asadbek064/hyparquet
1
Fork 0
mirror of https://github.com/asadbek064/hyparquet.git synced 2025-12-26 07:06:38 +00:00
Code Issues Actions Projects Releases Wiki Activity
4e9fbc095b
hyparquet/test/files/nonnullable.impala.metadata.json
Kenny Daniel 57ed66646d
Convert statistics based on column type
2024-05-04 01:11:46 -07:00

520 lines
14 KiB
JSON
Raw Blame History

{
"version": 1,
"created_by": "parquet-mr version 1.8.0 (build 0fda28af84b9746396014ad6a415b90592a98b3b)",
"key_value_metadata": [
{
"key": "parquet.avro.schema",
"value": "{\"type\":\"record\",\"name\":\"ComplexTypesTbl\",\"namespace\":\"org.apache.impala\",\"fields\":[{\"name\":\"ID\",\"type\":\"long\"},{\"name\":\"Int_Array\",\"type\":{\"type\":\"array\",\"items\":\"int\"}},{\"name\":\"int_array_array\",\"type\":{\"type\":\"array\",\"items\":{\"type\":\"array\",\"items\":\"int\"}}},{\"name\":\"Int_Map\",\"type\":{\"type\":\"map\",\"values\":\"int\"}},{\"name\":\"int_map_array\",\"type\":{\"type\":\"array\",\"items\":{\"type\":\"map\",\"values\":\"int\"}}},{\"name\":\"nested_Struct\",\"type\":{\"type\":\"record\",\"name\":\"r1\",\"fields\":[{\"name\":\"a\",\"type\":\"int\"},{\"name\":\"B\",\"type\":{\"type\":\"array\",\"items\":\"int\"}},{\"name\":\"c\",\"type\":{\"type\":\"record\",\"name\":\"r2\",\"fields\":[{\"name\":\"D\",\"type\":{\"type\":\"array\",\"items\":{\"type\":\"array\",\"items\":{\"type\":\"record\",\"name\":\"r3\",\"fields\":[{\"name\":\"e\",\"type\":\"int\"},{\"name\":\"f\",\"type\":\"string\"}]}}}}]}},{\"name\":\"G\",\"type\":{\"type\":\"map\",\"values\":{\"type\":\"record\",\"name\":\"r4\",\"fields\":[{\"name\":\"h\",\"type\":{\"type\":\"record\",\"name\":\"r5\",\"fields\":[{\"name\":\"i\",\"type\":{\"type\":\"array\",\"items\":\"double\"}}]}}]}}}]}}]}"
}
],
"metadata_length": 2544,
"num_rows": 1,
"row_groups": [
{
"columns": [
{
"file_offset": 4,
"meta_data": {
"codec": "UNCOMPRESSED",
"data_page_offset": 4,
"encodings": [ "PLAIN", "BIT_PACKED" ],
"num_values": 1,
"path_in_schema": [ "ID" ],
"statistics": {
"max": 8,
"min": 8,
"null_count": 0
},
"total_compressed_size": 49,
"total_uncompressed_size": 49,
"type": "INT64"
}
},
{
"file_offset": 53,
"meta_data": {
"codec": "UNCOMPRESSED",
"data_page_offset": 53,
"encodings": [ "PLAIN", "RLE" ],
"num_values": 1,
"path_in_schema": [ "Int_Array", "list", "element" ],
"statistics": {
"max": -1,
"min": -1,
"null_count": 0
},
"total_compressed_size": 49,
"total_uncompressed_size": 49,
"type": "INT32"
}
},
{
"file_offset": 102,
"meta_data": {
"codec": "UNCOMPRESSED",
"data_page_offset": 102,
"encodings": [ "PLAIN", "RLE" ],
"num_values": 3,
"path_in_schema": [
"int_array_array",
"list",
"element",
"list",
"element"
],
"statistics": {
"max": -1,
"min": -2,
"null_count": 1
},
"total_compressed_size": 55,
"total_uncompressed_size": 55,
"type": "INT32"
}
},
{
"file_offset": 157,
"meta_data": {
"codec": "UNCOMPRESSED",
"data_page_offset": 157,
"encodings": [ "PLAIN", "RLE" ],
"num_values": 1,
"path_in_schema": [ "Int_Map", "map", "key" ],
"statistics": {
"max": "k1",
"min": "k1",
"null_count": 0
},
"total_compressed_size": 47,
"total_uncompressed_size": 47,
"type": "BYTE_ARRAY"
}
},
{
"file_offset": 204,
"meta_data": {
"codec": "UNCOMPRESSED",
"data_page_offset": 204,
"encodings": [ "PLAIN", "RLE" ],
"num_values": 1,
"path_in_schema": [ "Int_Map", "map", "value" ],
"statistics": {
"max": -1,
"min": -1,
"null_count": 0
},
"total_compressed_size": 49,
"total_uncompressed_size": 49,
"type": "INT32"
}
},
{
"file_offset": 253,
"meta_data": {
"codec": "UNCOMPRESSED",
"data_page_offset": 253,
"encodings": [ "PLAIN", "RLE" ],
"num_values": 4,
"path_in_schema": [
"int_map_array",
"list",
"element",
"map",
"key"
],
"statistics": {
"max": "k1",
"min": "k1",
"null_count": 3
},
"total_compressed_size": 49,
"total_uncompressed_size": 49,
"type": "BYTE_ARRAY"
}
},
{
"file_offset": 302,
"meta_data": {
"codec": "UNCOMPRESSED",
"data_page_offset": 302,
"encodings": [ "PLAIN", "RLE" ],
"num_values": 4,
"path_in_schema": [
"int_map_array",
"list",
"element",
"map",
"value"
],
"statistics": {
"max": 1,
"min": 1,
"null_count": 3
},
"total_compressed_size": 51,
"total_uncompressed_size": 51,
"type": "INT32"
}
},
{
"file_offset": 353,
"meta_data": {
"codec": "UNCOMPRESSED",
"data_page_offset": 353,
"encodings": [ "PLAIN", "BIT_PACKED" ],
"num_values": 1,
"path_in_schema": [ "nested_Struct", "a" ],
"statistics": {
"max": -1,
"min": -1,
"null_count": 0
},
"total_compressed_size": 37,
"total_uncompressed_size": 37,
"type": "INT32"
}
},
{
"file_offset": 390,
"meta_data": {
"codec": "UNCOMPRESSED",
"data_page_offset": 390,
"encodings": [ "PLAIN", "RLE" ],
"num_values": 1,
"path_in_schema": [ "nested_Struct", "B", "list", "element" ],
"statistics": {
"max": -1,
"min": -1,
"null_count": 0
},
"total_compressed_size": 49,
"total_uncompressed_size": 49,
"type": "INT32"
}
},
{
"file_offset": 439,
"meta_data": {
"codec": "UNCOMPRESSED",
"data_page_offset": 439,
"encodings": [ "PLAIN", "RLE" ],
"num_values": 1,
"path_in_schema": [
"nested_Struct",
"c",
"D",
"list",
"element",
"list",
"element",
"e"
],
"statistics": {
"max": -1,
"min": -1,
"null_count": 0
},
"total_compressed_size": 51,
"total_uncompressed_size": 51,
"type": "INT32"
}
},
{
"file_offset": 490,
"meta_data": {
"codec": "UNCOMPRESSED",
"data_page_offset": 490,
"encodings": [ "PLAIN", "RLE" ],
"num_values": 1,
"path_in_schema": [
"nested_Struct",
"c",
"D",
"list",
"element",
"list",
"element",
"f"
],
"statistics": {
"max": "nonnullable",
"min": "nonnullable",
"null_count": 0
},
"total_compressed_size": 76,
"total_uncompressed_size": 76,
"type": "BYTE_ARRAY"
}
},
{
"file_offset": 566,
"meta_data": {
"codec": "UNCOMPRESSED",
"data_page_offset": 566,
"encodings": [ "PLAIN", "RLE" ],
"num_values": 1,
"path_in_schema": [ "nested_Struct", "G", "map", "key" ],
"statistics": {
"null_count": 1
},
"total_compressed_size": 33,
"total_uncompressed_size": 33,
"type": "BYTE_ARRAY"
}
},
{
"file_offset": 599,
"meta_data": {
"codec": "UNCOMPRESSED",
"data_page_offset": 599,
"encodings": [ "PLAIN", "RLE" ],
"num_values": 1,
"path_in_schema": [
"nested_Struct",
"G",
"map",
"value",
"h",
"i",
"list",
"element"
],
"statistics": {
"null_count": 1
},
"total_compressed_size": 35,
"total_uncompressed_size": 35,
"type": "DOUBLE"
}
}
],
"num_rows": 1,
"total_byte_size": 630
}
],
"schema": [
{
"name": "org.apache.impala.ComplexTypesTbl",
"num_children": 6
},
{
"name": "ID",
"repetition_type": "REQUIRED",
"type": "INT64"
},
{
"converted_type": "LIST",
"name": "Int_Array",
"num_children": 1,
"repetition_type": "REQUIRED"
},
{
"name": "list",
"num_children": 1,
"repetition_type": "REPEATED"
},
{
"name": "element",
"repetition_type": "REQUIRED",
"type": "INT32"
},
{
"converted_type": "LIST",
"name": "int_array_array",
"num_children": 1,
"repetition_type": "REQUIRED"
},
{
"name": "list",
"num_children": 1,
"repetition_type": "REPEATED"
},
{
"converted_type": "LIST",
"name": "element",
"num_children": 1,
"repetition_type": "REQUIRED"
},
{
"name": "list",
"num_children": 1,
"repetition_type": "REPEATED"
},
{
"name": "element",
"repetition_type": "REQUIRED",
"type": "INT32"
},
{
"converted_type": "MAP",
"name": "Int_Map",
"num_children": 1,
"repetition_type": "REQUIRED"
},
{
"converted_type": "MAP_KEY_VALUE",
"name": "map",
"num_children": 2,
"repetition_type": "REPEATED"
},
{
"converted_type": "UTF8",
"name": "key",
"repetition_type": "REQUIRED",
"type": "BYTE_ARRAY"
},
{
"name": "value",
"repetition_type": "REQUIRED",
"type": "INT32"
},
{
"converted_type": "LIST",
"name": "int_map_array",
"num_children": 1,
"repetition_type": "REQUIRED"
},
{
"name": "list",
"num_children": 1,
"repetition_type": "REPEATED"
},
{
"converted_type": "MAP",
"name": "element",
"num_children": 1,
"repetition_type": "REQUIRED"
},
{
"converted_type": "MAP_KEY_VALUE",
"name": "map",
"num_children": 2,
"repetition_type": "REPEATED"
},
{
"converted_type": "UTF8",
"name": "key",
"repetition_type": "REQUIRED",
"type": "BYTE_ARRAY"
},
{
"name": "value",
"repetition_type": "REQUIRED",
"type": "INT32"
},
{
"name": "nested_Struct",
"num_children": 4,
"repetition_type": "REQUIRED"
},
{
"name": "a",
"repetition_type": "REQUIRED",
"type": "INT32"
},
{
"converted_type": "LIST",
"name": "B",
"num_children": 1,
"repetition_type": "REQUIRED"
},
{
"name": "list",
"num_children": 1,
"repetition_type": "REPEATED"
},
{
"name": "element",
"repetition_type": "REQUIRED",
"type": "INT32"
},
{
"name": "c",
"num_children": 1,
"repetition_type": "REQUIRED"
},
{
"converted_type": "LIST",
"name": "D",
"num_children": 1,
"repetition_type": "REQUIRED"
},
{
"name": "list",
"num_children": 1,
"repetition_type": "REPEATED"
},
{
"converted_type": "LIST",
"name": "element",
"num_children": 1,
"repetition_type": "REQUIRED"
},
{
"name": "list",
"num_children": 1,
"repetition_type": "REPEATED"
},
{
"name": "element",
"num_children": 2,
"repetition_type": "REQUIRED"
},
{
"name": "e",
"repetition_type": "REQUIRED",
"type": "INT32"
},
{
"converted_type": "UTF8",
"name": "f",
"repetition_type": "REQUIRED",
"type": "BYTE_ARRAY"
},
{
"converted_type": "MAP",
"name": "G",
"num_children": 1,
"repetition_type": "REQUIRED"
},
{
"converted_type": "MAP_KEY_VALUE",
"name": "map",
"num_children": 2,
"repetition_type": "REPEATED"
},
{
"converted_type": "UTF8",
"name": "key",
"repetition_type": "REQUIRED",
"type": "BYTE_ARRAY"
},
{
"name": "value",
"num_children": 1,
"repetition_type": "REQUIRED"
},
{
"name": "h",
"num_children": 1,
"repetition_type": "REQUIRED"
},
{
"converted_type": "LIST",
"name": "i",
"num_children": 1,
"repetition_type": "REQUIRED"
},
{
"name": "list",
"num_children": 1,
"repetition_type": "REPEATED"
},
{
"name": "element",
"repetition_type": "REQUIRED",
"type": "DOUBLE"
}
]
}
Reference in New Issue View Git Blame Copy Permalink