class Eveuni_parser(): baseurl = 'http://wiki.eveuniversity.org/' ecm = '//div[@class="ecm-priority"]//div[contains(@class,"label2")]//text()'; faction = '//td[@class="faction"]/text()'; hull = '//td[@class="hull-type"]/a/text()'; def __init__(self): self.html_parser = None def fetch_ship_info(self, ship): self.ship_name = ship # html = (open("ashimmu.html","r").read()); html = urlopen(Eveuni_parser.baseurl + ship).read() self.html_parser = Html_parser(html) def get_bonuses(self): bonuses = Ship_bonuses(self.html_parser) bonuses.parse_bonuses() return bonuses def get_hull(self): return self.html_parser.parse_text(Eveuni_parser.hull).pop() def get_ecm(self): return self.html_parser.parse_text(Eveuni_parser.ecm).pop() def get_faction(self): return self.html_parser.parse_text(Eveuni_parser.faction).pop()
def main(): parser = argparse.ArgumentParser(description=__doc__) parser.add_argument("source_dir") parser.add_argument("dest_dir") #parser.add_argument("phrase_dir") parser.add_argument("--need_stem", "-m", action='store_true', default=False) parser.add_argument("--first_only", "-o", action='store_true', default=False) parser.add_argument("--top", "-t", type=int, default=10) args = parser.parse_args() parser = Html_parser(args.need_stem) sub_dirs = os.walk(args.source_dir).next()[1] for a_dir in sub_dirs: dest_dir = os.path.join(args.dest_dir, a_dir) source_dir = os.path.join(args.source_dir, a_dir) if not os.path.exists(dest_dir): os.makedirs(dest_dir) gene_text_single_dir(parser, source_dir, dest_dir, args.top, args.first_only)
class web_surfing(object): """description of class""" def __init__(self): """ Purpose: Initialize instance """ self.driver = webdriver.Firefox() self.html_parser = Html_parser("") def get_next_page(self, url): """ Purpose: Move on the website """ first_page = self.driver.page_source self.driver.get(url) current_page = None while current_page == first_page: time.sleep(10) first_page = self.driver.page_source def login(self, login_credentials): """ login(list) -> Purpose: Login into website """ # Update html_parser html document self.html_parser.update_parser_html(self.driver.page_source) # Get login fields login_fields = self.html_parser.get_login_fields(len(login_credentials)) # Number of login credentials can't be differnt from number of login_fields if len(login_credentials) != len(login_fields): return 0 # Insert Credentials for index in xrange(len(login_credentials)): element = self.driver.find_element_by_name(login_fields[index]) element.send_keys(login_credentials[index]) # Pass credentials element.send_keys(Keys.RETURN)
def __init__(self): """ Purpose: Initialize instance """ self.driver = webdriver.Firefox() self.html_parser = Html_parser("")
def fetch_ship_info(self, ship): self.ship_name = ship # html = (open("ashimmu.html","r").read()); html = urlopen(Eveuni_parser.baseurl + ship).read() self.html_parser = Html_parser(html)