zeroshade commented on code in PR #1607: URL: https://github.com/apache/iceberg-go/pull/1607#discussion_r3935764418
########## table/variant_residual.go: ########## @@ -0,0 +1,316 @@ +// 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" + "log/slog" + "strconv" + "strings" + + "github.com/apache/arrow-go/v18/arrow" + "github.com/apache/arrow-go/v18/arrow/array" + "github.com/apache/arrow-go/v18/arrow/compute" + "github.com/apache/arrow-go/v18/arrow/extensions" + "github.com/apache/arrow-go/v18/arrow/memory" + "github.com/apache/iceberg-go" + "github.com/google/uuid" +) + +// augmentSchemaWithExtracts returns fileSchema plus one primitive column per variant extract term. +func augmentSchemaWithExtracts(fileSchema *iceberg.Schema, cols []iceberg.VariantExtractColumn) *iceberg.Schema { + fields := fileSchema.Fields() + for _, c := range cols { + fields = append(fields, iceberg.NestedField{ + ID: c.FieldID, + Name: c.Name, + Type: c.Term.Type().(iceberg.PrimitiveType), + }) + } + + return iceberg.NewSchemaWithIdentifiers(fileSchema.ID, fileSchema.IdentifierFieldIDs, fields...) +} + +// buildExtractColumn materializes one variant extract term into a typed Arrow array over rec. +func buildExtractColumn(col iceberg.VariantExtractColumn, rec arrow.RecordBatch, mem memory.Allocator) (arrow.Array, arrow.Field, error) { + typ := col.Term.Type().(iceberg.PrimitiveType) + dt, err := TypeToArrowType(typ, false, false) + if err != nil { + return nil, arrow.Field{}, err + } + + bldr := array.NewBuilder(mem, dt) + defer bldr.Release() + + n := int(rec.NumRows()) + varName := col.Term.Ref().Field().Name + arr := resolveVariantSource(rec, col.Term.Ref().Field().ID, col.SourcePath) + if arr == nil { + return nil, arrow.Field{}, fmt.Errorf("%w: variant extract column %q not found in file", iceberg.ErrInvalidArgument, varName) + } + varr, ok := arr.(*extensions.VariantArray) + if !ok { + return nil, arrow.Field{}, fmt.Errorf("%w: variant extract column %q is not a VariantArray (got %T)", iceberg.ErrInvalidArgument, varName, arr) + } + + for i := range n { + if varr.IsNull(i) { + bldr.AppendNull() + + continue + } + + v, verr := varr.Value(i) + if verr != nil { + slog.Warn("variant extract: skipping undecodable variant value", "column", varName, "row", i, "err", verr) + bldr.AppendNull() + + continue + } + + lit, ok := col.Term.ExtractValue(v) + if !ok { + bldr.AppendNull() + + continue + } + + if aerr := appendExtractLiteral(bldr, lit); aerr != nil { + return nil, arrow.Field{}, aerr + } + } + + field := arrow.Field{ + Name: col.Name, + Type: dt, + Nullable: true, + Metadata: arrow.NewMetadata([]string{ArrowParquetFieldIDKey}, []string{strconv.Itoa(col.FieldID)}), + } + + return bldr.NewArray(), field, nil +} + +// resolveVariantSource locates the extract's source array by field id, descending +// nested structs; it falls back to the file-schema path when field ids are absent. +func resolveVariantSource(rec arrow.RecordBatch, fieldID int, sourcePath string) arrow.Array { + for i, f := range rec.Schema().Fields() { + if a := descendByFieldID(f, rec.Column(i), fieldID); a != nil { + return a + } + } + if sourcePath == "" { + return nil + } + + return descendByPath(rec.Schema(), rec.Columns(), strings.Split(sourcePath, ".")) +} + +func descendByFieldID(f arrow.Field, col arrow.Array, fieldID int) arrow.Array { + if v, ok := f.Metadata.GetValue(ArrowParquetFieldIDKey); ok { + if id, err := strconv.Atoi(v); err == nil && id == fieldID { + return col + } Review Comment: **minor** — Nested field-id descent is not pinned by any test and appears redundant Deleting the entire struct-descent loop in descendByFieldID leaves the full table package green, including TestBuildExtractColumnNested whose assertion message claims 'nested variant resolved by descending the struct' but which actually resolves via the descendByPath fallback. Because fileSchema is derived from the same pruned Arrow schema the record batches carry, the path walk always aligns, so the field-id recursion has no exercised code path. Either add a test that only the field-id descent can satisfy (field ids present, path unresolvable) or drop the recursion and keep descendByPath. ########## visitors.go: ########## @@ -802,54 +884,79 @@ const sanitizedLiteralMask = "(redacted)" // predicates (IN / NOT IN) keep their arity so the operation is not // misrepresented, but the members are masked. func SanitizeExpression(expr BooleanExpression) (BooleanExpression, error) { - return VisitExpr(expr, sanitizeVisitor{}) + res, err := VisitExpr(expr, sanitizeVisitor{}) + if err != nil { + return nil, err + } + + return res.expr, nil +} + +// sanitizedResult is the masked expression plus whether its subtree held a non-serializable term. +type sanitizedResult struct { + expr BooleanExpression + hasUnserializable bool Review Comment: **nit** — NOT over a mixed subtree discards the serializable conjuncts from the ScanReport filter VisitNot collapses the whole subtree to AlwaysTrue whenever any descendant was unserializable, so NOT(AND(extract=5, x<10)) reports AlwaysTrue and the surviving x<10 never reaches ScanReport.Filter. The direction is safe (over-broad, never AlwaysFalse) and matches the bbox convention, but the emitted report can under-report the executed filter. Worth a one-line comment on the tradeoff, or push the collapse down to the offending leaf and negate the rest. ########## table/variant_residual.go: ########## @@ -0,0 +1,316 @@ +// 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" + "log/slog" + "strconv" + "strings" + + "github.com/apache/arrow-go/v18/arrow" + "github.com/apache/arrow-go/v18/arrow/array" + "github.com/apache/arrow-go/v18/arrow/compute" + "github.com/apache/arrow-go/v18/arrow/extensions" + "github.com/apache/arrow-go/v18/arrow/memory" + "github.com/apache/iceberg-go" + "github.com/google/uuid" +) + +// augmentSchemaWithExtracts returns fileSchema plus one primitive column per variant extract term. +func augmentSchemaWithExtracts(fileSchema *iceberg.Schema, cols []iceberg.VariantExtractColumn) *iceberg.Schema { + fields := fileSchema.Fields() + for _, c := range cols { + fields = append(fields, iceberg.NestedField{ + ID: c.FieldID, + Name: c.Name, + Type: c.Term.Type().(iceberg.PrimitiveType), + }) + } + + return iceberg.NewSchemaWithIdentifiers(fileSchema.ID, fileSchema.IdentifierFieldIDs, fields...) +} + +// buildExtractColumn materializes one variant extract term into a typed Arrow array over rec. +func buildExtractColumn(col iceberg.VariantExtractColumn, rec arrow.RecordBatch, mem memory.Allocator) (arrow.Array, arrow.Field, error) { + typ := col.Term.Type().(iceberg.PrimitiveType) + dt, err := TypeToArrowType(typ, false, false) + if err != nil { + return nil, arrow.Field{}, err + } + + bldr := array.NewBuilder(mem, dt) + defer bldr.Release() + + n := int(rec.NumRows()) + varName := col.Term.Ref().Field().Name + arr := resolveVariantSource(rec, col.Term.Ref().Field().ID, col.SourcePath) + if arr == nil { + return nil, arrow.Field{}, fmt.Errorf("%w: variant extract column %q not found in file", iceberg.ErrInvalidArgument, varName) + } + varr, ok := arr.(*extensions.VariantArray) + if !ok { + return nil, arrow.Field{}, fmt.Errorf("%w: variant extract column %q is not a VariantArray (got %T)", iceberg.ErrInvalidArgument, varName, arr) + } + + for i := range n { + if varr.IsNull(i) { + bldr.AppendNull() + + continue + } + + v, verr := varr.Value(i) + if verr != nil { + slog.Warn("variant extract: skipping undecodable variant value", "column", varName, "row", i, "err", verr) + bldr.AppendNull() + + continue + } + + lit, ok := col.Term.ExtractValue(v) + if !ok { + bldr.AppendNull() + + continue + } + + if aerr := appendExtractLiteral(bldr, lit); aerr != nil { + return nil, arrow.Field{}, aerr + } + } + + field := arrow.Field{ + Name: col.Name, + Type: dt, + Nullable: true, + Metadata: arrow.NewMetadata([]string{ArrowParquetFieldIDKey}, []string{strconv.Itoa(col.FieldID)}), + } + + return bldr.NewArray(), field, nil +} + +// resolveVariantSource locates the extract's source array by field id, descending +// nested structs; it falls back to the file-schema path when field ids are absent. +func resolveVariantSource(rec arrow.RecordBatch, fieldID int, sourcePath string) arrow.Array { + for i, f := range rec.Schema().Fields() { + if a := descendByFieldID(f, rec.Column(i), fieldID); a != nil { + return a + } + } + if sourcePath == "" { + return nil + } Review Comment: **minor** — strings.Split on '.' misresolves Variant columns whose Iceberg name contains a dot resolveVariantSource re-splits the flattened column name returned by FindColumnName, so a legal Iceberg column literally named 'a.b' is looked up as top-level 'a' then child 'b'. On a name-mapped (field-id-less) file this aborts the scan. Resolve by walking the Iceberg schema's nested field structure (carrying the path segments through from FindColumnName) instead of re-splitting the flattened name. -- 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]
