Support JSON

cmd/scrub.go

@@ -9,6 +9,7 @@ import (
 	"github.com/spf13/cobra"
 	"github.com/xeger/sqlstream/nlp"
 	"github.com/xeger/sqlstream/scrubbing"
+	"github.com/xeger/sqlstream/scrubbing/json"
 	"github.com/xeger/sqlstream/scrubbing/mysql"
 )
 
@@ -67,13 +68,21 @@ func scrub(cmd *cobra.Command, args []string) {
 	}
 
 	switch format {
+	case "json":
+		scrubJson(models)
 	case "mysql":
 		scrubMysql(models)
 	default:
 		panic("unknown format: " + format)
 	}
 }
 
+func scrubJson(models []nlp.Model) {
+	sc := scrubbing.NewScrubber(salt, models, 0.95)
+	// TODO: parallelize JSON scrubbing (but not parsing)
+	json.Scrub(sc, os.Stdin, os.Stdout)
+}
+
 func scrubMysql(models []nlp.Model) {
 	N := parallelism
 

scrubbing/json/scrub.go

@@ -0,0 +1,19 @@
+package json
+
+import (
+	"encoding/json"
+	"io"
+
+	"github.com/xeger/sqlstream/scrubbing"
+)
+
+// Scrub sanitizes a single line, which may contain multiple SQL statements.
+func Scrub(sc *scrubbing.Scrubber, r io.Reader, w io.Writer) {
+	dec := json.NewDecoder(r)
+	enc := json.NewEncoder(w)
+	var v any
+	for err := dec.Decode(&v); err == nil; err = dec.Decode(&v) {
+		sc.ScrubData(v)
+		enc.Encode(v)
+	}
+}