new version of genericArticle
This commit is contained in:
parent
b84cae3431
commit
800796a2be
2 changed files with 28 additions and 10 deletions
|
|
@ -20,20 +20,24 @@ from weboob.tools.parsers.lxmlparser import select, SelectElementException
|
||||||
class NoAuthorElement(SelectElementException):
|
class NoAuthorElement(SelectElementException):
|
||||||
pass
|
pass
|
||||||
|
|
||||||
|
class NoneMainDiv(AttributeError):
|
||||||
|
pass
|
||||||
|
|
||||||
class Article(object):
|
class Article(object):
|
||||||
|
author = u''
|
||||||
|
|
||||||
def __init__(self, browser, _id):
|
def __init__(self, browser, _id):
|
||||||
self.browser = browser
|
self.browser = browser
|
||||||
self.id = _id
|
self.id = _id
|
||||||
self.title = u''
|
self.title = u''
|
||||||
self.body = u''
|
self.body = u''
|
||||||
self.url = u''
|
self.url = u''
|
||||||
self.author = u''
|
|
||||||
self.date = None
|
self.date = None
|
||||||
|
|
||||||
class GenericNewsPage(BasePage):
|
class GenericNewsPage(BasePage):
|
||||||
__element_body = NotImplementedError
|
__element_body = NotImplementedError
|
||||||
__article = Article
|
__article = Article
|
||||||
element_title_selector = "h1"
|
element_title_selector = NotImplementedError
|
||||||
main_div = NotImplementedError
|
main_div = NotImplementedError
|
||||||
element_body_selector = NotImplementedError
|
element_body_selector = NotImplementedError
|
||||||
element_author_selector = NotImplementedError
|
element_author_selector = NotImplementedError
|
||||||
|
|
@ -45,7 +49,7 @@ class GenericNewsPage(BasePage):
|
||||||
try:
|
try:
|
||||||
return self.get_element_author().text_content().strip()
|
return self.get_element_author().text_content().strip()
|
||||||
except NoAuthorElement:
|
except NoAuthorElement:
|
||||||
return None
|
return self.__article.author
|
||||||
|
|
||||||
def get_title(self):
|
def get_title(self):
|
||||||
return select(
|
return select(
|
||||||
|
|
@ -61,9 +65,14 @@ class GenericNewsPage(BasePage):
|
||||||
return select(self.main_div, self.element_author_selector, 1)
|
return select(self.main_div, self.element_author_selector, 1)
|
||||||
except SelectElementException:
|
except SelectElementException:
|
||||||
raise NoAuthorElement()
|
raise NoAuthorElement()
|
||||||
|
except AttributeError:
|
||||||
|
if self.main_div == None:
|
||||||
|
raise NoneMainDiv("main_div is none on %s" % (self.browser))
|
||||||
|
else:
|
||||||
|
raise
|
||||||
|
|
||||||
def get_article(self, id):
|
def get_article(self, _id):
|
||||||
__article = Article(self.browser, id)
|
__article = Article(self.browser, _id)
|
||||||
__article.author = self.get_author()
|
__article.author = self.get_author()
|
||||||
__article.title = self.get_title()
|
__article.title = self.get_title()
|
||||||
__article.url = self.url
|
__article.url = self.url
|
||||||
|
|
|
||||||
|
|
@ -20,20 +20,24 @@ from weboob.tools.parsers.lxmlparser import select, SelectElementException
|
||||||
class NoAuthorElement(SelectElementException):
|
class NoAuthorElement(SelectElementException):
|
||||||
pass
|
pass
|
||||||
|
|
||||||
|
class NoneMainDiv(AttributeError):
|
||||||
|
pass
|
||||||
|
|
||||||
class Article(object):
|
class Article(object):
|
||||||
|
author = u''
|
||||||
|
|
||||||
def __init__(self, browser, _id):
|
def __init__(self, browser, _id):
|
||||||
self.browser = browser
|
self.browser = browser
|
||||||
self.id = _id
|
self.id = _id
|
||||||
self.title = u''
|
self.title = u''
|
||||||
self.body = u''
|
self.body = u''
|
||||||
self.url = u''
|
self.url = u''
|
||||||
self.author = u''
|
|
||||||
self.date = None
|
self.date = None
|
||||||
|
|
||||||
class GenericNewsPage(BasePage):
|
class GenericNewsPage(BasePage):
|
||||||
__element_body = NotImplementedError
|
__element_body = NotImplementedError
|
||||||
__article = Article
|
__article = Article
|
||||||
element_title_selector = "h1"
|
element_title_selector = NotImplementedError
|
||||||
main_div = NotImplementedError
|
main_div = NotImplementedError
|
||||||
element_body_selector = NotImplementedError
|
element_body_selector = NotImplementedError
|
||||||
element_author_selector = NotImplementedError
|
element_author_selector = NotImplementedError
|
||||||
|
|
@ -45,7 +49,7 @@ class GenericNewsPage(BasePage):
|
||||||
try:
|
try:
|
||||||
return self.get_element_author().text_content().strip()
|
return self.get_element_author().text_content().strip()
|
||||||
except NoAuthorElement:
|
except NoAuthorElement:
|
||||||
return None
|
return self.__article.author
|
||||||
|
|
||||||
def get_title(self):
|
def get_title(self):
|
||||||
return select(
|
return select(
|
||||||
|
|
@ -61,9 +65,14 @@ class GenericNewsPage(BasePage):
|
||||||
return select(self.main_div, self.element_author_selector, 1)
|
return select(self.main_div, self.element_author_selector, 1)
|
||||||
except SelectElementException:
|
except SelectElementException:
|
||||||
raise NoAuthorElement()
|
raise NoAuthorElement()
|
||||||
|
except AttributeError:
|
||||||
|
if self.main_div == None:
|
||||||
|
raise NoneMainDiv("main_div is none on %s" % (self.browser))
|
||||||
|
else:
|
||||||
|
raise
|
||||||
|
|
||||||
def get_article(self, id):
|
def get_article(self, _id):
|
||||||
__article = Article(self.browser, id)
|
__article = Article(self.browser, _id)
|
||||||
__article.author = self.get_author()
|
__article.author = self.get_author()
|
||||||
__article.title = self.get_title()
|
__article.title = self.get_title()
|
||||||
__article.url = self.url
|
__article.url = self.url
|
||||||
|
|
|
||||||
Loading…
Add table
Add a link
Reference in a new issue