badalprasadsingh commented on code in PR #2487: URL: https://github.com/apache/iceberg-go/pull/2487#discussion_r4595040513
########## table/identity_partition_projection_internal_test.go: ########## @@ -0,0 +1,472 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information +// regarding copyright ownership. The ASF licenses this file +// to you under the Apache License, Version 2.0 (the +// "License"); you may not use this file except in compliance +// with the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, +// software distributed under the License is distributed on an +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +// KIND, either express or implied. See the License for the +// specific language governing permissions and limitations +// under the License. + +package table + +import ( + "context" + "fmt" + "iter" + "os" + "path/filepath" + "sort" + "testing" + + "github.com/apache/arrow-go/v18/arrow" + "github.com/apache/arrow-go/v18/arrow/array" + "github.com/apache/arrow-go/v18/arrow/decimal128" + "github.com/apache/arrow-go/v18/arrow/memory" + "github.com/apache/arrow-go/v18/parquet/pqarrow" + "github.com/apache/iceberg-go" + iceio "github.com/apache/iceberg-go/io" + "github.com/stretchr/testify/require" +) + +// A writer may leave an identity partition column out of its data files, as +// Delta and Hive-style layouts do. The spec's column projection reads such a +// field from the file's partition value (rule 1), required or not. +func TestScanFillsIdentityPartitionColumnMissingFromDataFile(t *testing.T) { + ctx := context.Background() + location := filepath.ToSlash(t.TempDir()) + schema := iceberg.NewSchema(0, + iceberg.NestedField{ID: 1, Name: "id", Type: iceberg.PrimitiveTypes.Int64, Required: true}, + iceberg.NestedField{ID: 2, Name: "region", Type: iceberg.PrimitiveTypes.String, Required: true}, + ) + spec := iceberg.NewPartitionSpec(iceberg.PartitionField{ + SourceIDs: []int{2}, FieldID: 1000, Name: "region", Transform: iceberg.IdentityTransform{}, + }) + meta, err := NewMetadata(schema, &spec, UnsortedSortOrder, location, nil) + require.NoError(t, err) + tbl := New(Identifier{"db", "identity"}, meta, location+"/metadata/v1.metadata.json", + func(context.Context) (iceio.IO, error) { return iceio.LocalFS{}, nil }, &countingCatalog{metadata: meta}) + + // Each file stores only id; its region lives in the manifest. + idOnly := arrow.NewSchema([]arrow.Field{{ + Name: "id", Type: arrow.PrimitiveTypes.Int64, + Metadata: arrow.NewMetadata([]string{ArrowParquetFieldIDKey}, []string{"1"}), + }}, nil) + files := make([]iceberg.DataFile, 0, 2) + for _, part := range []struct { + region string + ids string + }{{"east", `[{"id":1},{"id":2}]`}, {"west", `[{"id":3}]`}} { + data, err := array.TableFromJSON(memory.DefaultAllocator, idOnly, []string{part.ids}) + require.NoError(t, err) + path := location + "/data/" + part.region + ".parquet" + require.NoError(t, os.MkdirAll(filepath.Dir(path), 0o755)) + f, err := os.Create(path) + require.NoError(t, err) + require.NoError(t, pqarrow.WriteTable(data, f, 1024, nil, pqarrow.DefaultWriterProps())) + info, err := os.Stat(path) + require.NoError(t, err) + + b, err := iceberg.NewDataFileBuilder(spec, iceberg.EntryContentData, path, iceberg.ParquetFile, + map[int]any{1000: part.region}, nil, nil, data.NumRows(), info.Size()) + require.NoError(t, err) + files = append(files, b.Build()) + data.Release() + } + txn := tbl.NewTransaction() + require.NoError(t, txn.AddDataFiles(ctx, files, nil)) + tbl, err = txn.Commit(ctx) + require.NoError(t, err) + + result, err := tbl.Scan().ToArrowTable(ctx) + require.NoError(t, err) + defer result.Release() + + got := map[int64]string{} + reader := array.NewTableReader(result, -1) + defer reader.Release() + for reader.Next() { + rec := reader.RecordBatch() + ids := rec.Column(0).(*array.Int64) + regions := rec.Column(1).(*array.String) + for i := range int(rec.NumRows()) { + got[ids.Value(i)] = regions.Value(i) + } + } + require.Equal(t, map[int64]string{1: "east", 2: "east", 3: "west"}, got) + + // A filter on the partition column prunes by it and still fills it. + filtered, err := tbl.Scan(WithRowFilter(iceberg.EqualTo(iceberg.Reference("region"), "west"))).ToArrowTable(ctx) + require.NoError(t, err) + defer filtered.Release() + require.EqualValues(t, 1, filtered.NumRows()) +} + +// Partition values come back from the manifest as literals; a decimal one +// must still fill the column rather than fail the scan. +func TestScanFillsDecimalIdentityPartitionColumnMissingFromDataFile(t *testing.T) { + ctx := context.Background() + location := filepath.ToSlash(t.TempDir()) + schema := iceberg.NewSchema(0, + iceberg.NestedField{ID: 1, Name: "id", Type: iceberg.PrimitiveTypes.Int64, Required: true}, + iceberg.NestedField{ID: 2, Name: "amount", Type: iceberg.MustDecimalTypeOf(9, 2)}, + ) + spec := iceberg.NewPartitionSpec(iceberg.PartitionField{ + SourceIDs: []int{2}, FieldID: 1000, Name: "amount", Transform: iceberg.IdentityTransform{}, + }) + meta, err := NewMetadata(schema, &spec, UnsortedSortOrder, location, nil) + require.NoError(t, err) + tbl := New(Identifier{"db", "identity_decimal"}, meta, location+"/metadata/v1.metadata.json", + func(context.Context) (iceio.IO, error) { return iceio.LocalFS{}, nil }, &countingCatalog{metadata: meta}) + + idOnly := arrow.NewSchema([]arrow.Field{{ + Name: "id", Type: arrow.PrimitiveTypes.Int64, + Metadata: arrow.NewMetadata([]string{ArrowParquetFieldIDKey}, []string{"1"}), + }}, nil) + data, err := array.TableFromJSON(memory.DefaultAllocator, idOnly, []string{`[{"id":1},{"id":2}]`}) + require.NoError(t, err) + defer data.Release() + path := location + "/data/amount.parquet" + require.NoError(t, os.MkdirAll(filepath.Dir(path), 0o755)) + f, err := os.Create(path) + require.NoError(t, err) + require.NoError(t, pqarrow.WriteTable(data, f, 1024, nil, pqarrow.DefaultWriterProps())) + info, err := os.Stat(path) + require.NoError(t, err) + + value := iceberg.Decimal{Val: decimal128.FromI64(1234), Scale: 2} + b, err := iceberg.NewDataFileBuilder(spec, iceberg.EntryContentData, path, iceberg.ParquetFile, + map[int]any{1000: value}, nil, nil, data.NumRows(), info.Size()) + require.NoError(t, err) + txn := tbl.NewTransaction() + require.NoError(t, txn.AddDataFiles(ctx, []iceberg.DataFile{b.Build()}, nil)) + tbl, err = txn.Commit(ctx) + require.NoError(t, err) + + result, err := tbl.Scan().ToArrowTable(ctx) + require.NoError(t, err) + defer result.Release() + require.EqualValues(t, 2, result.NumRows()) + for _, chunk := range result.Column(1).Data().Chunks() { + amounts := chunk.(*array.Decimal128) + for i := range amounts.Len() { + require.False(t, amounts.IsNull(i)) + require.Equal(t, "12.34", amounts.ValueStr(i)) + } + } +} + +// identityPartitionTestTable returns an empty table at a fresh location. +func identityPartitionTestTable(t *testing.T, schema *iceberg.Schema, spec iceberg.PartitionSpec, props iceberg.Properties) *Table { + t.Helper() + + location := filepath.ToSlash(t.TempDir()) + meta, err := NewMetadata(schema, &spec, UnsortedSortOrder, location, props) + require.NoError(t, err) + + return New(Identifier{"db", "identity"}, meta, location+"/metadata/v1.metadata.json", + func(context.Context) (iceio.IO, error) { return iceio.LocalFS{}, nil }, &countingCatalog{metadata: meta}) +} + +// addIDOnlyFile commits a data file that stores only the id column (field ID +// 1), with partition as its partition data under the table's current spec. +func addIDOnlyFile(t *testing.T, tbl *Table, name string, partition map[int]any, ids string) *Table { + t.Helper() + + idOnly := arrow.NewSchema([]arrow.Field{{ + Name: "id", Type: arrow.PrimitiveTypes.Int64, + Metadata: arrow.NewMetadata([]string{ArrowParquetFieldIDKey}, []string{"1"}), + }}, nil) + data, err := array.TableFromJSON(memory.DefaultAllocator, idOnly, []string{ids}) + require.NoError(t, err) + defer data.Release() + + path := tbl.Location() + "/data/" + name + ".parquet" + require.NoError(t, os.MkdirAll(filepath.Dir(path), 0o755)) + f, err := os.Create(path) + require.NoError(t, err) + require.NoError(t, pqarrow.WriteTable(data, f, 1024, nil, pqarrow.DefaultWriterProps())) + info, err := os.Stat(path) + require.NoError(t, err) + + b, err := iceberg.NewDataFileBuilder(tbl.Spec(), iceberg.EntryContentData, path, iceberg.ParquetFile, + partition, nil, nil, data.NumRows(), info.Size()) + require.NoError(t, err) + if tbl.Metadata().FormatVersion() >= 3 { + b.FirstRowID(tbl.Metadata().NextRowID()) + } + txn := tbl.NewTransaction() + require.NoError(t, txn.AddDataFiles(t.Context(), []iceberg.DataFile{b.Build()}, nil)) + tbl, err = txn.Commit(t.Context()) + require.NoError(t, err) + + return tbl +} + +// scanIDAndColumn returns each row of a two-column (id, value) scan as +// "id:value", sorted, with "id:<null>" for a null value. +func scanIDAndColumn(t *testing.T, tbl *Table, opts ...ScanOption) []string { + t.Helper() + + _, records, err := tbl.Scan(opts...).ToArrowRecords(t.Context()) + require.NoError(t, err) + + return idAndColumnRows(t, records) Review Comment: without the widening, each record holds a "float32" column while the returned schema says "float64". "ValueStr" still prints 1.5 -> thus `TestScanWidensPromotedIdentityPartitionColumnMissingFromDataFile` passes -> "ToArrowTable" used to catch this better: can "idAndColumnRows" check each record against the returned schema? ########## table/identity_partition_projection_internal_test.go: ########## @@ -0,0 +1,472 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information +// regarding copyright ownership. The ASF licenses this file +// to you under the Apache License, Version 2.0 (the +// "License"); you may not use this file except in compliance +// with the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, +// software distributed under the License is distributed on an +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +// KIND, either express or implied. See the License for the +// specific language governing permissions and limitations +// under the License. + +package table + +import ( + "context" + "fmt" + "iter" + "os" + "path/filepath" + "sort" + "testing" + + "github.com/apache/arrow-go/v18/arrow" + "github.com/apache/arrow-go/v18/arrow/array" + "github.com/apache/arrow-go/v18/arrow/decimal128" + "github.com/apache/arrow-go/v18/arrow/memory" + "github.com/apache/arrow-go/v18/parquet/pqarrow" + "github.com/apache/iceberg-go" + iceio "github.com/apache/iceberg-go/io" + "github.com/stretchr/testify/require" +) + +// A writer may leave an identity partition column out of its data files, as +// Delta and Hive-style layouts do. The spec's column projection reads such a +// field from the file's partition value (rule 1), required or not. +func TestScanFillsIdentityPartitionColumnMissingFromDataFile(t *testing.T) { + ctx := context.Background() + location := filepath.ToSlash(t.TempDir()) + schema := iceberg.NewSchema(0, + iceberg.NestedField{ID: 1, Name: "id", Type: iceberg.PrimitiveTypes.Int64, Required: true}, + iceberg.NestedField{ID: 2, Name: "region", Type: iceberg.PrimitiveTypes.String, Required: true}, + ) + spec := iceberg.NewPartitionSpec(iceberg.PartitionField{ + SourceIDs: []int{2}, FieldID: 1000, Name: "region", Transform: iceberg.IdentityTransform{}, + }) + meta, err := NewMetadata(schema, &spec, UnsortedSortOrder, location, nil) + require.NoError(t, err) + tbl := New(Identifier{"db", "identity"}, meta, location+"/metadata/v1.metadata.json", + func(context.Context) (iceio.IO, error) { return iceio.LocalFS{}, nil }, &countingCatalog{metadata: meta}) + + // Each file stores only id; its region lives in the manifest. + idOnly := arrow.NewSchema([]arrow.Field{{ + Name: "id", Type: arrow.PrimitiveTypes.Int64, + Metadata: arrow.NewMetadata([]string{ArrowParquetFieldIDKey}, []string{"1"}), + }}, nil) + files := make([]iceberg.DataFile, 0, 2) + for _, part := range []struct { + region string + ids string + }{{"east", `[{"id":1},{"id":2}]`}, {"west", `[{"id":3}]`}} { + data, err := array.TableFromJSON(memory.DefaultAllocator, idOnly, []string{part.ids}) + require.NoError(t, err) + path := location + "/data/" + part.region + ".parquet" + require.NoError(t, os.MkdirAll(filepath.Dir(path), 0o755)) + f, err := os.Create(path) + require.NoError(t, err) + require.NoError(t, pqarrow.WriteTable(data, f, 1024, nil, pqarrow.DefaultWriterProps())) + info, err := os.Stat(path) + require.NoError(t, err) + + b, err := iceberg.NewDataFileBuilder(spec, iceberg.EntryContentData, path, iceberg.ParquetFile, + map[int]any{1000: part.region}, nil, nil, data.NumRows(), info.Size()) + require.NoError(t, err) + files = append(files, b.Build()) + data.Release() + } + txn := tbl.NewTransaction() + require.NoError(t, txn.AddDataFiles(ctx, files, nil)) + tbl, err = txn.Commit(ctx) + require.NoError(t, err) + + result, err := tbl.Scan().ToArrowTable(ctx) + require.NoError(t, err) + defer result.Release() + + got := map[int64]string{} + reader := array.NewTableReader(result, -1) + defer reader.Release() + for reader.Next() { + rec := reader.RecordBatch() + ids := rec.Column(0).(*array.Int64) + regions := rec.Column(1).(*array.String) + for i := range int(rec.NumRows()) { + got[ids.Value(i)] = regions.Value(i) + } + } + require.Equal(t, map[int64]string{1: "east", 2: "east", 3: "west"}, got) + + // A filter on the partition column prunes by it and still fills it. + filtered, err := tbl.Scan(WithRowFilter(iceberg.EqualTo(iceberg.Reference("region"), "west"))).ToArrowTable(ctx) + require.NoError(t, err) + defer filtered.Release() + require.EqualValues(t, 1, filtered.NumRows()) +} + +// Partition values come back from the manifest as literals; a decimal one +// must still fill the column rather than fail the scan. +func TestScanFillsDecimalIdentityPartitionColumnMissingFromDataFile(t *testing.T) { + ctx := context.Background() + location := filepath.ToSlash(t.TempDir()) + schema := iceberg.NewSchema(0, + iceberg.NestedField{ID: 1, Name: "id", Type: iceberg.PrimitiveTypes.Int64, Required: true}, + iceberg.NestedField{ID: 2, Name: "amount", Type: iceberg.MustDecimalTypeOf(9, 2)}, + ) + spec := iceberg.NewPartitionSpec(iceberg.PartitionField{ + SourceIDs: []int{2}, FieldID: 1000, Name: "amount", Transform: iceberg.IdentityTransform{}, + }) + meta, err := NewMetadata(schema, &spec, UnsortedSortOrder, location, nil) + require.NoError(t, err) + tbl := New(Identifier{"db", "identity_decimal"}, meta, location+"/metadata/v1.metadata.json", + func(context.Context) (iceio.IO, error) { return iceio.LocalFS{}, nil }, &countingCatalog{metadata: meta}) + + idOnly := arrow.NewSchema([]arrow.Field{{ + Name: "id", Type: arrow.PrimitiveTypes.Int64, + Metadata: arrow.NewMetadata([]string{ArrowParquetFieldIDKey}, []string{"1"}), + }}, nil) + data, err := array.TableFromJSON(memory.DefaultAllocator, idOnly, []string{`[{"id":1},{"id":2}]`}) + require.NoError(t, err) + defer data.Release() + path := location + "/data/amount.parquet" + require.NoError(t, os.MkdirAll(filepath.Dir(path), 0o755)) + f, err := os.Create(path) + require.NoError(t, err) + require.NoError(t, pqarrow.WriteTable(data, f, 1024, nil, pqarrow.DefaultWriterProps())) + info, err := os.Stat(path) + require.NoError(t, err) + + value := iceberg.Decimal{Val: decimal128.FromI64(1234), Scale: 2} + b, err := iceberg.NewDataFileBuilder(spec, iceberg.EntryContentData, path, iceberg.ParquetFile, + map[int]any{1000: value}, nil, nil, data.NumRows(), info.Size()) + require.NoError(t, err) + txn := tbl.NewTransaction() + require.NoError(t, txn.AddDataFiles(ctx, []iceberg.DataFile{b.Build()}, nil)) + tbl, err = txn.Commit(ctx) + require.NoError(t, err) + + result, err := tbl.Scan().ToArrowTable(ctx) + require.NoError(t, err) + defer result.Release() + require.EqualValues(t, 2, result.NumRows()) + for _, chunk := range result.Column(1).Data().Chunks() { + amounts := chunk.(*array.Decimal128) + for i := range amounts.Len() { + require.False(t, amounts.IsNull(i)) + require.Equal(t, "12.34", amounts.ValueStr(i)) + } + } +} + +// identityPartitionTestTable returns an empty table at a fresh location. +func identityPartitionTestTable(t *testing.T, schema *iceberg.Schema, spec iceberg.PartitionSpec, props iceberg.Properties) *Table { + t.Helper() + + location := filepath.ToSlash(t.TempDir()) + meta, err := NewMetadata(schema, &spec, UnsortedSortOrder, location, props) + require.NoError(t, err) + + return New(Identifier{"db", "identity"}, meta, location+"/metadata/v1.metadata.json", + func(context.Context) (iceio.IO, error) { return iceio.LocalFS{}, nil }, &countingCatalog{metadata: meta}) +} + +// addIDOnlyFile commits a data file that stores only the id column (field ID +// 1), with partition as its partition data under the table's current spec. +func addIDOnlyFile(t *testing.T, tbl *Table, name string, partition map[int]any, ids string) *Table { + t.Helper() + + idOnly := arrow.NewSchema([]arrow.Field{{ + Name: "id", Type: arrow.PrimitiveTypes.Int64, + Metadata: arrow.NewMetadata([]string{ArrowParquetFieldIDKey}, []string{"1"}), + }}, nil) + data, err := array.TableFromJSON(memory.DefaultAllocator, idOnly, []string{ids}) + require.NoError(t, err) + defer data.Release() + + path := tbl.Location() + "/data/" + name + ".parquet" + require.NoError(t, os.MkdirAll(filepath.Dir(path), 0o755)) + f, err := os.Create(path) + require.NoError(t, err) + require.NoError(t, pqarrow.WriteTable(data, f, 1024, nil, pqarrow.DefaultWriterProps())) + info, err := os.Stat(path) + require.NoError(t, err) + + b, err := iceberg.NewDataFileBuilder(tbl.Spec(), iceberg.EntryContentData, path, iceberg.ParquetFile, + partition, nil, nil, data.NumRows(), info.Size()) + require.NoError(t, err) + if tbl.Metadata().FormatVersion() >= 3 { + b.FirstRowID(tbl.Metadata().NextRowID()) + } + txn := tbl.NewTransaction() + require.NoError(t, txn.AddDataFiles(t.Context(), []iceberg.DataFile{b.Build()}, nil)) + tbl, err = txn.Commit(t.Context()) + require.NoError(t, err) + + return tbl +} + +// scanIDAndColumn returns each row of a two-column (id, value) scan as +// "id:value", sorted, with "id:<null>" for a null value. +func scanIDAndColumn(t *testing.T, tbl *Table, opts ...ScanOption) []string { + t.Helper() + + _, records, err := tbl.Scan(opts...).ToArrowRecords(t.Context()) + require.NoError(t, err) + + return idAndColumnRows(t, records) +} + +// idAndColumnRows formats each row of two-column (id, value) records as +// "id:value", sorted, with "id:<null>" for a null value. +func idAndColumnRows(t *testing.T, records iter.Seq2[arrow.RecordBatch, error]) []string { + t.Helper() + + rows := []string{} + for rec, err := range records { + require.NoError(t, err) + ids := rec.Column(0).(*array.Int64) + for i := range int(rec.NumRows()) { + value := "<null>" + if !rec.Column(1).IsNull(i) { + value = rec.Column(1).ValueStr(i) + } + rows = append(rows, fmt.Sprintf("%d:%s", ids.Value(i), value)) + } + rec.Release() + } + sort.Strings(rows) + + return rows +} + +// regionTestTable has id 1 in both partitions, in files that store only id. +func regionTestTable(t *testing.T, props iceberg.Properties) *Table { + t.Helper() + + schema := iceberg.NewSchema(0, + iceberg.NestedField{ID: 1, Name: "id", Type: iceberg.PrimitiveTypes.Int64, Required: true}, + iceberg.NestedField{ID: 2, Name: "region", Type: iceberg.PrimitiveTypes.String, Required: true}, + ) + spec := iceberg.NewPartitionSpec(iceberg.PartitionField{ + SourceIDs: []int{2}, FieldID: 1000, Name: "region", Transform: iceberg.IdentityTransform{}, + }) + tbl := identityPartitionTestTable(t, schema, spec, props) + tbl = addIDOnlyFile(t, tbl, "east", map[int]any{1000: "east"}, `[{"id":1},{"id":2}]`) + + return addIDOnlyFile(t, tbl, "west", map[int]any{1000: "west"}, `[{"id":1},{"id":3}]`) +} Review Comment: cow Delete and Overwrite pass without the fold: each file has its own commit, so east never gets rewritten, aloo, nothing covers v3 cow, which needs the row-group pruning fold. better: let "regionTestTable" add both files in one "AddDataFiles" call, and the delete test also cover v3 and region = 'east'? -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
