def partition_registration_list(self, conn, *, num_physical_shards):
"""Method to repartition registration_list for v47 upgrade."""
with conn.cursor() as cursor, utils.db_role_setter(
conn, role_name='dirbs_core_power_user'):
# Create parent partition
cursor.execute("""CREATE TABLE historic_registration_list_new (
LIKE historic_registration_list INCLUDING DEFAULTS
INCLUDING IDENTITY
INCLUDING CONSTRAINTS
INCLUDING STORAGE
INCLUDING COMMENTS
)
PARTITION BY RANGE (virt_imei_shard)
""")
part_utils._grant_perms_registration_list(
conn, part_name='historic_registration_list_new')
# Create child partitions
part_utils.create_imei_shard_partitions(
conn,
tbl_name='historic_registration_list_new',
num_physical_shards=num_physical_shards,
perms_func=part_utils._grant_perms_registration_list,
fillfactor=80)
# Insert data from original partition
cursor.execute("""INSERT INTO historic_registration_list_new
SELECT *
FROM historic_registration_list""")
# Add in indexes to each partition
idx_metadata = [
part_utils.IndexMetadatum(idx_cols=['imei_norm'],
is_unique=True,
partial_sql='WHERE end_date IS NULL')
]
part_utils.add_indices(conn,
tbl_name='historic_registration_list_new',
idx_metadata=idx_metadata)
# Drop old view + table, rename tables, indexes and constraints
cursor.execute('DROP VIEW registration_list')
cursor.execute('DROP TABLE historic_registration_list CASCADE')
part_utils.rename_table_and_indices(
conn,
old_tbl_name='historic_registration_list_new',
new_tbl_name='historic_registration_list',
idx_metadata=idx_metadata)
cursor.execute("""CREATE OR REPLACE VIEW registration_list AS
SELECT imei_norm, make, model, status, virt_imei_shard
FROM historic_registration_list
WHERE end_date IS NULL WITH CHECK OPTION"""
)
cursor.execute("""GRANT SELECT ON registration_list
TO dirbs_core_classify, dirbs_core_api, dirbs_core_import_registration_list"""
)
def _repartition_pairing_list(self, conn, *, num_physical_shards):
"""Repartition pairing list to implement change in structure."""
with conn.cursor() as cursor, utils.db_role_setter(
conn, role_name='dirbs_core_power_user'):
cursor.execute("""CREATE TABLE historic_pairing_list_new (
LIKE historic_pairing_list INCLUDING DEFAULTS
INCLUDING IDENTITY
INCLUDING CONSTRAINTS
INCLUDING STORAGE
INCLUDING COMMENTS
)
PARTITION BY RANGE (virt_imei_shard)
""")
cursor.execute(
"""ALTER TABLE historic_pairing_list_new ADD COLUMN msisdn TEXT NOT NULL"""
)
# grant permissions
part_utils._grant_perms_pairing_list(
conn, part_name='historic_pairing_list_new')
# create child partitions
part_utils.create_imei_shard_partitions(
conn,
tbl_name='historic_pairing_list_new',
num_physical_shards=num_physical_shards,
perms_func=part_utils._grant_perms_pairing_list,
fillfactor=80)
# copy data from original partition
cursor.execute("""INSERT INTO historic_pairing_list_new
SELECT p.imei_norm, p.imsi, p.start_date, p.end_date, p.virt_imei_shard, m.msisdn
FROM historic_pairing_list p
INNER JOIN monthly_network_triplets_country m ON p.imsi = m.imsi"""
)
# add indexes
idx_metadata = [
part_utils.IndexMetadatum(
idx_cols=['imei_norm', 'imsi', 'msisdn'],
is_unique=True,
partial_sql='WHERE end_date IS NULL')
]
part_utils.add_indices(conn,
tbl_name='historic_pairing_list_new',
idx_metadata=idx_metadata)
# drop old views, tables, indexes and constraints
cursor.execute('DROP VIEW pairing_list')
cursor.execute('DROP TABLE historic_pairing_list CASCADE')
part_utils.rename_table_and_indices(
conn,
old_tbl_name='historic_pairing_list_new',
new_tbl_name='historic_pairing_list',
idx_metadata=idx_metadata)
# create new view and grant permissions
cursor.execute("""CREATE VIEW pairing_list AS
SELECT imei_norm, imsi, msisdn, virt_imei_shard
FROM historic_pairing_list
WHERE end_date IS NULL WITH CHECK OPTION"""
)
cursor.execute("""GRANT SELECT ON pairing_list
TO dirbs_core_listgen, dirbs_core_report, dirbs_core_api, dirbs_core_import_pairing_list
""")
# drop and recreate staging data insert trigger
cursor.execute("""
DROP FUNCTION pairing_list_staging_data_insert_trigger_fn() CASCADE;
CREATE FUNCTION pairing_list_staging_data_insert_trigger_fn() RETURNS trigger
LANGUAGE plpgsql
AS $$
BEGIN
-- Clean/normalize data before inserting
NEW.imei_norm = normalize_imei(NULLIF(TRIM(NEW.imei), ''));
NEW.imsi = NULLIF(TRIM(new.imsi), '');
NEW.msisdn = NULLIF(TRIM(new.msisdn), '');
RETURN NEW;
END
$$;
""")
def upgrade(self, db_conn): # noqa: C901
"""Overrides AbstractMigrator upgrade method."""
logger = logging.getLogger('dirbs.db')
with db_conn.cursor() as cursor:
cursor.execute(
"""CREATE FUNCTION calc_virt_imei_shard(imei TEXT) RETURNS SMALLINT
LANGUAGE plpgsql IMMUTABLE PARALLEL SAFE
AS $$
BEGIN
RETURN SUBSTRING(COALESCE(imei, ''), 13, 2)::SMALLINT;
EXCEPTION WHEN OTHERS THEN
RETURN 0;
END;
$$""")
# By default, create 4 shards
num_initial_shards = 4
logger.info('Re-partitioning classification_state table...')
cursor.execute(
'ALTER TABLE classification_state ADD COLUMN virt_imei_shard SMALLINT'
)
cursor.execute(
'UPDATE classification_state SET virt_imei_shard = calc_virt_imei_shard(imei_norm)'
)
cursor.execute(
'ALTER TABLE classification_state ALTER COLUMN virt_imei_shard SET NOT NULL'
)
part_utils.repartition_classification_state(
db_conn, num_physical_shards=num_initial_shards)
logger.info('Re-partitioned classification_state table')
logger.info('Re-partitioning registration_list table...')
cursor.execute(
'ALTER TABLE historic_registration_list ADD COLUMN virt_imei_shard SMALLINT'
)
cursor.execute(
'UPDATE historic_registration_list SET virt_imei_shard = calc_virt_imei_shard(imei_norm)'
)
cursor.execute(
'ALTER TABLE historic_registration_list ALTER COLUMN virt_imei_shard SET NOT NULL'
)
self.partition_registration_list(
db_conn, num_physical_shards=num_initial_shards)
logger.info('Re-partitioned registration_list table')
logger.info('Re-partitioning pairing_list table...')
cursor.execute(
'ALTER TABLE historic_pairing_list ADD COLUMN virt_imei_shard SMALLINT'
)
cursor.execute(
'UPDATE historic_pairing_list SET virt_imei_shard = calc_virt_imei_shard(imei_norm)'
)
cursor.execute(
'ALTER TABLE historic_pairing_list ALTER COLUMN virt_imei_shard SET NOT NULL'
)
part_utils.repartition_pairing_list(
db_conn, num_physical_shards=num_initial_shards)
logger.info('Re-partitioned pairing_list table')
logger.info('Re-partitioning blacklist table...')
cursor.execute(
'ALTER TABLE blacklist ADD COLUMN virt_imei_shard SMALLINT')
cursor.execute(
'UPDATE blacklist SET virt_imei_shard = calc_virt_imei_shard(imei_norm)'
)
cursor.execute(
'ALTER TABLE blacklist ALTER COLUMN virt_imei_shard SET NOT NULL'
)
part_utils.repartition_blacklist(
db_conn, num_physical_shards=num_initial_shards)
logger.info('Re-partitioned blacklist table')
# Need to make sure owner of list tables is dirbs_core_listgen
logger.info('Re-partitioning notifications_lists table...')
# The original notifications_lists were not created with a single sequence for the IDs, so just do now
with utils.db_role_setter(db_conn, role_name='dirbs_core_listgen'):
cursor.execute(
"""CREATE UNLOGGED TABLE notifications_lists_new (
row_id BIGSERIAL NOT NULL,
operator_id TEXT NOT NULL,
imei_norm TEXT NOT NULL,
imsi TEXT NOT NULL,
msisdn TEXT NOT NULL,
block_date DATE NOT NULL,
reasons TEXT[] NOT NULL,
amnesty_granted BOOLEAN DEFAULT FALSE NOT NULL,
start_run_id BIGINT NOT NULL,
end_run_id BIGINT,
delta_reason TEXT NOT NULL CHECK (delta_reason IN ('new', 'resolved', 'blacklisted',
'no_longer_seen', 'changed')),
virt_imei_shard SMALLINT NOT NULL
) PARTITION BY LIST (operator_id)
""")
# Work out who the operators are
partitions = utils.child_table_names(db_conn,
'notifications_lists')
# Make sure that they are owned by dirbs_core_listgen (they can be owner by dirbs_core_power_user)
# due to bad previous migration scripts
with utils.db_role_setter(db_conn,
role_name='dirbs_core_power_user'):
for p in partitions:
cursor.execute(
sql.SQL('ALTER TABLE {0} OWNER TO dirbs_core_listgen').
format(sql.Identifier(p)))
operators = [
x.operator_id for x in utils.table_invariants_list(
db_conn, partitions, ['operator_id'])
]
# Create operator child partitions
for op_id in operators:
tbl_name = part_utils.per_mno_lists_partition(
operator_id=op_id,
suffix='_new',
list_type='notifications')
part_utils.create_per_mno_lists_partition(
db_conn,
operator_id=op_id,
parent_tbl_name='notifications_lists_new',
tbl_name=tbl_name,
num_physical_shards=1,
unlogged=True,
fillfactor=100)
cursor.execute(
"""INSERT INTO notifications_lists_new(operator_id, imei_norm, imsi, msisdn, block_date,
reasons, start_run_id, end_run_id, delta_reason,
virt_imei_shard)
SELECT operator_id, imei_norm, imsi, msisdn, block_date,
reasons, start_run_id, end_run_id, delta_reason, calc_virt_imei_shard(imei_norm)
FROM notifications_lists
""")
# Drop old table, rename tables, indexes and constraints
cursor.execute("""ALTER TABLE notifications_lists_new
RENAME CONSTRAINT notifications_lists_new_delta_reason_check
TO notifications_lists_delta_reason_check""")
cursor.execute('DROP TABLE notifications_lists CASCADE')
cursor.execute("""ALTER SEQUENCE notifications_lists_new_row_id_seq
RENAME TO notifications_lists_row_id_seq""")
part_utils.rename_table_and_indices(
db_conn,
old_tbl_name='notifications_lists_new',
new_tbl_name='notifications_lists')
part_utils.repartition_notifications_lists(
db_conn, num_physical_shards=num_initial_shards)
logger.info('Re-partitioned notifications_lists table')
logger.info('Re-partitioning exceptions_lists table...')
# The original exceptions_lists were not created with a single sequence for the IDs, so just do now
with utils.db_role_setter(db_conn, role_name='dirbs_core_listgen'):
cursor.execute("""CREATE UNLOGGED TABLE exceptions_lists_new (
row_id BIGSERIAL NOT NULL,
operator_id TEXT NOT NULL,
imei_norm TEXT NOT NULL,
imsi TEXT NOT NULL,
start_run_id BIGINT NOT NULL,
end_run_id BIGINT,
delta_reason TEXT NOT NULL CHECK (delta_reason IN ('added', 'removed')),
virt_imei_shard SMALLINT NOT NULL
) PARTITION BY LIST (operator_id)
""")
# Work out who the operators are
partitions = utils.child_table_names(db_conn, 'exceptions_lists')
# Make sure that they are owned by dirbs_core_listgen (they can be owner by dirbs_core_power_user)
# due to bad previous migration scripts
with utils.db_role_setter(db_conn,
role_name='dirbs_core_power_user'):
for p in partitions:
cursor.execute(
sql.SQL('ALTER TABLE {0} OWNER TO dirbs_core_listgen').
format(sql.Identifier(p)))
operators = [
x.operator_id for x in utils.table_invariants_list(
db_conn, partitions, ['operator_id'])
]
# Create operator child partitions
for op_id in operators:
tbl_name = part_utils.per_mno_lists_partition(
operator_id=op_id, suffix='_new', list_type='exceptions')
part_utils.create_per_mno_lists_partition(
db_conn,
operator_id=op_id,
parent_tbl_name='exceptions_lists_new',
tbl_name=tbl_name,
num_physical_shards=1,
unlogged=True,
fillfactor=100)
cursor.execute(
"""INSERT INTO exceptions_lists_new(operator_id, imei_norm, imsi, start_run_id,
end_run_id, delta_reason, virt_imei_shard)
SELECT operator_id, imei_norm, imsi, start_run_id, end_run_id, delta_reason,
calc_virt_imei_shard(imei_norm)
FROM exceptions_lists
""")
# Drop old table, rename tables, indexes and constraints
cursor.execute("""ALTER TABLE exceptions_lists_new
RENAME CONSTRAINT exceptions_lists_new_delta_reason_check
TO exceptions_lists_delta_reason_check""")
cursor.execute('DROP TABLE exceptions_lists CASCADE')
cursor.execute(
'ALTER SEQUENCE exceptions_lists_new_row_id_seq RENAME TO exceptions_lists_row_id_seq'
)
part_utils.rename_table_and_indices(
db_conn,
old_tbl_name='exceptions_lists_new',
new_tbl_name='exceptions_lists')
part_utils.repartition_exceptions_lists(
db_conn, num_physical_shards=num_initial_shards)
logger.info('Re-partitioned exceptions_lists table')
logger.info('Re-partitioning seen_imeis (network_imeis) table')
# First, just put everything in a temporary table so that we can call partutils
with utils.db_role_setter(db_conn,
role_name='dirbs_core_import_operator'):
cursor.execute("""CREATE UNLOGGED TABLE network_imeis (
first_seen DATE NOT NULL,
last_seen DATE NOT NULL,
seen_rat_bitmask INTEGER,
imei_norm TEXT NOT NULL,
virt_imei_shard SMALLINT NOT NULL
)
""")
#
# We disable index scans here as doing a merge append with index scans is much slower and involves
# a lot of seeks which kills performance on non-SSD drives. Better to use an append plan and sort
# the results by imei_norm
#
cursor.execute('SET enable_indexscan = false')
cursor.execute("""INSERT INTO network_imeis
SELECT MIN(first_seen),
MAX(last_seen),
bit_or(seen_rat_bitmask),
imei_norm,
calc_virt_imei_shard(imei_norm)
FROM seen_imeis
GROUP BY imei_norm
""")
cursor.execute('SET enable_indexscan = true')
part_utils.repartition_network_imeis(
db_conn, num_physical_shards=num_initial_shards)
cursor.execute('DROP TABLE seen_imeis CASCADE')
logger.info('Re-partitioned seen_imeis (network_imeis) table')
# First, just put all country-level triplets in a temporary table so that we can call partition_utils
with utils.db_role_setter(db_conn,
role_name='dirbs_core_import_operator'):
cursor.execute(
"""CREATE UNLOGGED TABLE monthly_network_triplets_country (
triplet_year SMALLINT NOT NULL,
triplet_month SMALLINT NOT NULL,
first_seen DATE NOT NULL,
last_seen DATE NOT NULL,
date_bitmask INTEGER NOT NULL,
triplet_hash UUID NOT NULL,
imei_norm TEXT,
imsi TEXT,
msisdn TEXT,
virt_imei_shard SMALLINT NOT NULL,
CHECK (last_seen >= first_seen),
CHECK (EXTRACT(month FROM last_seen) = triplet_month AND
EXTRACT(year FROM last_seen) = triplet_year),
CHECK (EXTRACT(month FROM first_seen) = triplet_month AND
EXTRACT(year FROM first_seen) = triplet_year)
) PARTITION BY RANGE (triplet_year, triplet_month)
""")
# Work out what partitions to create and create them
partitions = utils.child_table_names(db_conn, 'seen_triplets')
# Make sure that they are owned by dirbs_core_import_operator (they can be owner by dirbs_core_power_user)
# due to bad previous migration scripts
with utils.db_role_setter(db_conn,
role_name='dirbs_core_power_user'):
for p in partitions:
cursor.execute(
sql.SQL(
'ALTER TABLE {0} OWNER TO dirbs_core_import_operator'
).format(sql.Identifier(p)))
year_month_tuples = {
(x.triplet_year, x.triplet_month)
for x in utils.table_invariants_list(
db_conn, partitions, ['triplet_year', 'triplet_month'])
}
for year, month in year_month_tuples:
part_utils.create_monthly_network_triplets_country_partition(
db_conn, month=month, year=year, num_physical_shards=1)
with utils.db_role_setter(db_conn,
role_name='dirbs_core_import_operator'):
cursor.execute(
"""CREATE UNLOGGED TABLE monthly_network_triplets_per_mno (
LIKE monthly_network_triplets_country INCLUDING ALL,
operator_id TEXT NOT NULL
) PARTITION BY LIST (operator_id)
""")
# Work out what partitions to create and create them
op_year_month_tuples = {
(x.operator_id, x.triplet_year, x.triplet_month)
for x in utils.table_invariants_list(
db_conn, partitions,
['operator_id', 'triplet_year', 'triplet_month'])
}
# Create child partitions at per-MNO level
for op, year, month in op_year_month_tuples:
part_utils.create_monthly_network_triplets_per_mno_partition(
db_conn,
operator_id=op,
month=month,
year=year,
num_physical_shards=1)
# Create temporary monthly_network_triplets_per_mno table
for year, month in year_month_tuples:
logger.info(
'Generating temporary monthly_network_triplets_per_mno entries for {0:02d}/{1:d}...'
.format(month, year))
cursor.execute(
"""INSERT INTO monthly_network_triplets_per_mno
SELECT %(year)s,
%(month)s,
first_seen,
last_seen,
date_bitmask,
triplet_hash,
imei_norm,
imsi,
msisdn,
calc_virt_imei_shard(imei_norm),
operator_id
FROM seen_triplets
WHERE triplet_year = %(year)s
AND triplet_month = %(month)s
""", {
'year': year,
'month': month
})
logger.info(
'Generated temporary monthly_network_triplets_per_mno entries for {0:02d}/{1:d}'
.format(month, year))
# Create temporary monthly_network_triplets_country table. We need to do this monthly as we need
# to aggregate by triplets on a monthly basis
#
# We disable index scans here as doing a merge append with index scans is much slower and involves
# a lot of seeks which kills performance on non-SSD drives. Better to use an append plan and sort
# the results by imei_norm
#
cursor.execute('SET enable_indexscan = false')
for year, month in year_month_tuples:
logger.info(
'Generating temporary monthly_network_triplets_country entries for {0:02d}/{1:d}...'
.format(month, year))
cursor.execute(
"""INSERT INTO monthly_network_triplets_country
SELECT %(year)s,
%(month)s,
MIN(first_seen),
MAX(last_seen),
bit_or(date_bitmask),
triplet_hash,
FIRST(imei_norm),
FIRST(imsi),
FIRST(msisdn),
calc_virt_imei_shard(FIRST(imei_norm))
FROM seen_triplets
WHERE triplet_year = %(year)s
AND triplet_month = %(month)s
GROUP BY triplet_hash
""", {
'year': year,
'month': month
})
logger.info(
'Generated temporary monthly_network_triplets_country entries for {0:02d}/{1:d}'
.format(month, year))
cursor.execute('SET enable_indexscan = true')
logger.info(
'Re-partitioning temporary monthly_network_triplets tables...')
# Previously, the operator_data view was owned by dirbs_core_power_user but is now owned by the
# dirbs_core_import_operator since it must be re-created
with utils.db_role_setter(db_conn,
role_name='dirbs_core_power_user'):
cursor.execute(
'ALTER VIEW operator_data OWNER TO dirbs_core_import_operator'
)
part_utils.repartition_monthly_network_triplets(
db_conn, num_physical_shards=num_initial_shards)
cursor.execute('DROP TABLE seen_triplets CASCADE')
logger.info(
'Re-partitioned temporary monthly_network_triplets tables')
# Replace list generation function to include virt_imei_shard
cursor.execute("""
DROP FUNCTION gen_blacklist(run_id BIGINT);
DROP FUNCTION gen_notifications_list(op_id TEXT, run_id BIGINT);
DROP FUNCTION gen_exceptions_list(op_id TEXT, run_id BIGINT);
--
-- Create function to generate a full blacklist for a given run_id. A value of -1 means get the latest
-- list.
--
CREATE FUNCTION gen_blacklist(run_id BIGINT = -1)
RETURNS TABLE (
imei_norm TEXT,
virt_imei_shard SMALLINT,
block_date DATE,
reasons TEXT[]
)
LANGUAGE plpgsql STRICT STABLE PARALLEL SAFE
AS $$
DECLARE
query_run_id BIGINT;
BEGIN
--
-- If we don't specify a run_id, just set to the maximum run_id which will always return all rows
-- where end_run_id is NULL
--
IF run_id = -1 THEN
run_id := max_bigint();
END IF;
RETURN QUERY SELECT bl.imei_norm,
bl.virt_imei_shard,
bl.block_date,
bl.reasons
FROM blacklist bl
WHERE bl.delta_reason != 'unblocked'
AND run_id >= bl.start_run_id
AND (run_id < bl.end_run_id OR bl.end_run_id IS NULL);
END
$$;
--
-- Create function to generate a full notifications_list for a given run_id and operator ID. A value
-- of -1 means get the latest list.
--
CREATE FUNCTION gen_notifications_list(op_id TEXT, run_id BIGINT = -1)
RETURNS TABLE (
imei_norm TEXT,
virt_imei_shard SMALLINT,
imsi TEXT,
msisdn TEXT,
block_date DATE,
reasons TEXT[],
amnesty_granted BOOLEAN
)
LANGUAGE plpgsql STRICT STABLE PARALLEL SAFE
AS $$
BEGIN
--
-- If we don't specify a run_id, just set to the maximum run_id which will always return all rows
-- where end_run_id is NULL
--
IF run_id = -1 THEN
run_id := max_bigint();
END IF;
RETURN QUERY SELECT nl.imei_norm,
nl.virt_imei_shard,
nl.imsi,
nl.msisdn,
nl.block_date,
nl.reasons,
nl.amnesty_granted
FROM notifications_lists nl
WHERE nl.operator_id = op_id
AND nl.delta_reason NOT IN ('resolved', 'blacklisted')
AND run_id >= nl.start_run_id
AND (run_id < nl.end_run_id OR nl.end_run_id IS NULL);
END
$$;
--
-- Create function to generate a full exceptions_list for a given run_id and operator ID. A value
-- of -1 means get the latest list.
--
CREATE FUNCTION gen_exceptions_list(op_id TEXT, run_id BIGINT = -1)
RETURNS TABLE (
imei_norm TEXT,
virt_imei_shard SMALLINT,
imsi TEXT
)
LANGUAGE plpgsql STRICT STABLE PARALLEL SAFE
AS $$
BEGIN
--
-- If we don't specify a run_id, just set to the maximum run_id which will always return all
-- rows where end_run_id is NULL
--
IF run_id = -1 THEN
run_id := max_bigint();
END IF;
RETURN QUERY SELECT el.imei_norm,
el.virt_imei_shard,
el.imsi
FROM exceptions_lists el
WHERE el.operator_id = op_id
AND el.delta_reason != 'removed'
AND run_id >= el.start_run_id
AND (run_id < el.end_run_id OR el.end_run_id IS NULL);
END
$$;
""") # noqa: Q440, Q441
# Update schema metadata table
cursor.execute(
"""ALTER TABLE schema_metadata ADD COLUMN phys_shards SMALLINT NOT NULL
DEFAULT %s CHECK (phys_shards > 0 AND phys_shards <= 100)""",
[num_initial_shards])
cursor.execute(
'ALTER TABLE schema_metadata ALTER COLUMN phys_shards DROP DEFAULT'
)
# Drop obsolete columns
cursor.execute(
'ALTER TABLE schema_metadata DROP COLUMN potential_whitespace_imsis_msisdns'
)
cursor.execute(
'ALTER TABLE report_monthly_stats DROP COLUMN num_whitespace_imsi_records'
)
cursor.execute(
'ALTER TABLE report_monthly_stats DROP COLUMN num_whitespace_msisdn_records'
)
def _repartition_exceptions_lists(self, conn, *, num_physical_shards):
"""Repartition the exceptions lists to support msisdn."""
with conn.cursor() as cursor, utils.db_role_setter(
conn, role_name='dirbs_core_listgen'):
cursor.execute("""CREATE TABLE exceptions_lists_new (
LIKE exceptions_lists INCLUDING DEFAULTS
INCLUDING IDENTITY
INCLUDING CONSTRAINTS
INCLUDING STORAGE
INCLUDING COMMENTS
)
PARTITION BY LIST (operator_id);
ALTER TABLE exceptions_lists_new ADD COLUMN msisdn TEXT NOT NULL;
""")
part_utils._grant_perms_list(
conn,
part_name='exceptions_lists_new') # grant relevant permissions
imei_shard_names = utils.child_table_names(
conn, 'exceptions_lists') # determine the child table names
operators = [
o.operator_id for o in utils.table_invariants_list(
conn, imei_shard_names, ['operator_id'])
] # workout who the operators are
# create child partitions for new list (operator at top level, then IMEI sharded)
for op_id in operators:
tbl_name = part_utils.per_mno_lists_partition(
operator_id=op_id, suffix='_new', list_type='exceptions')
part_utils.create_per_mno_lists_partition(
conn,
parent_tbl_name='exceptions_lists_new',
tbl_name=tbl_name,
operator_id=op_id,
num_physical_shards=num_physical_shards)
# insert data into the new parent partition
cursor.execute("""INSERT INTO exceptions_lists_new
SELECT e.row_id, e.operator_id, e.imei_norm, e.imsi, e.start_run_id, e.end_run_id,
e.delta_reason, e.virt_imei_shard, p.msisdn
FROM exceptions_lists e
INNER JOIN historic_pairing_list p ON e.imsi = p.imsi"""
)
# add indexes in each partitions
part_utils.add_indices(
conn,
tbl_name='exceptions_lists_new',
idx_metadata=part_utils.exceptions_lists_indices())
# drop old table, after assigning sequence to new table
cursor.execute(
'ALTER SEQUENCE exceptions_lists_row_id_seq OWNED BY exceptions_lists_new.row_id'
)
cursor.execute('DROP TABLE exceptions_lists CASCADE')
# rename table, indexes and constraints
part_utils.rename_table_and_indices(
conn,
old_tbl_name='exceptions_lists_new',
new_tbl_name='exceptions_lists',
idx_metadata=part_utils.exceptions_lists_indices())
# recreating gen_exceptionlist function
with utils.db_role_setter(conn, role_name='dirbs_core_power_user'):
cursor.execute("""
DROP FUNCTION gen_exceptions_list(op_id TEXT, run_id BIGINT);
--
-- Recreate function to generate a full exceptions_list for a given
-- run_id and operator.
-- A value of -1 means get the latest list.
--
CREATE FUNCTION gen_exceptions_list(op_id TEXT, run_id BIGINT = -1)
RETURNS TABLE (
imei_norm TEXT,
virt_imei_shard SMALLINT,
imsi TEXT,
msisdn TEXT
)
LANGUAGE plpgsql STRICT STABLE PARALLEL SAFE
AS $$
BEGIN
--
-- If we don't specify a run_id, just set to the maximum run_id which will always
-- return all rows where end_run_id is NULL
--
IF run_id = -1 THEN
run_id := max_bigint();
END IF;
RETURN QUERY SELECT el.imei_norm,
el.virt_imei_shard,
el.imsi,
el.msisdn
FROM exceptions_lists el
WHERE el.operator_id = op_id
AND el.delta_reason != 'removed'
AND run_id >= el.start_run_id
AND (run_id < el.end_run_id OR el.end_run_id IS NULL);
END
$$;
DROP FUNCTION gen_delta_exceptions_list(op_id TEXT, base_run_id BIGINT, run_id BIGINT);
--
-- Create function to generate a per-MNO delta exceptions list for a run_id, operator
-- id and optional base_run_id.
--
-- If not base_run_id is supplied, this function will use the maximum run_id found in
-- the DB that it less than than the supplied run_id
--
CREATE FUNCTION gen_delta_exceptions_list(op_id TEXT,
base_run_id BIGINT,
run_id BIGINT = -1)
RETURNS TABLE (
imei_norm TEXT,
imsi TEXT,
msisdn TEXT,
delta_reason TEXT
)
LANGUAGE plpgsql STRICT STABLE PARALLEL SAFE
AS $$
BEGIN
--
-- If we don't specify a run_id, just set to the maximum run_id
--
IF run_id = -1 THEN
run_id := max_bigint();
END IF;
IF run_id < base_run_id THEN
RAISE EXCEPTION 'Parameter base_run_id % greater than run_id %',
base_run_id, run_id;
END IF;
RETURN QUERY SELECT *
FROM (SELECT el.imei_norm,
el.imsi,
el.msisdn,
overall_delta_reason(el.delta_reason
ORDER BY start_run_id DESC) AS delta_reason
FROM exceptions_lists el
WHERE operator_id = op_id
AND start_run_id > base_run_id
AND start_run_id <= run_id
GROUP BY el.imei_norm, el.imsi, el.msisdn) x
WHERE x.delta_reason IS NOT NULL;
END
$$;
""") # noqa: Q440, Q441