Initial checkin for final project

Author: gosunfish

Students/CarolynEvans/final_project/__init__.py

@@ -0,0 +1,42 @@
+import yaml
+
+class config():
+    """
+    This class parses config.yaml into a dictionary and provides callability.
+    """
+    _VALUES = dict()
+
+    def __init__(self):
+        with open('config.yaml', 'r') as f:
+            config = yaml.load(f)
+
+        self.flatten(config)
+
+
+    def flatten(self, config):
+        """
+        This method loads a dictionary using the data that was loaded from config.yaml in __init__.
+        The yaml dictionary 'values' may potentially be another dictionary.
+        This method is used recursively to explode all levels of the yaml into a flat dictionary.
+        :param config: A dict containing configuration values.
+        :return: There is no return value. This method is called during __init__.
+        """
+        # TODO: Rewrite for loop as comprehension
+
+        for key, val in config.items():
+            if type(val) is dict:
+                self.flatten(val)
+            else:
+                self._VALUES[key]=val
+
+
+
+    def get(self, key):
+        """
+        This method is provides a configuration value for the given 'key'.
+        :param key: A string containing the 'key' of the config item in the config dict.
+        :return: A string containing the value of the config item in the config dict.
+        """
+        val = self._VALUES[key]
+        return val
+

Students/CarolynEvans/final_project/config.py

@@ -0,0 +1,20 @@
+data_warehouse_config:
+  dw_dbname: "rightsidedw"
+  dw_user: "rouser"
+  dw_password: "RedshiftRO1"
+  dw_host: "rightside-dw.cwisduywvaeb.us-east-1.redshift.amazonaws.com"
+  dw_port: "5439"
+#
+#data_warehouse_config:
+#  dw_dbname: rightsidedw
+#  dw_user: carolyn.evans
+#  dw_password: PassW0rd
+#  dw_host: localhost
+#  dw_port: 5439
+
+hummingbird_config:
+  hummingbird_url: http://hummingbird.dizzyninja.co/
+
+email_config:
+  email_server: mail.notify-customer.com
+

Students/CarolynEvans/final_project/config.yaml

@@ -0,0 +1,54 @@
+import psycopg2
+from psycopg2._psycopg import ProgrammingError
+from config import config
+
+class data_warehouse(object):
+    '''
+    This class is used to get data from the Rightside data warehouse.
+    The complete data warehouse documentation can be found at
+        https://wiki.rightside.net/display/tnt/BI
+    '''
+
+    def __init__(self):
+        """
+        This method sets the data warehouse connection string from the config.
+        """
+        cnf = config()
+        dbname = cnf.get('dw_dbname')
+        user = cnf.get('dw_user')
+        password = cnf.get('dw_password')
+        host = cnf.get('dw_host')
+        port = cnf.get('dw_port')
+
+        self.dw_connection_string = \
+         "dbname='{}' user='{}' password='{}' host='{}' port='{}'".format(dbname,user,password,host,port)
+
+
+    def __call__(self, query):
+        """
+        This method calls the data warehouse API.
+        The complete data warehouse documentation can be found at
+            https://wiki.rightside.net/display/tnt/BI
+        :param query: A string containing a valid redshift data warehouse query.
+        :return: This depends on the query.
+                 If the query returns records, as in a 'select' query, the records are returned.
+                 If the query does not return records, as in an 'update' query, a unicode string is returned.
+        """
+        with psycopg2.connect(self.dw_connection_string) as conn:
+            with conn.cursor() as curs:
+                curs.execute(query)
+
+                try:
+                    records = curs.fetchall()
+                    return records
+                except ProgrammingError as e:
+                    # This error occurs when there are no records to return,
+                    # for example, queries such as inserts, updates, and unloads.
+                    message = unicode(e)
+                    if message == u'no results to fetch':
+                        return message
+                    else:
+                        raise message
+
+
+

Students/CarolynEvans/final_project/data_warehouse.py

@@ -0,0 +1,40 @@
+import urllib2
+import json
+
+from config import config
+
+class hummingbird(object):
+    """
+    This class is used to call the DizzyNinja Hummingbird Domain Recommendation API.
+    The complete hummingbird API documentation can be found at
+        https://wiki.rightside.net/display/tnt/API+Documentation
+    """
+
+    def __init__(self):
+        """
+        This method sets the hummingbird url from the config.
+        """
+        cnf = config()
+        self.url = cnf.get('hummingbird_url')
+
+
+    def __call__(self, command, **kwargs):
+        """
+        This method calls the hummingbird API.
+        The complete hummingbird API documentation can be found at
+        https://wiki.rightside.net/display/tnt/API+Documentation
+        :param command: A string containing a valid hummingbird command.
+        :param kwargs: Depends on 'command'. See hummingbird documentation for complete list.
+        :return: Returns JSON with results for the 'command'.
+        """
+
+        # build the query string parameter for the hummingbird API http request.
+        query_string = "".join(['{}={},'.format(key,val) for key, val in kwargs.items()])
+
+        #add the 'command' and the query string to the url.
+        url = '{}{}?{}'.format(self.url, command, query_string)
+
+        response = urllib2.urlopen(url).read()
+        return json.loads(response)
+
+

Students/CarolynEvans/final_project/hummingbird.py

@@ -0,0 +1,12 @@
+
+# TODO:  Get list of registered domain names from data warehouse
+# TODO:  Get list of premium domains from data warehouse
+# TODO:  Call word splitter on both lists and save each in a dict with
+#        original domain name as key and split words as values.
+# TODO:  Sort split words for each domain name.
+# TODO:  Compare split words for registered domain to split words for premium domains.
+# TODO:  Get registered owners contact info from data warehouse.
+# TODO:  Write file that can be opened in excel.
+
+
+

Students/CarolynEvans/final_project/lead_gen.py

@@ -0,0 +1,28 @@
+import smtplib
+from config import config
+
+
+def send_mail(email_address, subject, body=''):
+    """
+    This function sends email notifications to the current user.
+
+    :param email_address: This is the email address of the current user that was obtained from the ldap call.
+    :param subject: This parameter will be used as the subject line of the email.
+    :param body: This parameter will be used as the body of the email.
+    """
+
+    # If email is None, the task is being run from the command line and no email is sent.
+    if email_address is None:
+        return
+
+    cnf = config()
+    SERVER = cnf.get('email_server')
+    FROM = "[email protected]"
+    TO = [email_address]
+    TEXT = 'Subject: %s\n\n%s' % (subject, body)
+
+    # Send the mail
+    server = smtplib.SMTP(SERVER)
+    server.sendmail(FROM, TO, TEXT)
+
+

Students/CarolynEvans/final_project/send_mail.py

@@ -0,0 +1,51 @@
+def test_config():
+    # Test config.py
+    from config import config
+
+    cnf = config()
+    assert cnf.get('hummingbird_url') == 'http://hummingbird.dizzyninja.co/'
+    assert cnf.get('dw_dbname') == 'rightsidedw'
+    assert cnf.get('email_server') == 'mail.notify-customer.com'
+
+
+def test_hummingbird():
+    # Test hummingbird.py
+    # TODO: Add more hummingbird commands to the test.
+    from hummingbird import hummingbird
+
+    h = hummingbird()
+    result = h('word-split', input='toiletseat')
+    assert result['output']['split'] == 'toilet seat'
+
+
+def test_data_warehouse():
+    # Test data_warehouse.py
+    # This test dumps a database table to a file.
+    # TODO: Add a select query and an update query to the test.
+    from data_warehouse import data_warehouse
+
+    dw = data_warehouse()
+    # TODO: Make the following line readable.
+    query = "unload ('SELECT * FROM utld.tld') to 's3://rightside-data/whois/output/ActiveDomainStatus_20141210_ce.txt' with CREDENTIALS 'aws_access_key_id=AKIAI5ZAKRM5QRQ3LJXA;aws_secret_access_key=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' DELIMITER '\t' PARALLEL OFF ALLOWOVERWRITE GZIP;"
+    assert dw(query) == u'no results to fetch'
+    #result = dw(query)
+    #print result
+    x='x'
+
+
+
+def test_send_mail():
+    # Test send_mail.py
+    # The send_mail function only works inside the firewall.
+    from send_mail import send_mail
+    email_address = '[email protected]'
+    subject = 'test.py'
+    assert send_mail(email_address=email_address, subject=subject, body="") == None
+
+
+if __name__ == '__main__':
+    test_config()
+    test_data_warehouse()
+    test_hummingbird()
+    #test_send_mail()
+