pitrou commented on code in PR #35758:
URL: https://github.com/apache/arrow/pull/35758#discussion_r1211813681
##########
cpp/src/parquet/metadata.cc:
##########
@@ -1462,40 +1462,44 @@ class
ColumnChunkMetaDataBuilder::ColumnChunkMetaDataBuilderImpl {
column_chunk_->meta_data.__set_total_compressed_size(compressed_size);
std::vector<format::Encoding::type> thrift_encodings;
- if (has_dictionary) {
-
thrift_encodings.push_back(ToThrift(properties_->dictionary_index_encoding()));
- if (properties_->version() == ParquetVersion::PARQUET_1_0) {
- thrift_encodings.push_back(ToThrift(Encoding::PLAIN));
- } else {
-
thrift_encodings.push_back(ToThrift(properties_->dictionary_page_encoding()));
- }
- } else { // Dictionary not enabled
-
thrift_encodings.push_back(ToThrift(properties_->encoding(column_->path())));
- }
- thrift_encodings.push_back(ToThrift(Encoding::RLE));
- // Only PLAIN encoding is supported for fallback in V1
- // TODO(majetideepak): Use user specified encoding for V2
- if (dictionary_fallback) {
- thrift_encodings.push_back(ToThrift(Encoding::PLAIN));
- }
- column_chunk_->meta_data.__set_encodings(thrift_encodings);
std::vector<format::PageEncodingStats> thrift_encoding_stats;
+ auto add_encoding = [&thrift_encodings](format::Encoding::type value) {
+ auto it = std::find(thrift_encodings.begin(), thrift_encodings.end(),
value);
+ if (it == thrift_encodings.end()) {
+ thrift_encodings.push_back(value);
+ }
+ };
// Add dictionary page encoding stats
- for (const auto& entry : dict_encoding_stats) {
- format::PageEncodingStats dict_enc_stat;
- dict_enc_stat.__set_page_type(format::PageType::DICTIONARY_PAGE);
- dict_enc_stat.__set_encoding(ToThrift(entry.first));
- dict_enc_stat.__set_count(entry.second);
- thrift_encoding_stats.push_back(dict_enc_stat);
+ if (has_dictionary) {
+ for (const auto& entry : dict_encoding_stats) {
+ format::PageEncodingStats dict_enc_stat;
+ dict_enc_stat.__set_page_type(format::PageType::DICTIONARY_PAGE);
+ // Dictionary Encoding would be PLAIN_DICTIONARY in v1 and
+ // PLAIN in v2.
+ format::Encoding::type dict_encoding = ToThrift(entry.first);
+ dict_enc_stat.__set_encoding(dict_encoding);
+ dict_enc_stat.__set_count(entry.second);
+ thrift_encoding_stats.push_back(dict_enc_stat);
+ add_encoding(dict_encoding);
+ }
}
+ // Force add encoding for RL/DL.
+ // BIT_PACKED is supported in `LevelEncoder`, but would only be used
+ // in benchmark and testing.
+ // And even when RL/DL is empty(`column_->max_definition_level() == 0`),
+ // RLE would be added here.
Review Comment:
```suggestion
// And for now, we always add RLE even if there are no levels at all,
// while parquet-mr is more fine-grained.
```
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
