def test7():
prepareDataFolder("Data")
saveLinksFromTitlePages(KHOA_HOC, "01/01/2018", "05/01/2018", "Science_Jan2018_Titles.txt", "Data")
if not os.path.isdir("Data") or not os.path.isfile("Data/Science_Jan2018_Titles.txt"):
print("Test 7 not OK. Please make sure the file is in the folder.")
else:
pass
def test10():
prepareDataFolder("Data")
saveLinksFromTitlePages(KHOA_HOC, "01/01/2018", "02/01/2018", "Science_Jan2018_Titles.txt", "Data")
saveArticles("Science_Jan2018_Titles.txt", "Science_Jan2018_Articles.txt", "Data")
f = open("Data/Science_Jan2018_Articles.txt").readlines()
if len(f) >= 7:
print("Test 10 OK.")
else:
print("Test 10 not OK.")
def test13():
if not os.path.isfile("Data/Science_Jan2018_Articles.txt"):
prepareDataFolder("Data")
saveLinksFromTitlePages(KHOA_HOC, "01/01/2018", "02/01/2018", "Science_Jan2018_Titles.txt", "Data")
saveArticles("Science_Jan2018_Titles.txt", "Science_Jan2018_Articles.txt", "Data")
f = open("Data/Science_Jan2018_Articles.txt").readlines()
if len(f) < 7:
prepareDataFolder("Data")
saveLinksFromTitlePages(KHOA_HOC, "01/01/2018", "02/01/2018", "Science_Jan2018_Titles.txt", "Data")
saveArticles("Science_Jan2018_Titles.txt", "Science_Jan2018_Articles.txt", "Data")
r = readContent("Science_Jan2018_Articles.txt", "Data")
if getSimpleWordFrequency(r)["và"] >= 1:
print("Test 13 OK.")
else:
print("Test 13 not OK.")
def test11():
if not os.path.isfile("Data/Science_Jan2018_Articles.txt"):
prepareDataFolder("Data")
saveLinksFromTitlePages(KHOA_HOC, "01/01/2018", "02/01/2018", "Science_Jan2018_Titles.txt", "Data")
saveArticles("Science_Jan2018_Titles.txt", "Science_Jan2018_Articles.txt", "Data")
f = open("Data/Science_Jan2018_Articles.txt").readlines()
if len(f) < 7:
prepareDataFolder("Data")
saveLinksFromTitlePages(KHOA_HOC, "01/01/2018", "02/01/2018", "Science_Jan2018_Titles.txt", "Data")
saveArticles("Science_Jan2018_Titles.txt", "Science_Jan2018_Articles.txt", "Data")
r = readContent("Science_Jan2018_Articles.txt", "Data")
if len(r.iloc[0, 0]) > 0 or len(r.iloc[0, 1]) > 0 or len(r.iloc[0, 2]) > 0:
print("Test 11 OK")
else:
print("Test 11 not OK")
def test2():
if (os.path.isfile("clrscr")):
os.remove("clrscr")
elif (os.path.isdir("clrscr")):
shutil.rmtree("clrscr")
try:
prepareDataFolder("clrscr")
except Exception as e:
print("Test 2 not OK.")
print(e)
else:
if not (os.path.isdir("clrscr")):
print("Test 2 not OK.")
print("The folder has not been created.")
else:
os.rmdir("clrscr")
print("Test 2 OK.")
def test12():
if not os.path.isfile("Data/Science_Jan2018_Articles.txt"):
prepareDataFolder("Data")
saveLinksFromTitlePages(KHOA_HOC, "01/01/2018", "02/01/2018", "Science_Jan2018_Titles.txt", "Data")
saveArticles("Science_Jan2018_Titles.txt", "Science_Jan2018_Articles.txt", "Data")
f = open("Data/Science_Jan2018_Articles.txt").readlines()
if len(f) < 7:
prepareDataFolder("Data")
saveLinksFromTitlePages(KHOA_HOC, "01/01/2018", "02/01/2018", "Science_Jan2018_Titles.txt", "Data")
saveArticles("Science_Jan2018_Titles.txt", "Science_Jan2018_Articles.txt", "Data")
r = readContent("Science_Jan2018_Articles.txt", "Data")
s = addAuthorColumn(r)
tester = False
for e in s.loc[:, "author"]:
if e.find("Phương Hoa") >= 0:
tester = True
if tester:
print("Test 12 OK.")
else:
print("Test 12 not OK.")