Function IsColumnChunkFullyDictionaryEncoded

cpp/src/parquet/file_reader.cc:63–88

namespace {
bool IsColumnChunkFullyDictionaryEncoded(const ColumnChunkMetaData& col) {
  // Check the encoding_stats to see if all data pages are dictionary encoded.
  const std::vector<PageEncodingStats>& encoding_stats = col.encoding_stats();
  if (encoding_stats.empty()) {
    // Some parquet files may have empty encoding_stats. In this case we are
    // not sure whether all data pages are dictionary encoded.
    return false;
  }
  // The 1st page should be the dictionary page.
  if (encoding_stats[0].page_type != PageType::DICTIONARY_PAGE ||
      (encoding_stats[0].encoding != Encoding::PLAIN &&
       encoding_stats[0].encoding != Encoding::PLAIN_DICTIONARY)) {
    return false;
  }
  // The following pages should be dictionary encoded data pages.
  for (size_t idx = 1; idx < encoding_stats.size(); ++idx) {
    if (!IsDictionaryIndexEncoding(encoding_stats[idx].encoding) ||
        (encoding_stats[idx].page_type != PageType::DATA_PAGE &&
         encoding_stats[idx].page_type != PageType::DATA_PAGE_V2)) {
      // Return false if any following page is not a dictionary encoded data
      // page.
      return false;
    }
  }
  return true;
}
}  // namespace

static constexpr uint32_t kFooterSize = 8;