feat: add support for standalone mode in Python SDK

[HuilinWu2]

now users are able to use standalone mode in python SDK DBAPI through:

import openmldb
# cluster mode
db = openmldb.dbapi.connect("db_test","127.0.0.1:6181","/onebox")

# standalone mode
db = openmldb.dbapi.connect("db_test","127.0.0.1", 6527)

or through sqlalchemy API:

import sqlalchemy as db
# cluster
engine = db.create_engine("openmldb:///db_test?zk=127.0.0.1:6181&zkPath=/onebox")

# standalone
engine = db.create_engine("openmldb:///db_test?host=127.0.0.1&port=6527")

[github-actions]

Linux Test Report

     110 files       236 suites   43m 33s ⏱️
  8 613 tests   8 610 ✔️ 3 💤 0 ❌
12 745 runs  12 742 ✔️ 3 💤 0 ❌

Results for commit 6da0ec8.

♻️ This comment has been updated with latest results.

[codecov]

Codecov Report

Merging #1325 (5ecb1ad) into main (461d520) will increase coverage by 0.11%.
The diff coverage is n/a.

❗ Current head 5ecb1ad differs from pull request most recent head 6da0ec8. Consider uploading reports for the commit 6da0ec8 to get more accurate results

@@             Coverage Diff              @@
##               main    #1325      +/-   ##
============================================
+ Coverage     65.41%   65.53%   +0.11%     
+ Complexity      231      222       -9     
============================================
  Files           570      570              
  Lines        106540   106439     -101     
  Branches        872      857      -15     
============================================
+ Hits          69696    69752      +56     
+ Misses        36692    36541     -151     
+ Partials        152      146       -6     

Impacted Files	Coverage Δ
src/base/status.h	60.00% <0.00%> (-40.00%)	⬇️
hybridse/include/base/spin_lock.h	66.66% <0.00%> (-33.34%)	⬇️
hybridse/include/sdk/base.h	14.28% <0.00%> (-4.95%)	⬇️
..._4paradigm/openmldb/batch/nodes/LoadDataPlan.scala	53.68% <0.00%> (-4.81%)	⬇️
..._4paradigm/openmldb/batch/utils/SparkRowUtil.scala	58.53% <0.00%> (-2.44%)	⬇️
src/sdk/sql_cluster_router.cc	32.90% <0.00%> (-1.90%)	⬇️
..._4paradigm/openmldb/batch/utils/HybridseUtil.scala	60.90% <0.00%> (-1.51%)	⬇️
src/catalog/tablet_catalog.cc	67.92% <0.00%> (-0.95%)	⬇️
src/client/ns_client.cc	41.15% <0.00%> (-0.31%)	⬇️
src/sdk/db_sdk.cc	60.60% <0.00%> (-0.31%)	⬇️
... and 36 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update aac8167...6da0ec8. Read the comment docs.

[tobegit3hub]

LGTM

[dl239]

now users are able to use standalone mode in python SDK DBAPI through:

import openmldb
# cluster mode
db = openmldb.dbapi.connect("db_test","127.0.0.1:6181","/onebox")

# standalone mode
db = openmldb.dbapi.connect("db_test", host="127.0.0.1", port=6527, is_cluster_mode=False)

or through sqlalchemy API:

import sqlalchemy as db
# cluster
engine = db.create_engine("openmldb:///db_test?zk=127.0.0.1:6181&zkPath=/onebox")

# standalone
engine = db.create_engine("openmldb:///db_test?host=127.0.0.1&port=6527", connect_args={'is_cluster_mode':False})

why add is_cluster_mode option to host and port connect method. It's better to keep consist with CLI that zk as cluster and host/port as standalone

[HuilinWu2]

now users are able to use standalone mode in python SDK DBAPI through:

import openmldb
# cluster mode
db = openmldb.dbapi.connect("db_test","127.0.0.1:6181","/onebox")

# standalone mode
db = openmldb.dbapi.connect("db_test", host="127.0.0.1", port=6527, is_cluster_mode=False)

or through sqlalchemy API:

import sqlalchemy as db
# cluster
engine = db.create_engine("openmldb:///db_test?zk=127.0.0.1:6181&zkPath=/onebox")

# standalone
engine = db.create_engine("openmldb:///db_test?host=127.0.0.1&port=6527", connect_args={'is_cluster_mode':False})

why add is_cluster_mode option to host and port connect method. It's better to keep consist with CLI that zk as cluster and host/port as standalone

RE: I agree, we have also considered. However, if so , then we may know whether users want to use standalone/cluster mode only by identifying the type of the second input, namely zkPath string or port int, do you agree?