late-night-sexism_search-titles.py

census_surnames = pd.read_csv('app_c.csv')
census_firstnames = pd.read_csv('census-derived-all-first.csv')

firsts_lower = [str(name).lower() for name in list(census_firstnames.name)]
lasts_lower = [str(name).lower() for name in list(census_surnames.name)]

d = gender.Detector()

male_titles = []
female_titles = []
for i in tqdm(range(6647)):
    try:
        firstname = playlist[i].title.split()[0].lower()
        lastname = playlist[i].title.split()[1].lower()
        
        video_title = ' '.join(playlist[i].title.split()[2:])
        video_title = video_title.replace(" - CONAN on TBS", "")
        video_title = video_title.replace("Conan", "")
        # print(video_title)

        if firstname in firsts_lower and lastname in lasts_lower and "/" not in video_title:
            # print(firstname, d.get_gender(firstname.capitalize()))
            if 'female' in d.get_gender(firstname.capitalize()):
                female_titles.append(video_title)
            elif 'male' in d.get_gender(firstname.capitalize()):
                male_titles.append(video_title)
    except Exception:
        continue