S3Select: Handle array selection in from clause (#9076)

pkg/s3select/select.go

@@ -392,7 +392,6 @@ func (s3Select *S3Select) Evaluate(w http.ResponseWriter) {
 	}
 	writer := newMessageWriter(w, getProgressFunc)
 
-	var inputRecord sql.Record
 	var outputQueue []sql.Record
 
 	// Create queue based on the type.
@@ -433,6 +432,7 @@ func (s3Select *S3Select) Evaluate(w http.ResponseWriter) {
 	}
 
 	var rec sql.Record
+OuterLoop:
 	for {
 		if s3Select.statement.LimitReached() {
 			if !sendRecord() {
@@ -469,13 +469,15 @@ func (s3Select *S3Select) Evaluate(w http.ResponseWriter) {
 			break
 		}
 
-		if inputRecord, err = s3Select.statement.EvalFrom(s3Select.Input.format, rec); err != nil {
+		var inputRecords []*sql.Record
+		if inputRecords, err = s3Select.statement.EvalFrom(s3Select.Input.format, rec); err != nil {
 			break
 		}
 
+		for _, inputRecord := range inputRecords {
 			if s3Select.statement.IsAggregated() {
-			if err = s3Select.statement.AggregateRow(inputRecord); err != nil {
-				break
+				if err = s3Select.statement.AggregateRow(*inputRecord); err != nil {
+					break OuterLoop
 				}
 			} else {
 				var outputRecord sql.Record
@@ -492,13 +494,13 @@ func (s3Select *S3Select) Evaluate(w http.ResponseWriter) {
 					outputRecord = s3Select.outputRecord()
 					outputQueue[len(outputQueue)-1] = outputRecord
 				}
-			outputRecord, err = s3Select.statement.Eval(inputRecord, outputRecord)
+				outputRecord, err = s3Select.statement.Eval(*inputRecord, outputRecord)
 				if outputRecord == nil || err != nil {
 					// This should not be written.
 					// Remove it from the queue.
 					outputQueue = outputQueue[:len(outputQueue)-1]
 					if err != nil {
-					break
+						break OuterLoop
 					}
 					continue
 				}
@@ -509,7 +511,8 @@ func (s3Select *S3Select) Evaluate(w http.ResponseWriter) {
 				}
 
 				if !sendRecord() {
-				break
+					break OuterLoop
+				}
 			}
 		}
 	}

pkg/s3select/select_test.go

@@ -260,6 +260,68 @@ func TestJSONQueries(t *testing.T) {
 </SelectObjectContentRequest>`),
 			wantResult: `"[""foo"",""bar"",""whatever""]"`,
 		},
+		{
+			name:  "document",
+			query: "",
+			requestXML: []byte(`
+<?xml version="1.0" encoding="UTF-8"?>
+<SelectObjectContentRequest>
+    <Expression>select * from s3object[*].elements[*] s where s.element_type = '__elem__merfu'</Expression>
+    <ExpressionType>SQL</ExpressionType>
+    <InputSerialization>
+        <CompressionType>NONE</CompressionType>
+        <JSON>
+            <Type>DOCUMENT</Type>
+        </JSON>
+    </InputSerialization>
+    <OutputSerialization>
+        <JSON>
+        </JSON>
+    </OutputSerialization>
+    <RequestProgress>
+        <Enabled>FALSE</Enabled>
+    </RequestProgress>
+</SelectObjectContentRequest>`),
+			withJSON: `
+{
+  "name": "small_pdf1.pdf",
+  "lume_id": "9507193e-572d-4f95-bcf1-e9226d96be65",
+  "elements": [
+    {
+      "element_type": "__elem__image",
+      "element_id": "859d09c4-7cf1-4a37-9674-3a7de8b56abc",
+      "attributes": {
+        "__attr__image_dpi": 300,
+        "__attr__image_size": [
+          2550,
+          3299
+        ],
+        "__attr__image_index": 1,
+        "__attr__image_format": "JPEG",
+        "__attr__file_extension": "jpg",
+        "__attr__data": null
+      }
+    },
+    {
+      "element_type": "__elem__merfu",
+      "element_id": "d868aefe-ef9a-4be2-b9b2-c9fd89cc43eb",
+      "attributes": {
+        "__attr__image_dpi": 300,
+        "__attr__image_size": [
+          2550,
+          3299
+        ],
+        "__attr__image_index": 2,
+        "__attr__image_format": "JPEG",
+        "__attr__file_extension": "jpg",
+        "__attr__data": null
+      }
+    }
+  ],
+  "data": "asdascasdc1234e123erdasdas"
+}`,
+			wantResult: `{"element_type":"__elem__merfu","element_id":"d868aefe-ef9a-4be2-b9b2-c9fd89cc43eb","attributes":{"__attr__image_dpi":300,"__attr__image_size":[2550,3299],"__attr__image_index":2,"__attr__image_format":"JPEG","__attr__file_extension":"jpg","__attr__data":null}}`,
+		},
 	}
 
 	defRequest := `<?xml version="1.0" encoding="UTF-8"?>

pkg/s3select/sql/statement.go

@@ -140,9 +140,9 @@ func parseLimit(v *LitValue) (int64, error) {
 
 // EvalFrom evaluates the From clause on the input record. It only
 // applies to JSON input data format (currently).
-func (e *SelectStatement) EvalFrom(format string, input Record) (Record, error) {
+func (e *SelectStatement) EvalFrom(format string, input Record) ([]*Record, error) {
 	if !e.selectAST.From.HasKeypath() {
-		return input, nil
+		return []*Record{&input}, nil
 	}
 	_, rawVal := input.Raw()
 
@@ -160,6 +160,18 @@ func (e *SelectStatement) EvalFrom(format string, input Record) (Record, error)
 		switch v := txedRec.(type) {
 		case jstream.KVS:
 			kvs = v
+
+		case []interface{}:
+			recs := make([]*Record, len(v))
+			for i, val := range v {
+				tmpRec := input.Clone(nil)
+				if err = tmpRec.Replace(val); err != nil {
+					return nil, err
+				}
+				recs[i] = &tmpRec
+			}
+			return recs, nil
+
 		default:
 			kvs = jstream.KVS{jstream.KV{Key: "_1", Value: v}}
 		}
@@ -168,7 +180,7 @@ func (e *SelectStatement) EvalFrom(format string, input Record) (Record, error)
 			return nil, err
 		}
 
-		return input, nil
+		return []*Record{&input}, nil
 	case simdjson.Object:
 		txedRec, _, err := jsonpathEval(e.selectAST.From.Table.PathExpr[1:], rec)
 		if err != nil {
@@ -181,6 +193,18 @@ func (e *SelectStatement) EvalFrom(format string, input Record) (Record, error)
 			if err != nil {
 				return nil, err
 			}
+
+		case []interface{}:
+			recs := make([]*Record, len(v))
+			for i, val := range v {
+				tmpRec := input.Clone(nil)
+				if err = tmpRec.Replace(val); err != nil {
+					return nil, err
+				}
+				recs[i] = &tmpRec
+			}
+			return recs, nil
+
 		default:
 			input.Reset()
 			input, err = input.Set("_1", &Value{value: v})
@@ -188,7 +212,7 @@ func (e *SelectStatement) EvalFrom(format string, input Record) (Record, error)
 				return nil, err
 			}
 		}
-		return input, nil
+		return []*Record{&input}, nil
 	}
 	return nil, errDataSource(errors.New("unexpected non JSON input"))
 }