def connect(self):
config_xml = s3path.s3path().pbconfig
config = xml.etree.ElementTree.parse(config_xml)
host = config.find('database-hostname').text
db = config.find('database-name').text
user = config.find('database-username').text
passwd = 'dummypassword'
# retry the connect because mysql server at IA sometimes causes this
# exception:
# OperationalError: (2013,
# "Lost connection to MySQL server at 'reading authorization packet',system error: 0")
#
# this appears to be related to a config problem with mysqld and a loaded web server.
# see: http://bugs.mysql.com/bug.php?id=28359
try_count = 1
max_tries = 5
self.conn = None
while (not self.conn) and (try_count <= max_tries):
try:
try_count = try_count + 1
self.conn = MySQLdb.connect(
host=host,
db=db,
user=user,
passwd=passwd,
)
except MySQLdb.OperationalError, e:
if try_count > max_tries:
raise e
def bucket_in_host_header(self):
if self.headers.has_key('host'):
host = self.headers['host']
for regex in s3path.s3path().dns_bucket_regexs:
m = re.match(regex, host)
if m:
return m.group('bucket')
return ''
def __init__(self, uri, log_function=fshandler.FilesystemHandler._log_internal, verbose=False):
self.item_dirs = []
for n in range(0,23):
self.item_dirs.append('/%d/items' % n)
fshandler.FilesystemHandler.__init__(self, self.item_dirs[0], uri, log_function, verbose)
#for writes, should we sidestep contrib-submit, and write directly to the filesystem
self.use_fs_directly = False
p = s3path.s3path()
self.petabox_path = p.petabox
self.ias3_path = p.s3
def get_bucket_host(self, bucket):
try:
pipe = subprocess.Popen([
self.petabox_path + "/sw/bin/find_item.php",
bucket,
], stdout=subprocess.PIPE)
output = pipe.communicate()[0]
if not (pipe.wait() == 0):
raise S3_Error("NoSuchBucket", ("%s not found by find_item.php" % bucket))
except OSError:
raise S3_Error
locate_out = output.rstrip("\n")
(host, path) = locate_out.split(':')
#iaNNNNNN.us.archive.org
short_hostname = host.split('.')[0]
domain = host.split('.')[1:]
new_name = "%s.s3dns.%s:%d" % (short_hostname, '.'.join(domain), s3path.s3path().port)
return new_name
def do_GET(self):
"""Serve a GET request."""
dc=self.IFACE_CLASS
lm="Sun, 01 Dec 2014 00:00:00 GMT" # dummy!
headers={"Last-Modified":lm}
data_head = {};
query = urlparse.urlparse(self.path)[4]
try:
if not self.is_local_request():
return self.send_location_redirect()
(bucket, key) = self.get_bucket_key()
(accesskey, signature) = self.get_accesskey_signature()
if query == 'acl':
data = self.get_acl_xml()
headers['content-type'] = 'application/xml'
elif query == 'log':
p = s3path.s3path()
data = subprocess.Popen([p.s3+"/dumplog", "s3accesskey-"+accesskey], stdout=subprocess.PIPE).communicate()[0]
elif key:
data_head=dc.get_head(bucket, key)
data=dc.get_data(bucket, key)
elif bucket:
key_list = dc.get_key_list(bucket)
#dc.pp(key_list)
data=self.get_keys_xml(bucket, key_list)
data_head=dc.get_head(bucket, key)
headers['content-type'] = 'application/xml'
else:
bucket_list = dc.get_bucket_list(accesskey)
data = self.gen_bucket_list_xml(bucket_list)
data_head=dc.get_head(bucket, key)
headers['content-type'] = 'application/xml'
except S3_Error, err:
self.send_error(err)
return
def get_bucket_host(self, bucket):
self.assert_bucket_exists(bucket)
return "%s:%d" % (socket.gethostname(), s3path.s3path().port)
