add quotedValuesAreStrings option

bodkin.go

@@ -19,13 +19,14 @@ type (
 
 // Bodkin is a collection of field paths, describing the columns of a structured input(s).
 type Bodkin struct {
-	original       *fieldPos
-	old            *fieldPos
-	new            *fieldPos
-	inferTimeUnits bool
-	typeConversion bool
-	err            error
-	changes        error
+	original               *fieldPos
+	old                    *fieldPos
+	new                    *fieldPos
+	inferTimeUnits         bool
+	quotedValuesAreStrings bool
+	typeConversion         bool
+	err                    error
+	changes                error
 }
 
 // NewBodkin returns a new Bodkin value from a structured input.
@@ -106,6 +107,13 @@ func WithTypeConversion() Option {
 	}
 }
 
+// WithTypeConversion enables upgrading the column types to fix compatibilty conflicts.
+func WithQuotedValuesAreStrings() Option {
+	return func(cfg config) {
+		cfg.quotedValuesAreStrings = true
+	}
+}
+
 // Unify merges structured input's column definition with the previously input's schema.
 // Any uppopulated fields, empty objects or empty slices in JSON input are skipped.
 func (u *Bodkin) Unify(a any) {

cmd/main.go

@@ -104,7 +104,8 @@ var jsonS1 string = `{
 	"results": [{"id":7594}],
 	"arrayscalar":[],
 	"datefield":"1979-01-01",
-	"timefield":"01:02:03"
+	"timefield":"01:02:03",
+	"boolquotedfield":"true"
 	}`
 
 var jsonS2 string = `{

schema.go

@@ -44,14 +44,19 @@ var UpgradableTypes []arrow.Type = []arrow.Type{arrow.INT8,
 	arrow.STRING,
 }
 
+// Regular expressions and variables for type inference.
 var (
 	timestampMatchers []*regexp.Regexp
 	dateMatcher       *regexp.Regexp
 	timeMatcher       *regexp.Regexp
+	integerMatcher    *regexp.Regexp
+	floatMatcher      *regexp.Regexp
+	boolMatcher       []string
 )
 
 func init() {
 	registerTsMatchers()
+	registerQuotedStringValueMatchers()
 }
 
 func registerTsMatchers() {
@@ -64,6 +69,12 @@ func registerTsMatchers() {
 		regexp.MustCompile(`^\d{4}-\d{1,2}-\d{1,2}[T ]\d{1,2}:\d{1,2}:\d{1,2}(\.\d{1,6})? *(([+-]\d{1,2}(:\d{1,2})?)|Z|UTC)?$`))
 }
 
+func registerQuotedStringValueMatchers() {
+	integerMatcher = regexp.MustCompile(`^[-+]?\d+$`)
+	floatMatcher = regexp.MustCompile(`^[-+]?(?:\d+\.?\d*|\.\d+)(?:[eE][-+]?\d+)?$`)
+	boolMatcher = append(boolMatcher, "true", "false")
+}
+
 func newFieldPos(b *Bodkin) *fieldPos {
 	f := new(fieldPos)
 	f.owner = b
@@ -352,6 +363,17 @@ func goType2Arrow(f *fieldPos, gt any) arrow.DataType {
 				return arrow.FixedWidthTypes.Time64ns
 			}
 		}
+		if !f.owner.quotedValuesAreStrings {
+			if slices.Contains(boolMatcher, t) {
+				return arrow.FixedWidthTypes.Boolean
+			}
+			if integerMatcher.MatchString(t) {
+				return arrow.PrimitiveTypes.Int64
+			}
+			if floatMatcher.MatchString(t) {
+				return arrow.PrimitiveTypes.Float64
+			}
+		}
 		dt = arrow.BinaryTypes.String
 	case []byte:
 		dt = arrow.BinaryTypes.Binary