[
https://issues.apache.org/jira/browse/ARROW-13438?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17598859#comment-17598859
]
Laurent Erreca edited comment on ARROW-13438 at 9/1/22 10:08 AM:
-----------------------------------------------------------------
Hi,
I had a similar issue with timestamp field :
{code:c++}
...
// Schema definition
parquet::schema::NodeVector fields;
fields.push_back(PrimitiveNode::Make("posdate", Repetition::REQUIRED,
LogicalType::Timestamp(false, LogicalType::TimeUnit::MICROS, false, false),
Type::INT64));
std::shared_ptr<GroupNode> schema =
std::static_pointer_cast<GroupNode>(GroupNode::Make("schema",
Repetition::REQUIRED, fields));
...
// Write
parquet::StreamWriter out{
parquet::ParquetFileWriter::Open(outfile, schema, builder.build())};
// posdate is a timestamp converted to int64
out << posdate; // fails here
{code}
Exception:
{code:bash}
{code}
Column converted type mismatch. Column 'posdate' has converted type[NONE] not
'INT_64'
Could be a bug in
[ColumnCheck|https://github.com/apache/arrow/blob/8e43f23dcc6a9e630516228f110c48b64d13cec6/cpp/src/parquet/stream_writer.cc#L200]
as mentionned above?
{code}
I can bypass this issue by setting isAdjustedToUTC to false and by letting
posdate as type std::chrono::microseconds, but in this case the output parquet
file metadata has converted_type set to TIMESTAMP_MICROS.
was (Author: JIRAUSER295234):
Hi,
I had a similar issue with timestamp field :
{code:c++}
...
// Schema definition
parquet::schema::NodeVector fields;
fields.push_back(PrimitiveNode::Make("posdate", Repetition::REQUIRED,
LogicalType::Timestamp(false, LogicalType::TimeUnit::MICROS, false, false),
Type::INT64));
std::shared_ptr<GroupNode> schema =
std::static_pointer_cast<GroupNode>(GroupNode::Make("schema",
Repetition::REQUIRED, fields));
...
// Write
parquet::StreamWriter out{
parquet::ParquetFileWriter::Open(outfile, schema, builder.build())};
// posdate is a timestamp converted to int64
out << posdate; // fails here
{code}
Exception:
{code:bash}
Column converted type mismatch. Column 'posdate' has converted type[NONE] not
'INT_64'
Could be a bug in
[ColumnCheck|https://github.com/apache/arrow/blob/8e43f23dcc6a9e630516228f110c48b64d13cec6/cpp/src/parquet/stream_writer.cc#L200]
as mentionned above?
{}
I can bypass this issue by setting isAdjustedToUTC to false and by letting
posdate as type std::chrono::microseconds, but in this case the output parquet
file metadata has converted_type set to TIMESTAMP_MICROS.
> [C++] Can't use StreamWriter with ToParquetSchema schema
> --------------------------------------------------------
>
> Key: ARROW-13438
> URL: https://issues.apache.org/jira/browse/ARROW-13438
> Project: Apache Arrow
> Issue Type: Bug
> Components: C++
> Affects Versions: 4.0.1
> Reporter: Vasily Fomin
> Priority: Major
>
> Hi there,
> First of all, I'm not sure if I'm doing this correctly, as it took a bit of
> reverse engineering to figure this out.
> I'm using Arrow 4.0.1 on Ubuntu with C++.
> I followed the streaming example and created:
> {code:cpp}
> #include <cassert>
> #include <chrono>
> #include <cstdint>
> #include <cstring>
> #include <ctime>
> #include <iomanip>
> #include <iostream>
> #include <utility>
> #include "arrow/io/file.h"
> #include "parquet/exception.h"
> #include "parquet/stream_reader.h"
> #include "parquet/stream_writer.h"
> std::shared_ptr<parquet::schema::GroupNode> GetSchema() {
> parquet::schema::NodeVector fields;
> fields.push_back(parquet::schema::PrimitiveNode::Make(
> "int64_field", parquet::Repetition::OPTIONAL, parquet::Type::INT64,
> parquet::ConvertedType::NONE));
> return std::static_pointer_cast<parquet::schema::GroupNode>(
> parquet::schema::GroupNode::Make("schema",
> parquet::Repetition::REQUIRED, fields));
> }
> int main() {
> std::shared_ptr<arrow::io::FileOutputStream> outfile;
> PARQUET_ASSIGN_OR_THROW(
> outfile,
>
> arrow::io::FileOutputStream::Open("parquet-stream-api-example.parquet"));
> parquet::WriterProperties::Builder builder;
> parquet::StreamWriter os{parquet::ParquetFileWriter::Open(outfile,
> GetSchema(), builder.build())};
> os << int64_t(10);
> return 0;
> }
> {code}
> The code terminates with:
> {code:java}
> terminate called after throwing an instance of 'parquet::ParquetException'
> what(): Column converted type mismatch. Column 'int64_field' has
> converted type[NONE] not 'INT_64' {code}
> What I'm not sure about is {{parquet::ConvertedType::NONE}} part. The example
> provides this value even for primitives, while it's my understanding that
> it's necessary? If I do provide it, the code works.
> Now, to the reverse engineering part. I'm trying to write to Parquet using
> {{StreamWriter}}. {{StreamWriter}} requires
> {{parquet::schema::{{GroupNode}}}} as the schema, but I begin with
> {{arrow::Schema}} I
> [found|https://github.com/apache/arrow/blob/e990d177b1f1dec962315487682f613d46be573c/cpp/src/parquet/arrow/writer.cc#L442]
> that it can be converted to {{{{parquet::SchemaDescriptor}}}} using
> {{parquet::arrow::ToParquetSchema }}utility. Looking at the utility
> [implementation|https://github.com/apache/arrow/blob/85f192a45755b3f15653fdc0a8fbd788086e125f/cpp/src/parquet/arrow/schema.cc#L322]
> I can see that {{logical_type}} is set to {{None}} which equals to
> {{parquet::ConvertedType::None}} and hence the converted schema can't be used
> due to the issue I described above.
> # Do we need to provide {{ConvertedType}} even for primitives?
> # Is it a bug in the schema conversion utility or
> [ColumnCheck|https://github.com/apache/arrow/blob/8e43f23dcc6a9e630516228f110c48b64d13cec6/cpp/src/parquet/stream_writer.cc#L200]
> assert?
> # Or is it expected behavior, in this case, what's a suggested approach?
> Build Parquet schema instead of Arrow Schema?
> Thank you,
> Vasily.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
