def __init__(self,dataset, **kwargs):
self.id_ = kwargs.get("id",kwargs.get("id_",None))
self.name = kwargs.get("name",kwargs.get("name",None))
self.vname = kwargs.get("vname",None)
self.fqname = kwargs.get("fqname",None)
self.cache_key = kwargs.get("cache_key",None)
self.sequence_id = kwargs.get("sequence_id",None)
self.d_id = kwargs.get("d_id",None)
self.space = kwargs.get("space",None)
self.time = kwargs.get("time",None)
self.t_id = kwargs.get("t_id",None)
self.grain = kwargs.get('grain',None)
self.format = kwargs.get('format',None)
self.segment = kwargs.get('segment',None)
self.data = kwargs.get('data',None)
self.d_id = dataset.id_
self.d_vid = dataset.vid
# See before_insert for setting self.vid and self.id_
if self.t_id:
don = ObjectNumber.parse(self.d_vid)
ton = ObjectNumber.parse(self.t_id)
self.t_vid = str(ton.rev( don.revision))
assert self.cache_key is not None
if True: # Debugging
from partition import extension_for_format_name
ext = extension_for_format_name(self.format)
assert self.cache_key.endswith(ext)
def deref_tc_ref(ref):
"""Given a column or table, vid or id, return the object."""
from ambry.identity import ObjectNumber
on = ObjectNumber.parse(ref)
b = str(on.as_dataset)
try:
c = on
t = on.as_table
except AttributeError:
t = on
c = None
if not on.revision:
# The table does not have a revision, so we need to get one, just get the
# latest one
from . import renderer
r = renderer()
dc = r.doc_cache
tm = dc.table_version_map()
t_vid = reversed(sorted(tm.get(str(t)))).next()
t = ObjectNumber.parse(t_vid)
b = t.as_dataset
if c:
c = c.rev(t.revision)
return b, t, c
def update_number(target):
ton = ObjectNumber.parse(target.t_vid)
con = ColumnNumber(ton, target.sequence_id)
target.id = str(ton.rev(None))
target.vid = str(con)
target.id = str(con.rev(None))
target.d_vid = str(ObjectNumber.parse(target.t_vid).as_dataset)
def update_number(target):
ton = ObjectNumber.parse(target.t_vid)
con = ColumnNumber(ton, target.sequence_id)
target.id = str(ton.rev(None))
target.vid = str(con)
target.id = str(con.rev(None))
target.d_vid = str(ObjectNumber.parse(target.t_vid).as_dataset)
def test_id(self):
dnn = 1000000
rev = 100
dn = DatasetNumber(dnn)
self.assertEqual('d000004c92', str(dn))
dn = DatasetNumber(dnn, rev)
self.assertEqual('d000004c9201C', str(dn))
self.assertEqual('d000004c9201C', str(ObjectNumber.parse(str(dn))))
tn = TableNumber(dn, 1)
self.assertEqual('t000004c920101C', str(tn))
self.assertEqual('t000004c920101C', str(ObjectNumber.parse(str(tn))))
tnnr = tn.rev(None)
self.assertEqual('t000004c9201', str(tnnr))
self.assertEqual('t000004c9201004', str(tnnr.rev(4)))
# Other assignment classes
# dnn = 62 * 62 + 11
dn = DatasetNumber(62**3 - 1, None, 'authoritative')
self.assertEqual('dZZZ', str(dn))
dn = DatasetNumber(62**3 - 1, None, 'registered')
self.assertEqual('d00ZZZ', str(dn))
dn = DatasetNumber(62**3 - 1, None, 'unregistered')
self.assertEqual('d0000ZZZ', str(dn))
dn = DatasetNumber(62**3 - 1, None, 'self')
self.assertEqual('d000000ZZZ', str(dn))
tn = TableNumber(dn, 2)
self.assertEqual('t000000ZZZ02', str(tn))
cn = ColumnNumber(tn, 3)
self.assertEqual('c000000ZZZ02003', str(cn))
pn = dn.as_partition(5)
self.assertEqual('p000000ZZZ005', str(pn))
def test_id(self):
dnn = 1000000
rev = 100
dn = DatasetNumber(dnn)
self.assertEqual('d000004c92', str(dn))
dn = DatasetNumber(dnn, rev)
self.assertEqual('d000004c9201C', str(dn))
self.assertEqual('d000004c9201C', str(ObjectNumber.parse(str(dn))))
tn = TableNumber(dn, 1)
self.assertEqual('t000004c920101C', str(tn))
self.assertEqual('t000004c920101C', str(ObjectNumber.parse(str(tn))))
tnnr = tn.rev(None)
self.assertEqual('t000004c9201', str(tnnr))
self.assertEqual('t000004c9201004', str(tnnr.rev(4)))
# Other assignment classes
# dnn = 62 * 62 + 11
dn = DatasetNumber(62 ** 3 - 1, None, 'authoritative')
self.assertEqual('dZZZ', str(dn))
dn = DatasetNumber(62 ** 3 - 1, None, 'registered')
self.assertEqual('d00ZZZ', str(dn))
dn = DatasetNumber(62 ** 3 - 1, None, 'unregistered')
self.assertEqual('d0000ZZZ', str(dn))
dn = DatasetNumber(62 ** 3 - 1, None, 'self')
self.assertEqual('d000000ZZZ', str(dn))
tn = TableNumber(dn, 2)
self.assertEqual('t000000ZZZ02', str(tn))
cn = ColumnNumber(tn, 3)
self.assertEqual('c000000ZZZ02003', str(cn))
pn = dn.as_partition(5)
self.assertEqual('p000000ZZZ005', str(pn))
def table(self, ref):
""" Finds table by ref and returns it.
Args:
ref (str): id, vid (versioned id) or name of the table
Raises:
NotFoundError: if table with given ref not found.
Returns:
orm.Table
"""
try:
obj_number = ObjectNumber.parse(ref)
ds_obj_number = obj_number.as_dataset
dataset = self._db.dataset(
ds_obj_number
) # Could do it in on SQL query, but this is easier.
table = dataset.table(ref)
except NotObjectNumberError:
q = self.database.session.query(Table)\
.filter(Table.name == str(ref))\
.order_by(Table.vid.desc())
table = q.first()
if not table:
raise NotFoundError("No table for ref: '{}'".format(ref))
return table
def find_remote_bundle(self, ref, try_harder=None):
"""
Locate a bundle, by any reference, among the configured remotes. The routine will only look in the cache
directory lists stored in the remotes, which must be updated to be current.
:param vid: A bundle or partition reference, vid, or name
:param try_harder: If the reference isn't found, try parsing for an object id, or subsets of the name
:return: (remote,vname) or (None,None) if the ref is not found
"""
from ambry.identity import ObjectNumber
remote, vid = self._find_remote_bundle(ref)
if remote:
return (remote, vid)
if try_harder:
on = ObjectNumber.parse(vid)
if on:
raise NotImplementedError()
don = on.as_dataset
return self._find_remote_bundle(vid)
# Try subsets of a name, assuming it is a name
parts = ref.split('-')
for i in range(len(parts) - 1, 2, -1):
remote, vid = self._find_remote_bundle('-'.join(parts[:i]))
if remote:
return (remote, vid)
return (None, None)
def install(self, ref, table_name=None, index_columns=None,logger=None):
""" Finds partition by reference and installs it to warehouse db.
Args:
ref (str): id, vid (versioned id), name or vname (versioned name) of the partition.
"""
try:
obj_number = ObjectNumber.parse(ref)
if isinstance(obj_number, TableNumber):
table = self._library.table(ref)
connection = self._backend._get_connection()
return self._backend.install_table(connection, table, logger=logger)
else:
# assume partition
raise NotObjectNumberError
except NotObjectNumberError:
# assume partition.
partition = self._library.partition(ref)
connection = self._backend._get_connection()
return self._backend.install(
connection, partition, table_name=table_name, index_columns=index_columns,
logger=logger)
def search(self, search_phrase, limit=None):
"""Search for datasets, and expand to database records"""
from ambry.identity import ObjectNumber
from ambry.orm.exc import NotFoundError
from ambry.library.search_backends.base import SearchTermParser
results = []
stp = SearchTermParser()
# Because of the split between searching for partitions and bundles, some terms don't behave right.
# The source term should be a limit on everything, but it isn't part of the partition doc,
# so we check for it here.
parsed_terms = stp.parse(search_phrase)
for r in self.search_datasets(search_phrase, limit):
vid = r.vid or ObjectNumber.parse(next(iter(r.partitions))).as_dataset
r.vid = vid
try:
r.bundle = self.library.bundle(r.vid)
if 'source' not in parsed_terms or parsed_terms['source'] in r.bundle.dataset.source:
results.append(r)
except NotFoundError:
pass
return sorted(results, key=lambda r : r.score, reverse=True)
def before_update(mapper, conn, target):
'''Set the column id number based on the table number and the
sequence id for the column'''
if target.id_ is None:
table_on = ObjectNumber.parse(target.t_id)
target.id_ = str(ColumnNumber(table_on, target.sequence_id))
def __init__(self,table, **kwargs):
self.sequence_id = kwargs.get("sequence_id",len(table.columns)+1)
self.name = kwargs.get("name",None)
self.altname = kwargs.get("altname",None)
self.is_primary_key = _clean_flag(kwargs.get("is_primary_key",False))
self.datatype = kwargs.get("datatype",None)
self.size = kwargs.get("size",None)
self.precision = kwargs.get("precision",None)
self.width = kwargs.get("width",None)
self.sql = kwargs.get("sql",None)
self.flags = kwargs.get("flags",None)
self.description = kwargs.get("description",None)
self.keywords = kwargs.get("keywords",None)
self.measure = kwargs.get("measure",None)
self.units = kwargs.get("units",None)
self.universe = kwargs.get("universe",None)
self.scale = kwargs.get("scale",None)
self.data = kwargs.get("data",None)
# the table_name attribute is not stored. It is only for
# building the schema, linking the columns to tables.
self.table_name = kwargs.get("table_name",None)
if not self.name:
raise ValueError('Column must have a name')
self.t_id = table.id_
self.t_vid = table.vid
ton = ObjectNumber.parse(table.vid)
con = ColumnNumber(ton, self.sequence_id)
self.vid = str(con)
self.id = str(con.rev(None))
def test_parse_other(self):
dn = DatasetNumber(100, 5, 'authoritative')
self.assertEqual('d01C005', str(dn))
self.assertEqual('G01C001Z005', str(GeneralNumber1('G', dn, 123)))
self.assertEqual(
'G01C001Z005',
ObjectNumber.parse(str(GeneralNumber1('G', dn, 123)), 'other1'))
def __init__(self,**kwargs):
self.id_ = kwargs.get("oid",kwargs.get("id",kwargs.get("id_", None)) )
self.vid = kwargs.get("vid", None)
self.location = kwargs.get("location", self.LOCATION.LIBRARY)
self.name = kwargs.get("name",None)
self.vname = kwargs.get("vname",None)
self.fqname = kwargs.get("fqname",None)
self.cache_key = kwargs.get("cache_key",None)
self.source = kwargs.get("source",None)
self.dataset = kwargs.get("dataset",None)
self.subset = kwargs.get("subset",None)
self.variation = kwargs.get("variation",None)
self.btime = kwargs.get("btime", None)
self.bspace = kwargs.get("bspace", None)
self.creator = kwargs.get("creator",None)
self.revision = kwargs.get("revision",None)
self.version = kwargs.get("version",None)
if not self.id_:
dn = DatasetNumber(None, self.revision )
self.vid = str(dn)
self.id_ = str(dn.rev(None))
elif not self.vid:
try:
self.vid = str(ObjectNumber.parse(self.id_).rev(self.revision))
except ValueError as e:
print repr(self)
raise ValueError('Could not parse id value; '+e.message)
if self.cache_key is None:
self.cache_key = self.identity.cache_key
assert self.vid[0] == 'd'
def search(self, search_phrase, limit=None):
"""Search for datasets, and expand to database records"""
from ambry.identity import ObjectNumber
from ambry.orm.exc import NotFoundError
from ambry.library.search_backends.base import SearchTermParser
results = []
stp = SearchTermParser()
# Because of the split between searching for partitions and bundles, some terms don't behave right.
# The source term should be a limit on everything, but it isn't part of the partition doc,
# so we check for it here.
parsed_terms = stp.parse(search_phrase)
for r in self.search_datasets(search_phrase, limit):
vid = r.vid or ObjectNumber.parse(next(iter(
r.partitions))).as_dataset
r.vid = vid
try:
r.bundle = self.library.bundle(r.vid)
if 'source' not in parsed_terms or parsed_terms[
'source'] in r.bundle.dataset.source:
results.append(r)
except NotFoundError:
pass
return sorted(results, key=lambda r: r.score, reverse=True)
def tc_obj(ref):
"""Return an object for a table or column."""
dc = renderer().doc_cache
try:
b, t, c = deref_tc_ref(ref)
except NotFoundError:
return None
try:
table = dc.table(str(t))
except NotFoundError:
# This can happen when the table reference has a version id in it, and that version is not available.
# So, try it again without the version
table = dc.table(str(ObjectNumber.parse(str(t)).rev(None)))
if c:
try:
return table['columns'][str(c.rev(0))]
except KeyError:
return None
except TypeError:
return None
else:
return table
def find_remote_bundle(self, ref, try_harder=None):
"""
Locate a bundle, by any reference, among the configured remotes. The routine will only look in the cache
directory lists stored in the remotes, which must be updated to be current.
:param vid: A bundle or partition reference, vid, or name
:param try_harder: If the reference isn't found, try parsing for an object id, or subsets of the name
:return: (remote,vname) or (None,None) if the ref is not found
"""
from ambry.identity import ObjectNumber
remote, vid = self._find_remote_bundle(ref)
if remote:
return (remote, vid)
if try_harder:
on = ObjectNumber.parse(vid)
if on:
raise NotImplementedError()
don = on.as_dataset
return self._find_remote_bundle(vid)
# Try subsets of a name, assuming it is a name
parts = ref.split('-')
for i in range(len(parts) - 1, 2, -1):
remote, vid = self._find_remote_bundle('-'.join(parts[:i]))
if remote:
return (remote, vid)
return (None, None)
def table(self, ref):
""" Finds table by ref and returns it.
Args:
ref (str): id, vid (versioned id) or name of the table
Raises:
NotFoundError: if table with given ref not found.
Returns:
orm.Table
"""
try:
obj_number = ObjectNumber.parse(ref)
ds_obj_number = obj_number.as_dataset
dataset = self._db.dataset(ds_obj_number) # Could do it in on SQL query, but this is easier.
table = dataset.table(ref)
except NotObjectNumberError:
q = self.database.session.query(Table)\
.filter(Table.name == str(ref))\
.order_by(Table.vid.desc())
table = q.first()
if not table:
raise NotFoundError("No table for ref: '{}'".format(ref))
return table
def test_returns_identity_of_dataset_found_by_dataset_number(self):
ds1 = DatasetFactory()
ds_number = ObjectNumber.parse(ds1.vid)
assert isinstance(ds_number, DatasetNumber)
ret = self.sqlite_db.get(ds_number)
self.assertEquals(ret.vid, ds1.vid)
def install(self, ref, table_name=None, index_columns=None, logger=None):
""" Finds partition by reference and installs it to warehouse db.
Args:
ref (str): id, vid (versioned id), name or vname (versioned name) of the partition.
"""
try:
obj_number = ObjectNumber.parse(ref)
if isinstance(obj_number, TableNumber):
table = self._library.table(ref)
connection = self._backend._get_connection()
return self._backend.install_table(connection,
table,
logger=logger)
else:
# assume partition
raise NotObjectNumberError
except NotObjectNumberError:
# assume partition.
partition = self._library.partition(ref)
connection = self._backend._get_connection()
return self._backend.install(connection,
partition,
table_name=table_name,
index_columns=index_columns,
logger=logger)
def update_id(self, sequence_id=None, force=True):
"""Alter the sequence id, and all of the names and ids derived from it. This
often needs to be don after an IntegrityError in a multiprocessing run"""
from ..identity import ObjectNumber
if sequence_id:
self.sequence_id = sequence_id
assert self.d_vid
if self.id is None or force:
dataset_id = ObjectNumber.parse(self.d_vid).rev(None)
self.d_id = str(dataset_id)
self.id = str(TableNumber(dataset_id, self.sequence_id))
if self.vid is None or force:
dataset_vid = ObjectNumber.parse(self.d_vid)
self.vid = str(TableNumber(dataset_vid, self.sequence_id))
def before_update(mapper, conn, target):
'''Set the Table ID based on the dataset number and the sequence number
for the table '''
if isinstance(target,Column):
raise TypeError('Got a column instead of a table')
if target.id_ is None:
dataset_id = ObjectNumber.parse(target.d_id)
target.id_ = str(TableNumber(dataset_id, target.sequence_id))
def test_returns_identity_of_dataset_found_by_partition_number(self):
ds1 = DatasetFactory()
partition1 = PartitionFactory(dataset=ds1)
part_number = ObjectNumber.parse(partition1.vid)
assert isinstance(part_number, PartitionNumber)
ret = self.sqlite_db.get(part_number)
self.assertEquals(ret.vid, ds1.vid)
self.assertIn(partition1.vid, ret.partitions)
def update_id(self, sequence_id=None, force=True):
"""Alter the sequence id, and all of the names and ids derived from it. This
often needs to be don after an IntegrityError in a multiprocessing run"""
from ..identity import ObjectNumber
if sequence_id:
self.sequence_id = sequence_id
assert self.d_vid
if self.id is None or force:
dataset_id = ObjectNumber.parse(self.d_vid).rev(None)
self.d_id = str(dataset_id)
self.id = str(TableNumber(dataset_id, self.sequence_id))
if self.vid is None or force:
dataset_vid = ObjectNumber.parse(self.d_vid)
self.vid = str(TableNumber(dataset_vid, self.sequence_id))
def deref_tc_ref(ref):
"""Given a column or table, vid or id, return the object."""
from ambry.identity import ObjectNumber
from ambry.dbexceptions import NotFoundError
on = ObjectNumber.parse(ref)
b = str(on.as_dataset)
try:
c = on
t = on.as_table
except AttributeError:
t = on
c = None
if not on.revision:
# The table does not have a revision, so we need to get one, just get the
# latest one
from . import renderer
r = renderer()
dc = r.doc_cache
tm = dc.table_version_map()
if not str(t) in tm:
# This happens when the the referenced table is in a bundle that is not installed,
# often because it is private or restricted
raise NotFoundError('Table {} not in table_version_map'.format(str(t)))
t_vid = reversed(sorted(tm.get(str(t)))).next()
t = ObjectNumber.parse(t_vid)
b = t.as_dataset
if c:
c = c.rev(t.revision)
return b, t, c
def set_ids(self, sequence_id):
from identity import Identity
self.sequence_id = sequence_id
don = ObjectNumber.parse(self.d_vid)
pon = PartitionNumber(don, self.sequence_id)
self.vid = str(pon)
self.id_ = str(pon.rev(None))
self.fqname = Identity._compose_fqname(self.vname,self.vid)
def resolve_object_number(self, ref):
"""Resolve a variety of object numebrs to a dataset number"""
if not isinstance(ref, ObjectNumber):
on = ObjectNumber.parse(ref)
else:
on = ref
ds_on = on.as_dataset
return ds_on
def test_increment(self):
dnn = 1000000
rev = 100
dn = DatasetNumber(dnn, rev)
self.assertEqual('d000004c9201C', str(dn))
dn2 = ObjectNumber.increment(dn)
self.assertEqual(101, dn2.revision)
dn3 = ObjectNumber.increment(dn2)
self.assertEqual(102, dn3.revision)
tn = TableNumber(dn3, 1)
self.assertEqual(102, tn.revision)
self.assertEqual('t000004c920101E', str(tn))
tn2 = ObjectNumber.increment(tn)
self.assertEqual(103, tn2.revision)
def resolve_object_number(self, ref):
"""Resolve a variety of object numebrs to a dataset number"""
if not isinstance(ref, ObjectNumber):
on = ObjectNumber.parse(ref)
else:
on = ref
ds_on = on.as_dataset
return ds_on
def test_increment(self):
dnn = 1000000
rev = 100
dn = DatasetNumber(dnn, rev)
self.assertEqual('d000004c9201C', str(dn))
dn2 = ObjectNumber.increment(dn)
self.assertEqual(101, dn2.revision)
dn3 = ObjectNumber.increment(dn2)
self.assertEqual(102, dn3.revision)
tn = TableNumber(dn3, 1)
self.assertEqual(102, tn.revision)
self.assertEqual('t000004c920101E', str(tn))
tn2 = ObjectNumber.increment(tn)
self.assertEqual(103, tn2.revision)
def substitute_vids(library, statement):
""" Replace all of the references to tables and partitions with their vids.
This is a bit of a hack -- it ought to work with the parser, but instead it just looks for
common SQL tokens that indicate an identifier.
:param statement: an sqlstatement. String.
:return: tuple: new_statement, set of table vids, set of partition vids.
"""
from ambry.identity import ObjectNumber, TableNumber, NotObjectNumberError
from ambry.orm.exc import NotFoundError
try:
stmt_str = statement.to_unicode()
except AttributeError:
stmt_str = statement
parts = stmt_str.strip(';').split()
new_parts = []
tables = set()
partitions = set()
while parts:
token = parts.pop(0).strip()
if token.lower() in ('from', 'join', 'materialize', 'install'):
ident = parts.pop(0).strip(';')
new_parts.append(token)
try:
obj_number = ObjectNumber.parse(token)
if isinstance(obj_number, TableNumber):
table = library.table(ident)
tables.add(table.vid)
new_parts.append(table.vid)
else:
# Do not care about other object numbers. Assume partition.
raise NotObjectNumberError
except NotObjectNumberError:
# assume partition
try:
partition = library.partition(ident)
partitions.add(partition.vid)
new_parts.append(partition.vid)
except NotFoundError:
# Ok, maybe it is just a normal identifier...
new_parts.append(ident)
else:
new_parts.append(token)
return ' '.join(new_parts).strip(), tables, partitions
def partition_path(b, p=None):
if p is None:
from ambry.identity import ObjectNumber
p = b
on = ObjectNumber.parse(p)
try:
b = str(on.as_dataset)
except AttributeError:
b = str(on)
raise
return "/bundles/{}/partitions/{}.html".format(resolve(b), resolve(p))
def substitute_vids(library, statement):
""" Replace all of the references to tables and partitions with their vids.
This is a bit of a hack -- it ought to work with the parser, but instead it just looks for
common SQL tokens that indicate an identifier.
:param statement: an sqlstatement. String.
:return: tuple: new_statement, set of table vids, set of partition vids.
"""
from ambry.identity import ObjectNumber, TableNumber, NotObjectNumberError
from ambry.orm.exc import NotFoundError
try:
stmt_str = statement.to_unicode()
except AttributeError:
stmt_str = statement
parts = stmt_str.strip(';').split()
new_parts = []
tables = set()
partitions = set()
while parts:
token = parts.pop(0).strip()
if token.lower() in ('from', 'join', 'materialize', 'install'):
ident = parts.pop(0).strip(';')
new_parts.append(token)
try:
obj_number = ObjectNumber.parse(token)
if isinstance(obj_number, TableNumber):
table = library.table(ident)
tables.add(table.vid)
new_parts.append(table.vid)
else:
# Do not care about other object numbers. Assume partition.
raise NotObjectNumberError
except NotObjectNumberError:
# assume partition
try:
partition = library.partition(ident)
partitions.add(partition.vid)
new_parts.append(partition.vid)
except NotFoundError:
# Ok, maybe it is just a normal identifier...
new_parts.append(ident)
else:
new_parts.append(token)
return ' '.join(new_parts).strip(), tables, partitions
def __init__(self, *args, **kwargs):
super(Dataset, self).__init__(*args, **kwargs)
if self.vid and not self.id:
self.revision = ObjectNumber.parse(self.vid).revision
self.id = str(ObjectNumber.parse(self.vid).rev(None))
if not self.id:
dn = DatasetNumber(None, self.revision)
self.vid = str(dn)
self.id = str(dn.rev(None))
elif not self.vid:
try:
self.vid = str(ObjectNumber.parse(self.id).rev(self.revision))
except ValueError as e:
raise ValueError('Could not parse id value; ' + e.message)
if not self.revision:
self.revision = 1
if self.cache_key is None:
self.cache_key = self.identity.name.cache_key
if not self.name:
self.name = str(self.identity.name)
if not self.vname:
self.vname = str(self.identity.vname)
if not self.fqname:
self.fqname = str(self.identity.fqname)
if not self.version:
self.version = str(self.identity.version)
assert self.vid[0] == 'd'
def augment(path,o, parent=None):
"""Descend recursively"""
from ambry.identity import ObjectNumber
path_num = ObjectNumber.base62_decode(''.join( ObjectNumber.base62_encode(e[1]).zfill(2) for e in path ))
objects = []
if isinstance(o, dict):
for i,(k,v) in enumerate(o.items(),1):
objects += augment(path+[ (k,i,) ], v, k)
o['path'] = tuple(path)
o['path_num'] = path_num
o['parent'] = parent if parent else 'root'
objects += [{ k:v for k,v in o.items() if not isinstance(v,(list,dict))}]
elif isinstance(o, list):
for i, v in enumerate(o,1):
objects += augment(path+[ (i,i,) ], v, parent)
return objects
def partition_path(b, p=None):
if p is None:
p = b
try:
on = ObjectNumber.parse(p)
b = str(on.as_dataset)
except NotObjectNumberError as e:
return None
except AttributeError:
b = str(on)
raise
return '/bundles/{}/partitions/{}.html'.format(resolve(b), resolve(p))
def incver(o, prop_names):
"""Increment the version numbers of a set of properties and return a new object"""
from ambry.identity import ObjectNumber
d = {}
for p in o.__mapper__.attrs:
v = getattr(o, p.key)
if v is None:
d[p.key] = None
elif p.key in prop_names:
d[p.key] = str(ObjectNumber.increment(v))
else:
if not hasattr(v, '__mapper__'): # Only copy values, never objects
d[p.key] = v
return o.__class__(**d)
def incver(o, prop_names):
"""Increment the version numbers of a set of properties and return a new object"""
from ambry.identity import ObjectNumber
d = {}
for p in o.__mapper__.attrs:
v = getattr(o, p.key)
if v is None:
d[p.key] = None
elif p.key in prop_names:
d[p.key] = str(ObjectNumber.increment(v))
else:
if not hasattr(v, '__mapper__'): # Only copy values, never objects
d[p.key] = v
return o.__class__(**d)
def _set_ids(self, force=False):
if not self.sequence_id:
from .exc import DatabaseError
raise DatabaseError('Sequence ID must be set before insertion')
if not self.vid or force:
assert bool(self.d_vid)
assert bool(self.sequence_id)
don = ObjectNumber.parse(self.d_vid)
assert don.revision
on = don.as_partition(self.sequence_id)
self.vid = str(on.rev(don.revision))
self.id = str(on.rev(None))
if not self.data:
self.data = {}
def partition(self, ref, localize=False):
""" Finds partition by ref and converts to bundle partition.
:param ref: A partition reference
:param localize: If True, copy a remote partition to local filesystem. Defaults to False
:raises: NotFoundError: if partition with given ref not found.
:return: orm.Partition: found partition.
"""
if not ref:
raise NotFoundError("No partition for empty ref")
try:
on = ObjectNumber.parse(ref)
ds_on = on.as_dataset
ds = self._db.dataset(
ds_on) # Could do it in on SQL query, but this is easier.
# The refresh is required because in some places the dataset is loaded without the partitions,
# and if that persist, we won't have partitions in it until it is refreshed.
self.database.session.refresh(ds)
p = ds.partition(ref)
except NotObjectNumberError:
q = (self.database.session.query(Partition).filter(
or_(Partition.name == str(ref),
Partition.vname == str(ref))).order_by(
Partition.vid.desc()))
p = q.first()
if not p:
raise NotFoundError("No partition for ref: '{}'".format(ref))
b = self.bundle(p.d_vid)
p = b.wrap_partition(p)
if localize:
p.localize()
return p
def partition(self, ref, localize=False):
""" Finds partition by ref and converts to bundle partition.
:param ref: A partition reference
:param localize: If True, copy a remote partition to local filesystem. Defaults to False
:raises: NotFoundError: if partition with given ref not found.
:return: orm.Partition: found partition.
"""
if not ref:
raise NotFoundError("No partition for empty ref")
try:
on = ObjectNumber.parse(ref)
ds_on = on.as_dataset
ds = self._db.dataset(ds_on) # Could do it in on SQL query, but this is easier.
# The refresh is required because in some places the dataset is loaded without the partitions,
# and if that persist, we won't have partitions in it until it is refreshed.
self.database.session.refresh(ds)
p = ds.partition(ref)
except NotObjectNumberError:
q = (self.database.session.query(Partition)
.filter(or_(Partition.name == str(ref), Partition.vname == str(ref)))
.order_by(Partition.vid.desc()))
p = q.first()
if not p:
raise NotFoundError("No partition for ref: '{}'".format(ref))
b = self.bundle(p.d_vid)
p = b.wrap_partition(p)
if localize:
p.localize()
return p
def __init__(self,dataset, **kwargs):
self.sequence_id = kwargs.get("sequence_id",None)
self.name = kwargs.get("name",None)
self.vname = kwargs.get("vname",None)
self.altname = kwargs.get("altname",None)
self.description = kwargs.get("description",None)
self.universe = kwargs.get("universe", None)
self.keywords = kwargs.get("keywords",None)
self.data = kwargs.get("data",None)
self.d_id = dataset.id_
self.d_vid = dataset.vid
don = ObjectNumber.parse(dataset.vid)
ton = TableNumber(don, self.sequence_id)
self.vid = str(ton)
self.id_ = str(ton.rev(None))
if self.name:
self.name = self.mangle_name(self.name)
self.init_on_load()
def put_manifest(self, m, f):
"""WARNING!
This method must be run after all of the bundles are already
cached, or at least the bundles used in this manifest
"""
from ambry.identity import ObjectNumber
d = m.dict
d['file'] = f.dict
d['text'] = str(m)
#d['files'] = f.dict['data'].get('files')
#del d['file']['data']
# Update the partitions to include bundle references,
# then add bundle information.
partitions = {
pvid: str(
ObjectNumber.parse(pvid).as_dataset) for pvid in f.dict.get(
'partitions',
[])}
d["partitions"] = partitions
d['tables'] = {tvid: {
k: v for k, v in (self.get_table(tvid).items() + [('installed_names', [])]) if k != 'columns'
} for tvid in f.dict.get('tables', [])
}
d['bundles'] = {vid: self.get_bundle(vid)
for vid in partitions.values()}
for vid, b in d['bundles'].items():
b['installed_partitions'] = [
pvid for pvid,
pbvid in partitions.items() if vid == pbvid]
# Generate entries for the tables, using the names that they are installed with. These tables aren't
# nessiarily installed; this maps the instllation names to vids if they
# are installed.
installed_table_names = {}
def inst_table_entry(b, p, t):
return dict(
t_vid=t['vid'],
t_name=t['name'],
p_vid=p['vid'],
p_vname=p['vname'],
b_vid=b['identity']['vid'],
b_vname=b['identity']['vname']
)
for vid, b in d['bundles'].items():
for pvid, bvid in d['partitions'].items():
b = d['bundles'][bvid]
p = b['partitions'][pvid]
for tvid in p['table_vids']:
t = b['tables'][tvid]
e = inst_table_entry(b, p, t)
d['installed_table_names'] = installed_table_names
# Collect the views and mviews
views = {}
for s in d['sections']:
if s['tag'] in ('view', 'mview'):
views[s['args']] = dict(
tag=s['tag'],
tc_names=s.get('content', {}).get('tc_names'),
html=s.get('content', {}).get('html'),
text=s.get('content', {}).get('text'),
)
d['views'] = views
return self.put(self.manifest_relpath(m.uid), d)
def before_insert(mapper, conn, target):
target.d_vid = str(ObjectNumber.parse(target.c_vid).as_dataset)
def add_column(self, name, update_existing=False, **kwargs):
"""
Add a column to the table, or update an existing one.
:param name: Name of the new or existing column.
:param update_existing: If True, alter existing column values. Defaults to False
:param kwargs: Other arguments for the the Column() constructor
:return: a Column object
"""
from ..identity import ColumnNumber
try:
c = self.column(name)
extant = True
if not update_existing:
return c
except NotFoundError:
sequence_id = len(self.columns) + 1
assert sequence_id
c = Column(t_vid=self.vid,
sequence_id=sequence_id,
vid=str(
ColumnNumber(ObjectNumber.parse(self.vid),
sequence_id)),
name=name,
datatype='str')
extant = False
# Update possibly existing data
c.data = dict((list(c.data.items()) if c.data else []) +
list(kwargs.get('data', {}).items()))
for key, value in list(kwargs.items()):
if key[0] != '_' and key not in [
't_vid', 'name', 'sequence_id', 'data'
]:
# Don't update the type if the user has specfied a custom type
if key == 'datatype' and not c.type_is_builtin():
continue
# Don't change a datatype if the value is set and the new value is unknown
if key == 'datatype' and value == 'unknown' and c.datatype:
continue
# Don't change a datatype if the value is set and the new value is unknown
if key == 'description' and not value:
continue
try:
setattr(c, key, value)
except AttributeError:
raise AttributeError(
"Column record has no attribute {}".format(key))
if key == 'is_primary_key' and isinstance(value,
str) and len(value) == 0:
value = False
setattr(c, key, value)
# If the id column has a description and the table does not, add it to
# the table.
if c.name == 'id' and c.is_primary_key and not self.description:
self.description = c.description
if not extant:
self.columns.append(c)
return c
def put_manifest(self, m, f):
"""WARNING!
This method must be run after all of the bundles are already
cached, or at least the bundles used in this manifest
"""
from ambry.identity import ObjectNumber
d = m.dict
d['file'] = f.dict
d['text'] = str(m)
#d['files'] = f.dict['data'].get('files')
#del d['file']['data']
# Update the partitions to include bundle references,
# then add bundle information.
partitions = {
pvid: str(
ObjectNumber.parse(pvid).as_dataset) for pvid in f.dict.get(
'partitions',
[])}
d["partitions"] = partitions
d['tables'] = {tvid: {
k: v for k, v in (self.get_table(tvid).items() + [('installed_names', [])]) if k != 'columns'
} for tvid in f.dict.get('tables', [])
}
d['bundles'] = {vid: self.get_bundle(vid)
for vid in partitions.values()}
for vid, b in d['bundles'].items():
b['installed_partitions'] = [
pvid for pvid,
pbvid in partitions.items() if vid == pbvid]
# Generate entries for the tables, using the names that they are installed with. These tables aren't
# nessiarily installed; this maps the instllation names to vids if they
# are installed.
installed_table_names = {}
def inst_table_entry(b, p, t):
return dict(
t_vid=t['vid'],
t_name=t['name'],
p_vid=p['vid'],
p_vname=p['vname'],
b_vid=b['identity']['vid'],
b_vname=b['identity']['vname']
)
for vid, b in d['bundles'].items():
for pvid, bvid in d['partitions'].items():
b = d['bundles'][bvid]
p = b['partitions'][pvid]
for tvid in p['table_vids']:
t = b['tables'][tvid]
e = inst_table_entry(b, p, t)
d['installed_table_names'] = installed_table_names
# Collect the views and mviews
views = {}
for s in d['sections']:
if s['tag'] in ('view', 'mview'):
views[s['args']] = dict(
tag=s['tag'],
tc_names=s.get('content', {}).get('tc_names'),
html=s.get('content', {}).get('html'),
text=s.get('content', {}).get('text'),
)
d['views'] = views
return self.put(self.manifest_relpath(m.uid), d)
def test_parse_other(self):
dn = DatasetNumber(100, 5, 'authoritative')
self.assertEqual('d01C005', str(dn))
self.assertEqual('G01C001Z005', str(GeneralNumber1('G', dn, 123)))
self.assertEqual('G01C001Z005', ObjectNumber.parse(str(GeneralNumber1('G', dn, 123)), 'other1'))