added PDFcheck

modules/ConfigTree.py

@@ -5,7 +5,7 @@
 
 # TODO: Clean boolean in string.
 
-def ConfigTree(verifyApi,socialscan,leakeddb,breachedsites,hunterApi,dbdata,tcrwd,pastebindumps,googlesearch,dns, _verbose=None):
+def ConfigTree(verifyApi,socialscan,leakeddb,breachedsites,hunterApi,checkPDF,dbdata,tcrwd,pastebindumps,googlesearch,dns, _verbose=None):
 	if _verbose != None:
 		try:
 			fileshow = Node(f"{bcolors.BOLD}Config File{bcolors.ENDC} [Modules]")
@@ -43,6 +43,11 @@ def ConfigTree(verifyApi,socialscan,leakeddb,breachedsites,hunterApi,dbdata,tcrw
 				hntr11 = Node('\x1b[6;30;42m'+'True'+ '\x1b[0m', parent=hntr1)
 			else:
 				hntr11 = Node('\x1b[1;31;40m'+'False'+ '\x1b[0m', parent=hntr1)
+			pdf = Node("PDF Check", parent=fileshow)
+			if (checkPDF == "True" or checkPDF == "true"):
+				pdf1 = Node('\x1b[6;30;42m'+'True'+ '\x1b[0m', parent=pdf)
+			else:
+				pdf1 = Node('\x1b[1;31;40m'+'False'+ '\x1b[0m', parent=pdf)
 			dbdt1 = Node("Related Phone Numbers", parent=fileshow)
 			if (dbdata == "True" or dbdata == "true"):
 				dbdt11 = Node('\x1b[6;30;42m'+dbdata+ '\x1b[0m', parent=dbdt1)

modules/PDFcheck.py

@@ -0,0 +1,32 @@
+from googlesearch import search
+import requests,re,PyPDF2,tabula,os
+
+def PDFcheck(mail,_verbose=None):
+	domain = mail.split("@")[1]
+	term = "site:"+domain+" filetype:PDF intext:"+'"'+"email"+'"'
+	try:
+		data = search(term, num_results=5)
+		for i in data:
+			r = requests.get(i, stream=True)
+			with open('data.pdf', 'wb') as f:
+				f.write(r.content)
+			pdfFileObj = open('data.pdf', 'rb')
+			for pageNumber in range(1,3):
+				tabula.convert_into("data.pdf","out.txt",pages=pageNumber,silent=True)
+				file = open("out.txt","r",encoding="utf-8")
+				read = file.read()
+				findPDFs= re.findall('[\w\.-]+@[a-z0-9\.-]+', read)
+				try:
+					if(findPDFs[0] is not None):
+						for pdfs in findPDFs:
+							print(pdfs)
+				except:
+					pass
+			pdfFileObj.close()
+			file.close()
+			if os.path.exists("data.pdf"):
+				os.remove("data.pdf")
+			if os.path.exists("out.txt"):
+				os.remove("out.txt")
+	except:
+		print("PDF Search error!")

mosint.py

@@ -3,42 +3,48 @@
 from insides.bcolors import bcolors
 from insides.Header import Header
 
-with open('config.json', "r") as configFile:
-    conf = json.loads(configFile.read())
+def parse_args():
+    import argparse
+    parser = argparse.ArgumentParser()
+    parser.add_argument('-e', '--email', type=str, required=True, help="Email")
+    return parser.parse_args()
 
-for i in conf:
-    verifyApi = (i['verify-email.org API Key'])
-    socialscan = (i['Social Scan'])
-    leakeddb = (i['Leaked DB'])
-    breachedsites = (i['Breached Sites[leak-lookup.com API Key]'])
-    hunterApi = (i['hunter.io API Key'])
-    dbdata = (i['Related Phone Numbers'])
-    tcrwd = (i['Related Domains'])
-    pastebindumps = (i['Pastebin Dumps'])
-    googlesearch = (i['Google Search'])
-    dns = (i['DNS Lookup'])
 
-from insides.Banner import Banner
-Banner()
+# TODO: Clean boolean in string.
 
-from modules.ConfigTree import ConfigTree
-ConfigTree(verifyApi,socialscan,leakeddb,breachedsites,hunterApi,dbdata,tcrwd,pastebindumps,googlesearch,dns,_verbose=True)
+def main():
+    args = parse_args()
+    mail = args.email
 
-print("")
+    EMAIL_REGEX = r'(^[a-zA-Z0-9_.+-]+@[a-zA-Z0-9-]+\.[a-zA-Z0-9-.]+$)'
 
-# TODO: Clean boolean in string.
+    if not re.match(EMAIL_REGEX, mail):
+        print(f"{bcolors.FAIL}Email format is wrong!{bcolors.ENDC}")
+        exit()
 
-EMAIL_REGEX = r'(^[a-zA-Z0-9_.+-]+@[a-zA-Z0-9-]+\.[a-zA-Z0-9-.]+$)'
+    with open('config.json', "r") as configFile:
+        conf = json.loads(configFile.read())
 
-while True:
-    mail=input(f"{bcolors.OKBLUE}MAIL > {bcolors.ENDC}")
+    for i in conf:
+        verifyApi = (i['verify-email.org API Key'])
+        socialscan = (i['Social Scan'])
+        leakeddb = (i['Leaked DB'])
+        breachedsites = (i['Breached Sites[leak-lookup.com API Key]'])
+        hunterApi = (i['hunter.io API Key'])
+        checkPDF = (i['PDF Check for Related Emails'])
+        dbdata = (i['Related Phone Numbers'])
+        tcrwd = (i['Related Domains'])
+        pastebindumps = (i['Pastebin Dumps'])
+        googlesearch = (i['Google Search'])
+        dns = (i['DNS Lookup'])
 
-    if (mail == "q" or mail == "Q" or mail == "exit"):
-        print("Thank you for using "+f"{bcolors.BOLD}MOSINT{bcolors.ENDC}.")
-        break
-    elif not re.match(EMAIL_REGEX, mail):
-        print(f"{bcolors.FAIL}Email format is wrong!{bcolors.ENDC}")
-        continue
+    from insides.Banner import Banner
+    Banner()
+
+    from modules.ConfigTree import ConfigTree
+    ConfigTree(verifyApi,socialscan,leakeddb,breachedsites,hunterApi,checkPDF,dbdata,tcrwd,pastebindumps,googlesearch,dns,_verbose=True)
+
+    print("")
 
     if (verifyApi != ""): 
         from modules.VerifyMail import VerifyMail
@@ -70,6 +76,12 @@
         Header(title)
         Hunter(mail,hunterApi,_verbose=True)
 
+    if (checkPDF == "True" or checkPDF == "T" or checkPDF == "true"):
+        from modules.PDFcheck import PDFcheck
+        title = "RELATED EMAILS IN PDFs"
+        Header(title)
+        PDFcheck(mail,_verbose=True)
+
     if (dbdata == "True" or dbdata == "T" or dbdata == "true"):
         from modules.RelatedNumbers import RelatedNumbers
         title = "RELATED PHONE NUMBERS"
@@ -99,4 +111,6 @@
         title = "DNS LOOKUP"
         Header(title)
         DNS(mail,_verbose=True)
+
+main()