def test_partition_2(self):
bundle = Bundle()
bundle.clean()
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
table = self.bundle.schema.tables[0]
p = (('time', 'time2'), ('space', 'space3'), ('table', table.name),
('grain', 'grain4'))
p += p
pids = {}
for i in range(4):
for j in range(4):
pid = self.bundle.identity.as_partition(**dict(p[i:i + j + 1]))
pids[pid.fqname] = pid
for pid in pids.values():
part = bundle.partitions.new_db_partition(**pid.dict)
part.create()
parts = bundle.partitions._find_orm(
PartitionNameQuery(vid=pid.vid)).all()
self.assertIn(pid.sname, [p.name for p in parts])
def copy_or_build_bundle(self):
"""Set up a clean bundle build, either by re-building the bundle, or
by copying it from a saved bundle directory """
# For most cases, re-set the bundle by copying from a saved version. If
# the bundle doesn't exist and the saved version doesn't exist,
# build a new one.
bundle, marker, build_dir, save_dir = self.bundle_dirs()
idnt = bundle.identity
if str(idnt.name.version) != "0.0.1":
# Rebuild the bundle if the test_library.py:test_versions
# script didn't reset the bundle at the end
from ambry.util import rm_rf
rm_rf(build_dir)
rm_rf(save_dir)
idnt = Identity.from_dict(dict(bundle.metadata.identity))
bundle.metadata.identity = idnt.ident_dict
bundle.metadata.names = idnt.names_dict
bundle.metadata.write_to_dir()
if not os.path.exists(marker):
global_logger.info("Build dir marker ({}) is missing".format(marker))
# There is a good reason to create a seperate instance,
# but don't remember what it is ...
bundle.clean()
bundle = Bundle()
if not os.path.exists(save_dir):
global_logger.info("Save dir is missing; re-build bundle. ")
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
if str(bundle.identity.name.version) != '0.0.1':
raise Exception("Can only save bundle if version is 0.0.1. This one is version: {} ".format(
bundle.identity.name.version))
bundle.pre_build()
bundle.build()
bundle.post_build()
bundle.close()
with open(marker, 'w') as f:
f.write(str(time.time()))
# Copy the newly built bundle to the save directory
os.system("rm -rf {1}; rsync -arv {0} {1} > /dev/null ".format(build_dir, save_dir))
# Always copy, just to be safe.
# global_logger.info( "Copying bundle from {}".format(save_dir))
os.system("rm -rf {0}; rsync -arv {1} {0} > /dev/null ".format(build_dir, save_dir))
def x_test_rewrite(self):
from testbundle.bundle import Bundle
import json
from ambry.run import get_runconfig
# Prepare to rewrite the bundle.yaml file.
bundle = Bundle()
orig = os.path.join(bundle.bundle_dir,'bundle.yaml')
save = os.path.join(bundle.bundle_dir,'bundle.yaml.save')
try:
os.rename(orig,save)
print 'Write to ', orig
with open(orig,'w') as f:
f.write(json.dumps(
{
"identity":{
"dataset": "dataset1",
"id": "dfoo",
"revision": 100,
"source": "source1",
"subset": "subset1",
"variation": "variation1",
"version": "1.0.1",
"vid": "dfob001",
},
"about": {
"author": "*****@*****.**"
}
}
))
get_runconfig.clear() # clear config cache.
bundle = Bundle()
bundle.clean()
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare() # Does the rewrite, adding the 'names'
# Need to clear and reload one more time for the 'names' to appear
get_runconfig.clear() # clear config cache.
bundle = Bundle()
bundle.exit_on_fatal = False
self.assertEquals('dataset1', bundle.config.identity.dataset)
self.assertEquals('dfoo', bundle.config.identity.id)
self.assertEquals(100, bundle.config.identity.revision)
self.assertEquals("source1-dataset1-subset1-variation1-1.0.100~dfoo01C", bundle.config.names.fqname)
self.assertEquals("*****@*****.**", bundle.config.about.author)
finally:
os.rename(save, orig)
self.delete_bundle()
def test_build_bundle_hdf(self):
bundle = Bundle()
bundle.clean()
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
bundle.pre_build()
bundle.build_hdf()
bundle.post_build()
def test_simple_build(self):
import shutil
bundle = Bundle()
shutil.copyfile(bundle.filesystem.path('meta', 'schema-edit-me.csv'),
bundle.filesystem.path('meta', 'schema.csv'))
bundle.clean()
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
bundle.pre_build()
bundle.build()
bundle.post_build()
def test_simple_build(self):
import shutil
bundle = Bundle()
shutil.copyfile(
bundle.filesystem.path('meta', 'schema-edit-me.csv'),
bundle.filesystem.path('meta', 'schema.csv'))
bundle.clean()
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
bundle.pre_build()
bundle.build()
bundle.post_build()
def test_build_bundle(self):
import shutil
bundle = Bundle()
shutil.copyfile(
bundle.filesystem.path('meta','schema-edit-me.csv'),
bundle.filesystem.path('meta','schema.csv'))
#try:
bundle.database.enable_delete = True
bundle.clean()
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
bundle.pre_build()
bundle.build_db_inserter_codes()
bundle.post_build()
bundle.close()
# The second run will use the changes to the schema made in the
# first run, due to the types errors in the 'coding' table.
bundle.clean()
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
bundle.pre_build()
bundle.build_db_inserter_codes()
bundle.post_build()
bundle.close()
try:
pass
finally:
# Need to clean up to ensure that we're back to a good state.
# This runs the normal build, which will be used by the other
# tests.
shutil.copyfile(
bundle.filesystem.path('meta','schema-edit-me.csv'),
bundle.filesystem.path('meta','schema.csv'))
bundle.clean()
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
bundle.pre_build()
bundle.build()
bundle.post_build()
def test_build_bundle(self):
import shutil
bundle = Bundle()
shutil.copyfile(bundle.filesystem.path('meta', 'schema-edit-me.csv'),
bundle.filesystem.path('meta', 'schema.csv'))
#try:
bundle.database.enable_delete = True
bundle.clean()
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
bundle.pre_build()
bundle.build_db_inserter_codes()
bundle.post_build()
bundle.close()
# The second run will use the changes to the schema made in the
# first run, due to the types errors in the 'coding' table.
bundle.clean()
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
bundle.pre_build()
bundle.build_db_inserter_codes()
bundle.post_build()
bundle.close()
try:
pass
finally:
# Need to clean up to ensure that we're back to a good state.
# This runs the normal build, which will be used by the other
# tests.
shutil.copyfile(
bundle.filesystem.path('meta', 'schema-edit-me.csv'),
bundle.filesystem.path('meta', 'schema.csv'))
bundle.clean()
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
bundle.pre_build()
bundle.build()
bundle.post_build()
def test_partition_2(self):
bundle = Bundle()
bundle.clean()
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
table = self.bundle.schema.tables[0]
p = (('time', 'time2'), ('space', 'space3'), ('table', table.name), ('grain', 'grain4'))
p += p
pids = {}
for i in range(4):
for j in range(4):
pid = self.bundle.identity.as_partition(**dict(p[i:i + j + 1]))
pids[pid.fqname] = pid
for pid in pids.values():
part = bundle.partitions.new_db_partition(**pid.dict)
part.create()
parts = bundle.partitions._find_orm(PartitionNameQuery(vid=pid.vid)).all()
self.assertIn(pid.sname, [p.name for p in parts])
def test_bundle_build(self):
from ambry.dbexceptions import ConflictError
bundle = Bundle()
# Need to clear the library, or the Bundle's pre_prepare
# will cancel the build if this version is already installed
bundle.library.purge()
bundle.exit_on_fatal = False
bundle.clean()
bundle.database.create()
bundle.prepare()
bp = bundle.partitions
with bundle.session:
bp._new_orm_partition(PartialPartitionName(table = 'tone', time = 't1', space='s1'))
bp._new_orm_partition(PartialPartitionName(table = 'tone', time = 't1', space='s2'))
bp._new_orm_partition(PartialPartitionName(table = 'tone', time = 't1', space=None))
bp._new_orm_partition(PartialPartitionName(table = 'tone', time = 't2', space='s1'))
bp._new_orm_partition(PartialPartitionName(table = 'tone', time = 't2', space='s2'))
bp._new_orm_partition(PartialPartitionName(table = 'tone', time = 't2', space=None))
with self.assertRaises(ConflictError):
with bundle.session:
bp._new_orm_partition(PartialPartitionName(table = 'tone',time = 't1', space='s1'))
pnq = PartitionNameQuery(table = 'tone', time=NameQuery.ANY, space='s1')
names = [p.vname
for p in bp._find_orm(pnq).all()]
self.assertEqual({u'source-dataset-subset-variation-tone-t1-s1-0.0.1',
u'source-dataset-subset-variation-tone-t2-s1-0.0.1'},
set(names))
names = [p.vname
for p in bp._find_orm(PartitionNameQuery(space=NameQuery.ANY)).all()]
self.assertEqual(6, len(names))
names = [p.vname
for p in bp._find_orm(PartitionNameQuery(table = 'tone',time='t1',space=NameQuery.ANY)).all()]
self.assertEqual({'source-dataset-subset-variation-tone-t1-s2-0.0.1',
'source-dataset-subset-variation-tone-t1-0.0.1',
'source-dataset-subset-variation-tone-t1-s1-0.0.1'},
set(names))
names = [p.vname
for p in bp._find_orm(PartitionNameQuery(table = 'tone',time='t1',space=NameQuery.NONE)).all()]
self.assertEqual({'source-dataset-subset-variation-tone-t1-0.0.1'},
set(names))
# Start over, use a higher level function to create the partitions
bundle.close() # Or you'll get an OperationalError
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.clean()
bundle.database.create()
bundle.prepare()
bp = bundle.partitions
bp._new_partition(PartialPartitionName(table = 'tone',time = 't1', space='s1'))
self.assertEquals(1, len(bp.all))
bp._new_partition(PartialPartitionName(table = 'tone',time = 't1', space='s2'))
self.assertEquals(2, len(bp.all))
bp._new_partition(PartialPartitionName(table = 'tone',time = 't1', space=None))
bp._new_partition(PartialPartitionName(table = 'tone',time = 't2', space='s1'))
bp._new_partition(PartialPartitionName(table = 'tone',time = 't2', space='s2'))
bp._new_partition(PartialPartitionName(table = 'tone',time = 't2', space=None))
self.assertEquals(6, len(bp.all))
names = [p.vname
for p in bp._find_orm(PartitionNameQuery(time='t1', space=NameQuery.ANY)).all()]
self.assertEqual({'source-dataset-subset-variation-tone-t1-s2-0.0.1',
'source-dataset-subset-variation-tone-t1-0.0.1',
'source-dataset-subset-variation-tone-t1-s1-0.0.1'},
set(names))
# Start over, use a higher level function to create the partitions
bundle.close()
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.clean()
bundle.database.create()
bundle.prepare()
bp = bundle.partitions
p = bp.new_db_partition(table = 'tone',time = 't1', space='s1')
self.assertEquals('source-dataset-subset-variation-tone-t1-s1-0.0.1~piEGPXmDC8001001', p.identity.fqname)
p = bp.find_or_new(table = 'tone',time = 't1', space='s2')
self.assertEquals('source-dataset-subset-variation-tone-t1-s2-0.0.1~piEGPXmDC8002001', p.identity.fqname)
# Duplicate
p = bp.find_or_new(table = 'tone',time = 't1', space='s2')
self.assertEquals('source-dataset-subset-variation-tone-t1-s2-0.0.1~piEGPXmDC8002001', p.identity.fqname)
# Ok! Build!
bundle.close()
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.clean()
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
bundle.pre_build()
bundle.build_db_inserter_codes()
bundle.post_build()
self.assertEquals('diEGPXmDC8001', bundle.identity.vid)
self.assertEquals('source-dataset-subset-variation', bundle.identity.sname)
self.assertEquals('source-dataset-subset-variation-0.0.1', bundle.identity.vname)
self.assertEquals('source-dataset-subset-variation-0.0.1~diEGPXmDC8001', bundle.identity.fqname)
def test_bundle_build(self):
from ambry.dbexceptions import ConflictError
bundle = Bundle()
# Need to clear the library, or the Bundle's pre_prepare
# will cancel the build if this version is already installed
bundle.library.purge()
bundle.exit_on_fatal = False
bundle.clean()
bundle.database.create()
bp = bundle.partitions
with bundle.session:
bp._new_orm_partition(PartialPartitionName(time = 't1', space='s1'))
bp._new_orm_partition(PartialPartitionName(time = 't1', space='s2'))
bp._new_orm_partition(PartialPartitionName(time = 't1', space=None))
bp._new_orm_partition(PartialPartitionName(time = 't2', space='s1'))
bp._new_orm_partition(PartialPartitionName(time = 't2', space='s2'))
bp._new_orm_partition(PartialPartitionName(time = 't2', space=None))
with self.assertRaises(ConflictError):
with bundle.session:
bp._new_orm_partition(PartialPartitionName(time = 't1', space='s1'))
pnq = PartitionNameQuery(time=NameQuery.ANY, space='s1')
names = [p.vname
for p in bp._find_orm(pnq).all()]
self.assertEqual({u'source-dataset-subset-variation-t2-s1-0.0.1',
u'source-dataset-subset-variation-t1-s1-0.0.1'},
set(names))
names = [p.vname
for p in bp._find_orm(PartitionNameQuery(space=NameQuery.ANY)).all()]
self.assertEqual(6,len(names))
names = [p.vname
for p in bp._find_orm(PartitionNameQuery(time='t1',space=NameQuery.ANY)).all()]
self.assertEqual({'source-dataset-subset-variation-t1-s2-0.0.1',
'source-dataset-subset-variation-t1-0.0.1',
'source-dataset-subset-variation-t1-s1-0.0.1'},
set(names))
names = [p.vname
for p in bp._find_orm(PartitionNameQuery(time='t1',space=NameQuery.NONE)).all()]
self.assertEqual({'source-dataset-subset-variation-t1-0.0.1'},
set(names))
# Start over, use a higher level function to create the partitions
bundle.close() # Or you'll get an OperationalError
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.clean()
bundle.database.create()
bp = bundle.partitions
bp._new_partition(PartialPartitionName(time = 't1', space='s1'))
self.assertEquals(1, len(bp.all))
bp._new_partition(PartialPartitionName(time = 't1', space='s2'))
self.assertEquals(2, len(bp.all))
bp._new_partition(PartialPartitionName(time = 't1', space=None))
bp._new_partition(PartialPartitionName(time = 't2', space='s1'))
bp._new_partition(PartialPartitionName(time = 't2', space='s2'))
bp._new_partition(PartialPartitionName(time = 't2', space=None))
self.assertEquals(6, len(bp.all))
names = [p.vname
for p in bp._find_orm(PartitionNameQuery(time='t1',space=NameQuery.ANY)).all()]
self.assertEqual({'source-dataset-subset-variation-t1-s2-0.0.1',
'source-dataset-subset-variation-t1-0.0.1',
'source-dataset-subset-variation-t1-s1-0.0.1'},
set(names))
# Start over, use a higher level function to create the partitions
bundle.close()
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.clean()
bundle.database.create()
bp = bundle.partitions
p = bp.new_db_partition(time = 't1', space='s1')
self.assertEquals('source-dataset-subset-variation-t1-s1-0.0.1~piEGPXmDC8001001', p.identity.fqname)
p = bp.find_or_new(time = 't1', space='s2')
self.assertEquals('source-dataset-subset-variation-t1-s2-0.0.1~piEGPXmDC8002001', p.identity.fqname)
# Duplicate
p = bp.find_or_new(time = 't1', space='s2')
self.assertEquals('source-dataset-subset-variation-t1-s2-0.0.1~piEGPXmDC8002001', p.identity.fqname)
p = bp.find_or_new_geo(time = 't2', space='s1')
# Which it is depends on whether GDAL is installed.
self.assertIn(p.identity.fqname,[
'source-dataset-subset-variation-t2-s1-geo-0.0.1~piEGPXmDC8003001',
'source-dataset-subset-variation-t2-s1-0.0.1~piEGPXmDC8003001' ]
)
# Ok! Build!
bundle.close()
bundle = Bundle()
bundle.exit_on_fatal = False
bundle.clean()
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
bundle.pre_build()
bundle.build_db_inserter_codes()
bundle.post_build()
self.assertEquals('diEGPXmDC8001',bundle.identity.vid)
self.assertEquals('source-dataset-subset-variation',bundle.identity.sname)
self.assertEquals('source-dataset-subset-variation-0.0.1',bundle.identity.vname)
self.assertEquals('source-dataset-subset-variation-0.0.1~diEGPXmDC8001',bundle.identity.fqname)
def copy_or_build_bundle(self):
"""Set up a clean bundle build, either by re-building the bundle, or
by copying it from a saved bundle directory """
# For most cases, re-set the bundle by copying from a saved version. If
# the bundle doesn't exist and the saved version doesn't exist,
# build a new one.
bundle, marker, build_dir, save_dir = self.bundle_dirs()
idnt = bundle.identity
if str(idnt.name.version) != "0.0.1":
# Rebuild the bundle if the test_library.py:test_versions
# script didn't reset the bundle at the end
from ambry.util import rm_rf
rm_rf(build_dir)
rm_rf(save_dir)
idnt = Identity.from_dict(dict(bundle.metadata.identity))
bundle.metadata.identity = idnt.ident_dict
bundle.metadata.names = idnt.names_dict
bundle.metadata.write_to_dir()
if not os.path.exists(marker):
global_logger.info(
"Build dir marker ({}) is missing".format(marker))
# There is a good reason to create a seperate instance,
# but don't remember what it is ...
bundle.clean()
bundle = Bundle()
if not os.path.exists(save_dir):
global_logger.info("Save dir is missing; re-build bundle. ")
bundle.pre_prepare()
bundle.prepare()
bundle.post_prepare()
if str(bundle.identity.name.version) != '0.0.1':
raise Exception(
"Can only save bundle if version is 0.0.1. This one is version: {} "
.format(bundle.identity.name.version))
bundle.pre_build()
bundle.build()
bundle.post_build()
bundle.close()
with open(marker, 'w') as f:
f.write(str(time.time()))
# Copy the newly built bundle to the save directory
os.system("rm -rf {1}; rsync -arv {0} {1} > /dev/null ".format(
build_dir, save_dir))
# Always copy, just to be safe.
#global_logger.info( "Copying bundle from {}".format(save_dir))
os.system("rm -rf {0}; rsync -arv {1} {0} > /dev/null ".format(
build_dir, save_dir))
