|
1 | 1 | package operators |
2 | 2 |
|
3 | | -// This is what everything is going to be working off of |
| 3 | +import ( |
| 4 | + "fmt" |
| 5 | + "strings" |
| 6 | + |
| 7 | + "github.com/apache/arrow/go/v17/arrow" |
| 8 | + "github.com/apache/arrow/go/v17/arrow/array" |
| 9 | + "github.com/apache/arrow/go/v17/arrow/memory" |
| 10 | +) |
| 11 | + |
| 12 | +var ( |
| 13 | + ErrInvalidSchema = func(info string) error { |
| 14 | + return fmt.Errorf("invalid schema was provided. context: %s", info) |
| 15 | + } |
| 16 | +) |
| 17 | + |
| 18 | +type RecordBatch struct { |
| 19 | + Schema *arrow.Schema |
| 20 | + Columns []arrow.Array |
| 21 | +} |
| 22 | + |
| 23 | +type SchemaBuilder struct { |
| 24 | + fields []arrow.Field |
| 25 | +} |
| 26 | + |
| 27 | +type RecordBatchBuilder struct { |
| 28 | + SchemaBuilder *SchemaBuilder |
| 29 | +} |
| 30 | + |
| 31 | +func NewRecordBatchBuilder() *RecordBatchBuilder { |
| 32 | + return &RecordBatchBuilder{ |
| 33 | + SchemaBuilder: &SchemaBuilder{ |
| 34 | + fields: make([]arrow.Field, 0, 10), |
| 35 | + }, |
| 36 | + } |
| 37 | +} |
| 38 | + |
| 39 | +func (sb *SchemaBuilder) WithField(name string, dtype arrow.DataType, nullable bool) *SchemaBuilder { |
| 40 | + sb.fields = append(sb.fields, arrow.Field{ |
| 41 | + Name: name, |
| 42 | + Type: dtype, |
| 43 | + Nullable: nullable, |
| 44 | + }) |
| 45 | + return sb |
| 46 | +} |
| 47 | +func (sb *SchemaBuilder) WithoutField(names ...string) *SchemaBuilder { |
| 48 | + nameSet := make(map[string]struct{}, len(names)) |
| 49 | + for _, n := range names { |
| 50 | + nameSet[n] = struct{}{} |
| 51 | + } |
| 52 | + |
| 53 | + newFields := make([]arrow.Field, 0, len(sb.fields)) |
| 54 | + for _, field := range sb.fields { |
| 55 | + _, found := nameSet[field.Name] |
| 56 | + if !found { |
| 57 | + newFields = append(newFields, field) |
| 58 | + } |
| 59 | + } |
| 60 | + sb.fields = newFields |
| 61 | + return sb |
| 62 | + |
| 63 | +} |
| 64 | + |
| 65 | +func (sb *SchemaBuilder) Build() *arrow.Schema { |
| 66 | + return arrow.NewSchema(sb.fields, nil) |
| 67 | +} |
| 68 | +func (rbb *RecordBatchBuilder) Schema() *arrow.Schema { |
| 69 | + return arrow.NewSchema(rbb.SchemaBuilder.fields, nil) |
| 70 | +} |
| 71 | + |
| 72 | +// schema is always right in case of type mismatches |
| 73 | +func (rbb *RecordBatchBuilder) validate(schema *arrow.Schema, columns []arrow.Array) error { |
| 74 | + if len(schema.Fields()) != len(columns) { |
| 75 | + return ErrInvalidSchema("schema fields and column count do not match") |
| 76 | + } |
| 77 | + // make sure that the array data types line up with whats expected of the schema |
| 78 | + // Ensure array data types align with schema expectations. |
| 79 | + var errors []string |
| 80 | + for i := 0; i < len(columns); i++ { |
| 81 | + field := schema.Field(i) |
| 82 | + colType := columns[i].DataType() |
| 83 | + |
| 84 | + if !arrow.TypeEqual(colType, field.Type) { |
| 85 | + errors = append(errors, |
| 86 | + fmt.Sprintf("Type mismatch at position %d: column '%s' has type '%s', but schema expects '%s'.", |
| 87 | + i, field.Name, colType, field.Type)) |
| 88 | + } |
| 89 | + } |
| 90 | + if len(errors) > 0 { |
| 91 | + return ErrInvalidSchema(strings.Join(errors, " ")) |
| 92 | + } |
| 93 | + return nil |
| 94 | +} |
| 95 | +func (rbb *RecordBatchBuilder) NewRecordBatch(schema *arrow.Schema, columns []arrow.Array) (*RecordBatch, error) { |
| 96 | + if err := rbb.validate(schema, columns); err != nil { |
| 97 | + return nil, err |
| 98 | + } |
| 99 | + return &RecordBatch{ |
| 100 | + Schema: schema, |
| 101 | + Columns: columns, |
| 102 | + }, nil |
| 103 | +} |
| 104 | +func (rb *RecordBatch) DeepEqual(other *RecordBatch) bool { |
| 105 | + if !rb.Schema.Equal(other.Schema) { |
| 106 | + return false |
| 107 | + } |
| 108 | + if len(rb.Columns) != len(other.Columns) { |
| 109 | + return false |
| 110 | + } |
| 111 | + for i := 0; i < len(rb.Columns); i++ { |
| 112 | + if !array.Equal(rb.Columns[i], other.Columns[i]) { |
| 113 | + return false |
| 114 | + } |
| 115 | + } |
| 116 | + return true |
| 117 | +} |
| 118 | +func (rbb *RecordBatchBuilder) GenIntArray(values ...int) arrow.Array { |
| 119 | + mem := memory.NewGoAllocator() |
| 120 | + builder := array.NewInt32Builder(mem) |
| 121 | + defer builder.Release() |
| 122 | + for _, v := range values { |
| 123 | + builder.Append(int32(v)) |
| 124 | + } |
| 125 | + return builder.NewArray() |
| 126 | +} |
| 127 | + |
| 128 | +func (rbb *RecordBatchBuilder) GenFloatArray(values ...float64) arrow.Array { |
| 129 | + mem := memory.NewGoAllocator() |
| 130 | + builder := array.NewFloat64Builder(mem) |
| 131 | + defer builder.Release() |
| 132 | + for _, v := range values { |
| 133 | + builder.Append(v) |
| 134 | + } |
| 135 | + return builder.NewArray() |
| 136 | +} |
| 137 | + |
| 138 | +func (rbb *RecordBatchBuilder) GenStringArray(values ...string) arrow.Array { |
| 139 | + mem := memory.NewGoAllocator() |
| 140 | + builder := array.NewStringBuilder(mem) |
| 141 | + defer builder.Release() |
| 142 | + for _, v := range values { |
| 143 | + builder.Append(v) |
| 144 | + } |
| 145 | + return builder.NewArray() |
| 146 | +} |
| 147 | + |
| 148 | +func (rbb *RecordBatchBuilder) GenBoolArray(values ...bool) arrow.Array { |
| 149 | + mem := memory.NewGoAllocator() |
| 150 | + builder := array.NewBooleanBuilder(mem) |
| 151 | + defer builder.Release() |
| 152 | + for _, v := range values { |
| 153 | + builder.Append(v) |
| 154 | + } |
| 155 | + return builder.NewArray() |
| 156 | +} |
| 157 | + |
| 158 | +// GenInt8Array generates an Int8 array |
| 159 | +func (rbb *RecordBatchBuilder) GenInt8Array(values ...int8) arrow.Array { |
| 160 | + mem := memory.NewGoAllocator() |
| 161 | + builder := array.NewInt8Builder(mem) |
| 162 | + defer builder.Release() |
| 163 | + for _, v := range values { |
| 164 | + builder.Append(v) |
| 165 | + } |
| 166 | + return builder.NewArray() |
| 167 | +} |
| 168 | + |
| 169 | +// GenInt16Array generates an Int16 array |
| 170 | +func (rbb *RecordBatchBuilder) GenInt16Array(values ...int16) arrow.Array { |
| 171 | + mem := memory.NewGoAllocator() |
| 172 | + builder := array.NewInt16Builder(mem) |
| 173 | + defer builder.Release() |
| 174 | + for _, v := range values { |
| 175 | + builder.Append(v) |
| 176 | + } |
| 177 | + return builder.NewArray() |
| 178 | +} |
| 179 | + |
| 180 | +// GenInt64Array generates an Int64 array |
| 181 | +func (rbb *RecordBatchBuilder) GenInt64Array(values ...int64) arrow.Array { |
| 182 | + mem := memory.NewGoAllocator() |
| 183 | + builder := array.NewInt64Builder(mem) |
| 184 | + defer builder.Release() |
| 185 | + for _, v := range values { |
| 186 | + builder.Append(v) |
| 187 | + } |
| 188 | + return builder.NewArray() |
| 189 | +} |
| 190 | + |
| 191 | +// GenUint8Array generates a Uint8 array |
| 192 | +func (rbb *RecordBatchBuilder) GenUint8Array(values ...uint8) arrow.Array { |
| 193 | + mem := memory.NewGoAllocator() |
| 194 | + builder := array.NewUint8Builder(mem) |
| 195 | + defer builder.Release() |
| 196 | + for _, v := range values { |
| 197 | + builder.Append(v) |
| 198 | + } |
| 199 | + return builder.NewArray() |
| 200 | +} |
| 201 | + |
| 202 | +// GenUint16Array generates a Uint16 array |
| 203 | +func (rbb *RecordBatchBuilder) GenUint16Array(values ...uint16) arrow.Array { |
| 204 | + mem := memory.NewGoAllocator() |
| 205 | + builder := array.NewUint16Builder(mem) |
| 206 | + defer builder.Release() |
| 207 | + for _, v := range values { |
| 208 | + builder.Append(v) |
| 209 | + } |
| 210 | + return builder.NewArray() |
| 211 | +} |
| 212 | + |
| 213 | +// GenUint32Array generates a Uint32 array |
| 214 | +func (rbb *RecordBatchBuilder) GenUint32Array(values ...uint32) arrow.Array { |
| 215 | + mem := memory.NewGoAllocator() |
| 216 | + builder := array.NewUint32Builder(mem) |
| 217 | + defer builder.Release() |
| 218 | + for _, v := range values { |
| 219 | + builder.Append(v) |
| 220 | + } |
| 221 | + return builder.NewArray() |
| 222 | +} |
| 223 | + |
| 224 | +// GenUint64Array generates a Uint64 array |
| 225 | +func (rbb *RecordBatchBuilder) GenUint64Array(values ...uint64) arrow.Array { |
| 226 | + mem := memory.NewGoAllocator() |
| 227 | + builder := array.NewUint64Builder(mem) |
| 228 | + defer builder.Release() |
| 229 | + for _, v := range values { |
| 230 | + builder.Append(v) |
| 231 | + } |
| 232 | + return builder.NewArray() |
| 233 | +} |
| 234 | + |
| 235 | +// GenFloat32Array generates a Float32 array |
| 236 | +func (rbb *RecordBatchBuilder) GenFloat32Array(values ...float32) arrow.Array { |
| 237 | + mem := memory.NewGoAllocator() |
| 238 | + builder := array.NewFloat32Builder(mem) |
| 239 | + defer builder.Release() |
| 240 | + for _, v := range values { |
| 241 | + builder.Append(v) |
| 242 | + } |
| 243 | + return builder.NewArray() |
| 244 | +} |
| 245 | + |
| 246 | +// GenBinaryArray generates a Binary array |
| 247 | +func (rbb *RecordBatchBuilder) GenBinaryArray(values ...[]byte) arrow.Array { |
| 248 | + mem := memory.NewGoAllocator() |
| 249 | + builder := array.NewBinaryBuilder(mem, arrow.BinaryTypes.Binary) |
| 250 | + defer builder.Release() |
| 251 | + for _, v := range values { |
| 252 | + builder.Append(v) |
| 253 | + } |
| 254 | + return builder.NewArray() |
| 255 | +} |
| 256 | + |
| 257 | +// GenLargeStringArray generates a LargeString array |
| 258 | +func (rbb *RecordBatchBuilder) GenLargeStringArray(values ...string) arrow.Array { |
| 259 | + mem := memory.NewGoAllocator() |
| 260 | + builder := array.NewLargeStringBuilder(mem) |
| 261 | + defer builder.Release() |
| 262 | + for _, v := range values { |
| 263 | + builder.Append(v) |
| 264 | + } |
| 265 | + return builder.NewArray() |
| 266 | +} |
| 267 | + |
| 268 | +// GenLargeBinaryArray generates a LargeBinary array |
| 269 | +func (rbb *RecordBatchBuilder) GenLargeBinaryArray(values ...[]byte) arrow.Array { |
| 270 | + mem := memory.NewGoAllocator() |
| 271 | + builder := array.NewBinaryBuilder(mem, arrow.BinaryTypes.LargeBinary) |
| 272 | + defer builder.Release() |
| 273 | + for _, v := range values { |
| 274 | + builder.Append(v) |
| 275 | + } |
| 276 | + return builder.NewArray() |
| 277 | +} |
0 commit comments