From b0e8c104995808ee524127d946286caecb69c1ca Mon Sep 17 00:00:00 2001 From: Florent Date: Tue, 13 Mar 2012 17:31:24 +0100 Subject: [PATCH] Pep8 on ecrans module --- modules/ecrans/backend.py | 8 ++++---- modules/ecrans/browser.py | 1 - modules/ecrans/pages/article.py | 9 +++++---- modules/ecrans/tools.py | 14 +++++++++----- 4 files changed, 18 insertions(+), 14 deletions(-) diff --git a/modules/ecrans/backend.py b/modules/ecrans/backend.py index d83ffa7c..575486d6 100644 --- a/modules/ecrans/backend.py +++ b/modules/ecrans/backend.py @@ -23,6 +23,7 @@ from weboob.tools.capabilities.messages.GenericBackend import GenericNewspaperBa from .browser import NewspaperEcransBrowser from .tools import rssid, url2id + class NewspaperEcransBackend(GenericNewspaperBackend, ICapMessages): MAINTAINER = 'Julien Hebert' EMAIL = 'juke@free.fr' @@ -34,7 +35,6 @@ class NewspaperEcransBackend(GenericNewspaperBackend, ICapMessages): BROWSER = NewspaperEcransBrowser RSS_FEED = 'http://www.ecrans.fr/spip.php?page=backend' RSSID = staticmethod(rssid) - URL2ID = staticmethod(url2id) - # RSS Size is actually 10, but some articles are not sorted by publication date - RSSSIZE = 40 - + URL2ID = staticmethod(url2id) + # RSS Size is actually 10, but some articles are not sorted by publication date + RSSSIZE = 40 diff --git a/modules/ecrans/browser.py b/modules/ecrans/browser.py index 73b4507e..33ec83bb 100644 --- a/modules/ecrans/browser.py +++ b/modules/ecrans/browser.py @@ -22,7 +22,6 @@ from .pages.article import ArticlePage from weboob.tools.browser import BaseBrowser - class NewspaperEcransBrowser(BaseBrowser): "NewspaperEcransBrowser class" PAGES = { diff --git a/modules/ecrans/pages/article.py b/modules/ecrans/pages/article.py index 58fd2358..f33a97c7 100644 --- a/modules/ecrans/pages/article.py +++ b/modules/ecrans/pages/article.py @@ -19,19 +19,20 @@ # along with weboob. If not, see . from weboob.tools.capabilities.messages.genericArticle import GenericNewsPage, remove_from_selector_list, try_remove_from_selector_list, try_drop_tree + + class ArticlePage(GenericNewsPage): "ArticlePage object for inrocks" def on_loaded(self): self.main_div = self.document.getroot() self.element_title_selector = "title" - self.element_author_selector = "p.auteur>a" - self.element_body_selector = "div.bloc_article_01" + self.element_author_selector = "p.auteur>a" + self.element_body_selector = "div.bloc_article_01" def get_body(self): element_body = self.get_element_body() - remove_from_selector_list(self.parser, element_body, ["p.auteur", "h4" ]) + remove_from_selector_list(self.parser, element_body, ["p.auteur", "h4"]) try_remove_from_selector_list(self.parser, element_body, ["p.tag", "div.alire", self.element_title_selector, "h4"]) try_drop_tree(self.parser, element_body, "script") return self.parser.tostring(element_body) - diff --git a/modules/ecrans/tools.py b/modules/ecrans/tools.py index d5621d4d..8f9a94a7 100644 --- a/modules/ecrans/tools.py +++ b/modules/ecrans/tools.py @@ -20,25 +20,29 @@ import re + + def id2url(_id): "return an url from an id" regexp2 = re.compile("(\w+).([0-9]+).(.*$)") match = regexp2.match(_id) if match: - return 'http://www.20minutes.fr/%s/%s/%s' % ( match.group(1), - match.group(2), - match.group(3)) + return 'http://www.20minutes.fr/%s/%s/%s' % (match.group(1), + match.group(2), + match.group(3)) else: raise ValueError("id doesn't match") + def url2id(url): "return an id from an url" - regexp = re.compile("(^.*),([0-9]+)\.html$") + regexp = re.compile("(^.*),([0-9]+)\.html$") match = regexp.match(url) - if match: + if match: return match.group(2) else: raise ValueError("Can't find an id for the url") + def rssid(entry): return url2id(entry.id)