def testWriteTextToFile(self):
query = "python:math:how to use sine"
c = cw.Crawler(query)
c.writeTextToFile(c.getTextFromWebsite(c.getLibUrl()))
self.assertTrue(
os.path.isfile(c.handler.getSearchFilePath("python", "math.txt")))
file = open(c.handler.getSearchFilePath("python", "math.txt")).read()
self.assertTrue("sin" in file and "cos" in file
and "<html>" not in file)
def crawlQuery(self):
self.crawlerObject = cwl.Crawler(self.query)
self.crawlerObject.writeTextToFile(
self.crawlerObject.getTextFromWebsite(
self.crawlerObject.getLibUrl()))
def testInit(self):
query = "python:math:how to use sine"
c = cw.Crawler(query)
self.assertEqual(c.query, query)
def testGetTextFromWebsite(self):
query = "python:math:how to use sine"
c = cw.Crawler(query)
htmlCode = c.getTextFromWebsite(c.getLibUrl())
self.assertIs(type(htmlCode), BeautifulSoup)
def testGetLibUrl(self):
query, libUrl = "python:math:how to use sine", "https://docs.python.org/3/library/math.html"
c = cw.Crawler(query)
self.assertIs(type(c.getLibUrl()), type(libUrl))
self.assertEqual(c.getLibUrl(), libUrl)
def crawlQuery(self):
self.crawlerObject = cwl.Crawler(self.query)
lib = self.crawlerObject.getLibUrl()
text = self.crawlerObject.getTextFromWebsite(lib)
self.crawlerObject.writeTextToFile(text)