You are viewing a plain text version of this content. The canonical link for it is here.
Posted to notifications@superset.apache.org by GitBox <gi...@apache.org> on 2021/04/21 16:23:50 UTC
[GitHub] [superset] ktmud commented on a change in pull request #14240: fix(hive): Use parquet rather than textfile when uploading CSV files to Hive
ktmud commented on a change in pull request #14240:
URL: https://github.com/apache/superset/pull/14240#discussion_r617697523
##########
File path: superset/views/database/views.py
##########
@@ -156,48 +157,45 @@ def form_post(self, form: CsvToDatabaseForm) -> Response:
).name
try:
- utils.ensure_path_exists(config["UPLOAD_FOLDER"])
+ utils.ensure_path_exists(app.config["UPLOAD_FOLDER"])
upload_stream_write(form.csv_file.data, uploaded_tmp_file_path)
Review comment:
Not related to this PR, but I think it may be possible to let pandas read directly from the stream, without writing a tempfile:
```
chunks = pd.read_csv(form.csv_file.data.stream)
```
##########
File path: superset/views/database/views.py
##########
@@ -156,48 +157,45 @@ def form_post(self, form: CsvToDatabaseForm) -> Response:
).name
try:
- utils.ensure_path_exists(config["UPLOAD_FOLDER"])
+ utils.ensure_path_exists(app.config["UPLOAD_FOLDER"])
upload_stream_write(form.csv_file.data, uploaded_tmp_file_path)
con = form.data.get("con")
database = (
db.session.query(models.Database).filter_by(id=con.data.get("id")).one()
)
- # More can be found here:
- # https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.read_csv.html
- csv_to_df_kwargs = {
- "sep": form.sep.data,
- "header": form.header.data if form.header.data else 0,
- "index_col": form.index_col.data,
- "mangle_dupe_cols": form.mangle_dupe_cols.data,
- "skipinitialspace": form.skipinitialspace.data,
- "skiprows": form.skiprows.data,
- "nrows": form.nrows.data,
- "skip_blank_lines": form.skip_blank_lines.data,
- "parse_dates": form.parse_dates.data,
- "infer_datetime_format": form.infer_datetime_format.data,
- "chunksize": 1000,
- }
- if form.null_values.data:
- csv_to_df_kwargs["na_values"] = form.null_values.data
- csv_to_df_kwargs["keep_default_na"] = False
-
- # More can be found here:
- # https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.to_sql.html
- df_to_sql_kwargs = {
- "name": csv_table.table,
- "if_exists": form.if_exists.data,
- "index": form.index.data,
- "index_label": form.index_label.data,
- "chunksize": 1000,
- }
- database.db_engine_spec.create_table_from_csv(
+ chunks = pd.read_csv(
uploaded_tmp_file_path,
- csv_table,
+ chunksize=1000,
+ encoding="utf-8",
+ header=form.header.data if form.header.data else 0,
+ index_col=form.index_col.data,
+ infer_datetime_format=form.infer_datetime_format.data,
+ iterator=True,
+ keep_default_na=not form.null_values.data,
+ mangle_dupe_cols=form.mangle_dupe_cols.data,
+ na_values=form.null_values.data if form.null_values.data else None,
+ nrows=form.nrows.data,
+ parse_dates=form.parse_dates.data,
+ sep=form.sep.data,
+ skip_blank_lines=form.skip_blank_lines.data,
+ skipinitialspace=form.skipinitialspace.data,
+ skiprows=form.skiprows.data,
+ )
+
+ df = pd.concat(chunk for chunk in chunks)
+
+ database.db_engine_spec.df_to_sql(
Review comment:
It seems passing `database` and `table` to `db_engine_spec.df_to_sql` would introduce circular dependencies. Could we maybe create a database level instance method instead?
```python
database.fill_table(table, df)
class SqlaDatasource:
def fill_table(self, table, **to_sql_kwargs):
if table.is_virtual:
raise SupersetException(_("Cannot fill a virtual table"))
self.db_engine_spec.fill_table(
table.name, df, schema=table.schema, **to_sql_kwargs
)
```
##########
File path: superset/views/database/views.py
##########
@@ -156,48 +157,45 @@ def form_post(self, form: CsvToDatabaseForm) -> Response:
).name
try:
- utils.ensure_path_exists(config["UPLOAD_FOLDER"])
+ utils.ensure_path_exists(app.config["UPLOAD_FOLDER"])
upload_stream_write(form.csv_file.data, uploaded_tmp_file_path)
con = form.data.get("con")
database = (
db.session.query(models.Database).filter_by(id=con.data.get("id")).one()
)
- # More can be found here:
- # https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.read_csv.html
- csv_to_df_kwargs = {
- "sep": form.sep.data,
- "header": form.header.data if form.header.data else 0,
- "index_col": form.index_col.data,
- "mangle_dupe_cols": form.mangle_dupe_cols.data,
- "skipinitialspace": form.skipinitialspace.data,
- "skiprows": form.skiprows.data,
- "nrows": form.nrows.data,
- "skip_blank_lines": form.skip_blank_lines.data,
- "parse_dates": form.parse_dates.data,
- "infer_datetime_format": form.infer_datetime_format.data,
- "chunksize": 1000,
- }
- if form.null_values.data:
- csv_to_df_kwargs["na_values"] = form.null_values.data
- csv_to_df_kwargs["keep_default_na"] = False
-
- # More can be found here:
- # https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.to_sql.html
- df_to_sql_kwargs = {
- "name": csv_table.table,
- "if_exists": form.if_exists.data,
- "index": form.index.data,
- "index_label": form.index_label.data,
- "chunksize": 1000,
- }
- database.db_engine_spec.create_table_from_csv(
+ chunks = pd.read_csv(
uploaded_tmp_file_path,
- csv_table,
+ chunksize=1000,
+ encoding="utf-8",
+ header=form.header.data if form.header.data else 0,
+ index_col=form.index_col.data,
+ infer_datetime_format=form.infer_datetime_format.data,
+ iterator=True,
+ keep_default_na=not form.null_values.data,
+ mangle_dupe_cols=form.mangle_dupe_cols.data,
+ na_values=form.null_values.data if form.null_values.data else None,
+ nrows=form.nrows.data,
+ parse_dates=form.parse_dates.data,
+ sep=form.sep.data,
+ skip_blank_lines=form.skip_blank_lines.data,
+ skipinitialspace=form.skipinitialspace.data,
+ skiprows=form.skiprows.data,
+ )
+
+ df = pd.concat(chunk for chunk in chunks)
Review comment:
This should also work?
```python
df = pd.concat(chunks)
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@superset.apache.org
For additional commands, e-mail: notifications-help@superset.apache.org