':
+			result = descendantSelector(result, c)
+		case '>':
+			result = childSelector(result, c)
+		case '+':
+			result = siblingSelector(result, c, true)
+		case '~':
+			result = siblingSelector(result, c, false)
+		}
+	}
+
+	panic("unreachable")
+}
+
+// parseSelectorGroup parses a group of selectors, separated by commas.
+func (p *parser) parseSelectorGroup() (result Selector, err error) {
+	result, err = p.parseSelector()
+	if err != nil {
+		return
+	}
+
+	for p.i < len(p.s) {
+		if p.s[p.i] != ',' {
+			return result, nil
+		}
+		p.i++
+		c, err := p.parseSelector()
+		if err != nil {
+			return nil, err
+		}
+		result = unionSelector(result, c)
+	}
+
+	return
+}
diff --git a/vendor/github.com/andybalholm/cascadia/selector.go b/vendor/github.com/andybalholm/cascadia/selector.go
new file mode 100644
index 0000000000..9fb05ccb77
--- /dev/null
+++ b/vendor/github.com/andybalholm/cascadia/selector.go
@@ -0,0 +1,622 @@
+package cascadia
+
+import (
+	"bytes"
+	"fmt"
+	"regexp"
+	"strings"
+
+	"golang.org/x/net/html"
+)
+
+// the Selector type, and functions for creating them
+
+// A Selector is a function which tells whether a node matches or not.
+type Selector func(*html.Node) bool
+
+// hasChildMatch returns whether n has any child that matches a.
+func hasChildMatch(n *html.Node, a Selector) bool {
+	for c := n.FirstChild; c != nil; c = c.NextSibling {
+		if a(c) {
+			return true
+		}
+	}
+	return false
+}
+
+// hasDescendantMatch performs a depth-first search of n's descendants,
+// testing whether any of them match a. It returns true as soon as a match is
+// found, or false if no match is found.
+func hasDescendantMatch(n *html.Node, a Selector) bool {
+	for c := n.FirstChild; c != nil; c = c.NextSibling {
+		if a(c) || (c.Type == html.ElementNode && hasDescendantMatch(c, a)) {
+			return true
+		}
+	}
+	return false
+}
+
+// Compile parses a selector and returns, if successful, a Selector object
+// that can be used to match against html.Node objects.
+func Compile(sel string) (Selector, error) {
+	p := &parser{s: sel}
+	compiled, err := p.parseSelectorGroup()
+	if err != nil {
+		return nil, err
+	}
+
+	if p.i < len(sel) {
+		return nil, fmt.Errorf("parsing %q: %d bytes left over", sel, len(sel)-p.i)
+	}
+
+	return compiled, nil
+}
+
+// MustCompile is like Compile, but panics instead of returning an error.
+func MustCompile(sel string) Selector {
+	compiled, err := Compile(sel)
+	if err != nil {
+		panic(err)
+	}
+	return compiled
+}
+
+// MatchAll returns a slice of the nodes that match the selector,
+// from n and its children.
+func (s Selector) MatchAll(n *html.Node) []*html.Node {
+	return s.matchAllInto(n, nil)
+}
+
+func (s Selector) matchAllInto(n *html.Node, storage []*html.Node) []*html.Node {
+	if s(n) {
+		storage = append(storage, n)
+	}
+
+	for child := n.FirstChild; child != nil; child = child.NextSibling {
+		storage = s.matchAllInto(child, storage)
+	}
+
+	return storage
+}
+
+// Match returns true if the node matches the selector.
+func (s Selector) Match(n *html.Node) bool {
+	return s(n)
+}
+
+// MatchFirst returns the first node that matches s, from n and its children.
+func (s Selector) MatchFirst(n *html.Node) *html.Node {
+	if s.Match(n) {
+		return n
+	}
+
+	for c := n.FirstChild; c != nil; c = c.NextSibling {
+		m := s.MatchFirst(c)
+		if m != nil {
+			return m
+		}
+	}
+	return nil
+}
+
+// Filter returns the nodes in nodes that match the selector.
+func (s Selector) Filter(nodes []*html.Node) (result []*html.Node) {
+	for _, n := range nodes {
+		if s(n) {
+			result = append(result, n)
+		}
+	}
+	return result
+}
+
+// typeSelector returns a Selector that matches elements with a given tag name.
+func typeSelector(tag string) Selector {
+	tag = toLowerASCII(tag)
+	return func(n *html.Node) bool {
+		return n.Type == html.ElementNode && n.Data == tag
+	}
+}
+
+// toLowerASCII returns s with all ASCII capital letters lowercased.
+func toLowerASCII(s string) string {
+	var b []byte
+	for i := 0; i < len(s); i++ {
+		if c := s[i]; 'A' <= c && c <= 'Z' {
+			if b == nil {
+				b = make([]byte, len(s))
+				copy(b, s)
+			}
+			b[i] = s[i] + ('a' - 'A')
+		}
+	}
+
+	if b == nil {
+		return s
+	}
+
+	return string(b)
+}
+
+// attributeSelector returns a Selector that matches elements
+// where the attribute named key satisifes the function f.
+func attributeSelector(key string, f func(string) bool) Selector {
+	key = toLowerASCII(key)
+	return func(n *html.Node) bool {
+		if n.Type != html.ElementNode {
+			return false
+		}
+		for _, a := range n.Attr {
+			if a.Key == key && f(a.Val) {
+				return true
+			}
+		}
+		return false
+	}
+}
+
+// attributeExistsSelector returns a Selector that matches elements that have
+// an attribute named key.
+func attributeExistsSelector(key string) Selector {
+	return attributeSelector(key, func(string) bool { return true })
+}
+
+// attributeEqualsSelector returns a Selector that matches elements where
+// the attribute named key has the value val.
+func attributeEqualsSelector(key, val string) Selector {
+	return attributeSelector(key,
+		func(s string) bool {
+			return s == val
+		})
+}
+
+// attributeNotEqualSelector returns a Selector that matches elements where
+// the attribute named key does not have the value val.
+func attributeNotEqualSelector(key, val string) Selector {
+	key = toLowerASCII(key)
+	return func(n *html.Node) bool {
+		if n.Type != html.ElementNode {
+			return false
+		}
+		for _, a := range n.Attr {
+			if a.Key == key && a.Val == val {
+				return false
+			}
+		}
+		return true
+	}
+}
+
+// attributeIncludesSelector returns a Selector that matches elements where
+// the attribute named key is a whitespace-separated list that includes val.
+func attributeIncludesSelector(key, val string) Selector {
+	return attributeSelector(key,
+		func(s string) bool {
+			for s != "" {
+				i := strings.IndexAny(s, " \t\r\n\f")
+				if i == -1 {
+					return s == val
+				}
+				if s[:i] == val {
+					return true
+				}
+				s = s[i+1:]
+			}
+			return false
+		})
+}
+
+// attributeDashmatchSelector returns a Selector that matches elements where
+// the attribute named key equals val or starts with val plus a hyphen.
+func attributeDashmatchSelector(key, val string) Selector {
+	return attributeSelector(key,
+		func(s string) bool {
+			if s == val {
+				return true
+			}
+			if len(s) <= len(val) {
+				return false
+			}
+			if s[:len(val)] == val && s[len(val)] == '-' {
+				return true
+			}
+			return false
+		})
+}
+
+// attributePrefixSelector returns a Selector that matches elements where
+// the attribute named key starts with val.
+func attributePrefixSelector(key, val string) Selector {
+	return attributeSelector(key,
+		func(s string) bool {
+			if strings.TrimSpace(s) == "" {
+				return false
+			}
+			return strings.HasPrefix(s, val)
+		})
+}
+
+// attributeSuffixSelector returns a Selector that matches elements where
+// the attribute named key ends with val.
+func attributeSuffixSelector(key, val string) Selector {
+	return attributeSelector(key,
+		func(s string) bool {
+			if strings.TrimSpace(s) == "" {
+				return false
+			}
+			return strings.HasSuffix(s, val)
+		})
+}
+
+// attributeSubstringSelector returns a Selector that matches nodes where
+// the attribute named key contains val.
+func attributeSubstringSelector(key, val string) Selector {
+	return attributeSelector(key,
+		func(s string) bool {
+			if strings.TrimSpace(s) == "" {
+				return false
+			}
+			return strings.Contains(s, val)
+		})
+}
+
+// attributeRegexSelector returns a Selector that matches nodes where
+// the attribute named key matches the regular expression rx
+func attributeRegexSelector(key string, rx *regexp.Regexp) Selector {
+	return attributeSelector(key,
+		func(s string) bool {
+			return rx.MatchString(s)
+		})
+}
+
+// intersectionSelector returns a selector that matches nodes that match
+// both a and b.
+func intersectionSelector(a, b Selector) Selector {
+	return func(n *html.Node) bool {
+		return a(n) && b(n)
+	}
+}
+
+// unionSelector returns a selector that matches elements that match
+// either a or b.
+func unionSelector(a, b Selector) Selector {
+	return func(n *html.Node) bool {
+		return a(n) || b(n)
+	}
+}
+
+// negatedSelector returns a selector that matches elements that do not match a.
+func negatedSelector(a Selector) Selector {
+	return func(n *html.Node) bool {
+		if n.Type != html.ElementNode {
+			return false
+		}
+		return !a(n)
+	}
+}
+
+// writeNodeText writes the text contained in n and its descendants to b.
+func writeNodeText(n *html.Node, b *bytes.Buffer) {
+	switch n.Type {
+	case html.TextNode:
+		b.WriteString(n.Data)
+	case html.ElementNode:
+		for c := n.FirstChild; c != nil; c = c.NextSibling {
+			writeNodeText(c, b)
+		}
+	}
+}
+
+// nodeText returns the text contained in n and its descendants.
+func nodeText(n *html.Node) string {
+	var b bytes.Buffer
+	writeNodeText(n, &b)
+	return b.String()
+}
+
+// nodeOwnText returns the contents of the text nodes that are direct
+// children of n.
+func nodeOwnText(n *html.Node) string {
+	var b bytes.Buffer
+	for c := n.FirstChild; c != nil; c = c.NextSibling {
+		if c.Type == html.TextNode {
+			b.WriteString(c.Data)
+		}
+	}
+	return b.String()
+}
+
+// textSubstrSelector returns a selector that matches nodes that
+// contain the given text.
+func textSubstrSelector(val string) Selector {
+	return func(n *html.Node) bool {
+		text := strings.ToLower(nodeText(n))
+		return strings.Contains(text, val)
+	}
+}
+
+// ownTextSubstrSelector returns a selector that matches nodes that
+// directly contain the given text
+func ownTextSubstrSelector(val string) Selector {
+	return func(n *html.Node) bool {
+		text := strings.ToLower(nodeOwnText(n))
+		return strings.Contains(text, val)
+	}
+}
+
+// textRegexSelector returns a selector that matches nodes whose text matches
+// the specified regular expression
+func textRegexSelector(rx *regexp.Regexp) Selector {
+	return func(n *html.Node) bool {
+		return rx.MatchString(nodeText(n))
+	}
+}
+
+// ownTextRegexSelector returns a selector that matches nodes whose text
+// directly matches the specified regular expression
+func ownTextRegexSelector(rx *regexp.Regexp) Selector {
+	return func(n *html.Node) bool {
+		return rx.MatchString(nodeOwnText(n))
+	}
+}
+
+// hasChildSelector returns a selector that matches elements
+// with a child that matches a.
+func hasChildSelector(a Selector) Selector {
+	return func(n *html.Node) bool {
+		if n.Type != html.ElementNode {
+			return false
+		}
+		return hasChildMatch(n, a)
+	}
+}
+
+// hasDescendantSelector returns a selector that matches elements
+// with any descendant that matches a.
+func hasDescendantSelector(a Selector) Selector {
+	return func(n *html.Node) bool {
+		if n.Type != html.ElementNode {
+			return false
+		}
+		return hasDescendantMatch(n, a)
+	}
+}
+
+// nthChildSelector returns a selector that implements :nth-child(an+b).
+// If last is true, implements :nth-last-child instead.
+// If ofType is true, implements :nth-of-type instead.
+func nthChildSelector(a, b int, last, ofType bool) Selector {
+	return func(n *html.Node) bool {
+		if n.Type != html.ElementNode {
+			return false
+		}
+
+		parent := n.Parent
+		if parent == nil {
+			return false
+		}
+
+		if parent.Type == html.DocumentNode {
+			return false
+		}
+
+		i := -1
+		count := 0
+		for c := parent.FirstChild; c != nil; c = c.NextSibling {
+			if (c.Type != html.ElementNode) || (ofType && c.Data != n.Data) {
+				continue
+			}
+			count++
+			if c == n {
+				i = count
+				if !last {
+					break
+				}
+			}
+		}
+
+		if i == -1 {
+			// This shouldn't happen, since n should always be one of its parent's children.
+			return false
+		}
+
+		if last {
+			i = count - i + 1
+		}
+
+		i -= b
+		if a == 0 {
+			return i == 0
+		}
+
+		return i%a == 0 && i/a >= 0
+	}
+}
+
+// simpleNthChildSelector returns a selector that implements :nth-child(b).
+// If ofType is true, implements :nth-of-type instead.
+func simpleNthChildSelector(b int, ofType bool) Selector {
+	return func(n *html.Node) bool {
+		if n.Type != html.ElementNode {
+			return false
+		}
+
+		parent := n.Parent
+		if parent == nil {
+			return false
+		}
+
+		if parent.Type == html.DocumentNode {
+			return false
+		}
+
+		count := 0
+		for c := parent.FirstChild; c != nil; c = c.NextSibling {
+			if c.Type != html.ElementNode || (ofType && c.Data != n.Data) {
+				continue
+			}
+			count++
+			if c == n {
+				return count == b
+			}
+			if count >= b {
+				return false
+			}
+		}
+		return false
+	}
+}
+
+// simpleNthLastChildSelector returns a selector that implements
+// :nth-last-child(b). If ofType is true, implements :nth-last-of-type
+// instead.
+func simpleNthLastChildSelector(b int, ofType bool) Selector {
+	return func(n *html.Node) bool {
+		if n.Type != html.ElementNode {
+			return false
+		}
+
+		parent := n.Parent
+		if parent == nil {
+			return false
+		}
+
+		if parent.Type == html.DocumentNode {
+			return false
+		}
+
+		count := 0
+		for c := parent.LastChild; c != nil; c = c.PrevSibling {
+			if c.Type != html.ElementNode || (ofType && c.Data != n.Data) {
+				continue
+			}
+			count++
+			if c == n {
+				return count == b
+			}
+			if count >= b {
+				return false
+			}
+		}
+		return false
+	}
+}
+
+// onlyChildSelector returns a selector that implements :only-child.
+// If ofType is true, it implements :only-of-type instead.
+func onlyChildSelector(ofType bool) Selector {
+	return func(n *html.Node) bool {
+		if n.Type != html.ElementNode {
+			return false
+		}
+
+		parent := n.Parent
+		if parent == nil {
+			return false
+		}
+
+		if parent.Type == html.DocumentNode {
+			return false
+		}
+
+		count := 0
+		for c := parent.FirstChild; c != nil; c = c.NextSibling {
+			if (c.Type != html.ElementNode) || (ofType && c.Data != n.Data) {
+				continue
+			}
+			count++
+			if count > 1 {
+				return false
+			}
+		}
+
+		return count == 1
+	}
+}
+
+// inputSelector is a Selector that matches input, select, textarea and button elements.
+func inputSelector(n *html.Node) bool {
+	return n.Type == html.ElementNode && (n.Data == "input" || n.Data == "select" || n.Data == "textarea" || n.Data == "button")
+}
+
+// emptyElementSelector is a Selector that matches empty elements.
+func emptyElementSelector(n *html.Node) bool {
+	if n.Type != html.ElementNode {
+		return false
+	}
+
+	for c := n.FirstChild; c != nil; c = c.NextSibling {
+		switch c.Type {
+		case html.ElementNode, html.TextNode:
+			return false
+		}
+	}
+
+	return true
+}
+
+// descendantSelector returns a Selector that matches an element if
+// it matches d and has an ancestor that matches a.
+func descendantSelector(a, d Selector) Selector {
+	return func(n *html.Node) bool {
+		if !d(n) {
+			return false
+		}
+
+		for p := n.Parent; p != nil; p = p.Parent {
+			if a(p) {
+				return true
+			}
+		}
+
+		return false
+	}
+}
+
+// childSelector returns a Selector that matches an element if
+// it matches d and its parent matches a.
+func childSelector(a, d Selector) Selector {
+	return func(n *html.Node) bool {
+		return d(n) && n.Parent != nil && a(n.Parent)
+	}
+}
+
+// siblingSelector returns a Selector that matches an element
+// if it matches s2 and in is preceded by an element that matches s1.
+// If adjacent is true, the sibling must be immediately before the element.
+func siblingSelector(s1, s2 Selector, adjacent bool) Selector {
+	return func(n *html.Node) bool {
+		if !s2(n) {
+			return false
+		}
+
+		if adjacent {
+			for n = n.PrevSibling; n != nil; n = n.PrevSibling {
+				if n.Type == html.TextNode || n.Type == html.CommentNode {
+					continue
+				}
+				return s1(n)
+			}
+			return false
+		}
+
+		// Walk backwards looking for element that matches s1
+		for c := n.PrevSibling; c != nil; c = c.PrevSibling {
+			if s1(c) {
+				return true
+			}
+		}
+
+		return false
+	}
+}
+
+// rootSelector implements :root
+func rootSelector(n *html.Node) bool {
+	if n.Type != html.ElementNode {
+		return false
+	}
+	if n.Parent == nil {
+		return false
+	}
+	return n.Parent.Type == html.DocumentNode
+}
diff --git a/vendor/vendor.json b/vendor/vendor.json
index bbd759c12c..a28e71f3d3 100644
--- a/vendor/vendor.json
+++ b/vendor/vendor.json
@@ -15,6 +15,12 @@
 			"revisionTime": "2017-05-04T17:35:58Z"
 		},
 		{
+			"checksumSHA1": "bOODD4Gbw3GfcuQPU2dI40crxxk=",
+			"path": "github.com/PuerkitoBio/goquery",
+			"revision": "ed7d758e9a34ba1f55e8084e0d731448b46921a8",
+			"revisionTime": "2017-03-24T13:54:48Z"
+		},
+		{
 			"checksumSHA1": "IyfS7Rbl6OgR83QR7TOfKdDCq+M=",
 			"path": "github.com/Unknwon/cae",
 			"revision": "c6aac99ea2cae2ebaf23f26f76b04fe3fcfc9f8c",
@@ -45,6 +51,12 @@
 			"revisionTime": "2015-11-04T15:16:17Z"
 		},
 		{
+			"checksumSHA1": "2rWSos1D+xqNSGkvEMOG3pntod8=",
+			"path": "github.com/andybalholm/cascadia",
+			"revision": "349dd0209470eabd9514242c688c403c0926d266",
+			"revisionTime": "2016-12-24T14:14:13Z"
+		},
+		{
 			"checksumSHA1": "Ehq5j/CeBqurZOP1vXMDLEl+RZo=",
 			"path": "github.com/blevesearch/bleve",
 			"revision": "3b2bc30b544cd4589cf3ad8c04030329b143dce9",