from __future__ import generators """ httplib2 A caching http interface that supports ETags and gzip to conserve bandwidth. Requires Python 2.3 or later Changelog: 2007-08-18, Rick: Modified so it's able to use a socks proxy if needed. """ __author__ = "Joe Gregorio (joe@bitworking.org)" __copyright__ = "Copyright 2006, Joe Gregorio" __contributors__ = ["Thomas Broyer (t.broyer@ltgt.net)", "James Antill", "Xavier Verges Farrero", "Jonathan Feinberg", "Blair Zajac", "Sam Ruby", "Louis Nyffenegger"] __license__ = "MIT" __version__ = "$Rev: 259 $" import re import sys import md5 import email import email.Utils import email.Message import StringIO import gzip import zlib import httplib import urlparse import base64 import os import copy import calendar import time import random import sha import hmac from gettext import gettext as _ import socket try: import socks except ImportError: socks = None if sys.version_info >= (2,3): from iri2uri import iri2uri else: def iri2uri(uri): return uri __all__ = ['Http', 'Response', 'ProxyInfo', 'HttpLib2Error', 'RedirectMissingLocation', 'RedirectLimit', 'FailedToDecompressContent', 'UnimplementedDigestAuthOptionError', 'UnimplementedHmacDigestAuthOptionError', 'debuglevel'] # The httplib debug level, set to a non-zero value to get debug output debuglevel = 0 # Python 2.3 support if sys.version_info < (2,4): def sorted(seq): seq.sort() return seq # Python 2.3 support def HTTPResponse__getheaders(self): """Return list of (header, value) tuples.""" if self.msg is None: raise httplib.ResponseNotReady() return self.msg.items() if not hasattr(httplib.HTTPResponse, 'getheaders'): httplib.HTTPResponse.getheaders = HTTPResponse__getheaders # All exceptions raised here derive from HttpLib2Error class HttpLib2Error(Exception): pass # Some exceptions can be caught and optionally # be turned back into responses. class HttpLib2ErrorWithResponse(HttpLib2Error): def __init__(self, desc, response, content): self.response = response self.content = content HttpLib2Error.__init__(self, desc) class RedirectMissingLocation(HttpLib2ErrorWithResponse): pass class RedirectLimit(HttpLib2ErrorWithResponse): pass class FailedToDecompressContent(HttpLib2ErrorWithResponse): pass class UnimplementedDigestAuthOptionError(HttpLib2ErrorWithResponse): pass class UnimplementedHmacDigestAuthOptionError(HttpLib2ErrorWithResponse): pass class RelativeURIError(HttpLib2Error): pass class ServerNotFoundError(HttpLib2Error): pass # Open Items: # ----------- # Proxy support # Are we removing the cached content too soon on PUT (only delete on 200 Maybe?) # Pluggable cache storage (supports storing the cache in # flat files by default. We need a plug-in architecture # that can support Berkeley DB and Squid) # == Known Issues == # Does not handle a resource that uses conneg and Last-Modified but no ETag as a cache validator. # Does not handle Cache-Control: max-stale # Does not use Age: headers when calculating cache freshness. # The number of redirections to follow before giving up. # Note that only GET redirects are automatically followed. # Will also honor 301 requests by saving that info and never # requesting that URI again. DEFAULT_MAX_REDIRECTS = 5 # Which headers are hop-by-hop headers by default HOP_BY_HOP = ['connection', 'keep-alive', 'proxy-authenticate', 'proxy-authorization', 'te', 'trailers', 'transfer-encoding', 'upgrade'] def _get_end2end_headers(response): hopbyhop = list(HOP_BY_HOP) hopbyhop.extend([x.strip() for x in response.get('connection', '').split(',')]) return [header for header in response.keys() if header not in hopbyhop] URI = re.compile(r"^(([^:/?#]+):)?(//([^/?#]*))?([^?#]*)(\?([^#]*))?(#(.*))?") def parse_uri(uri): """Parses a URI using the regex given in Appendix B of RFC 3986. (scheme, authority, path, query, fragment) = parse_uri(uri) """ groups = URI.match(uri).groups() return (groups[1], groups[3], groups[4], groups[6], groups[8]) def urlnorm(uri): (scheme, authority, path, query, fragment) = parse_uri(uri) if not scheme or not authority: raise RelativeURIError("Only absolute URIs are allowed. uri = %s" % uri) authority = authority.lower() scheme = scheme.lower() if not path: path = "/" # Could do syntax based normalization of the URI before # computing the digest. See Section 6.2.2 of Std 66. request_uri = query and "?".join([path, query]) or path scheme = scheme.lower() defrag_uri = scheme + "://" + authority + request_uri return scheme, authority, request_uri, defrag_uri # Cache filename construction (original borrowed from Venus http://intertwingly.net/code/venus/) re_url_scheme = re.compile(r'^\w+://') re_slash = re.compile(r'[?/:|]+') def safename(filename): """Return a filename suitable for the cache. Strips dangerous and common characters to create a filename we can use to store the cache in. """ try: if re_url_scheme.match(filename): if isinstance(filename,str): filename = filename.decode('utf-8') filename = filename.encode('idna') else: filename = filename.encode('idna') except UnicodeError: pass if isinstance(filename,unicode): filename=filename.encode('utf-8') filemd5 = md5.new(filename).hexdigest() filename = re_url_scheme.sub("", filename) filename = re_slash.sub(",", filename) # limit length of filename if len(filename)>200: filename=filename[:200] return ",".join((filename, filemd5)) NORMALIZE_SPACE = re.compile(r'(?:\r\n)?[ \t]+') def _normalize_headers(headers): return dict([ (key.lower(), NORMALIZE_SPACE.sub(value, ' ').strip()) for (key, value) in headers.iteritems()]) def _parse_cache_control(headers): retval = {} if headers.has_key('cache-control'): parts = headers['cache-control'].split(',') parts_with_args = [tuple([x.strip() for x in part.split("=")]) for part in parts if -1 != part.find("=")] parts_wo_args = [(name.strip(), 1) for name in parts if -1 == name.find("=")] retval = dict(parts_with_args + parts_wo_args) return retval # Whether to use a strict mode to parse WWW-Authenticate headers # Might lead to bad results in case of ill-formed header value, # so disabled by default, falling back to relaxed parsing. # Set to true to turn on, usefull for testing servers. USE_WWW_AUTH_STRICT_PARSING = 0 # In regex below: # [^\0-\x1f\x7f-\xff()<>@,;:\\\"/[\]?={} \t]+ matches a "token" as defined by HTTP # "(?:[^\0-\x08\x0A-\x1f\x7f-\xff\\\"]|\\[\0-\x7f])*?" matches a "quoted-string" as defined by HTTP, when LWS have already been replaced by a single space # Actually, as an auth-param value can be either a token or a quoted-string, they are combined in a single pattern which matches both: # \"?((?<=\")(?:[^\0-\x1f\x7f-\xff\\\"]|\\[\0-\x7f])*?(?=\")|(?@,;:\\\"/[\]?={} \t]+(?!\"))\"? WWW_AUTH_STRICT = re.compile(r"^(?:\s*(?:,\s*)?([^\0-\x1f\x7f-\xff()<>@,;:\\\"/[\]?={} \t]+)\s*=\s*\"?((?<=\")(?:[^\0-\x08\x0A-\x1f\x7f-\xff\\\"]|\\[\0-\x7f])*?(?=\")|(?@,;:\\\"/[\]?={} \t]+(?!\"))\"?)(.*)$") WWW_AUTH_RELAXED = re.compile(r"^(?:\s*(?:,\s*)?([^ \t\r\n=]+)\s*=\s*\"?((?<=\")(?:[^\\\"]|\\.)*?(?=\")|(? current_age: retval = "FRESH" return retval def _decompressContent(response, new_content): content = new_content try: encoding = response.get('content-encoding', None) if encoding in ['gzip', 'deflate']: if encoding == 'gzip': content = gzip.GzipFile(fileobj=StringIO.StringIO(new_content)).read() if encoding == 'deflate': content = zlib.decompress(content) response['content-length'] = str(len(content)) del response['content-encoding'] except IOError: content = "" raise FailedToDecompressContent(_("Content purported to be compressed with %s but failed to decompress.") % response.get('content-encoding'), response, content) return content def _updateCache(request_headers, response_headers, content, cache, cachekey): if cachekey: cc = _parse_cache_control(request_headers) cc_response = _parse_cache_control(response_headers) if cc.has_key('no-store') or cc_response.has_key('no-store'): cache.delete(cachekey) else: info = email.Message.Message() for key, value in response_headers.iteritems(): if key not in ['status','content-encoding','transfer-encoding']: info[key] = value status = response_headers.status if status == 304: status = 200 status_header = 'status: %d\r\n' % response_headers.status header_str = info.as_string() header_str = re.sub("\r(?!\n)|(? 0: service = "cl" # No point in guessing Base or Spreadsheet #elif request_uri.find("spreadsheets") > 0: # service = "wise" auth = dict(Email=credentials[0], Passwd=credentials[1], service=service, source=headers['user-agent']) resp, content = self.http.request("https://www.google.com/accounts/ClientLogin", method="POST", body=urlencode(auth), headers={'Content-Type': 'application/x-www-form-urlencoded'}) lines = content.split('\n') d = dict([tuple(line.split("=", 1)) for line in lines if line]) if resp.status == 403: self.Auth = "" else: self.Auth = d['Auth'] def request(self, method, request_uri, headers, content): """Modify the request headers to add the appropriate Authorization header.""" headers['authorization'] = 'GoogleLogin Auth=' + self.Auth AUTH_SCHEME_CLASSES = { "basic": BasicAuthentication, "wsse": WsseAuthentication, "digest": DigestAuthentication, "hmacdigest": HmacDigestAuthentication, "googlelogin": GoogleLoginAuthentication } AUTH_SCHEME_ORDER = ["hmacdigest", "googlelogin", "digest", "wsse", "basic"] def _md5(s): return class FileCache(object): """Uses a local directory as a store for cached files. Not really safe to use if multiple threads or processes are going to be running on the same cache. """ def __init__(self, cache, safe=safename): # use safe=lambda x: md5.new(x).hexdigest() for the old behavior self.cache = cache self.safe = safe if not os.path.exists(cache): os.makedirs(self.cache) def get(self, key): retval = None cacheFullPath = os.path.join(self.cache, self.safe(key)) try: f = file(cacheFullPath, "r") retval = f.read() f.close() except IOError: pass return retval def set(self, key, value): cacheFullPath = os.path.join(self.cache, self.safe(key)) f = file(cacheFullPath, "w") f.write(value) f.close() def delete(self, key): cacheFullPath = os.path.join(self.cache, self.safe(key)) if os.path.exists(cacheFullPath): os.remove(cacheFullPath) class Credentials(object): def __init__(self): self.credentials = [] def add(self, name, password, domain=""): self.credentials.append((domain.lower(), name, password)) def clear(self): self.credentials = [] def iter(self, domain): for (cdomain, name, password) in self.credentials: if cdomain == "" or domain == cdomain: yield (name, password) class KeyCerts(Credentials): """Identical to Credentials except that name/password are mapped to key/cert.""" pass class ProxyInfo(object): """Collect information required to use a proxy.""" def __init__(self, proxy_type, proxy_host, proxy_port, proxy_rdns=None, proxy_user=None, proxy_pass=None): """The parameter proxy_type must be set to one of socks.PROXY_TYPE_XXX constants. For example: p = ProxyInfo(proxy_type=socks.PROXY_TYPE_HTTP, proxy_host='localhost', proxy_port=8000) """ self.proxy_type, self.proxy_host, self.proxy_port, self.proxy_rdns, self.proxy_user, self.proxy_pass = proxy_type, proxy_host, proxy_port, proxy_rdns, proxy_user, proxy_pass def astuple(self): return (self.proxy_type, self.proxy_host, self.proxy_port, self.proxy_rdns, self.proxy_user, self.proxy_pass) def isgood(self): return socks and (self.proxy_host != None) and (self.proxy_port != None) class HTTPConnectionWithTimeout(httplib.HTTPConnection): """HTTPConnection subclass that supports timeouts""" def __init__(self, host, port=None, strict=None, timeout=None, proxy_info=None): httplib.HTTPConnection.__init__(self, host, port, strict) self.timeout = timeout self.proxy_info = proxy_info def connect(self): """Connect to the host and port specified in __init__.""" # Mostly verbatim from httplib.py. msg = "getaddrinfo returns an empty list" for res in socket.getaddrinfo(self.host, self.port, 0, socket.SOCK_STREAM): af, socktype, proto, canonname, sa = res try: if self.proxy_info and self.proxy_info.isgood(): self.sock = socks.socksocket(af, socktype, proto) self.sock.setproxy(*self.proxy_info.astuple()) else: self.sock = socket.socket(af, socktype, proto) # Different from httplib: support timeouts. if self.timeout is not None: self.sock.settimeout(self.timeout) # End of difference from httplib. if self.debuglevel > 0: print "connect: (%s, %s)" % (self.host, self.port) self.sock.connect(sa) except socket.error, msg: if self.debuglevel > 0: print 'connect fail:', (self.host, self.port) if self.sock: self.sock.close() self.sock = None continue break if not self.sock: raise socket.error, msg class HTTPSConnectionWithTimeout(httplib.HTTPSConnection): "This class allows communication via SSL." def __init__(self, host, port=None, key_file=None, cert_file=None, strict=None, timeout=None, proxy_info=None): self.timeout = timeout self.proxy_info = proxy_info httplib.HTTPSConnection.__init__(self, host, port=port, key_file=key_file, cert_file=cert_file, strict=strict) def connect(self): "Connect to a host on a given (SSL) port." if self.proxy_info and self.proxy_info.isgood(): self.sock.setproxy(*self.proxy_info.astuple()) sock.setproxy(*self.proxy_info.astuple()) else: sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM) if self.timeout is not None: sock.settimeout(self.timeout) sock.connect((self.host, self.port)) ssl = socket.ssl(sock, self.key_file, self.cert_file) self.sock = httplib.FakeSocket(sock, ssl) class Http(object): """An HTTP client that handles: - all methods - caching - ETags - compression, - HTTPS - Basic - Digest - WSSE and more. """ def __init__(self, cache=None, timeout=None, proxy_info=None): """The value of proxy_info is a ProxyInfo instance. If 'cache' is a string then it is used as a directory name for a disk cache. Otherwise it must be an object that supports the same interface as FileCache.""" self.proxy_info = proxy_info # Map domain name to an httplib connection self.connections = {} # The location of the cache, for now a directory # where cached responses are held. if cache and isinstance(cache, str): self.cache = FileCache(cache) else: self.cache = cache # Name/password self.credentials = Credentials() # Key/cert self.certificates = KeyCerts() # authorization objects self.authorizations = [] # If set to False then no redirects are followed, even safe ones. self.follow_redirects = True # If 'follow_redirects' is True, and this is set to True then # all redirecs are followed, including unsafe ones. self.follow_all_redirects = False self.ignore_etag = False self.force_exception_to_status_code = False self.timeout = timeout def _auth_from_challenge(self, host, request_uri, headers, response, content): """A generator that creates Authorization objects that can be applied to requests. """ challenges = _parse_www_authenticate(response, 'www-authenticate') for cred in self.credentials.iter(host): for scheme in AUTH_SCHEME_ORDER: if challenges.has_key(scheme): yield AUTH_SCHEME_CLASSES[scheme](cred, host, request_uri, headers, response, content, self) def add_credentials(self, name, password, domain=""): """Add a name and password that will be used any time a request requires authentication.""" self.credentials.add(name, password, domain) def add_certificate(self, key, cert, domain): """Add a key and cert that will be used any time a request requires authentication.""" self.certificates.add(key, cert, domain) def clear_credentials(self): """Remove all the names and passwords that are used for authentication""" self.credentials.clear() self.authorizations = [] def _conn_request(self, conn, request_uri, method, body, headers): for i in range(2): try: conn.request(method, request_uri, body, headers) response = conn.getresponse() except socket.gaierror: conn.close() raise ServerNotFoundError("Unable to find the server at %s" % conn.host) except httplib.HTTPException, e: if i == 0: conn.close() conn.connect() continue else: raise else: content = response.read() response = Response(response) if method != "HEAD": content = _decompressContent(response, content) break; return (response, content) def _request(self, conn, host, absolute_uri, request_uri, method, body, headers, redirections, cachekey): """Do the actual request using the connection object and also follow one level of redirects if necessary""" auths = [(auth.depth(request_uri), auth) for auth in self.authorizations if auth.inscope(host, request_uri)] auth = auths and sorted(auths)[0][1] or None if auth: auth.request(method, request_uri, headers, body) (response, content) = self._conn_request(conn, request_uri, method, body, headers) if auth: if auth.response(response, body): auth.request(method, request_uri, headers, body) (response, content) = self._conn_request(conn, request_uri, method, body, headers ) response._stale_digest = 1 if response.status == 401: for authorization in self._auth_from_challenge(host, request_uri, headers, response, content): authorization.request(method, request_uri, headers, body) (response, content) = self._conn_request(conn, request_uri, method, body, headers, ) if response.status != 401: self.authorizations.append(authorization) authorization.response(response, body) break if (self.follow_all_redirects or (method in ["GET", "HEAD"]) or response.status == 303): if self.follow_redirects and response.status in [300, 301, 302, 303, 307]: # Pick out the location header and basically start from the beginning # remembering first to strip the ETag header and decrement our 'depth' if redirections: if not response.has_key('location') and response.status != 300: raise RedirectMissingLocation( _("Redirected but the response is missing a Location: header."), response, content) # Fix-up relative redirects (which violate an RFC 2616 MUST) if response.has_key('location'): location = response['location'] (scheme, authority, path, query, fragment) = parse_uri(location) if authority == None: response['location'] = urlparse.urljoin(absolute_uri, location) if response.status == 301 and method in ["GET", "HEAD"]: response['-x-permanent-redirect-url'] = response['location'] if not response.has_key('content-location'): response['content-location'] = absolute_uri _updateCache(headers, response, content, self.cache, cachekey) if headers.has_key('if-none-match'): del headers['if-none-match'] if headers.has_key('if-modified-since'): del headers['if-modified-since'] if response.has_key('location'): location = response['location'] old_response = copy.deepcopy(response) if not old_response.has_key('content-location'): old_response['content-location'] = absolute_uri redirect_method = ((response.status == 303) and (method not in ["GET", "HEAD"])) and "GET" or method (response, content) = self.request(location, redirect_method, body=body, headers = headers, redirections = redirections - 1) response.previous = old_response else: raise RedirectLimit( _("Redirected more times than rediection_limit allows."), response, content) elif response.status in [200, 203] and method == "GET": # Don't cache 206's since we aren't going to handle byte range requests if not response.has_key('content-location'): response['content-location'] = absolute_uri _updateCache(headers, response, content, self.cache, cachekey) return (response, content) # Need to catch and rebrand some exceptions # Then need to optionally turn all exceptions into status codes # including all socket.* and httplib.* exceptions. def request(self, uri, method="GET", body=None, headers=None, redirections=DEFAULT_MAX_REDIRECTS, connection_type=None): """ Performs a single HTTP request. The 'uri' is the URI of the HTTP resource and can begin with either 'http' or 'https'. The value of 'uri' must be an absolute URI. The 'method' is the HTTP method to perform, such as GET, POST, DELETE, etc. There is no restriction on the methods allowed. The 'body' is the entity body to be sent with the request. It is a string object. Any extra headers that are to be sent with the request should be provided in the 'headers' dictionary. The maximum number of redirect to follow before raising an exception is 'redirections. The default is 5. The return value is a tuple of (response, content), the first being and instance of the 'Response' class, the second being a string that contains the response entity body. """ try: if headers is None: headers = {} else: headers = _normalize_headers(headers) if not headers.has_key('user-agent'): headers['user-agent'] = "Python-httplib2/%s" % __version__ uri = iri2uri(uri) (scheme, authority, request_uri, defrag_uri) = urlnorm(uri) conn_key = scheme+":"+authority if conn_key in self.connections: conn = self.connections[conn_key] else: if not connection_type: connection_type = (scheme == 'https') and HTTPSConnectionWithTimeout or HTTPConnectionWithTimeout certs = list(self.certificates.iter(authority)) if scheme == 'https' and certs: conn = self.connections[conn_key] = connection_type(authority, key_file=certs[0][0], cert_file=certs[0][1], timeout=self.timeout, proxy_info=self.proxy_info) else: conn = self.connections[conn_key] = connection_type(authority, timeout=self.timeout, proxy_info=self.proxy_info) conn.set_debuglevel(debuglevel) if method in ["GET", "HEAD"] and 'range' not in headers: headers['accept-encoding'] = 'compress, gzip' info = email.Message.Message() cached_value = None if self.cache: cachekey = defrag_uri cached_value = self.cache.get(cachekey) if cached_value: info = email.message_from_string(cached_value) try: content = cached_value.split('\r\n\r\n', 1)[1] except IndexError: self.cache.delete(cachekey) cachekey = None cached_value = None else: cachekey = None if method in ["PUT"] and self.cache and info.has_key('etag') and not self.ignore_etag and 'if-match' not in headers: # http://www.w3.org/1999/04/Editing/ headers['if-match'] = info['etag'] if method not in ["GET", "HEAD"] and self.cache and cachekey: # RFC 2616 Section 13.10 self.cache.delete(cachekey) if cached_value and method in ["GET", "HEAD"] and self.cache and 'range' not in headers: if info.has_key('-x-permanent-redirect-url'): # Should cached permanent redirects be counted in our redirection count? For now, yes. (response, new_content) = self.request(info['-x-permanent-redirect-url'], "GET", headers = headers, redirections = redirections - 1) response.previous = Response(info) response.previous.fromcache = True else: # Determine our course of action: # Is the cached entry fresh or stale? # Has the client requested a non-cached response? # # There seems to be three possible answers: # 1. [FRESH] Return the cache entry w/o doing a GET # 2. [STALE] Do the GET (but add in cache validators if available) # 3. [TRANSPARENT] Do a GET w/o any cache validators (Cache-Control: no-cache) on the request entry_disposition = _entry_disposition(info, headers) if entry_disposition == "FRESH": if not cached_value: info['status'] = '504' content = "" response = Response(info) if cached_value: response.fromcache = True return (response, content) if entry_disposition == "STALE": if info.has_key('etag') and not self.ignore_etag and not 'if-none-match' in headers: headers['if-none-match'] = info['etag'] if info.has_key('last-modified') and not 'last-modified' in headers: headers['if-modified-since'] = info['last-modified'] elif entry_disposition == "TRANSPARENT": pass (response, new_content) = self._request(conn, authority, uri, request_uri, method, body, headers, redirections, cachekey) if response.status == 304 and method == "GET": # Rewrite the cache entry with the new end-to-end headers # Take all headers that are in response # and overwrite their values in info. # unless they are hop-by-hop, or are listed in the connection header. for key in _get_end2end_headers(response): info[key] = response[key] merged_response = Response(info) if hasattr(response, "_stale_digest"): merged_response._stale_digest = response._stale_digest _updateCache(headers, merged_response, content, self.cache, cachekey) response = merged_response response.status = 200 response.fromcache = True elif response.status == 200: content = new_content else: self.cache.delete(cachekey) content = new_content else: (response, content) = self._request(conn, authority, uri, request_uri, method, body, headers, redirections, cachekey) except Exception, e: if self.force_exception_to_status_code: if isinstance(e, HttpLib2ErrorWithResponse): response = e.response content = e.content response.status = 500 response.reason = str(e) elif isinstance(e, socket.timeout): content = "Request Timeout" response = Response( { "content-type": "text/plain", "status": "408", "content-length": len(content) }) response.reason = "Request Timeout" else: content = str(e) response = Response( { "content-type": "text/plain", "status": "400", "content-length": len(content) }) response.reason = "Bad Request" else: raise return (response, content) class Response(dict): """An object more like email.Message than httplib.HTTPResponse.""" """Is this response from our local cache""" fromcache = False """HTTP protocol version used by server. 10 for HTTP/1.0, 11 for HTTP/1.1. """ version = 11 "Status code returned by server. " status = 200 """Reason phrase returned by server.""" reason = "Ok" previous = None def __init__(self, info): # info is either an email.Message or # an httplib.HTTPResponse object. if isinstance(info, httplib.HTTPResponse): for key, value in info.getheaders(): self[key] = value self.status = info.status self['status'] = str(self.status) self.reason = info.reason self.version = info.version elif isinstance(info, email.Message.Message): for key, value in info.items(): self[key] = value self.status = int(self['status']) else: for key, value in info.iteritems(): self[key] = value self.status = int(self.get('status', self.status)) def __getattr__(self, name): if name == 'dict': return self else: raise AttributeError, name