fix: use title to check sha1 file names

pkg/crawler/crawler.go

@@ -256,10 +256,14 @@ func (c *Crawler) sha1Urls(ctx context.Context, url string) ([]string, error) {
 	// We need to take all links.
 	var sha1URLs []string
 	d.Find("a").Each(func(i int, selection *goquery.Selection) {
-		link := selection.Text()
-		if strings.HasSuffix(link, ".jar.sha1") && !strings.HasSuffix(link, "sources.jar.sha1") &&
-			!strings.HasSuffix(link, "test.jar.sha1") && !strings.HasSuffix(link, "javadoc.jar.sha1") {
-			sha1URLs = append(sha1URLs, url+link)
+		// There are times when the file name is very long.
+		// e.g. https://repo.maven.apache.org/maven2/africa/absa/inception-oauth2-resource-server/1.0.0/
+		// We need to use `title` to make sure we use the correct filename
+		if fileName, ok := selection.Attr("title"); ok {
+			if strings.HasSuffix(fileName, ".jar.sha1") && !strings.HasSuffix(fileName, "sources.jar.sha1") &&
+				!strings.HasSuffix(fileName, "test.jar.sha1") && !strings.HasSuffix(fileName, "javadoc.jar.sha1") {
+				sha1URLs = append(sha1URLs, url+fileName)
+			}
 		}
 	})
 	return sha1URLs, nil

pkg/crawler/testdata/abbot_abbot_0.12.3.html

@@ -19,7 +19,7 @@ <h1>abbot/abbot/0.12.3</h1>
 		<pre id="contents"><a href="https://repo.maven.apache.org/maven2/abbot/abbot/">../</a>
 <a href="https://repo.maven.apache.org/maven2/abbot/abbot/0.12.3/abbot-0.12.3.jar" title="abbot-0.12.3.jar">abbot-0.12.3.jar</a>                                  2005-09-20 05:44    689791      
 <a href="https://repo.maven.apache.org/maven2/abbot/abbot/0.12.3/abbot-0.12.3.jar.md5" title="abbot-0.12.3.jar.md5">abbot-0.12.3.jar.md5</a>                              2005-09-20 05:44        32      
-<a href="https://repo.maven.apache.org/maven2/abbot/abbot/0.12.3/abbot-0.12.3.jar.sha1" title="abbot-0.12.3.jar.sha1">abbot-0.12.3.jar.sha1</a>                             2005-09-20 05:44        40      
+<a href="https://repo.maven.apache.org/maven2/abbot/abbot/0.12.3/abbot-0.12.3.jar.sha1" title="abbot-0.12.3.jar.sha1">abbot-0.12.3.j...</a>                             2005-09-20 05:44        40
 <a href="https://repo.maven.apache.org/maven2/abbot/abbot/0.12.3/abbot-0.12.3.pom" title="abbot-0.12.3.pom">abbot-0.12.3.pom</a>                                  2005-09-20 05:44       166      
 <a href="https://repo.maven.apache.org/maven2/abbot/abbot/0.12.3/abbot-0.12.3.pom.md5" title="abbot-0.12.3.pom.md5">abbot-0.12.3.pom.md5</a>                              2005-09-20 05:44       128      
 <a href="https://repo.maven.apache.org/maven2/abbot/abbot/0.12.3/abbot-0.12.3.pom.sha1" title="abbot-0.12.3.pom.sha1">abbot-0.12.3.pom.sha1</a>                             2005-09-20 05:44       136