From 19c0a3c2eda10392b56ed49b514075b1a483ba7a Mon Sep 17 00:00:00 2001
From: calvin <calvin@e7d03fd6-7b0d-0410-9947-9c21f3af8025>
Date: Sun, 18 Dec 2005 08:19:11 +0000
Subject: [PATCH] use new cookie parsing

git-svn-id: https://linkchecker.svn.sourceforge.net/svnroot/linkchecker/trunk/linkchecker@2983 e7d03fd6-7b0d-0410-9947-9c21f3af8025
---
 linkcheck/checker/cache.py   | 53 ++++++++++--------------------------
 linkcheck/checker/httpurl.py | 17 +++++++++---
 2 files changed, 28 insertions(+), 42 deletions(-)

diff --git a/linkcheck/checker/cache.py b/linkcheck/checker/cache.py
index 9e3b723d..7fc7c721 100644
--- a/linkcheck/checker/cache.py
+++ b/linkcheck/checker/cache.py
@@ -19,6 +19,7 @@ Store cached data during checking.
 """
 
 import Cookie
+import time
 import collections
 
 import linkcheck
@@ -26,28 +27,11 @@ import linkcheck.log
 import linkcheck.lock
 import linkcheck.containers
 import linkcheck.configuration
+import linkcheck.cookies
 import linkcheck.threader
 import linkcheck.checker.pool
 
 
-def check_morsel (m, host, path):
-    """
-    Check given cookie morsel against the desired host and path.
-    """
-    # check domain (if its stored)
-    if m["domain"] and not host.endswith(m["domain"]):
-        return None
-    # check path (if its stored)
-    if m["path"] and not path.startswith(m["path"]):
-        return None
-    # check expiry date (if its stored)
-    if m["expires"]:
-        linkcheck.log.debug(linkcheck.LOG_CACHE, "Cookie expires %s",
-                            m["expires"])
-        # XXX check cookie expiration
-    return m.output(header='').strip()
-
-
 class Cache (object):
     """
     Store and provide routines for cached data. Currently there are
@@ -61,17 +45,19 @@ class Cache (object):
         """
         super(Cache, self).__init__()
         # already checked URLs
-        # {cache key (string) -> cache data (dict)}
+        # format: {cache key (string) -> cache data (dict)}
         self.checked = {}
         # URLs that are being checked
-        # {cache key (string) -> urldata (UrlData)}
+        # format: {cache key (string) -> urldata (UrlData)}
         self.in_progress = {}
         # to-be-checked URLs
-        # [urldata (UrlData)]
+        # format: [urldata (UrlData)]
         self.incoming = collections.deque()
         # downloaded robots.txt files
+	# format: {cache key (string) -> robots.txt content (RobotFileParser)}
         self.robots_txt = {}
         # stored cookies
+	# format: {cache key (string) -> cookie jar (linkcheck.cookielib.CookieJar)}
         self.cookies = {}
         # pooled connections
         self.pool = linkcheck.checker.pool.ConnectionPool()
@@ -221,30 +207,21 @@ class Cache (object):
         """
         self.pool.release_connection(key)
 
-    def store_cookies (self, headers, host):
+    def store_cookies (self, headers, scheme, host, path):
         """
         Thread-safe cookie cache setter function. Can raise the
         exception Cookie.CookieError.
         """
-        output = []
-        for h in headers.getallmatchingheaders("Set-Cookie"):
-            output.append(h)
-            linkcheck.log.debug(linkcheck.LOG_CACHE, "Store cookie %s", h)
-            c = self.cookies.setdefault(host, Cookie.SimpleCookie())
-            c.load(h)
-        return output
+        jar = self.cookies.setdefault(host, linkcheck.cookies.CookieJar())
+        return jar.add_cookies(headers, scheme, host, path)
 
-    def get_cookies (self, host, path):
+    def get_cookies (self, scheme, host, port, path):
         """
         Thread-safe cookie cache getter function.
         """
         linkcheck.log.debug(linkcheck.LOG_CACHE,
                             "Get cookies for host %r path %r", host, path)
-        if not self.cookies.has_key(host):
-            return []
-        cookievals = []
-        for m in self.cookies[host].values():
-            val = check_morsel(m, host, path)
-            if val:
-                cookievals.append(val)
-        return cookievals
+        jar = self.cookies.setdefault(host, linkcheck.cookies.CookieJar())
+        jar.remove_expired()
+        return [x for x in jar if x.is_valid_for(scheme, host, port, path)]
+
diff --git a/linkcheck/checker/httpurl.py b/linkcheck/checker/httpurl.py
index fffc2c34..d324490b 100644
--- a/linkcheck/checker/httpurl.py
+++ b/linkcheck/checker/httpurl.py
@@ -19,6 +19,7 @@ Handle http links.
 """
 
 import urlparse
+import urllib
 import time
 import re
 import zlib
@@ -408,7 +409,9 @@ class HttpUrl (internpaturl.InternPatternUrl, proxysupport.ProxySupport):
                     self.add_info(_("Store cookie: %s.") % c)
                 try:
                     out = self.consumer.store_cookies(self.headers,
-                                                      self.urlparts[1])
+                                                      self.urlparts[0],
+                                                      self.urlparts[1],
+                                                      self.urlparts[2])
                     for h in out:
                         self.add_info(linkcheck.strformat.unicode_safe(h))
                 except Cookie.CookieError, msg:
@@ -466,10 +469,16 @@ class HttpUrl (internpaturl.InternPatternUrl, proxysupport.ProxySupport):
         self.url_connection.putheader("Accept-Encoding",
                                   "gzip;q=1.0, deflate;q=0.9, identity;q=0.5")
         if self.consumer.config('cookies'):
-            self.cookies = self.consumer.get_cookies(self.urlparts[1],
-                                                     self.urlparts[2])
+            scheme = self.urlparts[0]
+            host = self.urlparts[1]
+            port = linkcheck.url.default_ports.get(scheme, 80)
+            host, port = urllib.splitnport(host, port)
+            path = self.urlparts[2]
+            self.cookies = self.consumer.get_cookies(scheme, host, port, path)
             for c in self.cookies:
-                self.url_connection.putheader("Cookie", c)
+                name = c.client_header_name()
+                value = c.client_header_value()
+                self.url_connection.putheader(name, value)
         self.url_connection.endheaders()
         response = self.url_connection.getresponse()
         self.persistent = headers.http_persistent(response)