def setUp(self):
'''The test case level setup.'''
for index, board in enumerate(self.board_list):
board.set_url(self.BOARD_NAME)
board.parser = dp.build_parser(dp.PageType.board,
self.page_list[index])
board.latest_page = self.expect_list[index]['latest_page']
def setUpClass(cls):
'''The class level setup.'''
cls.pages, cls.expects = tests.board_helper.setup()
cls.parsers = [
dp.build_parser(dp.PageType.board, board_page)
for board_page in cls.pages
]
cls.last_page = [
board_expect['latest_page']
for board_expect in cls.expects
]
def setUpClass(cls):
'''The class level setup.'''
cls.path = 'test_path'
cls.pages, cls.meta, _ = tests.article_helper.setup()
cls.expects = load_ranged_json(len(cls.pages), 'expect_crawler_')
cls.contents = {}
cls.articles = {}
for index, article_meta in enumerate(cls.meta):
article_meta = article_meta['article_meta']
cls.articles[index] = ptt.Article(**article_meta)
cls.articles[index].parser = dp.build_parser(
dp.PageType.article,
cls.pages[index]
)
cls.contents[index] = cls.articles[index].parse_content()
def _get_content(self, page):
'''Transfer HTML content to BeautifulSoup object'''
with suppress(TypeError):
self.parser = dp.build_parser(dp.PageType.board, page)
def _get_content(self, page):
'''Get complete article content.'''
with suppress(TypeError):
self.parser = dp.build_parser(dp.PageType.article, page)
def retrieve_dom(self, pagetype, page):
'''A monkey patch for AbstractPage.retrieve_dom.'''
try:
self.parser = dp.build_parser(pagetype, page)
except (KeyError, TypeError):
raise ValueError
def setUp(self):
'''The test case level setup.'''
for index, article in enumerate(self.article_list):
article.set_url(self.meta_list[index]['article_meta']['href'])
article.parser = dp.build_parser(dp.PageType.article,
self.page_list[index])
def setUp(self):
'''The test case level setup.'''
self.parsers = [
dp.build_parser(dp.PageType.article, page)
for page in self.pages
]
