WeichenXu123 commented on a change in pull request #29284:
URL: https://github.com/apache/spark/pull/29284#discussion_r462672906
##########
File path: python/pyspark/sql/pandas/conversion.py
##########
@@ -404,8 +404,10 @@ def _create_from_pandas_with_arrow(self, pdf, schema,
timezone):
for t in pdf.dtypes]
# Slice the DataFrame to be batched
- step = -(-len(pdf) // self.sparkContext.defaultParallelism) # round
int up
- pdf_slices = (pdf.iloc[start:start + step] for start in range(0,
len(pdf), step))
+ length = len(pdf)
+ num_slices = self.sparkContext.defaultParallelism
+ pdf_slices = (pdf.iloc[i * length // num_slices: (i + 1) * length //
num_slices]
+ for i in range(0, num_slices))
Review comment:
let's use
```
start_end_pairs = ((i * length // num_slices, (i + 1) * length //
num_slices) for i in range(0, num_slices))
pdf_slices = (pdf.iloc[start, end] for start, end in start_end_pairs if
start < end)
```
to filter out empty slices (which my leads to error).
##########
File path: python/pyspark/sql/pandas/conversion.py
##########
@@ -404,8 +404,10 @@ def _create_from_pandas_with_arrow(self, pdf, schema,
timezone):
for t in pdf.dtypes]
# Slice the DataFrame to be batched
- step = -(-len(pdf) // self.sparkContext.defaultParallelism) # round
int up
- pdf_slices = (pdf.iloc[start:start + step] for start in range(0,
len(pdf), step))
+ length = len(pdf)
+ num_slices = self.sparkContext.defaultParallelism
+ pdf_slices = (pdf.iloc[i * length // num_slices: (i + 1) * length //
num_slices]
+ for i in range(0, num_slices))
Review comment:
let's use
```
start_end_pairs = ((i * length // num_slices, (i + 1) * length //
num_slices) for i in range(0, num_slices))
pdf_slices = (pdf.iloc[start, end] for start, end in start_end_pairs if
start < end)
```
to filter out empty slices (it will happen when len(pdf) < num_slices,
which my leads to error).
----------------------------------------------------------------
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:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
