make the hoot code amenable to testing

hoot.go

@@ -35,10 +35,7 @@ var authorregex = regexp.MustCompile("twitter.com/([^/]+)")
 
 var re_hoots = regexp.MustCompile(`hoot: ?https://\S+`)
 
-func hooterize(noise string) string {
+func hootextractor(r io.Reader, url string, seen map[string]bool) string {
-	seen := make(map[string]bool)
-
-	hootfixer := func(r io.Reader, url string) string {
 	root, err := html.Parse(r)
 	if err != nil {
 		log.Printf("error parsing hoot: %s", err)
@@ -46,15 +43,15 @@ func hooterize(noise string) string {
 	}
 	divs := tweetsel.MatchAll(root)
 
+	var wanted string
 	wantmatch := authorregex.FindStringSubmatch(url)
-		if len(wantmatch) < 2 {
+	if len(wantmatch) == 2 {
-			log.Printf("no wanted author?")
+		wanted = wantmatch[1]
 	}
-		wanted := wantmatch[1]
 	var buf strings.Builder
 
-		var htf htfilter.Filter
 	fmt.Fprintf(&buf, "%s\n", url)
+	var htf htfilter.Filter
 	for _, div := range divs {
 		twp := div.Parent.Parent.Parent
 		alink := linksel.MatchFirst(twp)
@@ -69,6 +66,9 @@ func hooterize(noise string) string {
 			continue
 		}
 		author := authormatch[1]
+		if wanted == "" {
+			wanted = author
+		}
 		if author != wanted {
 			continue
 		}
@@ -84,7 +84,10 @@ func hooterize(noise string) string {
 		seen[text] = true
 	}
 	return buf.String()
-	}
+}
+
+func hooterize(noise string) string {
+	seen := make(map[string]bool)
 
 	hootfetcher := func(hoot string) string {
 		url := hoot[5:]
@@ -113,7 +116,7 @@ func hooterize(noise string) string {
 		}
 		ld, _ := os.Create("lasthoot.html")
 		r := io.TeeReader(resp.Body, ld)
-		return hootfixer(r, url)
+		return hootextractor(r, url, seen)
 	}
 
 	return re_hoots.ReplaceAllStringFunc(noise, hootfetcher)

hoot_test.go

@@ -0,0 +1,17 @@
+package main
+
+import (
+	"fmt"
+	"os"
+	"testing"
+)
+
+func TestHooterize(t *testing.T) {
+	fd, err := os.Open("lasthoot.html")
+	if err != nil {
+		return
+	}
+	seen := make(map[string]bool)
+	hoots := hootextractor(fd, "lasthoot.html", seen)
+	fmt.Printf("hoots: %s\n", hoots)
+}