Refactor codebase into modules

2 files changed, 74 insertions, 12 deletions

diff --git a/modules/followlinks/followlinks.go b/modules/followlinks/followlinks.go
index 99d6cee..c53f167 100644
--- a/modules/followlinks/followlinks.go
+++ b/modules/followlinks/followlinks.go
@@ -5,19 +5,71 @@
 package followlinks
 
 import (
+	"net/url"
+	"strings"
+
+	"github.com/PuerkitoBio/goquery"
 	"github.com/philippta/flyscrape"
 )
 
 func init() {
-	flyscrape.RegisterModule(new(Module))
+	flyscrape.RegisterModule(Module{})
 }
 
 type Module struct{}
 
-func (m *Module) OnResponse(resp *flyscrape.Response) {
-	for _, link := range flyscrape.ParseLinks(string(resp.Body), resp.Request.URL) {
+func (Module) ModuleInfo() flyscrape.ModuleInfo {
+	return flyscrape.ModuleInfo{
+		ID:  "followlinks",
+		New: func() flyscrape.Module { return new(Module) },
+	}
+}
+
+func (m *Module) ReceiveResponse(resp *flyscrape.Response) {
+	for _, link := range parseLinks(string(resp.Body), resp.Request.URL) {
 		resp.Visit(link)
 	}
 }
 
-var _ flyscrape.OnResponse = (*Module)(nil)
+func parseLinks(html string, origin string) []string {
+	var links []string
+	doc, err := goquery.NewDocumentFromReader(strings.NewReader(html))
+	if err != nil {
+		return nil
+	}
+
+	originurl, err := url.Parse(origin)
+	if err != nil {
+		return nil
+	}
+
+	uniqueLinks := make(map[string]bool)
+	doc.Find("a").Each(func(i int, s *goquery.Selection) {
+		link, _ := s.Attr("href")
+
+		parsedLink, err := originurl.Parse(link)
+
+		if err != nil || !isValidLink(parsedLink) {
+			return
+		}
+
+		absLink := parsedLink.String()
+
+		if !uniqueLinks[absLink] {
+			links = append(links, absLink)
+			uniqueLinks[absLink] = true
+		}
+	})
+
+	return links
+}
+
+func isValidLink(link *url.URL) bool {
+	if link.Scheme != "" && link.Scheme != "http" && link.Scheme != "https" {
+		return false
+	}
+
+	return true
+}
+
+var _ flyscrape.ResponseReceiver = (*Module)(nil)
diff --git a/modules/followlinks/followlinks_test.go b/modules/followlinks/followlinks_test.go
index 18c8ceb..0a628c3 100644
--- a/modules/followlinks/followlinks_test.go
+++ b/modules/followlinks/followlinks_test.go
@@ -5,27 +5,37 @@
 package followlinks_test
 
 import (
+	"net/http"
+	"sync"
 	"testing"
 
 	"github.com/philippta/flyscrape"
 	"github.com/philippta/flyscrape/modules/followlinks"
+	"github.com/philippta/flyscrape/modules/hook"
 	"github.com/philippta/flyscrape/modules/starturl"
 	"github.com/stretchr/testify/require"
 )
 
 func TestFollowLinks(t *testing.T) {
+	var urls []string
+	var mu sync.Mutex
+
 	scraper := flyscrape.NewScraper()
 	scraper.LoadModule(&starturl.Module{URL: "http://www.example.com/foo/bar"})
 	scraper.LoadModule(&followlinks.Module{})
 
-	scraper.SetTransport(flyscrape.MockTransport(200, `
-        <a href="/baz">Baz</a>
-        <a href="baz">Baz</a>
-        <a href="http://www.google.com">Google</a>`))
-
-	var urls []string
-	scraper.OnRequest(func(req *flyscrape.Request) {
-		urls = append(urls, req.URL)
+	scraper.LoadModule(hook.Module{
+		AdaptTransportFn: func(rt http.RoundTripper) http.RoundTripper {
+			return flyscrape.MockTransport(200, `
+				<a href="/baz">Baz</a>
+				<a href="baz">Baz</a>
+				<a href="http://www.google.com">Google</a>`)
+		},
+		ReceiveResponseFn: func(r *flyscrape.Response) {
+			mu.Lock()
+			urls = append(urls, r.Request.URL)
+			mu.Unlock()
+		},
 	})
 
 	scraper.Run()