Exercised the API endpoints with Unicode domains

dnstwister/api/__init__.py

@@ -67,7 +67,8 @@ def whois(hexdomain):
         )
     payload = standard_api_values(domain, skip='whois')
     try:
-        payload['whois_text'] = whois_mod.whois(domain).text.strip()
+        idna_domain = domain.encode('idna')
+        payload['whois_text'] = whois_mod.whois(idna_domain).text.strip()
         if payload['whois_text'] == '':
             raise Exception('No whois data retrieved')
     except Exception as ex:

dnstwister/api/checks/parked.py

@@ -32,12 +32,13 @@
 
 def _domain_redirects(domain, path=''):
     """Returns whether a domain (and optional path) redirects to another."""
+    idna_domain = domain.encode('idna')
     req = requests.get(
-        'http://{}/{}'.format(domain, path),
+        'http://{}/{}'.format(idna_domain, path),
         **shared.REQ_KWARGS
     )
     landed_domain = shared.get_domain(req.url)
-    return landed_domain != domain, landed_domain, req.content[:CONTENT_MAX]
+    return landed_domain != idna_domain, landed_domain, req.content[:CONTENT_MAX]
 
 
 def get_text(score):

dnstwister/api/checks/safebrowsing.py

@@ -14,8 +14,9 @@ def get_report(domain):
 
     Returns a count of matches.
     """
+    idna_domain = domain.encode('idna')
     data = {
-        'site': domain
+        'site': idna_domain
     }
 
     result = requests.get(API_URL, params=data)

tests/api/test_api.py

@@ -2,6 +2,8 @@
 import pytest
 import webtest.app
 
+from dnstwister import tools
+
 
 def test_api_root(webapp):
     """Test the API root."""
@@ -32,3 +34,15 @@ def test_api_domain_validation(webapp):
         with pytest.raises(webtest.app.AppError) as err:
             webapp.get('/api/{}/{}'.format(endpoint, malformed_domain))
         assert '400 BAD REQUEST' in err.value.message
+
+
+def test_unicode_basics(webapp):
+    """Test that Unicode domains work on all endpoints."""
+    unicode_domain = 'xn--sterreich-z7a.icom.museum'.decode('idna')
+    endpoints = ('fuzz', 'ip', 'parked', 'safebrowsing', 'whois')
+    for endpoint in endpoints:
+        webapp.get('/api/{}/{}'.format(
+            endpoint,
+            tools.encode_domain(unicode_domain),
+        ))
+    webapp.get('/api/to_hex/{}'.format(unicode_domain.encode('idna')))