nojhan/weboob-devel
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

use assessor

Browse source
This commit is contained in:
Juke 2011-02-17 10:38:57 +01:00 • committed by Romain Bignon
commit 9dad89bf7f
4 changed files with 45 additions and 28 deletions
Download patch file Download diff file Expand all files Collapse all files

2
weboob/backends/minutes20/browser.py
View file

@ -44,4 +44,4 @@ class Newspaper20minutesBrowser(BaseBrowser):
raise ValueError("thread id is empty") raise ValueError("thread id is empty")
else: else:
raise raise
return self.page.article return self.page.get_article()

12
weboob/backends/minutes20/pages/article.py
View file

@ -28,7 +28,11 @@ def try_remove(base_element, selector):
class ArticlePage(Minutes20Page): class ArticlePage(Minutes20Page):
"ArticlePage object for minutes20" "ArticlePage object for minutes20"
def get_body(self): def get_body(self):
try_remove(self.element_body, "div.mna-tools") element_body = self.get_element_body()
try_remove(self.element_body, "div.mna-comment-call") try_remove(element_body, "div.mna-tools")
try_remove(self.element_body, self.element_author_selector) try_remove(element_body, "div.mna-comment-call")
return self.browser.parser.tostring(self.element_body) try :
element_body.remove(self.get_element_author())
except NoAuthorElement:
pass
return self.browser.parser.tostring(element_body)

57
weboob/backends/minutes20/pages/minutes20.py
View file

@ -19,7 +19,7 @@ from weboob.tools.parsers.lxmlparser import select, SelectElementException
from weboob.backends.minutes20.tools import url2id from weboob.backends.minutes20.tools import url2id
__all__ = ['Minutes20Page', 'Article', 'NoAuthorElement'] __all__ = ['Minutes20Page', 'Article', 'NoAuthorElement']
class NoAuthorElement(Exception): class NoAuthorElement(SelectElementException):
pass pass
class Article(object): class Article(object):
@ -33,34 +33,47 @@ class Article(object):
self.date = None self.date = None
class Minutes20Page(BasePage): class Minutes20Page(BasePage):
main_div = NotImplementedError __main_div = NotImplementedError
element_body = NotImplementedError __element_body = NotImplementedError
article = Article __article = Article
element_author_selector = ValueError __element_author_selector = ValueError
element_title_selector = ValueError __element_title_selector = ValueError
element_body_selector = ValueError __element_body_selector = ValueError
def get_body(self): def get_body(self):
return self.browser.parser.tostring(self.element_body) return self.browser.parser.tostring(self.get_element_body())
def get_author(self): def get_author(self):
return select(self.main_div, self.element_author_selector, 1).text_content().strip() try:
return self.get_element_author().text_content().strip()
except NoAuthorElement:
return None
def get_title(self): def get_title(self):
return select(self.main_div, self.element_title_selector, 1).text_content().strip() return select(self.__main_div, self.__element_title_selector, 1).text_content().strip()
def get_element_body(self):
return select(self.__main_div, self.__element_body_selector, 1)
def get_element_author(self):
try:
return select(self.__main_div, self.__element_author_selector, 1)
except SelectElementException:
raise NoAuthorElement()
def get_article(self):
__article = Article(self.browser, url2id(self.url) )
__article.author = self.get_author()
__article.title = self.get_title()
__article.url = self.url
__article.body = self.get_body()
return __article
def on_loaded(self): def on_loaded(self):
self.article = Article(self.browser, url2id(self.url) ) self.__main_div = self.document.getroot()
self.main_div = self.document.getroot()
self.element_author_selector = "div.mna-signature" self.__element_author_selector = "div.mna-signature"
self.element_title_selector = "h1" self.__element_title_selector = "h1"
self.element_body_selector = "div.mna-body" self.__element_body_selector = "div.mna-body"
self.element_body = select(self.main_div, self.element_body_selector, 1)
self.article.author = self.get_author()
self.article.title = self.get_title()
self.article.url = self.url
self.article.body = self.get_body()