"filters" is now propagated through from pipeline component to the job
[arvados.git] / sdk / python / arvados / keep.py
index f5014a404c25cee8de6ab699885a3c9480ff3ed5..82c04ea61bed7d06734eb015469eeb2061d4c540 100644 (file)
@@ -18,12 +18,89 @@ import fcntl
 import time
 import threading
 import timer
+import datetime
 
 global_client_object = None
 
 from api import *
 import config
 import arvados.errors
+import arvados.util
+
+class KeepLocator(object):
+    EPOCH_DATETIME = datetime.datetime.utcfromtimestamp(0)
+
+    def __init__(self, locator_str):
+        self.size = None
+        self.loc_hint = None
+        self._perm_sig = None
+        self._perm_expiry = None
+        pieces = iter(locator_str.split('+'))
+        self.md5sum = next(pieces)
+        for hint in pieces:
+            if hint.startswith('A'):
+                self.parse_permission_hint(hint)
+            elif hint.startswith('K'):
+                self.loc_hint = hint  # FIXME
+            elif hint.isdigit():
+                self.size = int(hint)
+            else:
+                raise ValueError("unrecognized hint data {}".format(hint))
+
+    def __str__(self):
+        return '+'.join(
+            str(s) for s in [self.md5sum, self.size, self.loc_hint,
+                             self.permission_hint()]
+            if s is not None)
+
+    def _make_hex_prop(name, length):
+        # Build and return a new property with the given name that
+        # must be a hex string of the given length.
+        data_name = '_{}'.format(name)
+        def getter(self):
+            return getattr(self, data_name)
+        def setter(self, hex_str):
+            if not arvados.util.is_hex(hex_str, length):
+                raise ValueError("{} must be a {}-digit hex string: {}".
+                                 format(name, length, hex_str))
+            setattr(self, data_name, hex_str)
+        return property(getter, setter)
+
+    md5sum = _make_hex_prop('md5sum', 32)
+    perm_sig = _make_hex_prop('perm_sig', 40)
+
+    @property
+    def perm_expiry(self):
+        return self._perm_expiry
+
+    @perm_expiry.setter
+    def perm_expiry(self, value):
+        if not arvados.util.is_hex(value, 1, 8):
+            raise ValueError(
+                "permission timestamp must be a hex Unix timestamp: {}".
+                format(value))
+        self._perm_expiry = datetime.datetime.utcfromtimestamp(int(value, 16))
+
+    def permission_hint(self):
+        data = [self.perm_sig, self.perm_expiry]
+        if None in data:
+            return None
+        data[1] = int((data[1] - self.EPOCH_DATETIME).total_seconds())
+        return "A{}@{:08x}".format(*data)
+
+    def parse_permission_hint(self, s):
+        try:
+            self.perm_sig, self.perm_expiry = s[1:].split('@', 1)
+        except IndexError:
+            raise ValueError("bad permission hint {}".format(s))
+
+    def permission_expired(self, as_of_dt=None):
+        if self.perm_expiry is None:
+            return False
+        elif as_of_dt is None:
+            as_of_dt = datetime.datetime.now()
+        return self.perm_expiry <= as_of_dt
+
 
 class Keep:
     @staticmethod
@@ -55,6 +132,7 @@ class KeepClient(object):
         def __init__(self, todo):
             self._todo = todo
             self._done = 0
+            self._response = None
             self._todo_lock = threading.Semaphore(todo)
             self._done_lock = threading.Lock()
 
@@ -73,12 +151,23 @@ class KeepClient(object):
             with self._done_lock:
                 return (self._done < self._todo)
 
-        def increment_done(self):
+        def save_response(self, response_body, replicas_stored):
+            """
+            Records a response body (a locator, possibly signed) returned by
+            the Keep server.  It is not necessary to save more than
+            one response, since we presume that any locator returned
+            in response to a successful request is valid.
+            """
+            with self._done_lock:
+                self._done += replicas_stored
+                self._response = response_body
+
+        def response(self):
             """
-            Report that the current thread was successful.
+            Returns the body from the response to a PUT request.
             """
             with self._done_lock:
-                self._done += 1
+                return self._response
 
         def done(self):
             """
@@ -89,9 +178,9 @@ class KeepClient(object):
 
     class KeepWriterThread(threading.Thread):
         """
-        Write a blob of data to the given Keep server. Call
-        increment_done() of the given ThreadLimiter if the write
-        succeeds.
+        Write a blob of data to the given Keep server. On success, call
+        save_response() of the given ThreadLimiter to save the returned
+        locator.
         """
         def __init__(self, **kwargs):
             super(KeepClient.KeepWriterThread, self).__init__()
@@ -136,19 +225,18 @@ class KeepClient(object):
                                       (str(threading.current_thread()),
                                        self.args['data_hash'],
                                        self.args['service_root']))
-
+                        replicas_stored = 1
                         if 'x-keep-replicas-stored' in resp:
                             # Tick the 'done' counter for the number of replica
                             # reported stored by the server, for the case that
                             # we're talking to a proxy or other backend that
                             # stores to multiple copies for us.
-                            replicas = int(resp['x-keep-replicas-stored'])
-                            while replicas > 0:
-                                limiter.increment_done()
-                                replicas -= 1
-                        else:
-                            limiter.increment_done()
-                        return
+                            try:
+                                replicas_stored = int(resp['x-keep-replicas-stored'])
+                            except ValueError:
+                                pass
+                        return limiter.save_response(content.strip(), replicas_stored)
+
                     logging.warning("Request fail: PUT %s => %s %s" %
                                     (url, resp['status'], content))
                 except (httplib2.HttpLib2Error, httplib.HTTPException) as e:
@@ -172,6 +260,7 @@ class KeepClient(object):
             # configuration.
             keep_proxy_env = config.get("ARVADOS_KEEP_PROXY")
             if keep_proxy_env != None and len(keep_proxy_env) > 0:
+
                 if keep_proxy_env[-1:] != '/':
                     keep_proxy_env += "/"
                 self.service_roots = [keep_proxy_env]
@@ -233,8 +322,6 @@ class KeepClient(object):
             # selected server.
             probe = int(seed[0:8], 16) % len(pool)
 
-            print seed[0:8], int(seed[0:8], 16), len(pool), probe
-
             # Append the selected server to the probe sequence and remove it
             # from the pool.
             pseq += [pool[probe]]
@@ -317,7 +404,7 @@ class KeepClient(object):
 
         try:
             for service_root in self.shuffled_service_roots(expect_hash):
-                url = service_root + expect_hash
+                url = service_root + locator
                 api_token = config.get('ARVADOS_API_TOKEN')
                 headers = {'Authorization': "OAuth2 %s" % api_token,
                            'Accept': 'application/octet-stream'}
@@ -329,7 +416,7 @@ class KeepClient(object):
 
             for location_hint in re.finditer(r'\+K@([a-z0-9]+)', locator):
                 instance = location_hint.group(1)
-                url = 'http://keep.' + instance + '.arvadosapi.com/' + expect_hash
+                url = 'http://keep.' + instance + '.arvadosapi.com/' + locator
                 blob = self.get_url(url, {}, expect_hash)
                 if blob:
                     slot.set(blob)
@@ -390,8 +477,9 @@ class KeepClient(object):
         for t in threads:
             t.join()
         have_copies = thread_limiter.done()
+        # If we're done, return the response from Keep
         if have_copies >= want_copies:
-            return (data_hash + '+' + str(len(data)))
+            return thread_limiter.response()
         raise arvados.errors.KeepWriteError(
             "Write fail for %s: wanted %d but wrote %d" %
             (data_hash, want_copies, have_copies))