main_test.py

import time
import unittest
from urllib.parse import urlparse, parse_qs

from bs4 import element
from selenium import webdriver

from main import request_html, parse_pub_id, web_driver
from main import parse_all_citations
from main import parse_cluster_id
from main import delete_none_value
from main import parse_publication_metadata
from selenium.common.exceptions import WebDriverException

link = "https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=object+happy&btnG="


class Test(unittest.TestCase):
    def setUp(self):
        self.web_driver = webdriver.Chrome("/Users/shichengyi/Desktop/chromedriver" )
        self.web_driver.get(link)
        time.sleep(3)


    def test_request_html(self):
        a = self.web_driver.title
        self.assertTrue("object" in a)

    def test_parse_all_citations(self):
        visited = set()
        visited.add(link)
        self.assertTrue(link in visited)

    def test_parse_all_citations2(self):

        element_a = self.web_driver.find_element_by_id("nQDDiju8FWYJ")
        self.assertTrue("object" in element_a.text)
        self.assertTrue("happy" in element_a.text)

    def test_parse_cluster_id(self):
        results = parse_qs(urlparse(link).query).get("cluster", [])
        self.assertEqual(results, [])


    def test_parse_cluster_id2(self):
        results2 = parse_qs(urlparse(link).query).get("cites", [])
        self.assertEqual(results2, [])


    def test_delete_none_value(self):
        original_dict = {}
        result_dict = {}
        for key in original_dict.keys():
            self.assertEqual(result_dict[key], original_dict[key])

    def test_parse_publication_metadata(self):

        article_title = self.web_driver.find_element_by_id("gwXKF-MPl_8J")
        self.assertTrue("object" in article_title.text)


    def test_parse_pub_id(self):

        element_b = self.web_driver.find_element_by_xpath("//*[@id='gs_res_ccl_mid']/div[2]/div/div[3]/a[3]")
        self.assertTrue("Cited by" in element_b.text)
        element_c = self.web_driver.find_element_by_xpath("//*[@id='gs_res_ccl_mid']/div[2]/div/div[3]/a[5]")
        self.assertTrue("versions" in element_c.text)
    def tearDown(self):
       self.web_driver.quit()


if __name__ == '__main__':
    unittest.main()