Function MakeGeoArrowCrsMetadata

cpp/src/parquet/geospatial/util_json_internal.cc:89–125

std::string EscapeCrsAsJsonIfRequired(std::string_view crs);

::arrow::Result<std::string> MakeGeoArrowCrsMetadata(
    std::string_view crs,
    const std::shared_ptr<const ::arrow::KeyValueMetadata>& metadata) {
  const std::string kSridPrefix{"srid:"};
  const std::string kProjjsonPrefix{"projjson:"};

  // Two recommendations are explicitly mentioned in the Parquet format for the
  // LogicalType crs:
  //
  // - "srid:XXXX" as a way to encode an application-specific integer identifier
  // - "projjson:some_field_name" as a way to avoid repeating PROJJSON strings
  //   unnecessarily (with a suggestion to place them in the file metadata)
  //
  // While we don't currently generate those values to reduce the complexity
  // of the writer, we do interpret these values according to the suggestion in
  // the format and pass on this information to GeoArrow.
  if (crs.empty()) {
    return R"("crs": "OGC:CRS84", "crs_type": "authority_code")";
  } else if (crs.starts_with(kSridPrefix)) {
    return R"("crs": ")" + std::string(crs.substr(kSridPrefix.size())) +
           R"(", "crs_type": "srid")";
  } else if (crs.starts_with(kProjjsonPrefix)) {
    std::string_view metadata_field = crs.substr(kProjjsonPrefix.size());
    if (metadata && metadata->Contains(metadata_field)) {
      ARROW_ASSIGN_OR_RAISE(std::string projjson_value, metadata->Get(metadata_field));
      // This value should be valid JSON, but if it is not, we escape it as a string such
      // that it can be inspected by the consumer of GeoArrow.
      return R"("crs": )" + EscapeCrsAsJsonIfRequired(projjson_value) +
             R"(, "crs_type": "projjson")";
    }
  }

  // Pass on the string directly to GeoArrow. If the string is already valid JSON,
  // insert it directly into GeoArrow's "crs" field. Otherwise, escape it and pass it as a
  // string value.
  return R"("crs": )" + EscapeCrsAsJsonIfRequired(crs);
}

std::string EscapeCrsAsJsonIfRequired(std::string_view crs) {
  namespace rj = ::arrow::rapidjson;