def test_toCheckEpisodeAddsEpisodeObjectID(self):
self.ep.setEpisodeID("EA1004H")
self.assertEqual(Episode.getShowID(self.ep.episodeID), '1004')
self.ep.setEpisodeID("EASP06H")
self.assertEqual(Episode.getShowID(self.ep.episodeID), '1506')
self.ep.setEpisodeID("EA1G02")
self.assertEqual(Episode.getShowID(self.ep.episodeID), '0702')
def test_toCheckEpisodeForSeason(self):
self.ep.setEpisodeID("EA1004H")
self.assertEqual(Episode.getSeasonNumber(self.ep.episodeID), 10)
self.ep.setEpisodeID("EASP06H")
self.assertEqual(Episode.getSeasonNumber(self.ep.episodeID), 15)
self.ep.setEpisodeID("EA1G02")
self.assertEqual(Episode.getSeasonNumber(self.ep.episodeID), 7)
self.ep.setEpisodeID("")
self.assertEqual(Episode.getSeasonNumber(self.ep.episodeID), 0)
def test_toCheckEpisodeForSeasonEpisodeNumber(self):
self.ep.setEpisodeID("EA1004H")
self.assertEqual(Episode.getSeasonEpisodeNumber(self.ep.episodeID), 4)
self.ep.setEpisodeID("EASP06H")
self.assertEqual(Episode.getSeasonEpisodeNumber(self.ep.episodeID), 6)
self.ep.setEpisodeID("EA1G02")
self.assertEqual(Episode.getSeasonEpisodeNumber(self.ep.episodeID), 2)
self.ep.setEpisodeID("")
self.assertEqual(Episode.getSeasonEpisodeNumber(self.ep.episodeID), None)
def setUp(self):
self.ep = Episode('dummyTitle')
class TestEpisode(unittest.TestCase):
"""docstring for TestEpisode"""
def setUp(self):
self.ep = Episode('dummyTitle')
# Sanity check test
def test_Sanity(self):
self.assertEqual(True, True)
# Check Episode Constructor
# def test_EpisodeForID(self):
# self.assertEqual(self.ep.showID, "dummyID")
# Check for Episode Title
def test_EpisodeForTitle(self):
self.assertEqual(self.ep.showTitle, "dummyTitle")
# Check for Show Number
def test_toCheckEpisodeNumberRemovesOrdinalEnding(self):
# check removes 'st' from 'x1st'
self.ep.setEpisodeNumber('21st')
self.assertEqual(self.ep.episodeNumber, '21')
self.ep.setEpisodeNumber('1ST')
self.assertEqual(self.ep.episodeNumber, '1')
# check removes 'nd' from 'x2nd'
self.ep.setEpisodeNumber('92nd')
self.assertEqual(self.ep.episodeNumber, '92')
# check removes 'rd' from 'x3rd'
self.ep.setEpisodeNumber('33rd')
self.assertEqual(self.ep.episodeNumber, '33')
# check removes 'th' from 'x[4-9]th'
self.ep.setEpisodeNumber('24th')
self.assertEqual(self.ep.episodeNumber, '24')
# Check for xxx(st|nd|rd|th) followed by other text
self.ep.setEpisodeNumber('199th Note: If the cranberry episode, Cran Opening, had aired in the US during the same season as it was created, this would have been the 200th episode in the USA.')
self.assertEqual(self.ep.episodeNumber, '199')
# Check for topics list
def test_toCreateListFromAddedTopics(self):
self.ep.addTopics("Turkeys, Brining, Electric Knives, Stuffing, Basting, Probe Thermometers")
self.assertEqual(self.ep.topicList, sorted(["Turkeys","Brining","Electric", "Knives","Stuffing","Basting","Probe", "Thermometers"]))
self.ep.addTopics("Bread Pudding, Bread Stalling, Amylose, Half & Half")
self.assertEqual(self.ep.topicList, sorted(["Bread", "Pudding", "Stalling", "Amylose", "Half"]))
# Check for recipe list
def test_toCreateListfromAddedRecipes(self):
self.ep.addRecipes("1) Beef Tenderloin in Salt Crust 2) Perfect Fingerling Potatoes 3) Salt Roasted Shrimp 4) Sauerkraut")
self.assertEqual(self.ep.recipeList, ["Beef Tenderloin in Salt Crust","Perfect Fingerling Potatoes","Salt Roasted Shrimp","Sauerkraut"])
self.ep.addRecipes("1. Dan Dan Noodles 2. Ants in Trees 3. Thai Shrimp Spring Rolls")
self.assertEqual(self.ep.recipeList, ["Dan Dan Noodles","Ants in Trees","Thai Shrimp Spring Rolls"])
self.ep.addRecipes("None given at FN.com")
self.assertEqual(self.ep.recipeList,[])
self.ep.addRecipes("--none--")
self.assertEqual(self.ep.recipeList,[])
self.ep.addRecipes("")
self.assertEqual(self.ep.recipeList,[])
self.ep.addRecipes(".")
self.assertEqual(self.ep.recipeList,[])
# Check for air date
def test_toCheckEpisodeForOriginalAirDate(self):
self.ep.addAirDate("02.10.2011")
self.assertEqual(self.ep.airDate, "2011-02-10T00:00:00")
self.ep.addAirDate("7.11.2001")
self.assertEqual(self.ep.airDate, "2001-07-11T00:00:00")
self.ep.addAirDate("07.??.1998: WTTW, Ch. 11, Chicago, IL 07.07.1999: FoodTV")
self.assertEqual(self.ep.airDate, "1999-07-07T00:00:00")
self.ep.addAirDate("11.14.1999 (During Season 1)")
self.assertEqual(self.ep.airDate, "1999-11-14T00:00:00")
self.ep.addAirDate("12.08.2008 in the US 05.26.2008 (maybe) in Canada")
self.assertEqual(self.ep.airDate, "2008-12-08T00:00:00")
self.ep.addAirDate("10.10.2009 at 10:00 pm")
self.assertEqual(self.ep.airDate, "2009-10-10T00:00:00")
self.ep.addAirDate("10.26..2009")
self.assertEqual(self.ep.airDate, "2009-10-26T00:00:00")
# Check for transcript ID
def test_toCheckEpisodeForTranscriptID(self):
self.ep.addTranscriptID("Turn On the Dark")
self.assertEqual(self.ep.transcriptID, "Turn On the Dark")
self.ep.addTranscriptID("")
self.assertEqual(self.ep.transcriptID, "")
# Check if correct season is set
def test_toCheckEpisodeForEpisodeID(self):
self.ep.setEpisodeID("EA1004H")
self.assertEqual(self.ep.episodeID, "EA1004H")
# self.ep.setEpisodeID("EASP06H")
# self.ep.setEpisodeID("EA1G02")
def test_toCheckEpisodeForSeason(self):
self.ep.setEpisodeID("EA1004H")
self.assertEqual(Episode.getSeasonNumber(self.ep.episodeID), 10)
self.ep.setEpisodeID("EASP06H")
self.assertEqual(Episode.getSeasonNumber(self.ep.episodeID), 15)
self.ep.setEpisodeID("EA1G02")
self.assertEqual(Episode.getSeasonNumber(self.ep.episodeID), 7)
self.ep.setEpisodeID("")
self.assertEqual(Episode.getSeasonNumber(self.ep.episodeID), 0)
def test_toCheckEpisodeForSeasonEpisodeNumber(self):
self.ep.setEpisodeID("EA1004H")
self.assertEqual(Episode.getSeasonEpisodeNumber(self.ep.episodeID), 4)
self.ep.setEpisodeID("EASP06H")
self.assertEqual(Episode.getSeasonEpisodeNumber(self.ep.episodeID), 6)
self.ep.setEpisodeID("EA1G02")
self.assertEqual(Episode.getSeasonEpisodeNumber(self.ep.episodeID), 2)
self.ep.setEpisodeID("")
self.assertEqual(Episode.getSeasonEpisodeNumber(self.ep.episodeID), None)
def test_AddingEpisodeTranscript(self):
self.ep.addTranscript("This is some placeholder text")
self.assertEqual(self.ep.transcriptContent, "This is some placeholder text")
def test_toCheckEpisodeAddsEpisodeObjectID(self):
self.ep.setEpisodeID("EA1004H")
self.assertEqual(Episode.getShowID(self.ep.episodeID), '1004')
self.ep.setEpisodeID("EASP06H")
self.assertEqual(Episode.getShowID(self.ep.episodeID), '1506')
self.ep.setEpisodeID("EA1G02")
self.assertEqual(Episode.getShowID(self.ep.episodeID), '0702')
delimiter=',')
readData = csv.reader(open('/'.join([filePath[0], inputFileName[1]]),
'r',
encoding='utf-8'),
delimiter=',')
writeData = open('/'.join([filePath[1], outputFileName[0]]),
'w',
encoding='utf-8')
# Data Structure to hold network node and connectivity information
nodesList, linksList, nodeIDs = [], [], []
episodeList, jsonEpisodeList = {}, {}
# Create Episode instances and store into dict using showID
for line in readMetadata:
episode = Episode(line['Title'])
episode.setEpisodeID(line['Show #'])
episode.addTranscriptID(line['Transcript'])
# print(episode.__dict__)
episodeList[Episode.getShowID(line['Show #'])] = episode
nodeIDs.append(Episode.getShowID(line['Show #']))
# break
# Add 'another show' as an extra show as it is always referenced
# in the show but never really links anywhere, so put it as a dead-end link
episode = Episode('Referenced Mystery Show')
episode.setEpisodeID('EA0000')
episode.addTranscriptID('Another Show')
episodeList[Episode.getShowID('EA0000')] = episode
nodeIDs.append(Episode.getShowID('EA0000'))
print(nodeIDs)
import csv
from EpisodeClass import Episode
from collections import OrderedDict, defaultdict
filePath = ['../Data', '../DataCleaned']
readFiles = ['goodEatsMetadata.csv', 'goodEatsData.csv']
writeFiles = ['goodEatsJSONData.json']
metadataReader = csv.DictReader(open('/'.join([filePath[0], readFiles[0]]), 'r', encoding='utf-8'), delimiter=',')
dataReader = csv.reader(open('/'.join([filePath[0], readFiles[1]]), 'r', encoding='utf-8'), delimiter=',')
jsonWriter = open('/'.join([filePath[1], writeFiles[0]]), 'w', encoding='utf-8')
# tempvars for debugging and short termination of loops
count = 0
episodeList = defaultdict(list)
for line in metadataReader:
episode = Episode(line['Title'])
episode.setEpisodeID(line['Show #'])
episode.setEpisodeNumber(line['Show No.'])
episode.addAirDate(line['Original Air Date'])
episode.addTopics(line['Topics'])
episode.addRecipes(line['Recipes'])
episode.addTranscriptID(line['Transcript'])
if episode.getSeasonNumber(line['Show #']) in episodeList:
episodeList[episode.getSeasonNumber(line['Show #'])].append(episode.__dict__)
else :
episodeList[episode.getSeasonNumber(line['Show #'])] = [episode.__dict__]
json.dump(episodeList, jsonWriter)
encoding='utf-8'),
delimiter=',')
dataReader = csv.reader(open('/'.join([filePath[0], readFiles[1]]),
'r',
encoding='utf-8'),
delimiter=',')
jsonWriter = open('/'.join([filePath[1], writeFiles[0]]),
'w',
encoding='utf-8')
# tempvars for debugging and short termination of loops
count = 0
episodeList = defaultdict(list)
for line in metadataReader:
episode = Episode(line['Title'])
episode.setEpisodeID(line['Show #'])
episode.setEpisodeNumber(line['Show No.'])
episode.addAirDate(line['Original Air Date'])
episode.addTopics(line['Topics'])
episode.addRecipes(line['Recipes'])
episode.addTranscriptID(line['Transcript'])
if episode.getSeasonNumber(line['Show #']) in episodeList:
episodeList[episode.getSeasonNumber(line['Show #'])].append(
episode.__dict__)
else:
episodeList[episode.getSeasonNumber(
line['Show #'])] = [episode.__dict__]
json.dump(episodeList, jsonWriter)
""" MetaData operations for force graph"""
# for line in metadataReader:
# rNodesList, nodesList = [],[]
# linksList = []
# rNodesList.append({'name': Episode.getShowID(line['Show #']), 'group' : Episode.getSeasonNumber(line['Show #'])})
# nodesListLength = len(rNodesList)
# for i in range(0, nodesListLength):
# nodesList.append(rNodesList[nodesListLength - i - 1])
# if i > 0:
# linksList.append({'source' : i - 1, 'target' : i, 'value' : 1})
# json.dump({ 'nodes' : nodesList, 'links' : linksList}, jsonWriter)
""" CSV Data-Subset for visualization """
dataset = {}
csvWriter.writeheader()
for line in metadataReader:
episode = Episode(line['Title'])
episode.setEpisodeID(line['Show #'])
episode.setEpisodeNumber(line['Show No.'])
episode.addAirDate(line['Original Air Date'])
csvWriter.writerow(episode.__dict__)
# tempvar for debugging
count = 0
# Use @staticmethod getShowID as key for EpisodeList and TranscriptList Dictionaries
# episodeList = OrderedDict()
# transcriptList = {}
""" MetaData operations for force graph"""
# for line in metadataReader:
# rNodesList, nodesList = [],[]
# linksList = []
# rNodesList.append({'name': Episode.getShowID(line['Show #']), 'group' : Episode.getSeasonNumber(line['Show #'])})
# nodesListLength = len(rNodesList)
# for i in range(0, nodesListLength):
# nodesList.append(rNodesList[nodesListLength - i - 1])
# if i > 0:
# linksList.append({'source' : i - 1, 'target' : i, 'value' : 1})
# json.dump({ 'nodes' : nodesList, 'links' : linksList}, jsonWriter)
""" CSV Data-Subset for visualization """
dataset = {}
csvWriter.writeheader()
for line in metadataReader:
episode = Episode(line['Title'])
episode.setEpisodeID(line['Show #'])
episode.setEpisodeNumber(line['Show No.'])
episode.addAirDate(line['Original Air Date'])
csvWriter.writerow(episode.__dict__)