Merge branch 'release/0.1.7'

presser/exceptions.py

@@ -1,15 +1,18 @@
 
-class PresserJavaScriptParseError(Exception):
+class PresserError(Exception):
+	pass
+
+class PresserJavaScriptParseError(PresserError):
     pass
 
-class PresserInvalidVineIdError(Exception):
+class PresserInvalidVineIdError(PresserError):
     pass
 
-class PresserURLError(Exception):
+class PresserURLError(PresserError):
     pass
 
-class Presser404Error(Exception):
+class Presser404Error(PresserError):
     pass
 
-class PresserRequestError(Exception):
+class PresserRequestError(PresserError):
     pass

presser/presser.py

@@ -7,17 +7,14 @@
 from .exceptions import PresserJavaScriptParseError, PresserURLError, Presser404Error, PresserRequestError, PresserInvalidVineIdError
 
 class Presser:
-    def get_data_for_vine_id(self, vine_id):
+    def get_data_for_vine_id(self, vine_id, timeout=30):
         try:
-            page = requests.get("https://vine.co/v/{}".format(vine_id))
+            page = requests.get("https://vine.co/v/{}".format(vine_id), timeout=timeout)
         except requests.exceptions.RequestException as e:
             error_message = "Problem with comminicating with vine page - {}".format(e)
             raise PresserRequestError(error_message)
         if page.ok:
             content = BeautifulSoup(page.content)
-            if content.find("title").text == u'Vine':
-                if not content.find("body").text.count(u"Video:"):
-                    raise Presser404Error("Could not find Vine Id {}".format(vine_id))
             all_script_tags = content.find_all("script")
             potential_script_tags = [script for script in all_script_tags if not script.has_attr("src")]
             script_lines = []
@@ -38,16 +35,18 @@ def get_data_for_vine_id(self, vine_id):
             except execjs.RuntimeError as e:
                 error_message = "Problem with parsing, check parsing logic. {}".format(e)
                 raise PresserJavaScriptParseError(error_message)
+        elif page.status_code == 404:
+            raise Presser404Error("{} could not be found".format(page.url))
         else:
             raise PresserURLError("{} could not be accessed {} - {}".format(page.url, page.status_code,page.content))
 
-    def get_data_for_vine_from_url(self, url):
+    def get_data_for_vine_from_url(self, url, timeout=30):
         parsed_url = urllib.parse.urlparse(url)
         if parsed_url.netloc == "vine.co":
             results = re.search('/v/(?P<vine_id>\w+)',parsed_url.path)
             if results:
                 vine_id = results.group("vine_id")
-                return self.get_data_for_vine_id(vine_id)
+                return self.get_data_for_vine_id(vine_id, timeout=timeout)
             else:
                 raise PresserInvalidVineIdError("{} does not contain a valid vine id".format(parsed_url.path))    
         else:

setup.py

@@ -4,7 +4,7 @@
     long_description = readme_file.read()
 
 setup(name="Presser",
-    version="0.1.6",
+    version="0.1.7",
     packages=["presser",],
     license="GNU GPL v3.0",
     description="Extracts data from vine, in lieu of an API",
@@ -38,9 +38,9 @@
         "Programming Language :: Python :: 3.3",
         "Programming Language :: Python :: 3.4",
         "Programming Language :: Python :: 2.7",
-        "Programming Language :: Python :: Implementation",
+        # "Programming Language :: Python :: Implementation",
         "Programming Language :: Python :: Implementation :: CPython",
-        # "Programming Language :: Python :: Implementation :: PyPy",
+        "Programming Language :: Python :: Implementation :: PyPy",
         "Programming Language :: Python",
         "Intended Audience :: Developers",
     ]

tests/unit.py

@@ -2,7 +2,7 @@
 import responses
 import requests
 
-from mock import patch
+from mock import patch, MagicMock
 
 from presser.presser import Presser
 from presser.exceptions import Presser404Error, PresserURLError, PresserInvalidVineIdError, PresserJavaScriptParseError, PresserRequestError
@@ -30,7 +30,7 @@ def test_not_a_valid_vine_id(self):
     @patch('presser.presser.Presser.get_data_for_vine_id')
     def test_vine_id_extraction(self, vine_response):
         vine = self.presser.get_data_for_vine_from_url(VINE_URL)
-        self.presser.get_data_for_vine_id.assert_called_with(VINE_ID)
+        self.presser.get_data_for_vine_id.assert_called_with(VINE_ID, timeout=30)
 
     @responses.activate
     def test_vine_data_extraction(self):
@@ -47,7 +47,7 @@ def test_404_detection_logic(self):
         with open("tests/404.html") as not_found_html:
             body = not_found_html.read()
         responses.add(responses.GET, NOT_FOUND_URL,
-                    body=body, status=200,
+                    body=body, status=404,
                     content_type='text/html')
         self.assertRaises(Presser404Error, self.presser.get_data_for_vine_from_url, NOT_FOUND_URL)
 
@@ -84,4 +84,10 @@ def test_error_request(self):
 
     @patch("requests.models.Response.ok", False)
     def test_page_not_okay(self):
-        self.assertRaises(PresserURLError, self.presser.get_data_for_vine_from_url, VINE_URL)
+        self.assertRaises(PresserURLError, self.presser.get_data_for_vine_from_url, VINE_URL)
+
+    @patch("requests.get")
+    def test_timeout_passed_through(self, request_mock):
+        #Yes this is hacky, BUT responses doesn't record the timeout parameter
+        self.assertRaises(TypeError, self.presser.get_data_for_vine_from_url, VINE_URL, timeout=5)
+        requests.get.assert_called_with(VINE_URL, timeout=5)