zeroshade commented on code in PR #330:
URL: https://github.com/apache/iceberg-go/pull/330#discussion_r1991759367
##########
table/table_test.go:
##########
@@ -128,3 +138,236 @@ func (t *TableTestSuite) TestSnapshotByName() {
t.True(testSnapshot.Equals(*t.tbl.SnapshotByName("test")))
}
+
+type TableWritingTestSuite struct {
+ suite.Suite
+
+ tableSchema *iceberg.Schema
+ arrSchema *arrow.Schema
+ arrTbl arrow.Table
+ arrSchemaWithIDs *arrow.Schema
+ arrTblWithIDs arrow.Table
+ arrSchemaUpdated *arrow.Schema
+ arrTblUpdated arrow.Table
+
+ location string
+ formatVersion int
+}
+
+func (t *TableWritingTestSuite) SetupSuite() {
+ mem := memory.DefaultAllocator
+
+ t.tableSchema = iceberg.NewSchema(0,
+ iceberg.NestedField{ID: 1, Name: "foo", Type:
iceberg.PrimitiveTypes.Bool},
+ iceberg.NestedField{ID: 2, Name: "bar", Type:
iceberg.PrimitiveTypes.String},
+ iceberg.NestedField{ID: 4, Name: "baz", Type:
iceberg.PrimitiveTypes.Int32},
+ iceberg.NestedField{ID: 10, Name: "qux", Type:
iceberg.PrimitiveTypes.Date})
+
+ t.arrSchema = arrow.NewSchema([]arrow.Field{
+ {Name: "foo", Type: arrow.FixedWidthTypes.Boolean, Nullable:
true},
+ {Name: "bar", Type: arrow.BinaryTypes.String, Nullable: true},
+ {Name: "baz", Type: arrow.PrimitiveTypes.Int32, Nullable: true},
+ {Name: "qux", Type: arrow.PrimitiveTypes.Date32, Nullable:
true},
+ }, nil)
+
+ var err error
+ t.arrTbl, err = array.TableFromJSON(mem, t.arrSchema, []string{
+ `[{"foo": true, "bar": "bar_string", "baz": 123, "qux":
"2024-03-07"}]`,
+ })
+ t.Require().NoError(err)
+
+ t.arrSchemaWithIDs = arrow.NewSchema([]arrow.Field{
+ {
+ Name: "foo", Type: arrow.FixedWidthTypes.Boolean,
+ Metadata:
arrow.MetadataFrom(map[string]string{"PARQUET:field_id": "1"}),
+ },
+ {
+ Name: "bar", Type: arrow.BinaryTypes.String,
+ Metadata:
arrow.MetadataFrom(map[string]string{"PARQUET:field_id": "2"}),
+ },
+ {
+ Name: "baz", Type: arrow.PrimitiveTypes.Int32,
+ Metadata:
arrow.MetadataFrom(map[string]string{"PARQUET:field_id": "3"}),
+ },
+ {
+ Name: "qux", Type: arrow.PrimitiveTypes.Date32,
+ Metadata:
arrow.MetadataFrom(map[string]string{"PARQUET:field_id": "4"}),
+ },
+ }, nil)
+
+ t.arrTblWithIDs, err = array.TableFromJSON(mem, t.arrSchemaWithIDs,
[]string{
+ `[{"foo": true, "bar": "bar_string", "baz": 123, "qux":
"2024-03-07"}]`,
+ })
+ t.Require().NoError(err)
+
+ t.arrSchemaUpdated = arrow.NewSchema([]arrow.Field{
+ {Name: "foo", Type: arrow.FixedWidthTypes.Boolean, Nullable:
true},
+ {Name: "baz", Type: arrow.PrimitiveTypes.Int32, Nullable: true},
+ {Name: "qux", Type: arrow.PrimitiveTypes.Date32, Nullable:
true},
+ {Name: "quux", Type: arrow.PrimitiveTypes.Int32, Nullable:
true},
+ }, nil)
+
+ t.arrTblUpdated, err = array.TableFromJSON(mem, t.arrSchemaUpdated,
[]string{
+ `[{"foo": true, "baz": 123, "qux": "2024-03-07", "quux": 234}]`,
+ })
+ t.Require().NoError(err)
+}
+
+func (t *TableWritingTestSuite) SetupTest() {
+ t.location = t.T().TempDir()
+}
+
+func (t *TableWritingTestSuite) TearDownSuite() {
+ t.arrTbl.Release()
+ t.arrTblUpdated.Release()
+ t.arrTblWithIDs.Release()
+}
+
+func (t *TableWritingTestSuite) getMetadataLoc() string {
+ return fmt.Sprintf("%s/metadata/%05d-%s.metadata.json",
+ t.location, 1, uuid.New().String())
+}
+
+func (t *TableWritingTestSuite) writeParquet(fio iceio.WriteFileIO, filePath
string, arrTbl arrow.Table) {
+ fo, err := fio.Create(filePath)
+ t.Require().NoError(err)
+
+ t.Require().NoError(pqarrow.WriteTable(arrTbl, fo, arrTbl.NumRows(),
+ nil, pqarrow.DefaultWriterProps()))
+}
+
+func (t *TableWritingTestSuite) createTable(identifier table.Identifier,
formatVersion int, spec iceberg.PartitionSpec, sc *iceberg.Schema) *table.Table
{
+ meta, err := table.NewMetadata(sc, &spec, table.UnsortedSortOrder,
+ t.getMetadataLoc(), iceberg.Properties{"format-version":
strconv.Itoa(formatVersion)})
+ t.Require().NoError(err)
+
+ return table.New(identifier, meta, t.location, iceio.LocalFS{}, nil)
+}
+
+func (t *TableWritingTestSuite) TestAddFilesUnpartitioned() {
+ ident := table.Identifier{"default", "unpartitioned_table_v" +
strconv.Itoa(t.formatVersion)}
+ tbl := t.createTable(ident, t.formatVersion,
+ *iceberg.UnpartitionedSpec, t.tableSchema)
+
+ t.NotNil(tbl)
+
+ files := make([]string, 0)
+ for i := range 5 {
+ filePath := fmt.Sprintf("%s/unpartitioned/test-%d.parquet",
t.location, i)
+ t.writeParquet(tbl.FS().(iceio.WriteFileIO), filePath, t.arrTbl)
Review Comment:
by default, the parquet writer will use -1 for the field IDs because we are
constructing them from Arrow tables which don't have the `PARQUET::field_id`
key in the metadata for each field of the schema. So the files produced for the
tests will not have field ids.
--
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]
