From 8dd2c168d2d915f8487dc1176198845523bf2a01 Mon Sep 17 00:00:00 2001 From: "Jip J. Dekker" Date: Sun, 16 Mar 2014 23:14:59 +0100 Subject: [PATCH] Added the basic structure for the first two spiders --- Scrapy/spiders/Chemspider.py | 11 +++++++++++ Scrapy/spiders/Wikipedia.py | 11 +++++++++++ 2 files changed, 22 insertions(+) create mode 100644 Scrapy/spiders/Chemspider.py create mode 100644 Scrapy/spiders/Wikipedia.py diff --git a/Scrapy/spiders/Chemspider.py b/Scrapy/spiders/Chemspider.py new file mode 100644 index 0000000..3fc74a0 --- /dev/null +++ b/Scrapy/spiders/Chemspider.py @@ -0,0 +1,11 @@ +from scrapy.spider import Spider + +class ChemspiderSpider(Spider): + name = "Chemspider" + allowed_domains = ["chemspider.com"] + start_urls = ( + 'http://www.chemspider.com/', + ) + + def parse(self, response): + pass diff --git a/Scrapy/spiders/Wikipedia.py b/Scrapy/spiders/Wikipedia.py new file mode 100644 index 0000000..03b202b --- /dev/null +++ b/Scrapy/spiders/Wikipedia.py @@ -0,0 +1,11 @@ +from scrapy.spider import Spider + +class WikipediaSpider(Spider): + name = "Wikipedia" + allowed_domains = ["wikipedia.org"] + start_urls = ( + 'http://www.wikipedia.org/', + ) + + def parse(self, response): + pass