Attribute extractor added. HTML extractor added. Outer HTML Extractor added.

exporter package renamed to export, extractor package renamed to extract for simplicity.

extract/attr.go

@@ -0,0 +1,24 @@
+package extract
+
+import (
+	"errors"
+	"github.com/PuerkitoBio/goquery"
+)
+
+var ErrAttrNotExists = errors.New("attribute not exist")
+
+// Attr returns HTML attribute value of provided selector
+type Attr struct {
+	Name     string
+	Selector string
+	Attr     string
+}
+
+// Extract returns HTML attribute value of provided selector
+func (e *Attr) Extract(doc *goquery.Document) (interface{}, error) {
+	attr, exists := doc.Find(e.Selector).Attr(e.Attr)
+	if !exists {
+		return nil, ErrAttrNotExists
+	}
+	return map[string]string{e.Name: attr}, nil
+}

extract/html.go

@@ -0,0 +1,53 @@
+package extract
+
+import (
+	"bytes"
+	"github.com/PuerkitoBio/goquery"
+	"golang.org/x/net/html"
+)
+
+// HTML extracts and returns the HTML from inside each element of the given selection.
+type HTML struct {
+	Name     string
+	Selector string
+}
+
+// Extract extracts and returns the HTML from inside each element of the given selection.
+func (e *HTML) Extract(doc *goquery.Document) (interface{}, error) {
+	var ret, h string
+	var err error
+
+	doc.Find(e.Selector).EachWithBreak(func(i int, s *goquery.Selection) bool {
+		h, err = s.Html()
+		if err != nil {
+			return false
+		}
+
+		ret += h
+		return true
+	})
+
+	if err != nil {
+		return nil, err
+	}
+
+	return map[string]string{e.Name: ret}, nil
+}
+
+// OuterHTML extracts and returns the HTML of each element of the given selection.
+type OuterHTML struct {
+	Name     string
+	Selector string
+}
+
+// Extract extracts and returns the HTML of each element of the given selection.
+func (e *OuterHTML) Extract(doc *goquery.Document) (interface{}, error) {
+	output := bytes.NewBufferString("")
+	for _, node := range doc.Find(e.Selector).Nodes {
+		if err := html.Render(output, node); err != nil {
+			return nil, err
+		}
+	}
+
+	return map[string]string{e.Name: output.String()}, nil
+}

extract/text.go

@@ -0,0 +1,14 @@
+package extract
+
+import "github.com/PuerkitoBio/goquery"
+
+// Text returns the combined text contents of provided selector.
+type Text struct {
+	Name     string
+	Selector string
+}
+
+// Extract returns the combined text contents of provided selector.
+func (e *Text) Extract(doc *goquery.Document) (interface{}, error) {
+	return map[string]string{e.Name: doc.Find(e.Selector).Text()}, nil
+}