def ReadLatestPathInfosWithHashBlobReferences(self,
client_paths,
max_timestamp=None,
cursor=None):
"""Returns PathInfos that have corresponding HashBlobReferences."""
path_infos = {client_path: None for client_path in client_paths}
path_id_components = {}
for client_path in client_paths:
path_id = rdf_objects.PathID.FromComponents(client_path.components)
path_id_components[path_id] = client_path.components
params = []
query = """
SELECT t.client_id, t.path_type, t.path_id, unix_timestamp(t.timestamp),
s.stat_entry, h.hash_entry
FROM (SELECT h.client_id, h.path_type, h.path_id,
MAX(h.timestamp) AS timestamp
FROM client_path_hash_entries AS h
INNER JOIN hash_blob_references AS b
ON b.hash_id = h.sha256
WHERE {conditions}
GROUP BY client_id, path_type, path_id) AS t
LEFT JOIN client_path_stat_entries AS s
ON s.client_id = t.client_id
AND s.path_type = t.path_type
AND s.path_id = t.path_id
AND s.timestamp = t.timestamp
LEFT JOIN client_path_hash_entries AS h
ON h.client_id = t.client_id
AND h.path_type = t.path_type
AND h.path_id = t.path_id
AND h.timestamp = t.timestamp
"""
path_conditions = []
for client_path in client_paths:
path_id = rdf_objects.PathID.FromComponents(client_path.components)
path_conditions.append("""
(client_id = %s AND path_type = %s AND path_id = %s)
""")
params.append(db_utils.ClientIDToInt(client_path.client_id))
params.append(int(client_path.path_type))
params.append(path_id.AsBytes())
conditions = " OR ".join(path_conditions)
if max_timestamp is not None:
conditions = "({}) AND unix_timestamp(timestamp) <= %s".format(conditions)
params.append(mysql_utils.RDFDatetimeToTimestamp(max_timestamp))
cursor.execute(query.format(conditions=conditions), params)
for row in cursor.fetchall():
# pyformat: disable
(client_id, path_type, path_id_bytes, timestamp,
stat_entry_bytes, hash_entry_bytes) = row
# pyformat: enable
path_id = rdf_objects.PathID.FromBytes(path_id_bytes)
components = path_id_components[path_id]
if stat_entry_bytes is not None:
stat_entry = rdf_client_fs.StatEntry.FromSerializedString(
stat_entry_bytes)
else:
stat_entry = None
hash_entry = rdf_crypto.Hash.FromSerializedString(hash_entry_bytes)
client_path = db.ClientPath(
client_id=db_utils.IntToClientID(client_id),
path_type=path_type,
components=path_id_components[path_id])
path_info = rdf_objects.PathInfo(
path_type=path_type,
components=components,
stat_entry=stat_entry,
hash_entry=hash_entry,
timestamp=mysql_utils.TimestampToRDFDatetime(timestamp))
path_infos[client_path] = path_info
return path_infos
def ReadPathInfosHistories(self,
client_id,
path_type,
components_list,
cursor=None):
"""Reads a collection of hash and stat entries for given paths."""
# MySQL does not handle well empty `IN` clauses so we guard against that.
if not components_list:
return {}
path_infos = {components: [] for components in components_list}
path_id_components = {}
for components in components_list:
path_id = rdf_objects.PathID.FromComponents(components)
path_id_components[path_id] = components
# MySQL does not support full outer joins, so we emulate them with a union.
query = """
SELECT s.path_id, s.stat_entry, unix_timestamp(s.timestamp),
h.path_id, h.hash_entry, unix_timestamp(h.timestamp)
FROM client_path_stat_entries AS s
LEFT JOIN client_path_hash_entries AS h
ON s.client_id = h.client_id
AND s.path_type = h.path_type
AND s.path_id = h.path_id
AND s.timestamp = h.timestamp
WHERE s.client_id = %(client_id)s
AND s.path_type = %(path_type)s
AND s.path_id IN %(path_ids)s
UNION
SELECT s.path_id, s.stat_entry, unix_timestamp(s.timestamp),
h.path_id, h.hash_entry, unix_timestamp(h.timestamp)
FROM client_path_hash_entries AS h
LEFT JOIN client_path_stat_entries AS s
ON h.client_id = s.client_id
AND h.path_type = s.path_type
AND h.path_id = s.path_id
AND h.timestamp = s.timestamp
WHERE h.client_id = %(client_id)s
AND h.path_type = %(path_type)s
AND h.path_id IN %(path_ids)s
"""
params = {
"client_id": db_utils.ClientIDToInt(client_id),
"path_type": int(path_type),
"path_ids": [path_id.AsBytes() for path_id in path_id_components]
}
cursor.execute(query, params)
for row in cursor.fetchall():
# pyformat: disable
(stat_entry_path_id_bytes, stat_entry_bytes, stat_entry_timestamp,
hash_entry_path_id_bytes, hash_entry_bytes, hash_entry_timestamp) = row
# pyformat: enable
path_id_bytes = stat_entry_path_id_bytes or hash_entry_path_id_bytes
path_id = rdf_objects.PathID.FromBytes(path_id_bytes)
components = path_id_components[path_id]
timestamp = stat_entry_timestamp or hash_entry_timestamp
if stat_entry_bytes is not None:
stat_entry = rdf_client_fs.StatEntry.FromSerializedString(
stat_entry_bytes)
else:
stat_entry = None
if hash_entry_bytes is not None:
hash_entry = rdf_crypto.Hash.FromSerializedString(hash_entry_bytes)
else:
hash_entry = None
path_info = rdf_objects.PathInfo(
path_type=path_type,
components=components,
stat_entry=stat_entry,
hash_entry=hash_entry,
timestamp=mysql_utils.TimestampToRDFDatetime(timestamp))
path_infos[components].append(path_info)
for components in components_list:
path_infos[components].sort(key=lambda path_info: path_info.timestamp)
return path_infos
def ListDescendentPathInfos(self,
client_id,
path_type,
components,
timestamp=None,
max_depth=None,
cursor=None):
"""Lists path info records that correspond to descendants of given path."""
path_infos = []
query = ""
path = mysql_utils.ComponentsToPath(components)
values = {
"client_id": db_utils.ClientIDToInt(client_id),
"path_type": int(path_type),
"path": db_utils.EscapeWildcards(path),
}
query += """
SELECT path, directory, unix_timestamp(p.timestamp),
stat_entry, unix_timestamp(last_stat_entry_timestamp),
hash_entry, unix_timestamp(last_hash_entry_timestamp)
FROM client_paths AS p
"""
if timestamp is None:
query += """
LEFT JOIN client_path_stat_entries AS s ON
(p.client_id = s.client_id AND
p.path_type = s.path_type AND
p.path_id = s.path_id AND
p.last_stat_entry_timestamp = s.timestamp)
LEFT JOIN client_path_hash_entries AS h ON
(p.client_id = h.client_id AND
p.path_type = h.path_type AND
p.path_id = h.path_id AND
p.last_hash_entry_timestamp = h.timestamp)
"""
only_explicit = False
else:
query += """
LEFT JOIN (SELECT sr.client_id, sr.path_type, sr.path_id, sr.stat_entry
FROM client_path_stat_entries AS sr
INNER JOIN (SELECT client_id, path_type, path_id,
MAX(timestamp) AS max_timestamp
FROM client_path_stat_entries
WHERE unix_timestamp(timestamp) <= %(timestamp)s
GROUP BY client_id, path_type, path_id) AS st
ON sr.client_id = st.client_id
AND sr.path_type = st.path_type
AND sr.path_id = st.path_id
AND sr.timestamp = st.max_timestamp) AS s
ON (p.client_id = s.client_id AND
p.path_type = s.path_type AND
p.path_id = s.path_id)
LEFT JOIN (SELECT hr.client_id, hr.path_type, hr.path_id, hr.hash_entry
FROM client_path_hash_entries AS hr
INNER JOIN (SELECT client_id, path_type, path_id,
MAX(timestamp) AS max_timestamp
FROM client_path_hash_entries
WHERE unix_timestamp(timestamp) <= %(timestamp)s
GROUP BY client_id, path_type, path_id) AS ht
ON hr.client_id = ht.client_id
AND hr.path_type = ht.path_type
AND hr.path_id = ht.path_id
AND hr.timestamp = ht.max_timestamp) AS h
ON (p.client_id = h.client_id AND
p.path_type = h.path_type AND
p.path_id = h.path_id)
"""
values["timestamp"] = mysql_utils.RDFDatetimeToTimestamp(timestamp)
only_explicit = True
query += """
WHERE p.client_id = %(client_id)s
AND p.path_type = %(path_type)s
AND path LIKE concat(%(path)s, '/%%')
"""
if max_depth is not None:
query += """
AND depth <= %(depth)s
"""
values["depth"] = len(components) + max_depth
cursor.execute(query, values)
for row in cursor.fetchall():
# pyformat: disable
(path, directory, timestamp,
stat_entry_bytes, last_stat_entry_timestamp,
hash_entry_bytes, last_hash_entry_timestamp) = row
# pyformat: enable
components = mysql_utils.PathToComponents(path)
if stat_entry_bytes is not None:
stat_entry = rdf_client_fs.StatEntry.FromSerializedString(
stat_entry_bytes)
else:
stat_entry = None
if hash_entry_bytes is not None:
hash_entry = rdf_crypto.Hash.FromSerializedString(hash_entry_bytes)
else:
hash_entry = None
datetime = mysql_utils.TimestampToRDFDatetime
path_info = rdf_objects.PathInfo(
path_type=path_type,
components=components,
timestamp=datetime(timestamp),
last_stat_entry_timestamp=datetime(last_stat_entry_timestamp),
last_hash_entry_timestamp=datetime(last_hash_entry_timestamp),
directory=directory,
stat_entry=stat_entry,
hash_entry=hash_entry)
path_infos.append(path_info)
path_infos.sort(key=lambda _: tuple(_.components))
# For specific timestamp, we return information only about explicit paths
# (paths that have associated stat or hash entry or have an ancestor that is
# explicit).
if not only_explicit:
return path_infos
explicit_path_infos = []
has_explicit_ancestor = set()
# This list is sorted according to the keys component, so by traversing it
# in the reverse order we make sure that we process deeper paths first.
for path_info in reversed(path_infos):
components = tuple(path_info.components)
if (path_info.HasField("stat_entry") or
path_info.HasField("hash_entry") or
components in has_explicit_ancestor):
explicit_path_infos.append(path_info)
has_explicit_ancestor.add(components[:-1])
# Since we collected explicit paths in reverse order, we need to reverse it
# again to conform to the interface.
return list(reversed(explicit_path_infos))
def ReadPathInfo(self,
client_id,
path_type,
components,
timestamp=None,
cursor=None):
"""Retrieves a path info record for a given path."""
if timestamp is None:
path_infos = self.ReadPathInfos(client_id, path_type, [components])
path_info = path_infos[components]
if path_info is None:
raise db.UnknownPathError(
client_id=client_id, path_type=path_type, components=components)
return path_info
query = """
SELECT directory, unix_timestamp(p.timestamp),
stat_entry, unix_timestamp(last_stat_entry_timestamp),
hash_entry, unix_timestamp(last_hash_entry_timestamp)
FROM client_paths as p
LEFT JOIN (SELECT client_id, path_type, path_id, stat_entry
FROM client_path_stat_entries
WHERE client_id = %(client_id)s
AND path_type = %(path_type)s
AND path_id = %(path_id)s
AND unix_timestamp(timestamp) <= %(timestamp)s
ORDER BY timestamp DESC
LIMIT 1) AS s
ON p.client_id = s.client_id
AND p.path_type = s.path_type
AND p.path_id = s.path_id
LEFT JOIN (SELECT client_id, path_type, path_id, hash_entry
FROM client_path_hash_entries
WHERE client_id = %(client_id)s
AND path_type = %(path_type)s
AND path_id = %(path_id)s
AND unix_timestamp(timestamp) <= %(timestamp)s
ORDER BY timestamp DESC
LIMIT 1) AS h
ON p.client_id = h.client_id
AND p.path_type = h.path_type
AND p.path_id = h.path_id
WHERE p.client_id = %(client_id)s
AND p.path_type = %(path_type)s
AND p.path_id = %(path_id)s
"""
values = {
"client_id": db_utils.ClientIDToInt(client_id),
"path_type": int(path_type),
"path_id": rdf_objects.PathID.FromComponents(components).AsBytes(),
"timestamp": mysql_utils.RDFDatetimeToTimestamp(timestamp),
}
cursor.execute(query, values)
row = cursor.fetchone()
if row is None:
raise db.UnknownPathError(
client_id=client_id, path_type=path_type, components=components)
# pyformat: disable
(directory, timestamp,
stat_entry_bytes, last_stat_entry_timestamp,
hash_entry_bytes, last_hash_entry_timestamp) = row
# pyformat: enable
if stat_entry_bytes is not None:
stat_entry = rdf_client_fs.StatEntry.FromSerializedString(
stat_entry_bytes)
else:
stat_entry = None
if hash_entry_bytes is not None:
hash_entry = rdf_crypto.Hash.FromSerializedString(hash_entry_bytes)
else:
hash_entry = None
datetime = mysql_utils.TimestampToRDFDatetime
return rdf_objects.PathInfo(
path_type=path_type,
components=components,
timestamp=datetime(timestamp),
last_stat_entry_timestamp=datetime(last_stat_entry_timestamp),
last_hash_entry_timestamp=datetime(last_hash_entry_timestamp),
directory=directory,
stat_entry=stat_entry,
hash_entry=hash_entry)
def _MultiWritePathInfos(self, path_infos, cursor=None):
"""Writes a collection of path info records for specified clients."""
query = ""
path_info_count = 0
path_info_values = []
parent_path_info_count = 0
parent_path_info_values = []
has_stat_entries = False
has_hash_entries = False
for client_id, client_path_infos in iteritems(path_infos):
for path_info in client_path_infos:
path = mysql_utils.ComponentsToPath(path_info.components)
path_info_values.append(db_utils.ClientIDToInt(client_id))
path_info_values.append(int(path_info.path_type))
path_info_values.append(path_info.GetPathID().AsBytes())
path_info_values.append(path)
path_info_values.append(bool(path_info.directory))
path_info_values.append(len(path_info.components))
if path_info.HasField("stat_entry"):
path_info_values.append(path_info.stat_entry.SerializeToString())
has_stat_entries = True
else:
path_info_values.append(None)
if path_info.HasField("hash_entry"):
path_info_values.append(path_info.hash_entry.SerializeToString())
path_info_values.append(path_info.hash_entry.sha256.AsBytes())
has_hash_entries = True
else:
path_info_values.append(None)
path_info_values.append(None)
path_info_count += 1
# TODO(hanuszczak): Implement a trie in order to avoid inserting
# duplicated records.
for parent_path_info in path_info.GetAncestors():
path = mysql_utils.ComponentsToPath(parent_path_info.components)
parent_path_info_values.append(db_utils.ClientIDToInt(client_id))
parent_path_info_values.append(int(parent_path_info.path_type))
parent_path_info_values.append(parent_path_info.GetPathID().AsBytes())
parent_path_info_values.append(path)
parent_path_info_values.append(len(parent_path_info.components))
parent_path_info_count += 1
query += """
CREATE TEMPORARY TABLE client_path_infos(
client_id BIGINT UNSIGNED NOT NULL,
path_type INT UNSIGNED NOT NULL,
path_id BINARY(32) NOT NULL,
path TEXT NOT NULL,
directory BOOLEAN NOT NULL,
depth INT NOT NULL,
stat_entry MEDIUMBLOB NULL,
hash_entry MEDIUMBLOB NULL,
sha256 BINARY(32) NULL,
timestamp TIMESTAMP(6) NOT NULL DEFAULT now(6)
);"""
if path_info_count > 0:
query += """
INSERT INTO client_path_infos(client_id, path_type, path_id,
path, directory, depth,
stat_entry, hash_entry, sha256)
VALUES {};
""".format(mysql_utils.Placeholders(num=9, values=path_info_count))
query += """
INSERT INTO client_paths(client_id, path_type, path_id, path,
directory, depth)
SELECT client_id, path_type, path_id, path, directory, depth
FROM client_path_infos
ON DUPLICATE KEY UPDATE
client_paths.directory = client_paths.directory OR VALUES(client_paths.directory),
client_paths.timestamp = now(6);
"""
if parent_path_info_count > 0:
placeholders = ["(%s, %s, %s, %s, TRUE, %s)"] * parent_path_info_count
query += """
INSERT INTO client_paths(client_id, path_type, path_id, path,
directory, depth)
VALUES {}
ON DUPLICATE KEY UPDATE
directory = TRUE,
timestamp = now();
""".format(", ".join(placeholders))
if has_stat_entries:
query += """
INSERT INTO client_path_stat_entries(client_id, path_type, path_id,
stat_entry, timestamp)
SELECT client_id, path_type, path_id, stat_entry, timestamp
FROM client_path_infos
WHERE stat_entry IS NOT NULL;
"""
query += """
UPDATE client_paths, client_path_infos
SET client_paths.last_stat_entry_timestamp = client_path_infos.timestamp
WHERE client_paths.client_id = client_path_infos.client_id
AND client_paths.path_type = client_path_infos.path_type
AND client_paths.path_id = client_path_infos.path_id
AND client_path_infos.stat_entry IS NOT NULL;
"""
if has_hash_entries:
query += """
INSERT INTO client_path_hash_entries(client_id, path_type, path_id,
hash_entry, sha256, timestamp)
SELECT client_id, path_type, path_id, hash_entry, sha256, timestamp
FROM client_path_infos
WHERE hash_entry IS NOT NULL;
"""
query += """
UPDATE client_paths, client_path_infos
SET client_paths.last_hash_entry_timestamp = client_path_infos.timestamp
WHERE client_paths.client_id = client_path_infos.client_id
AND client_paths.path_type = client_path_infos.path_type
AND client_paths.path_id = client_path_infos.path_id
AND client_path_infos.hash_entry IS NOT NULL;
"""
query += """
DROP TEMPORARY TABLE client_path_infos;
"""
cursor.execute(query, path_info_values + parent_path_info_values)
def ReadPathInfos(self, client_id, path_type, components_list, cursor=None):
"""Retrieves path info records for given paths."""
if not components_list:
return {}
path_ids = list(map(rdf_objects.PathID.FromComponents, components_list))
path_infos = {components: None for components in components_list}
query = """
SELECT path, directory, unix_timestamp(client_paths.timestamp),
stat_entry, unix_timestamp(last_stat_entry_timestamp),
hash_entry, unix_timestamp(last_hash_entry_timestamp)
FROM client_paths
LEFT JOIN client_path_stat_entries ON
(client_paths.client_id = client_path_stat_entries.client_id AND
client_paths.path_type = client_path_stat_entries.path_type AND
client_paths.path_id = client_path_stat_entries.path_id AND
client_paths.last_stat_entry_timestamp = client_path_stat_entries.timestamp)
LEFT JOIN client_path_hash_entries ON
(client_paths.client_id = client_path_hash_entries.client_id AND
client_paths.path_type = client_path_hash_entries.path_type AND
client_paths.path_id = client_path_hash_entries.path_id AND
client_paths.last_hash_entry_timestamp = client_path_hash_entries.timestamp)
WHERE client_paths.client_id = %(client_id)s
AND client_paths.path_type = %(path_type)s
AND client_paths.path_id IN %(path_ids)s
"""
values = {
"client_id": db_utils.ClientIDToInt(client_id),
"path_type": int(path_type),
"path_ids": [path_id.AsBytes() for path_id in path_ids]
}
cursor.execute(query, values)
for row in cursor.fetchall():
# pyformat: disable
(path, directory, timestamp,
stat_entry_bytes, last_stat_entry_timestamp,
hash_entry_bytes, last_hash_entry_timestamp) = row
# pyformat: enable
components = mysql_utils.PathToComponents(path)
if stat_entry_bytes is not None:
stat_entry = rdf_client_fs.StatEntry.FromSerializedString(
stat_entry_bytes)
else:
stat_entry = None
if hash_entry_bytes is not None:
hash_entry = rdf_crypto.Hash.FromSerializedString(hash_entry_bytes)
else:
hash_entry = None
datetime = mysql_utils.TimestampToRDFDatetime
path_info = rdf_objects.PathInfo(
path_type=path_type,
components=components,
timestamp=datetime(timestamp),
last_stat_entry_timestamp=datetime(last_stat_entry_timestamp),
last_hash_entry_timestamp=datetime(last_hash_entry_timestamp),
directory=directory,
stat_entry=stat_entry,
hash_entry=hash_entry)
path_infos[components] = path_info
return path_infos