new branch inrocks

weboob/backends/inrocks/backend.py

@@ -25,8 +25,6 @@ from weboob.tools.newsfeed import Newsfeed
 from .tools import url2id
 from .browser import NewspaperInrocksBrowser
 
-__all__ = ['NewspaperInrocksBackend']
-
 class NewspaperInrocksBackend(BaseBackend, ICapMessages):
     MAINTAINER = 'Julien Hebert'
     EMAIL = 'juke@free.fr'
@@ -89,7 +87,6 @@ class NewspaperInrocksBackend(BaseBackend, ICapMessages):
                 if msg.flags & msg.IS_UNREAD:
                     yield msg
 
-
     def set_message_read(self, message):
         self.storage.set(
             'seen',

weboob/backends/inrocks/browser.py

@@ -19,7 +19,6 @@ from .pages.article import ArticlePage
 from weboob.tools.browser import BaseBrowser
 from .tools import id2url
 
-__all__ = ['NewspaperInrocksBrowser']
 
 
 class NewspaperInrocksBrowser(BaseBrowser):
@@ -35,4 +34,4 @@ class NewspaperInrocksBrowser(BaseBrowser):
     def get_content(self, _id):
         url = _id
         self.location(url)
-        return self.page.article
+        return self.page.get_article(_id)

weboob/backends/inrocks/pages/article.py

@@ -17,7 +17,7 @@
 # Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
 
 from weboob.tools.parsers.lxmlparser import select, SelectElementException
-from .inrocks import InrocksPage
+from .genericArticle import GenericNewsPage
 
 def try_remove(base_element, selector):
     try :
@@ -25,18 +25,22 @@ def try_remove(base_element, selector):
     except (SelectElementException, ValueError):
         pass
 
-class ArticlePage(InrocksPage):
+class ArticlePage(GenericNewsPage):
     "ArticlePage object for inrocks"
+    def on_loaded(self):
+        self.main_div = self.document.getroot()
+        self.element_author_selector    = "div.name>span"
+        self.element_body_selector      = "div.maincol"
+
     def get_body(self):
-        try_remove(self.element_body, "div.sidebar")
-        details = select(self.element_body, "div.details", 1)
+        element_body = self.get_element_body()
+        try_remove(element_body, "div.sidebar")
+        details = select(element_body, "div.details", 1)
         try_remove(details, "div.footer")
-        header = select(self.element_body, "div.header", 1)
+        header = select(element_body, "div.header", 1)
         for selector in ["h1", "div.picture", "div.date", "div.news-single-img",
                          "div.metas_img", "strong"]:
             try_remove(header, selector)
- 
-        return self.browser.parser.tostring(self.element_body)
-
+        return self.browser.parser.tostring(element_body)
 
 

weboob/backends/inrocks/pages/genericArticle.py

@@ -16,10 +16,8 @@
 # Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
 from weboob.tools.browser import BasePage
 from weboob.tools.parsers.lxmlparser import select, SelectElementException
-from weboob.backends.inrocks.tools import url2id
-__all__ = ['InrocksPage', 'Article', 'NoAuthorElement']
 
-class NoAuthorElement(Exception):
+class NoAuthorElement(SelectElementException):
     pass
 
 class Article(object):
@@ -32,40 +30,43 @@ class Article(object):
         self.author = u''
         self.date = None
 
-class InrocksPage(BasePage):
+class GenericNewsPage(BasePage):
+    __element_body = NotImplementedError
+    __article = Article
+    __element_title_selector  = "h1"
     main_div = NotImplementedError
-    element_body = NotImplementedError
-    article = Article
-    element_author_selector = ValueError
-    element_title_selector  = ValueError
-    element_body_selector   = ValueError
+    element_body_selector = NotImplementedError
+    element_author_selector = NotImplementedError
 
     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):
-        try :
-            return select(self.main_div, self.element_author_selector, 1).text_content().strip()
-        except SelectElementException:
-            #TODO: test nombre d'element en retour
-            pass
+        try:
+            return self.get_element_author().text_content().strip()
+        except NoAuthorElement:
+            return None
 
     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 on_loaded(self):
-        self.article = Article(self.browser, url2id(self.url) )
-        self.main_div = self.document.getroot()
+    def get_element_body(self):
+        return select(self.main_div, self.element_body_selector, 1)
 
-        self.element_author_selector    = "div.name>span"
-        self.element_title_selector     = "h1"
-        self.element_body_selector      = "div.maincol"
+    def get_element_author(self):
+        try:
+            return select(self.main_div, self.element_author_selector, 1)
+        except SelectElementException:
+            raise NoAuthorElement()
 
-        self.element_body = select(self.main_div, self.element_body_selector, 1)
+    def get_article(self, id):
+        __article = Article(self.browser, id)
+        __article.author = self.get_author()
+        __article.title  = self.get_title()
+        __article.url    = self.url
+        __article.body   = self.get_body()
 
-        self.article.author = self.get_author()
-        self.article.title  = self.get_title()
-        self.article.url    = self.url
-        self.article.body   = self.get_body()
-
-    
+        return __article