mirror of
https://github.com/apache/impala.git
synced 2026-01-16 00:00:42 -05:00
bcc73a36da
This patch modifies the Parquet scanner to resolve nested schemas, and read and materialize collection types. The high-level modification is to create a CollectionColumnReader that recursively materializes map- and array-type slots. This patch also adds many tests, most of which query a new table called complextypestbl. This table contains hand-generated data that is meant to expose edge cases in the scanner. The tests mostly test the scanner, with a few tests of other functionality (e.g. array serialization). I ran a local benchmark comparing this scanner code to the original scanner code on an expanded version of tpch_parquet.lineitem with 48009720 rows. My benchmark involved selecting different numbers of columns with a single scanner thread, and I looked at the HDFS scan node time in the query profiles. This code introduces a 10%-20% regression in single-threaded scan time. Change-Id: Id27fb728934e8346444f61752c9278d8010e5f3a Reviewed-on: http://gerrit.cloudera.org:8080/576 Reviewed-by: Alex Behm <alex.behm@cloudera.com> Tested-by: Internal Jenkins
9.5 KiB
9.5 KiB
| 1 | # Table level constraints: |
|---|---|
| 2 | # Allows for defining constraints on which file formats to generate for an individual |
| 3 | # table. The table name should match the base table name defined in the schema template |
| 4 | # file. |
| 5 | table_name:stringids, constraint:restrict_to, table_format:hbase/none/none |
| 6 | table_name:hbasecolumnfamilies, constraint:restrict_to, table_format:hbase/none/none |
| 7 | table_name:insertalltypesagg, constraint:restrict_to, table_format:hbase/none/none |
| 8 | table_name:alltypessmallbinary, constraint:restrict_to, table_format:hbase/none/none |
| 9 | table_name:insertalltypesaggbinary, constraint:restrict_to, table_format:hbase/none/none |
| 10 | table_name:hbasealltypeserror, constraint:restrict_to, table_format:hbase/none/none |
| 11 | table_name:hbasealltypeserrornonulls, constraint:restrict_to, table_format:hbase/none/none |
| 12 | table_name:alltypesinsert, constraint:restrict_to, table_format:text/none/none |
| 13 | table_name:alltypesnopart_insert, constraint:restrict_to, table_format:text/none/none |
| 14 | table_name:insert_overwrite_nopart, constraint:restrict_to, table_format:text/none/none |
| 15 | table_name:insert_overwrite_partitioned, constraint:restrict_to, table_format:text/none/none |
| 16 | table_name:insert_string_partitioned, constraint:restrict_to, table_format:text/none/none |
| 17 | table_name:alltypesinsert, constraint:restrict_to, table_format:parquet/none/none |
| 18 | table_name:alltypesnopart_insert, constraint:restrict_to, table_format:parquet/none/none |
| 19 | table_name:alltypesinsert, constraint:restrict_to, table_format:text/none/none |
| 20 | table_name:alltypesnopart_insert, constraint:restrict_to, table_format:text/none/none |
| 21 | table_name:insert_overwrite_nopart, constraint:restrict_to, table_format:text/none/none |
| 22 | table_name:insert_overwrite_partitioned, constraint:restrict_to, table_format:text/none/none |
| 23 | table_name:insert_string_partitioned, constraint:restrict_to, table_format:text/none/none |
| 24 | table_name:alltypesinsert, constraint:restrict_to, table_format:parquet/none/none |
| 25 | table_name:alltypesnopart_insert, constraint:restrict_to, table_format:parquet/none/none |
| 26 | table_name:insert_overwrite_nopart, constraint:restrict_to, table_format:parquet/none/none |
| 27 | table_name:insert_overwrite_partitioned, constraint:restrict_to, table_format:parquet/none/none |
| 28 | table_name:insert_string_partitioned, constraint:restrict_to, table_format:parquet/none/none |
| 29 | table_name:old_rcfile_table, constraint:restrict_to, table_format:rc/none/none |
| 30 | table_name:bad_text_lzo, constraint:restrict_to, table_format:text/lzo/block |
| 31 | table_name:bad_text_gzip, constraint:restrict_to, table_format:text/gzip/block |
| 32 | table_name:bad_seq_snap, constraint:restrict_to, table_format:seq/snap/block |
| 33 | table_name:bad_parquet, constraint:restrict_to, table_format:parquet/none/none |
| 34 | table_name:bad_magic_number, constraint:restrict_to, table_format:parquet/none/none |
| 35 | table_name:bad_metadata_len, constraint:restrict_to, table_format:parquet/none/none |
| 36 | table_name:bad_dict_page_offset, constraint:restrict_to, table_format:parquet/none/none |
| 37 | table_name:bad_compressed_size, constraint:restrict_to, table_format:parquet/none/none |
| 38 | table_name:alltypesagg_hive_13_1, constraint:restrict_to, table_format:parquet/none/none |
| 39 | table_name:kite_required_fields, constraint:restrict_to, table_format:parquet/none/none |
| 40 | # TODO: Support Avro. Data loading currently fails for Avro because complex types |
| 41 | # cannot be converted to the corresponding Avro types yet. |
| 42 | table_name:allcomplextypes, constraint:restrict_to, table_format:text/none/none |
| 43 | table_name:allcomplextypes, constraint:restrict_to, table_format:parquet/none/none |
| 44 | table_name:allcomplextypes, constraint:restrict_to, table_format:hbase/none/none |
| 45 | table_name:functional, constraint:restrict_to, table_format:text/none/none |
| 46 | table_name:complextypes_fileformat, constraint:restrict_to, table_format:text/none/none |
| 47 | table_name:complextypes_fileformat, constraint:restrict_to, table_format:parquet/none/none |
| 48 | table_name:complextypes_fileformat, constraint:restrict_to, table_format:avro/snap/block |
| 49 | table_name:complextypes_fileformat, constraint:restrict_to, table_format:rc/snap/block |
| 50 | table_name:complextypes_fileformat, constraint:restrict_to, table_format:seq/snap/block |
| 51 | table_name:complextypes_multifileformat, constraint:restrict_to, table_format:text/none/none |
| 52 | # TODO: Avro |
| 53 | table_name:complextypestbl, constraint:restrict_to, table_format:parquet/none/none |
| 54 | table_name:alltypeserror, constraint:exclude, table_format:parquet/none/none |
| 55 | table_name:alltypeserrornonulls, constraint:exclude, table_format:parquet/none/none |
| 56 | table_name:unsupported_types, constraint:exclude, table_format:parquet/none/none |
| 57 | table_name:escapechartesttable, constraint:exclude, table_format:parquet/none/none |
| 58 | table_name:TblWithRaggedColumns, constraint:exclude, table_format:parquet/none/none |
| 59 | # the text_ tables are for testing test delimiters and escape chars in text files |
| 60 | table_name:text_comma_backslash_newline, constraint:restrict_to, table_format:text/none/none |
| 61 | table_name:text_dollar_hash_pipe, constraint:restrict_to, table_format:text/none/none |
| 62 | table_name:text_thorn_ecirc_newline, constraint:restrict_to, table_format:text/none/none |
| 63 | table_name:bad_serde, constraint:restrict_to, table_format:text/none/none |
| 64 | table_name:rcfile_lazy_binary_serde, constraint:restrict_to, table_format:rc/none/none |
| 65 | table_name:unsupported_partition_types, constraint:restrict_to, table_format:text/none/none |
| 66 | table_name:nullformat_custom, constraint:exclude, table_format:parquet/none/none |
| 67 | table_name:alltypes_view, constraint:restrict_to, table_format:text/none/none |
| 68 | table_name:allcomplextypes_view, constraint:restrict_to, table_format:text/none/none |
| 69 | table_name:alltypes_view, constraint:restrict_to, table_format:seq/snap/block |
| 70 | table_name:alltypes_hive_view, constraint:restrict_to, table_format:text/none/none |
| 71 | table_name:alltypes_view_sub, constraint:restrict_to, table_format:text/none/none |
| 72 | table_name:alltypes_view_sub, constraint:restrict_to, table_format:seq/snap/block |
| 73 | table_name:alltypes_parens, constraint:restrict_to, table_format:text/none/none |
| 74 | table_name:complex_view, constraint:restrict_to, table_format:text/none/none |
| 75 | table_name:complex_view, constraint:restrict_to, table_format:seq/snap/block |
| 76 | table_name:view_view, constraint:restrict_to, table_format:text/none/none |
| 77 | table_name:view_view, constraint:restrict_to, table_format:seq/snap/block |
| 78 | # liketbl and tblwithraggedcolumns all have |
| 79 | # NULLs in primary key columns. hbase does not support |
| 80 | # writing NULLs to primary key columns. |
| 81 | table_name:liketbl, constraint:exclude, table_format:hbase/none/none |
| 82 | table_name:tblwithraggedcolumns, constraint:exclude, table_format:hbase/none/none |
| 83 | # Tables with only one column are not supported in hbase. |
| 84 | table_name:greptiny, constraint:exclude, table_format:hbase/none/none |
| 85 | table_name:tinyinttable, constraint:exclude, table_format:hbase/none/none |
| 86 | # overflow has a bigint that's too big. hbase may lose precision, hence this |
| 87 | # table cannot be loaded. |
| 88 | table_name:overflow, constraint:exclude, table_format:hbase/none/none |
| 89 | # widerow has a single column with a single row containing a 10MB string. hbase doesn't |
| 90 | # seem to like this. |
| 91 | table_name:widerow, constraint:exclude, table_format:hbase/none/none |
| 92 | # nullformat_custom is used in null-insert tests, which user insert overwrite, |
| 93 | # which is not supported in hbase. The schema is also specified in HIVE_CREATE |
| 94 | # with no corresponding LOAD statement. |
| 95 | table_name:nullformat_custom, constraint:exclude, table_format:hbase/none/none |
| 96 | table_name:unsupported_types, constraint:exclude, table_format:hbase/none/none |
| 97 | # On CDH4, decimal can only be tested on formats Impala can write to (text and parquet) |
| 98 | # TODO: add Avro once Hive or Impala can write Avro decimals |
| 99 | table_name:decimal_tbl, constraint:restrict_to, table_format:text/none/none |
| 100 | table_name:decimal_tiny, constraint:restrict_to, table_format:text/none/none |
| 101 | table_name:decimal_tbl, constraint:restrict_to, table_format:parquet/none/none |
| 102 | table_name:decimal_tiny, constraint:restrict_to, table_format:parquet/none/none |
| 103 | table_name:avro_decimal_tbl, constraint:restrict_to, table_format:avro/snap/block |
| 104 | # TODO first set of tests are for text/none/none |
| 105 | table_name:chars_tiny, constraint:restrict_to, table_format:text/none/none |
| 106 | # invalid_decimal_part_tbl[1,2,3] tables are used for testing invalid decimal |
| 107 | # partition key values (see IMPALA-1040) |
| 108 | table_name:invalid_decimal_part_tbl1, constraint:restrict_to, table_format:text/none/none |
| 109 | table_name:invalid_decimal_part_tbl2, constraint:restrict_to, table_format:text/none/none |
| 110 | table_name:invalid_decimal_part_tbl3, constraint:restrict_to, table_format:text/none/none |
| 111 | table_name:avro_decimal_tbl, constraint:restrict_to, table_format:avro/snap/block |
| 112 | # testescape tables are used for testing text scanner delimiter handling |
| 113 | table_name:table_no_newline, constraint:restrict_to, table_format:text/none/none |
| 114 | table_name:table_no_newline_part, constraint:restrict_to, table_format:text/none/none |
| 115 | table_name:testescape_16_lf, constraint:restrict_to, table_format:text/none/none |
| 116 | table_name:testescape_16_crlf, constraint:restrict_to, table_format:text/none/none |
| 117 | table_name:testescape_17_lf, constraint:restrict_to, table_format:text/none/none |
| 118 | table_name:testescape_17_crlf, constraint:restrict_to, table_format:text/none/none |
| 119 | table_name:testescape_32_lf, constraint:restrict_to, table_format:text/none/none |
| 120 | table_name:testescape_32_crlf, constraint:restrict_to, table_format:text/none/none |
| 121 | # alltimezones is used to verify that impala properly deals with timezones |
| 122 | table_name:alltimezones, constraint:restrict_to, table_format:text/none/none |
| 123 | # Avro schema is inferred from the column definitions (IMPALA-1136) |
| 124 | table_name:no_avro_schema, constraint:restrict_to, table_format:avro/snap/block |
| 125 | table_name:avro_unicode_nulls, constraint:restrict_to, table_format:avro/snap/block |