Encode mixed numeric/alphanumeric data segments as one alphanumeric d…

…ata segment if that's more efficient. Also add error check to NewWithForcedVersion(): Check if the content is too large for the requested QR code version.

encoder.go

@@ -172,7 +172,7 @@ func (d *dataEncoder) encode(data []byte) (*bitset.Bitset, error) {
 	}
 
 	// Classify data into unoptimised segments.
-	d.classifyDataModes()
+	highestRequiredMode := d.classifyDataModes()
 
 	// Optimise segments.
 	err := d.optimiseDataModes()
@@ -190,13 +190,13 @@ func (d *dataEncoder) encode(data []byte) (*bitset.Bitset, error) {
 		optimizedLength += length
 	}
 
-	singleByteSegmentLength, err := d.encodedLength(dataModeByte, len(d.data))
+	singleByteSegmentLength, err := d.encodedLength(highestRequiredMode, len(d.data))
 	if err != nil {
 		return nil, err
 	}
 
 	if singleByteSegmentLength <= optimizedLength {
-		d.optimised = []segment{segment{dataMode: dataModeByte, data: d.data}}
+		d.optimised = []segment{segment{dataMode: highestRequiredMode, data: d.data}}
 	}
 
 	// Encode data.
@@ -211,9 +211,15 @@ func (d *dataEncoder) encode(data []byte) (*bitset.Bitset, error) {
 // classifyDataModes classifies the raw data into unoptimised segments.
 // e.g. "123ZZ#!#!" =>
 // [numeric, 3, "123"] [alphanumeric, 2, "ZZ"] [byte, 4, "#!#!"].
-func (d *dataEncoder) classifyDataModes() {
+//
+// Returns the highest data mode needed to encode the data. e.g. for a mixed
+// numeric/alphanumeric input, the highest is alphanumeric.
+//
+// dataModeNone < dataModeNumeric < dataModeAlphanumeric < dataModeByte
+func (d *dataEncoder) classifyDataModes() dataMode {
 	var start int
 	mode := dataModeNone
+	highestRequiredMode := mode
 
 	for i, v := range d.data {
 		newMode := dataModeNone
@@ -236,9 +242,15 @@ func (d *dataEncoder) classifyDataModes() {
 
 			mode = newMode
 		}
+
+		if newMode > highestRequiredMode {
+			highestRequiredMode = newMode
+		}
 	}
 
 	d.actual = append(d.actual, segment{dataMode: mode, data: d.data[start:len(d.data)]})
+
+	return highestRequiredMode
 }
 
 // optimiseDataModes optimises the list of segments to reduce the overall output

encoder_test.go

@@ -267,6 +267,57 @@ func TestOptimiseEncoding(t *testing.T) {
 				{dataModeByte, 8},
 			},
 		},
+		// HTTPS://ABC.DE/Q/393AABB6998877XYZ0518AUQCRVJN25
+		// AAAAAAAAAAAAAAAAANNNAAAANNNNNNNAAANNNNAAAAAAAANN
+		// different to below---------^--------------------
+		{
+			dataEncoderType1To9,
+			[]testModeSegment{
+				{dataModeAlphanumeric, 17},
+				{dataModeNumeric, 3},
+				{dataModeAlphanumeric, 4},
+				{dataModeNumeric, 7},
+				{dataModeAlphanumeric, 3},
+				{dataModeNumeric, 4},
+				{dataModeAlphanumeric, 8},
+				{dataModeNumeric, 2},
+			},
+			[]testModeSegment{
+				{dataModeAlphanumeric, 48},
+			},
+		},
+		// HTTPS://ABC.DE/Q/393AABB699E877XYZ0518AUQCRVJN25
+		// AAAAAAAAAAAAAAAAANNNAAAANNNANNNAAANNNNAAAAAAAANN
+		// different to above---------^--------------------
+		{
+			dataEncoderType1To9,
+			[]testModeSegment{
+				{dataModeAlphanumeric, 17},
+				{dataModeNumeric, 3},
+				{dataModeAlphanumeric, 4},
+				{dataModeNumeric, 3},
+				{dataModeAlphanumeric, 1},
+				{dataModeNumeric, 3},
+				{dataModeAlphanumeric, 3},
+				{dataModeNumeric, 4},
+				{dataModeAlphanumeric, 8},
+				{dataModeNumeric, 2},
+			},
+			[]testModeSegment{
+				{dataModeAlphanumeric, 48},
+			},
+		},
+		// 0123456789
+		// NNNNNNNNNN
+		{
+			dataEncoderType1To9,
+			[]testModeSegment{
+				{dataModeNumeric, 10},
+			},
+			[]testModeSegment{
+				{dataModeNumeric, 10},
+			},
+		},
 	}
 
 	for _, test := range tests {

qrcode.go

@@ -233,6 +233,13 @@ func NewWithForcedVersion(content string, version int, level RecoveryLevel) (*QR
 		return nil, errors.New("cannot find QR Code version")
 	}
 
+	if encoded.Len() > chosenVersion.numDataBits() {
+		return nil, fmt.Errorf("Cannot encode QR code: content too large for fixed size QR Code version %d (encoded length is %d bits, maximum length is %d bits)",
+			version,
+			encoded.Len(),
+			chosenVersion.numDataBits())
+	}
+
 	q := &QRCode{
 		Content: content,