First tests

test_urlchecker_status.py

+import unittest
+from urlchecker_db import LogDB
+from urlchecker_status import logStatus
+
+
+class UrlCheckerStatusTestCase(unittest.TestCase):
+    def setUp(self):
+        self.db = LogDB(":memory:")
+        self.db.createTables()
+
+    def test_logStatus_insert(self):
+        result = logStatus(self.db, "foo")
+        assert self.db.Status.select().count() == 1
+        assert self.db.Status.get(self.db.Status.text == "foo").id == result
+
+    def test_logStatus_insertTwice(self):
+        result1 = logStatus(self.db, "foo")
+        result2 = logStatus(self.db, "foo")
+        assert self.db.Status.select().count() == 2
+        assert result1 < result2
+
+
+def suite():
+    suite = unittest.TestSuite()
+    suite.addTest(unittest.makeSuite(UrlCheckerStatusTestCase))
+    return suite
+
+
+if __name__ == "__main__":
+    unittest.main(defaultTest="suite")

urlchecker_db.py

@@ -6,7 +6,7 @@ from playhouse.sqlite_ext import SqliteExtDatabase
 class LogDB:
     def __init__(self, sqlite_path):
         self._db = SqliteExtDatabase(
-            sqlite_path, pragmas=(("journal_mode", "WAL"),)
+            sqlite_path, pragmas=(("journal_mode", "WAL"), ("foreign_keys", 1))
         )
         self._db.connect()
 
@@ -62,6 +62,8 @@ class LogDB:
             ip = peewee.TextField(index=True)
             url = peewee.TextField(index=True)
             status_code = peewee.IntegerField()
+            class Meta:
+                primary_key = peewee.CompositeKey("status", "ip", "url")
 
         self.Status = Status
         self.ConfigurationChange = ConfigurationChange

urlchecker_http.py

@@ -13,7 +13,7 @@ def getUrlHostname(url):
     return urlparse(url).hostname
 
 
-def getUserAgent(self, version="0"):
+def getUserAgent(version):
     return "%s/%s (+%s)" % (
         "URLCHECKER",
         version,
@@ -22,16 +22,10 @@ def getUserAgent(self, version="0"):
 
 
 def request(
-    method,
     url,
     headers=None,
-    stream=False,
-    timeout=TIMEOUT,
-    allow_redirects=False,
-    verify=True,
     session=requests,
-    version=None,
-    **kwargs,
+    version=0
 ):
 
     if headers is None:
@@ -42,11 +36,12 @@ def request(
         # XXX user agent
         headers["User-Agent"] = getUserAgent(version)
 
-    kwargs["stream"] = stream
-    kwargs["timeout"] = timeout
-    kwargs["allow_redirects"] = allow_redirects
-    kwargs["verify"] = verify
-    args = [method, url]
+    kwargs = {}
+    kwargs["stream"] = False
+    kwargs["timeout"] = TIMEOUT
+    kwargs["allow_redirects"] = False
+    kwargs["verify"] = True
+    args = ["GET", url]
 
     kwargs["headers"] = headers
 
@@ -89,33 +84,34 @@ def logHttpStatus(db, ip, url, code, status_id):
             previous_entry = db.HttpCodeChange.create(
                 status=status_id, ip=ip, url=url, status_code=code
             )
-        return previous_entry.id
+        return previous_entry.status
 
 
 def checkHttpStatus(db, status_id, url, ip, bot_version):
     parsed_url = urlparse(url)
     hostname = parsed_url.hostname
-
-    session = requests.Session()
+    request_kw = {}
     # SNI Support
     if parsed_url.scheme == "https":
         # Provide SNI support
         base_url = urlunsplit(
             (parsed_url.scheme, parsed_url.netloc, "", "", "")
         )
+        session = requests.Session()
         session.mount(base_url, ForcedIPHTTPSAdapter(dest_ip=ip))
+        request_kw['session'] = session
         ip_url = url
     elif parsed_url.scheme == "http":
         # Force IP location
-
         parsed_url = parsed_url._replace(netloc=ip)
         ip_url = parsed_url.geturl()
+    else:
+        raise NotImplementedError('Unhandled url: %s' % url)
 
     response = request(
-        "GET",
         ip_url,
         headers={"Host": hostname},
-        session=session,
         version=bot_version,
+        **request_kw
     )
     logHttpStatus(db, ip, url, response.status_code, status_id)