gitea/geziyor
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Extractors refactored to support pass by value. Documentation added for request and response.

Browse Source
This commit is contained in:
Musab Gültekin
2019-07-04 02:13:29 +03:00
parent 71683ec6de
commit da03567fae
9 changed files with 51 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
extract/attr.go
View File

@ -15,8 +15,8 @@ type Attr struct {
}
// 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)
func (e Attr) Extract(sel *goquery.Selection) (interface{}, error) {
attr, exists := sel.Find(e.Selector).Attr(e.Attr)
if !exists {
return nil, ErrAttrNotExists
}

2
extract/extract.go
View File

@ -4,5 +4,5 @@ import "github.com/PuerkitoBio/goquery"
// Extractor interface is for extracting data from HTML document
type Extractor interface {
Extract(doc *goquery.Document) (interface{}, error)
Extract(sel *goquery.Selection) (interface{}, error)
}

8
extract/html.go
View File

@ -13,11 +13,11 @@ type HTML struct {
}
// Extract extracts and returns the HTML from inside each element of the given selection.
func (e *HTML) Extract(doc *goquery.Document) (interface{}, error) {
func (e HTML) Extract(sel *goquery.Selection) (interface{}, error) {
var ret, h string
var err error
doc.Find(e.Selector).EachWithBreak(func(i int, s *goquery.Selection) bool {
sel.Find(e.Selector).EachWithBreak(func(i int, s *goquery.Selection) bool {
h, err = s.Html()
if err != nil {
return false
@ -40,9 +40,9 @@ type OuterHTML struct {
}
// Extract extracts and returns the HTML of each element of the given selection.
func (e *OuterHTML) Extract(doc *goquery.Document) (interface{}, error) {
func (e OuterHTML) Extract(sel *goquery.Selection) (interface{}, error) {
output := bytes.NewBufferString("")
for _, node := range doc.Find(e.Selector).Nodes {
for _, node := range sel.Find(e.Selector).Nodes {
if err := html.Render(output, node); err != nil {
return nil, err
}

18
extract/text.go
View File

@ -1,14 +1,22 @@
package extract
import "github.com/PuerkitoBio/goquery"
import (
"github.com/PuerkitoBio/goquery"
"strings"
)
// Text returns the combined text contents of provided selector.
type Text struct {
Name string
Selector string
Name string
Selector string
TrimSpace bool
}
// 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
func (e Text) Extract(sel *goquery.Selection) (interface{}, error) {
text := sel.Find(e.Selector).Text()
if e.TrimSpace {
text = strings.TrimSpace(text)
}
return map[string]string{e.Name: text}, nil
}