licht1stein
diff --git a/‎pandas_to_postgres/__init__.py
Copy file name to clipboardExpand all lines: pandas_to_postgres/__init__.py
+5Lines changed: 5 additions & 0 deletions b/‎pandas_to_postgres/__init__.py
Copy file name to clipboardExpand all lines: pandas_to_postgres/__init__.py
+5Lines changed: 5 additions & 0 deletions
diff --git a/‎pandas_to_postgres/hdf_to_postgres.py
Copy file name to clipboard
+92Lines changed: 92 additions & 0 deletions b/‎pandas_to_postgres/hdf_to_postgres.py
Copy file name to clipboard
+92Lines changed: 92 additions & 0 deletions
diff --git a/‎pandas_to_postgres/utilities.py
Copy file name to clipboardExpand all lines: pandas_to_postgres/utilities.py
+2-1Lines changed: 2 additions & 1 deletion b/‎pandas_to_postgres/utilities.py
Copy file name to clipboardExpand all lines: pandas_to_postgres/utilities.py
+2-1Lines changed: 2 additions & 1 deletion
@@ -1,5 +1,10 @@
 from .copy_df import DataFrameCopy
 from .copy_hdf import HDFTableCopy, SmallHDFTableCopy, BigHDFTableCopy
+from .hdf_to_postgres import (
+    hdf_to_postgres,
+    multiprocess_hdf_to_postgres,
+    create_hdf_table_objects,
+)
 from .utilities import (
     logger,
     HDFMetadata,
 
@@ -0,0 +1,92 @@
+from typing import List
+from multiprocessing import Pool
+import SQLAlchemy
+from pandas_to_postgres import HDFTableCopy, HDFMetadata
+
+
+def create_hdf_table_objects(hdf_meta, csv_chunksize=10 ** 6):
+    tables = []
+
+    for sql_table, hdf_tables in hdf_meta.sql_to_hdf.items():
+        tables.append(
+            HDFTableCopy(
+                hdf_tables,
+                hdf_meta,
+                defer_sql_objs=True,
+                sql_table=sql_table,
+                csv_chunksize=csv_chunksize,
+            )
+        )
+
+    return tables
+
+
+def copy_worker(db: SQLAlchemy, copy_obj: HDFTableCopy, defer_sql_objs: bool = True):
+    db.engine.dispose()
+    with db.engine.connect() as conn:
+        conn.execution_options(autocommit=True)
+        conn.execute("SET maintenance_work_mem TO 1000000;")
+
+        if defer_sql_objs:
+            table_obj = db.metadata.tables[copy_obj.sql_table]
+            copy_obj.instantiate_sql_objs(conn, table_obj)
+
+        copy_obj.copy()
+
+
+def hdf_to_postgres(
+    file_name: str, db: SQLAlchemy, keys: List[str] = [], csv_chunksize: int = 10 ** 6
+):
+    """
+    Copy tables in a HDF file to PostgreSQL database
+
+    Parameters
+    ----------
+    file_name: name of file or path to file of HDF to use to copy
+    db: SQLAlchemy object of destination database
+    keys: list of HDF keys to copy
+    csv_chunksize: maximum number of StringIO CSV rows to keep in memory at a time
+    """
+    hdf = HDFMetadata(
+        file_name, keys, metadata_attr="atlas_metadata", metadata_keys=["levels"]
+    )
+
+    tables = create_hdf_table_objects(hdf, csv_chunksize=csv_chunksize)
+
+    for table in tables:
+        copy_worker(table, defer_sql_objs=True)
+
+
+def multiprocess_hdf_to_postgres(
+    file_name: str,
+    db: SQLAlchemy,
+    keys: List[str] = [],
+    processes: int = 4,
+    csv_chunksize: int = 10 ** 6,
+):
+    """
+    Copy tables in a HDF file to PostgreSQL database using a multiprocessing Pool
+
+    Parameters
+    ----------
+    file_name: name of file or path to file of HDF to use to copy
+    db: SQLAlchemy object of destination database
+    keys: list of HDF keys to copy
+    processes: number of processes in the Pool
+    csv_chunksize: maximum number of StringIO CSV rows to keep in memory at a time
+    """
+
+    hdf = HDFMetadata(
+        file_name, keys, metadata_attr="atlas_metadata", metadata_keys=["levels"]
+    )
+
+    tables = create_hdf_table_objects(hdf, csv_chunksize=csv_chunksize)
+
+    try:
+        p = Pool(processes)
+        p.map(copy_worker, tables, chunksize=1)
+    finally:
+        del tables
+        del hdf
+        p.close()
+        p.join()
@@ -34,7 +34,8 @@ def __init__(
         ----------
         file_name: path to hdf file to copy from
         keys: list of hdf keys to copy data from
-        chunksize: maximum rows read from an hdf file into a pandas dataframe
+        chunksize: maximum rows read from an hdf file into a pandas dataframe when using
+            the BigTable protocol
         metadata_attr: location of relevant metadata in store.get_storer().attrs
         metadata_keys: list of keys to get from metadata store
         """