From 335ee5c8d1453b8fd62a513261efaf1fcfa6d3ca Mon Sep 17 00:00:00 2001
From: theblackturtle <tuan.tv1966@sinhvien.hoasen.edu.vn>
Date: Tue, 11 Feb 2020 12:05:16 +0700
Subject: [PATCH] add alienvault.com source and new version

---
 core/crawler.go          | 30 +++++++++++++---------------
 core/othersource.go      | 43 ++++++++++++++++++++++++++++++++++++++++
 core/othersource_test.go |  9 +++++++++
 core/version.go          |  2 +-
 4 files changed, 67 insertions(+), 17 deletions(-)

diff --git a/core/crawler.go b/core/crawler.go
index e86b92c..2db5743 100644
--- a/core/crawler.go
+++ b/core/crawler.go
@@ -256,25 +256,23 @@ func (crawler *Crawler) Start() {
 		}
 
 		fileExt := GetExtType(jsFileUrl)
-		if fileExt != ".js" || fileExt != ".xml" || fileExt != ".json" {
-			return
-		}
+		if fileExt == ".js" || fileExt == ".xml" || fileExt == ".json" {
+			if !crawler.jsSet.Duplicate(jsFileUrl) {
+				outputFormat := fmt.Sprintf("[javascript] - %s", jsFileUrl)
+				fmt.Println(outputFormat)
+				if crawler.Output != nil {
+					crawler.Output.WriteToFile(outputFormat)
+				}
 
-		if !crawler.jsSet.Duplicate(jsFileUrl) {
-			outputFormat := fmt.Sprintf("[javascript] - %s", jsFileUrl)
-			fmt.Println(outputFormat)
-			if crawler.Output != nil {
-				crawler.Output.WriteToFile(outputFormat)
-			}
+				// If JS file is minimal format. Try to find original format
+				if strings.Contains(jsFileUrl, ".min.js") {
+					originalJS := strings.ReplaceAll(jsFileUrl, ".min.js", ".js")
+					crawler.linkFinder(originalJS)
+				}
 
-			// If JS file is minimal format. Try to find original format
-			if strings.Contains(jsFileUrl, ".min.js") {
-				originalJS := strings.ReplaceAll(jsFileUrl, ".min.js", ".js")
-				crawler.linkFinder(originalJS)
+				// Request and Get JS link
+				crawler.linkFinder(jsFileUrl)
 			}
-
-			// Request and Get JS link
-			crawler.linkFinder(jsFileUrl)
 		}
 	})
 
diff --git a/core/othersource.go b/core/othersource.go
index 764079d..ac26b86 100644
--- a/core/othersource.go
+++ b/core/othersource.go
@@ -21,6 +21,7 @@ func OtherSources(domain string, includeSubs bool) []string {
 		getWaybackURLs,
 		getCommonCrawlURLs,
 		getVirusTotalURLs,
+		getOtxUrls,
 	}
 
 	var wg sync.WaitGroup
@@ -170,3 +171,45 @@ func getVirusTotalURLs(domain string, noSubs bool) ([]wurl, error) {
 
 	return out, nil
 }
+
+func getOtxUrls(domain string, noSubs bool) ([]wurl, error) {
+	var urls []wurl
+	page := 0
+	for {
+		r, err := http.Get(fmt.Sprintf("https://otx.alienvault.com/api/v1/indicators/hostname/%s/url_list?limit=50&page=%d", domain, page))
+		if err != nil {
+			return []wurl{}, err
+		}
+		bytes, err := ioutil.ReadAll(r.Body)
+		if err != nil {
+			return []wurl{}, err
+		}
+		r.Body.Close()
+
+		wrapper := struct {
+			HasNext    bool `json:"has_next"`
+			ActualSize int  `json:"actual_size"`
+			URLList    []struct {
+				Domain   string `json:"domain"`
+				URL      string `json:"url"`
+				Hostname string `json:"hostname"`
+				Httpcode int    `json:"httpcode"`
+				PageNum  int    `json:"page_num"`
+				FullSize int    `json:"full_size"`
+				Paged    bool   `json:"paged"`
+			} `json:"url_list"`
+		}{}
+		err = json.Unmarshal(bytes, &wrapper)
+		if err != nil {
+			return []wurl{}, err
+		}
+		for _, url := range wrapper.URLList {
+			urls = append(urls, wurl{url: url.URL})
+		}
+		if !wrapper.HasNext {
+			break
+		}
+		page++
+	}
+	return urls, nil
+}
diff --git a/core/othersource_test.go b/core/othersource_test.go
index 4fbcc2b..45f6501 100644
--- a/core/othersource_test.go
+++ b/core/othersource_test.go
@@ -36,3 +36,12 @@ func TestGetWaybackURLs(t *testing.T) {
 	t.Log(len(urls))
 	t.Log(urls)
 }
+
+func TestGetOtxUrls(t *testing.T) {
+	urls, err := getOtxUrls(domain, false)
+	if err != nil {
+		t.Fatal(err)
+	}
+	t.Log(len(urls))
+	t.Log(urls)
+}
diff --git a/core/version.go b/core/version.go
index b8a6c7c..9d3d840 100644
--- a/core/version.go
+++ b/core/version.go
@@ -3,5 +3,5 @@ package core
 const (
 	CLIName = "gospider"
 	AUTHOR  = "@theblackturtle"
-	VERSION = "v1.0.7"
+	VERSION = "v1.0.8"
 )