try:
client.delete_database(db)
except Exception as E:
pass
client.create_database(db,
account,
label="Schemaless to Schema",
description="Two branch database one with schema",
include_schema=False)
# Create a schemaless branch
client.branch('schemaless')
# Switch back to main
client.checkout('main')
# Create the schema graph
client.create_graph("schema", "main", "Adding schema")
# Add the schema
WQ().woql_and(WQ().doctype("scm:BankAccount").label(
"Bank Account").description("A bank account").property(
"scm:owner", "xsd:string").label("owner").cardinality(1).property(
"scm:balance", "xsd:nonNegativeInteger").label("balance")).execute(
client, "Adding bank account object to schema")
# Add some data which meets the schema
WQ().woql_and(WQ().insert("doc:mike", "scm:BankAccount").property(
"scm:owner", "mike").property("scm:balance",
123)).execute(client, "Add mike")
except:
pass
times = []
directory = 'wordnet_chunks'
for f in os.listdir(directory):
filename = f'{directory}/{f}'
ttl_file = open(filename)
contents = ttl_file.read()
ttl_file.close()
before = time.time()
print(f"Loading WordNet in chunks ({f})")
client.insert_triples("instance", "main", contents,
f"Adding WordNet in chunks ({f})")
after = time.time()
total = (after - before)
times.append(total)
print(f"Update took {total} seconds")
print(f"Squashing main")
before = time.time()
client.checkout("main")
result = client.squash('Squash commit of properties and types')
after = time.time()
total = (after - before)
print(f"Squash took {total} seconds")
commit = result['api:commit']
print(f"Branch reset to {commit}")
client.reset(commit)
try:
client.delete_database(db)
except Exception as E:
print(E)
client.create_database(db,
account,
label=db_label,
description=db_comment,
include_schema=False)
for branch in branches:
# Load the perties branch
client.branch(branch, empty=True)
client.checkout(branch)
print(f"Importing from {branch}")
for f in os.listdir(branch):
filename = f'{branch}/{f}'
ttl_file = open(filename)
contents = ttl_file.read()
ttl_file.close()
# start the chunk work
before = time.time()
client.insert_triples("instance", "main", contents,
f"Adding {branch} in 200k chunk ({f})")
after = time.time()
total = (after - before)
client.optimize(f'{account}/{db}/local/_commits')
try:
client.delete_database(db)
except Exception as E:
print(E)
client.create_database(db,
account,
label=db_label,
description=db_comment,
include_schema=False)
client.branch('properties')
client.checkout('properties')
times = []
directory = 'properties_100k'
for f in os.listdir(directory):
optimizer(client)
filename = f'{directory}/{f}'
ttl_file = open(filename)
contents = ttl_file.read()
ttl_file.close()
before = time.time()
client.insert_triples("instance", "main", contents,
f"Adding properties in 100k chunk ({f})")
after = time.time()
total = (after - before)
times.append(total)
print(f"Update took {total} seconds")
#!/usr/bin/python3
import os
import time
from terminusdb_client import WOQLClient
from terminusdb_client import WOQLQuery as WQ
server_url = "https://127.0.0.1:6363"
db = "DBPedia_squash"
user = "******"
account = "admin"
key = "root"
client = WOQLClient(server_url)
client.connect(user=user, account=account, key=key, db=db)
# client.branch("squash", empty=True)
client.checkout("squash")
query = WQ().select().woql_and(
WQ().using(f"{account}/{db}").triple("v:X","v:Y","v:Z"),
WQ().add_triple("v:X","v:Y","v:Z"))
client.query(query, "Squash commit")
account,
label=db_label,
description=db_comment,
include_schema=False)
print("##########################################################")
print("# #")
print(
f"# Starting run with optimization == {optimize}{space} #")
print("# #")
print("##########################################################")
print("")
# Load the perties branch
client.branch('literal_properties')
client.checkout('literal_properties')
times = []
print(f"Importing literal properties from {literal_properties_directory}")
for f in os.listdir(literal_properties_directory):
filename = f'{literal_properties_directory}/{f}'
ttl_file = open(filename)
contents = ttl_file.read()
ttl_file.close()
# start the chunk work
before = time.time()
if optimize == True:
optimizer(client)
client.insert_triples("instance", "main", contents,
f"Adding literal properties in 100k chunk ({f})")
after = time.time()
new_db = "DBPedia_squash"
client.remote_auth({"type": "basic", "user": user, "key": key})
client.clonedb(
{
"remote_url": f"{server_url}/{user}/{db}",
"comment": "Squash test",
"label": "Squash test"
}, new_db)
new_db = "DBPedia_squash"
client = WOQLClient(server_url)
client.remote_auth({"type": "basic", "user": user, "key": key})
client.connect(user=user, account=account, key=key, db=new_db)
client.branch("types")
client.checkout("types")
client.pull({"remote": "origin", "remote_branch": "types"})
client.branch("properties")
client.checkout("properties")
client.pull({"remote": "origin", "remote_branch": "properties"})
client.checkout('main')
# Rebase valid data to schema branch
client.rebase({
"rebase_from": f'{user}/{new_db}/local/branch/types',
"author": user,
"message": "Merging types into main"
})
client.checkout('main')
