From 068d2278a37b6570bcf98a103bff312d122a6ea9 Mon Sep 17 00:00:00 2001 From: elpolilla Date: Tue, 2 Dec 2008 13:29:20 +0000 Subject: [PATCH] Removed silly bug from the previous revision --HG-- extra : convert_revision : svn%3Ab85faa78-f9eb-468e-a121-7cced6da292c%40460 --- scrapy/trunk/scrapy/link/__init__.py | 2 +- scrapy/trunk/scrapy/link/extractors.py | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/scrapy/trunk/scrapy/link/__init__.py b/scrapy/trunk/scrapy/link/__init__.py index 61592820c..bdcf209d9 100644 --- a/scrapy/trunk/scrapy/link/__init__.py +++ b/scrapy/trunk/scrapy/link/__init__.py @@ -44,7 +44,7 @@ class LinkExtractor(FixedSGMLParser): base_url = self.base_url if self.base_url else response.url ret = [] for link in self.links: - link.url = urljoin(base_url, link.url) + link.url = urljoin(base_url, link.url).strip() ret.append(link) return ret diff --git a/scrapy/trunk/scrapy/link/extractors.py b/scrapy/trunk/scrapy/link/extractors.py index 70cc7e772..f222f82f9 100644 --- a/scrapy/trunk/scrapy/link/extractors.py +++ b/scrapy/trunk/scrapy/link/extractors.py @@ -55,7 +55,7 @@ class RegexLinkExtractor(LinkExtractor): response = new_response_from_xpaths(response, self.restrict_xpaths) links = LinkExtractor.extract_urls(self, response, unique) - links = [link.strip() for link in links if _is_valid_url(link.url)] + links = [link for link in links if _is_valid_url(link.url)] if self.allow_res: links = [link for link in links if _matches(link.url, self.allow_res)]