MCPcopy Create free account
hub / github.com/apache/arrow / serialize_pandas

Function serialize_pandas

python/pyarrow/ipc.py:234–259  ·  view source on GitHub ↗

Serialize a pandas DataFrame into a buffer protocol compatible object. Parameters ---------- df : pandas.DataFrame nthreads : int, default None Number of threads to use for conversion to Arrow, default all CPUs. preserve_index : bool, default None The defaul

(df, *, nthreads=None, preserve_index=None)

Source from the content-addressed store, hash-verified

232
233
234def serialize_pandas(df, *, nthreads=None, preserve_index=None):
235 """
236 Serialize a pandas DataFrame into a buffer protocol compatible object.
237
238 Parameters
239 ----------
240 df : pandas.DataFrame
241 nthreads : int, default None
242 Number of threads to use for conversion to Arrow, default all CPUs.
243 preserve_index : bool, default None
244 The default of None will store the index as a column, except for
245 RangeIndex which is stored as metadata only. If True, always
246 preserve the pandas index data as a column. If False, no index
247 information is saved and the result will have a default RangeIndex.
248
249 Returns
250 -------
251 buf : buffer
252 An object compatible with the buffer protocol.
253 """
254 batch = pa.RecordBatch.from_pandas(df, nthreads=nthreads,
255 preserve_index=preserve_index)
256 sink = pa.BufferOutputStream()
257 with pa.RecordBatchStreamWriter(sink, batch.schema) as writer:
258 writer.write_batch(batch)
259 return sink.getvalue()
260
261
262def deserialize_pandas(buf, *, use_threads=True):

Callers

nothing calls this directly

Calls 3

BufferOutputStreamMethod · 0.80
write_batchMethod · 0.80

Tested by

no test coverage detected