add read_url debug output

This commit is contained in:
j 2012-08-21 08:41:25 +02:00
parent 21dd1c7903
commit 61b14e9537
2 changed files with 5 additions and 0 deletions

View file

@ -82,6 +82,8 @@ def read_url(url, data=None, headers=DEFAULT_HEADERS, timeout=cache_timeout, val
valid - function to check if result is ok, its passed result and headers valid - function to check if result is ok, its passed result and headers
if this function fails, InvalidResult will be raised deal with it in your code if this function fails, InvalidResult will be raised deal with it in your code
''' '''
if net.DEBUG:
print 'ox.cache.read_url', url
#FIXME: send last-modified / etag from cache and only update if needed #FIXME: send last-modified / etag from cache and only update if needed
url = _fix_unicode_url(url) url = _fix_unicode_url(url)
result = store.get(url, data, headers, timeout) result = store.get(url, data, headers, timeout)

View file

@ -11,6 +11,7 @@ import urllib2
from chardet.universaldetector import UniversalDetector from chardet.universaldetector import UniversalDetector
DEBUG = False
# Default headers for HTTP requests. # Default headers for HTTP requests.
DEFAULT_HEADERS = { DEFAULT_HEADERS = {
'User-Agent': 'Mozilla/5.0 (X11; Linux i686; rv:7.0.1) Gecko/20100101 Firefox/7.0.1', 'User-Agent': 'Mozilla/5.0 (X11; Linux i686; rv:7.0.1) Gecko/20100101 Firefox/7.0.1',
@ -51,6 +52,8 @@ def open_url(url, data=None, headers=DEFAULT_HEADERS):
return urllib2.urlopen(req) return urllib2.urlopen(req)
def read_url(url, data=None, headers=DEFAULT_HEADERS, return_headers=False, unicode=False): def read_url(url, data=None, headers=DEFAULT_HEADERS, return_headers=False, unicode=False):
if DEBUG:
print 'ox.net.read_url', url
f = open_url(url, data, headers) f = open_url(url, data, headers)
result = f.read() result = f.read()
f.close() f.close()