meerschaum 2.7.1__py3-none-any.whl → 2.7.3__py3-none-any.whl

meerschaum/utils/sql.py

@@ -45,12 +45,12 @@ DROP_IF_EXISTS_FLAVORS = {
 }
 SKIP_AUTO_INCREMENT_FLAVORS = {'citus', 'duckdb'}
 COALESCE_UNIQUE_INDEX_FLAVORS = {'timescaledb', 'postgresql', 'citus'}
-update_queries = {
+UPDATE_QUERIES = {
     'default': """
         UPDATE {target_table_name} AS f
         {sets_subquery_none}
         FROM {target_table_name} AS t
-        INNER JOIN (SELECT DISTINCT {patch_cols_str} FROM {patch_table_name}) AS p
+        INNER JOIN (SELECT {patch_cols_str} FROM {patch_table_name}) AS p
             ON
                 {and_subquery_t}
         WHERE
@@ -84,7 +84,7 @@ update_queries = {
     """,
     'mysql': """
         UPDATE {target_table_name} AS f
-        JOIN (SELECT DISTINCT {patch_cols_str} FROM {patch_table_name}) AS p
+        JOIN (SELECT {patch_cols_str} FROM {patch_table_name}) AS p
         ON
             {and_subquery_f}
         {sets_subquery_f}
@@ -100,7 +100,7 @@ update_queries = {
     """,
     'mariadb': """
         UPDATE {target_table_name} AS f
-        JOIN (SELECT DISTINCT {patch_cols_str} FROM {patch_table_name}) AS p
+        JOIN (SELECT {patch_cols_str} FROM {patch_table_name}) AS p
         ON
             {and_subquery_f}
         {sets_subquery_f}
@@ -179,13 +179,13 @@ update_queries = {
         WHERE ROWID IN (
             SELECT t.ROWID
             FROM {target_table_name} AS t
-            INNER JOIN (SELECT DISTINCT * FROM {patch_table_name}) AS p
+            INNER JOIN (SELECT * FROM {patch_table_name}) AS p
                 ON {and_subquery_t}
         );
         """,
         """
         INSERT INTO {target_table_name} AS f
-        SELECT DISTINCT {patch_cols_str} FROM {patch_table_name} AS p
+        SELECT {patch_cols_str} FROM {patch_table_name} AS p
         """,
     ],
 }
@@ -573,21 +573,21 @@ def dateadd_str(
     Examples
     --------
     >>> dateadd_str(
-    ...     flavor = 'mssql',
-    ...     begin = datetime(2022, 1, 1, 0, 0),
-    ...     number = 1,
+    ...     flavor='mssql',
+    ...     begin=datetime(2022, 1, 1, 0, 0),
+    ...     number=1,
     ... )
     "DATEADD(day, 1, CAST('2022-01-01 00:00:00' AS DATETIME2))"
     >>> dateadd_str(
-    ...     flavor = 'postgresql',
-    ...     begin = datetime(2022, 1, 1, 0, 0),
-    ...     number = 1,
+    ...     flavor='postgresql',
+    ...     begin=datetime(2022, 1, 1, 0, 0),
+    ...     number=1,
     ... )
     "CAST('2022-01-01 00:00:00' AS TIMESTAMP) + INTERVAL '1 day'"
 
     """
     from meerschaum.utils.packages import attempt_import
-    from meerschaum.utils.dtypes.sql import get_db_type_from_pd_type
+    from meerschaum.utils.dtypes.sql import get_db_type_from_pd_type, get_pd_type_from_db_type
     dateutil_parser = attempt_import('dateutil.parser')
     if 'int' in str(type(begin)).lower():
         return str(begin)
@@ -619,7 +619,14 @@ def dateadd_str(
             else f"'{begin}'"
         )
 
-    dt_is_utc = begin_time.tzinfo is not None if begin_time is not None else '+' in str(begin)
+    dt_is_utc = (
+        begin_time.tzinfo is not None
+        if begin_time is not None
+        else ('+' in str(begin) or '-' in str(begin).split(':', maxsplit=1)[-1])
+    )
+    if db_type:
+        db_type_is_utc = 'utc' in get_pd_type_from_db_type(db_type).lower()
+        dt_is_utc = dt_is_utc or db_type_is_utc
     db_type = db_type or get_db_type_from_pd_type(
         ('datetime64[ns, UTC]' if dt_is_utc else 'datetime64[ns]'),
         flavor=flavor,
@@ -629,22 +636,32 @@ def dateadd_str(
     if flavor in ('postgresql', 'timescaledb', 'cockroachdb', 'citus'):
         begin = (
             f"CAST({begin} AS {db_type})" if begin != 'now'
-            else "CAST(NOW() AT TIME ZONE 'utc' AS {db_type})"
+            else f"CAST(NOW() AT TIME ZONE 'utc' AS {db_type})"
         )
+        if dt_is_utc:
+            begin += " AT TIME ZONE 'UTC'"
         da = begin + (f" + INTERVAL '{number} {datepart}'" if number != 0 else '')
 
     elif flavor == 'duckdb':
         begin = f"CAST({begin} AS {db_type})" if begin != 'now' else 'NOW()'
+        if dt_is_utc:
+            begin += " AT TIME ZONE 'UTC'"
         da = begin + (f" + INTERVAL '{number} {datepart}'" if number != 0 else '')
 
     elif flavor in ('mssql',):
         if begin_time and begin_time.microsecond != 0 and not dt_is_utc:
             begin = begin[:-4] + "'"
         begin = f"CAST({begin} AS {db_type})" if begin != 'now' else 'GETUTCDATE()'
+        if dt_is_utc:
+            begin += " AT TIME ZONE 'UTC'"
         da = f"DATEADD({datepart}, {number}, {begin})" if number != 0 else begin
 
     elif flavor in ('mysql', 'mariadb'):
-        begin = f"CAST({begin} AS DATETIME(6))" if begin != 'now' else 'UTC_TIMESTAMP(6)'
+        begin = (
+            f"CAST({begin} AS DATETIME(6))"
+            if begin != 'now'
+            else 'UTC_TIMESTAMP(6)'
+        )
         da = (f"DATE_ADD({begin}, INTERVAL {number} {datepart})" if number != 0 else begin)
 
     elif flavor == 'sqlite':
@@ -653,10 +670,10 @@ def dateadd_str(
     elif flavor == 'oracle':
         if begin == 'now':
             begin = str(
-                datetime.now(timezone.utc).replace(tzinfo=None).strftime('%Y:%m:%d %M:%S.%f')
+                datetime.now(timezone.utc).replace(tzinfo=None).strftime(r'%Y:%m:%d %M:%S.%f')
             )
         elif begin_time:
-            begin = str(begin_time.strftime('%Y-%m-%d %H:%M:%S.%f'))
+            begin = str(begin_time.strftime(r'%Y-%m-%d %H:%M:%S.%f'))
         dt_format = 'YYYY-MM-DD HH24:MI:SS.FF'
         _begin = f"'{begin}'" if begin_time else begin
         da = (
@@ -691,7 +708,7 @@ def test_connection(
         warnings.filterwarnings('ignore', 'Could not')
         try:
             return retry_connect(**_default_kw)
-        except Exception as e:
+        except Exception:
             return False
 
 
@@ -784,7 +801,7 @@ def sql_item_name(item: str, flavor: str, schema: Optional[str] = None) -> str:
         ### NOTE: System-reserved words must be quoted.
         if truncated_item.lower() in (
             'float', 'varchar', 'nvarchar', 'clob',
-            'boolean', 'integer', 'table',
+            'boolean', 'integer', 'table', 'row',
         ):
             wrappers = ('"', '"')
         else:
@@ -1494,9 +1511,9 @@ def get_update_queries(
     ):
         flavor = 'sqlite_delete_insert'
     flavor_key = (f'{flavor}-upsert' if upsert else flavor)
-    base_queries = update_queries.get(
+    base_queries = UPDATE_QUERIES.get(
         flavor_key,
-        update_queries['default']
+        UPDATE_QUERIES['default']
     )
     if not isinstance(base_queries, list):
         base_queries = [base_queries]
@@ -1577,6 +1594,12 @@ def get_update_queries(
         if not value_cols:
             return ''
 
+        utc_value_cols = {
+            c_name
+            for c_name, c_type in value_cols
+            if ('utc' in get_pd_type_from_db_type(c_type).lower())
+        } if flavor not in TIMEZONE_NAIVE_FLAVORS else set()
+
         cast_func_cols = {
             c_name: (
                 ('', '', '')
@@ -1585,7 +1608,11 @@ def get_update_queries(
                     and are_dtypes_equal(get_pd_type_from_db_type(c_type), 'bytes')
                 )
                 else (
-                    ('CAST(', f" AS {c_type.replace('_', ' ')}", ')')
+                    ('CAST(', f" AS {c_type.replace('_', ' ')}", ')' + (
+                        " AT TIME ZONE 'UTC'"
+                        if c_name in utc_value_cols
+                        else ''
+                    ))
                     if flavor != 'sqlite'
                     else ('', '', '')
                 )
@@ -1865,6 +1892,7 @@ def get_create_table_queries(
     flavor: str,
     schema: Optional[str] = None,
     primary_key: Optional[str] = None,
+    primary_key_db_type: Optional[str] = None,
     autoincrement: bool = False,
     datetime_column: Optional[str] = None,
 ) -> List[str]:
@@ -1889,6 +1917,9 @@ def get_create_table_queries(
     primary_key: Optional[str], default None
         If provided, designate this column as the primary key in the new table.
 
+    primary_key_db_type: Optional[str], default None
+        If provided, alter the primary key to this type (to set NOT NULL constraint).
+
     autoincrement: bool, default False
         If `True` and `primary_key` is provided, create the `primary_key` column
         as an auto-incrementing integer column.
@@ -1915,6 +1946,7 @@ def get_create_table_queries(
         flavor,
         schema=schema,
         primary_key=primary_key,
+        primary_key_db_type=primary_key_db_type,
         autoincrement=(autoincrement and flavor not in SKIP_AUTO_INCREMENT_FLAVORS),
         datetime_column=datetime_column,
     )
@@ -1926,6 +1958,7 @@ def _get_create_table_query_from_dtypes(
     flavor: str,
     schema: Optional[str] = None,
     primary_key: Optional[str] = None,
+    primary_key_db_type: Optional[str] = None,
     autoincrement: bool = False,
     datetime_column: Optional[str] = None,
 ) -> List[str]:
@@ -2015,6 +2048,7 @@ def _get_create_table_query_from_cte(
     flavor: str,
     schema: Optional[str] = None,
     primary_key: Optional[str] = None,
+    primary_key_db_type: Optional[str] = None,
     autoincrement: bool = False,
     datetime_column: Optional[str] = None,
 ) -> List[str]:
@@ -2035,7 +2069,11 @@ def _get_create_table_query_from_cte(
         if primary_key
         else None
     )
-    primary_key_clustered = "CLUSTERED" if not datetime_column else "NONCLUSTERED"
+    primary_key_clustered = (
+        "CLUSTERED"
+        if not datetime_column or datetime_column == primary_key
+        else "NONCLUSTERED"
+    )
     datetime_column_name = (
         sql_item_name(datetime_column, flavor)
         if datetime_column
@@ -2045,80 +2083,106 @@ def _get_create_table_query_from_cte(
         query = query.lstrip()
         if query.lower().startswith('with '):
             final_select_ix = query.lower().rfind('select')
-            create_table_query = (
-                query[:final_select_ix].rstrip() + ',\n'
-                + f"{create_cte_name} AS (\n"
-                + query[final_select_ix:]
-                + "\n)\n"
-                + f"SELECT *\nINTO {new_table_name}\nFROM {create_cte_name}"
-            )
+            create_table_queries = [
+                (
+                    query[:final_select_ix].rstrip() + ',\n'
+                    + f"{create_cte_name} AS (\n"
+                    + textwrap.indent(query[final_select_ix:], '    ')
+                    + "\n)\n"
+                    + f"SELECT *\nINTO {new_table_name}\nFROM {create_cte_name}"
+                ),
+            ]
         else:
-            create_table_query = f"""
-                SELECT *
-                INTO {new_table_name}
-                FROM ({query}) AS {create_cte_name}
-            """
-
-        alter_type_query = f"""
-            ALTER TABLE {new_table_name}
-            ADD CONSTRAINT {primary_key_constraint_name} PRIMARY KEY {primary_key_clustered} ({primary_key_name})
-        """
+            create_table_queries = [
+                (
+                    "SELECT *\n"
+                    f"INTO {new_table_name}\n"
+                    f"FROM (\n{textwrap.indent(query, '    ')}\n) AS {create_cte_name}"
+                ),
+            ]
+
+        alter_type_queries = []
+        if primary_key_db_type:
+            alter_type_queries.extend([
+                (
+                    f"ALTER TABLE {new_table_name}\n"
+                    f"ALTER COLUMN {primary_key_name} {primary_key_db_type} NOT NULL"
+                ),
+            ])
+        alter_type_queries.extend([
+            (
+                f"ALTER TABLE {new_table_name}\n"
+                f"ADD CONSTRAINT {primary_key_constraint_name} "
+                f"PRIMARY KEY {primary_key_clustered} ({primary_key_name})"
+            ),
+        ])
     elif flavor in (None,):
-        create_table_query = f"""
-            WITH {create_cte_name} AS ({query})
-            CREATE TABLE {new_table_name} AS
-            SELECT *
-            FROM {create_cte_name}
-        """
+        create_table_queries = [
+            (
+                f"WITH {create_cte_name} AS (\n{textwrap.index(query, '    ')}\n)\n"
+                f"CREATE TABLE {new_table_name} AS\n"
+                "SELECT *\n"
+                f"FROM {create_cte_name}"
+            ),
+        ]
 
-        alter_type_query = f"""
-            ALTER TABLE {new_table_name}
-            ADD PRIMARY KEY ({primary_key_name})
-        """
+        alter_type_queries = [
+            (
+                f"ALTER TABLE {new_table_name}\n"
+                f"ADD PRIMARY KEY ({primary_key_name})"
+            ),
+        ]
     elif flavor in ('sqlite', 'mysql', 'mariadb', 'duckdb', 'oracle'):
-        create_table_query = f"""
-            CREATE TABLE {new_table_name} AS
-            SELECT *
-            FROM ({query})""" + (f""" AS {create_cte_name}""" if flavor != 'oracle' else '') + """
-        """
+        create_table_queries = [
+            (
+                f"CREATE TABLE {new_table_name} AS\n"
+                "SELECT *\n"
+                f"FROM (\n{textwrap.indent(query, '    ')}\n)"
+                + (f" AS {create_cte_name}" if flavor != 'oracle' else '')
+            ),
+        ]
 
-        alter_type_query = f"""
-            ALTER TABLE {new_table_name}
-            ADD PRIMARY KEY ({primary_key_name})
-        """
+        alter_type_queries = [
+            (
+                f"ALTER TABLE {new_table_name}\n"
+                "ADD PRIMARY KEY ({primary_key_name})"
+            ),
+        ]
     elif flavor == 'timescaledb' and datetime_column and datetime_column != primary_key:
-        create_table_query = f"""
-            SELECT *
-            INTO {new_table_name}
-            FROM ({query}) AS {create_cte_name}
-        """
+        create_table_queries = [
+            (
+                "SELECT *\n"
+                f"INTO {new_table_name}\n"
+                f"FROM (\n{textwrap.indent(query, '    ')}\n) AS {create_cte_name}\n"
+            ),
+        ]
 
-        alter_type_query = f"""
-            ALTER TABLE {new_table_name}
-            ADD PRIMARY KEY ({datetime_column_name}, {primary_key_name})
-        """
+        alter_type_queries = [
+            (
+                f"ALTER TABLE {new_table_name}\n"
+                f"ADD PRIMARY KEY ({datetime_column_name}, {primary_key_name})"
+            ),
+        ]
     else:
-        create_table_query = f"""
-            SELECT *
-            INTO {new_table_name}
-            FROM ({query}) AS {create_cte_name}
-        """
+        create_table_queries = [
+            (
+                "SELECT *\n"
+                f"INTO {new_table_name}\n"
+                f"FROM (\n{textwrap.indent(query, '    ')}\n) AS {create_cte_name}"
+            ),
+        ]
 
-        alter_type_query = f"""
-            ALTER TABLE {new_table_name}
-            ADD PRIMARY KEY ({primary_key_name})
-        """
+        alter_type_queries = [
+            (
+                f"ALTER TABLE {new_table_name}\n"
+                f"ADD PRIMARY KEY ({primary_key_name})"
+            ),
+        ]
 
-    create_table_query = textwrap.dedent(create_table_query).lstrip().rstrip()
     if not primary_key:
-        return [create_table_query]
-
-    alter_type_query = textwrap.dedent(alter_type_query).lstrip().rstrip()
+        return create_table_queries
 
-    return [
-        create_table_query,
-        alter_type_query,
-    ]
+    return create_table_queries + alter_type_queries
 
 
 def wrap_query_with_cte(
@@ -2167,6 +2231,7 @@ def wrap_query_with_cte(
     ```
 
     """
+    import textwrap
     sub_query = sub_query.lstrip()
     cte_name_quoted = sql_item_name(cte_name, flavor, None)
 
@@ -2190,7 +2255,7 @@ def wrap_query_with_cte(
 
     return (
         f"WITH {cte_name_quoted} AS (\n"
-        f"    {sub_query}\n"
+        f"{textwrap.indent(sub_query, '    ')}\n"
         f")\n{parent_query}"
     )
 
@@ -2269,6 +2334,8 @@ def session_execute(
         queries = [queries]
     successes, msgs, results = [], [], []
     for query in queries:
+        if debug:
+            dprint(query)
         query_text = sqlalchemy.text(query)
         fail_msg = "Failed to execute queries."
         try:
@@ -2283,6 +2350,8 @@ def session_execute(
         msgs.append(query_msg)
         results.append(result)
         if not query_success:
+            if debug:
+                dprint("Rolling back session.")
             session.rollback()
             break
     success, msg = all(successes), '\n'.join(msgs)

meerschaum/utils/venv/__init__.py

@@ -67,7 +67,6 @@ def activate_venv(
             return True
     import sys
     import os
-    from meerschaum.config._paths import VIRTENV_RESOURCES_PATH
     if venv is not None:
         init_venv(venv=venv, debug=debug)
     with LOCKS['active_venvs']:
@@ -379,22 +378,47 @@ def init_venv(
     import os
     import pathlib
     import shutil
+    import time
 
     from meerschaum.config.static import STATIC_CONFIG
-    from meerschaum.config._paths import VIRTENV_RESOURCES_PATH, VENVS_CACHE_RESOURCES_PATH
+    from meerschaum.config._paths import (
+        VIRTENV_RESOURCES_PATH,
+        VENVS_CACHE_RESOURCES_PATH,
+    )
     from meerschaum.utils.packages import is_uv_enabled
 
     venv_path = VIRTENV_RESOURCES_PATH / venv
     vtp = venv_target_path(venv=venv, allow_nonexistent=True, debug=debug)
     docker_home_venv_path = pathlib.Path('/home/meerschaum/venvs/mrsm')
-
+    lock_path = VENVS_CACHE_RESOURCES_PATH / (venv + '.lock')
     work_dir_env_var = STATIC_CONFIG['environment']['work_dir']
+
+    def update_lock(active: bool):
+        try:
+            if active:
+                lock_path.unlink()
+            else:
+                lock_path.touch()
+        except Exception:
+            pass
+
+    def wait_for_lock():
+        max_lock_seconds = 1.0
+        step_sleep_seconds = 0.1
+        init_venv_check_start = time.perf_counter()
+        while (time.perf_counter() - init_venv_check_start < max_lock_seconds):
+            if not lock_path.exists():
+                continue
+            time.sleep(step_sleep_seconds)
+        update_lock(False)
+
     if (
         not force
         and venv == 'mrsm'
         and os.environ.get(work_dir_env_var, None) is not None
         and docker_home_venv_path.exists()
     ):
+        wait_for_lock()
         shutil.move(docker_home_venv_path, venv_path)
         if verify:
             verify_venv(venv, debug=debug)
@@ -422,6 +446,9 @@ def init_venv(
         except FileExistsError:
             pass
 
+    wait_for_lock()
+    update_lock(True)
+
     if uv is not None:
         _venv_success = run_python_package(
             'uv',
@@ -437,7 +464,9 @@ def init_venv(
             _venv_success = run_python_package(
                 'venv',
                 [venv_path.as_posix()] + (
-                    ['--symlinks'] if platform.system() != 'Windows' else []
+                    ['--symlinks']
+                    if platform.system() != 'Windows'
+                    else []
                 ),
                 venv=None, debug=debug
             ) == 0
@@ -447,8 +476,14 @@ def init_venv(
             _venv = None
     if not _venv_success:
         virtualenv = attempt_import(
-            'virtualenv', venv=None, lazy=False, install=(not tried_virtualenv), warn=False,
-            check_update=False, color=False, debug=debug,
+            'virtualenv',
+            venv=None,
+            lazy=False,
+            install=(not tried_virtualenv),
+            warn=False,
+            check_update=False,
+            color=False,
+            debug=debug,
         )
         if virtualenv is None:
             print(
@@ -457,6 +492,7 @@ def init_venv(
             )
             if rename_vtp and temp_vtp.exists():
                 temp_vtp.rename(vtp)
+            update_lock(False)
             return False
 
         tried_virtualenv = True
@@ -495,6 +531,7 @@ def init_venv(
             traceback.print_exc()
             if rename_vtp and temp_vtp.exists():
                 temp_vtp.rename(vtp)
+            update_lock(False)
             return False
     if verify:
         verify_venv(venv, debug=debug)
@@ -503,6 +540,7 @@ def init_venv(
     if rename_vtp and temp_vtp.exists():
         temp_vtp.rename(vtp)
 
+    update_lock(False)
     return True
 
 

{meerschaum-2.7.1.dist-info → meerschaum-2.7.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: meerschaum
-Version: 2.7.1
+Version: 2.7.3
 Summary: Sync Time-Series Pipes with Meerschaum
 Home-page: https://meerschaum.io
 Author: Bennett Meares