Merge pull request #475 from dangra/473-do-not-send-header

[MRG] Do not set Referer by default when its value is None

scrapy/http/headers.py

@@ -17,7 +17,9 @@ class Headers(CaselessDict):
 
     def normvalue(self, value):
         """Headers must not be unicode"""
-        if not hasattr(value, '__iter__'):
+        if value is None:
+            value = []
+        elif not hasattr(value, '__iter__'):
             value = [value]
         return [x.encode(self.encoding) if isinstance(x, unicode) else x \
             for x in value]

scrapy/tests/mockserver.py

@@ -1,5 +1,5 @@
 from __future__ import print_function
-import sys, time, random, urllib, os
+import sys, time, random, urllib, os, json
 from subprocess import Popen, PIPE
 from twisted.web.server import Site, NOT_DONE_YET
 from twisted.web.resource import Resource
@@ -119,6 +119,16 @@ class Raw(LeafResource):
         request.finish()
 
 
+class Echo(LeafResource):
+
+    def render_GET(self, request):
+        output = {
+            'headers': dict(request.requestHeaders.getAllRawHeaders()),
+            'body': request.content.read(),
+        }
+        return json.dumps(output)
+
+
 class Partial(LeafResource):
 
     def render_GET(self, request):
@@ -156,6 +166,7 @@ class Root(Resource):
         self.putChild("partial", Partial())
         self.putChild("drop", Drop())
         self.putChild("raw", Raw())
+        self.putChild("echo", Echo())
 
     def getChild(self, name, request):
         return self

scrapy/tests/spiders.py

@@ -132,3 +132,22 @@ class BrokenStartRequestsSpider(FollowAllSpider):
         self.seedsseen.append(response.meta.get('seed'))
         for req in super(BrokenStartRequestsSpider, self).parse(response):
             yield req
+
+
+class SingleRequestSpider(MetaSpider):
+
+    seed = None
+
+    def start_requests(self):
+        if isinstance(self.seed, Request):
+            yield self.seed.replace(callback=self.parse, errback=self.on_error)
+        else:
+            yield Request(self.seed, callback=self.parse, errback=self.on_error)
+
+    def parse(self, response):
+        self.meta.setdefault('responses', []).append(response)
+        if 'next' in response.meta:
+            return response.meta['next']
+
+    def on_error(self, failure):
+        self.meta['failure'] = failure

scrapy/tests/test_crawl.py

@@ -1,9 +1,11 @@
+import json
 from twisted.internet import defer
 from twisted.trial.unittest import TestCase
 from scrapy.utils.test import get_crawler, get_testlog
 from scrapy.tests.spiders import FollowAllSpider, DelaySpider, SimpleSpider, \
-    BrokenStartRequestsSpider
+    BrokenStartRequestsSpider, SingleRequestSpider
 from scrapy.tests.mockserver import MockServer
+from scrapy.http import Request
 
 
 def docrawl(spider, settings=None):
@@ -158,3 +160,31 @@ with multiples lines
         log = get_testlog()
         self.assertEqual(log.count("Retrying"), 2)
         self.assertEqual(log.count("Gave up retrying"), 1)
+
+    @defer.inlineCallbacks
+    def test_referer_header(self):
+        """Referer header is set by RefererMiddleware unless it is already set"""
+        req0 = Request('http://localhost:8998/echo?headers=1&body=0', dont_filter=1)
+        req1 = req0.replace()
+        req2 = req0.replace(headers={'Referer': None})
+        req3 = req0.replace(headers={'Referer': 'http://example.com'})
+        req0.meta['next'] = req1
+        req1.meta['next'] = req2
+        req2.meta['next'] = req3
+        spider = SingleRequestSpider(seed=req0)
+        yield docrawl(spider)
+        # basic asserts in case of weird communication errors
+        self.assertIn('responses', spider.meta)
+        self.assertNotIn('failures', spider.meta)
+        # start requests doesn't set Referer header
+        echo0 = json.loads(spider.meta['responses'][2].body)
+        self.assertNotIn('Referer', echo0['headers'])
+        # following request sets Referer to start request url
+        echo1 = json.loads(spider.meta['responses'][1].body)
+        self.assertEqual(echo1['headers'].get('Referer'), [req0.url])
+        # next request avoids Referer header
+        echo2 = json.loads(spider.meta['responses'][2].body)
+        self.assertNotIn('Referer', echo2['headers'])
+        # last request explicitly sets a Referer header
+        echo3 = json.loads(spider.meta['responses'][3].body)
+        self.assertEqual(echo3['headers'].get('Referer'), ['http://example.com'])

scrapy/tests/test_http_headers.py

@@ -119,3 +119,11 @@ class HeadersTest(unittest.TestCase):
         h1.setlistdefault('header2', ['value2', 'value3'])
         self.assertEqual(h1.getlist('header1'), ['value1'])
         self.assertEqual(h1.getlist('header2'), ['value2', 'value3'])
+
+    def test_none_value(self):
+        h1 = Headers()
+        h1['foo'] = 'bar'
+        h1['foo'] = None
+        h1.setdefault('foo', 'bar')
+        self.assertEqual(h1.get('foo'), None)
+        self.assertEqual(h1.getlist('foo'), [])