optimize read csv

benchmark/src/test/java/com/alibaba/fastjson2/benchmark/CSVPersonTest.java

@@ -15,8 +15,8 @@ public class CSVPersonTest {
             long millis = System.currentTimeMillis() - start;
             System.out.println("fastjson2 millis : " + millis);
             // zulu8.68.0.21 : 3621 3270 2788 2424 2431 2123
-            // zulu11.62.17 : 2576 1848 2155 1567 1522
+            // zulu11.62.17 : 2576 1848 2155 1567 1522 1307
-            // zulu17.40.19 : 3501 3134 2586 2788 2733 2682
+            // zulu17.40.19 : 3501 3134 2586 2788 2733 2682 2494
         }
     }
 

core/src/main/java/com/alibaba/fastjson2/stream/JSONStreamReaderUTF8.java

@@ -106,7 +106,7 @@ class JSONStreamReaderUTF8<T>
             }
 
             if (!lineTerminated) {
-                if (input != null && !inputEnd) {
+                if (input != null && !(inputEnd && off >= end)) {
                     int len = end - off;
                     if (off > 0) {
                         if (len > 0) {
@@ -145,7 +145,7 @@ class JSONStreamReaderUTF8<T>
 
     public <T> T readLineObject() {
         try {
-            if (inputEnd) {
+            if (inputEnd && off >= end) {
                 return null;
             }
 

core/src/main/java/com/alibaba/fastjson2/stream/StreamReader.java

@@ -584,7 +584,7 @@ public abstract class StreamReader<T> {
                 throw new IllegalArgumentException("action must not be null");
             }
             T object = next();
-            if (streamReader.inputEnd || object == null) {
+            if ((streamReader.inputEnd && streamReader.off >= streamReader.end) || object == null) {
                 return false;
             } else {
                 action.accept(object);
@@ -611,7 +611,7 @@ public abstract class StreamReader<T> {
 
         @Override
         public long estimateSize() {
-            return streamReader.inputEnd ? 0L : Long.MAX_VALUE;
+            return (streamReader.inputEnd && streamReader.off >= streamReader.end) ? 0L : Long.MAX_VALUE;
         }
 
         @Override

core/src/main/java/com/alibaba/fastjson2/support/csv/CSVReader.java

@@ -345,7 +345,7 @@ public abstract class CSVReader<T>
         if (!objectSupport) {
             throw new UnsupportedOperationException("this method should not be called, try specify objectClass or method readLineValues instead ?");
         }
-        if (inputEnd) {
+        if (inputEnd && off == end) {
             return null;
         }
 

core/src/main/java/com/alibaba/fastjson2/support/csv/CSVReaderUTF8.java

@@ -24,6 +24,9 @@ import static com.alibaba.fastjson2.util.DateUtils.DEFAULT_ZONE_ID;
 
 final class CSVReaderUTF8<T>
         extends CSVReader<T> {
+    static final int ESCAPE_INDEX_NOT_SET = -2;
+    int nextEscapeIndex = ESCAPE_INDEX_NOT_SET;
+
     static final Map<Long, Function<Consumer, ByteArrayValueConsumer>> valueConsumerCreators
             = new ConcurrentHashMap<>();
 
@@ -107,34 +110,99 @@ final class CSVReaderUTF8<T>
         return ((x22 | x0a | x0d) & 0x8080808080808080L) != 0;
     }
 
-    protected boolean seekLine() throws IOException {
+    private byte[] getBuf() throws IOException {
         byte[] buf = this.buf;
-        int off = this.off;
+        if (buf != null) {
-        if (buf == null) {
+            return buf;
-            if (input != null) {
+        }
-                buf = this.buf = new byte[SIZE_512K];
+        return initBuf();
-                int cnt = input.read(buf);
+    }
+
+    private byte[] initBuf() throws IOException {
+        if (input == null) {
+            throw new JSONException("init buf error");
+        }
+        byte[] buf = new byte[SIZE_512K];
+        int end = 0;
+        while (end < buf.length) {
+            int cnt = input.read(buf, off, buf.length - off);
             if (cnt == -1) {
                 inputEnd = true;
-                    return false;
+                break;
             }
-                this.end = cnt;
+            end += cnt;
-
+        }
+        this.end = end;
         if (end > 4 && IOUtils.isUTF8BOM(buf, 0)) {
             off = 3;
-                    lineNextStart = off;
+            lineNextStart = 3;
-                }
         }
+        this.buf = buf;
+        return buf;
     }
 
+    protected boolean seekLine() throws IOException {
+        byte[] buf = getBuf();
+        int off = this.off, end = this.end;
+
+        int slashIndex = this.nextEscapeIndex;
+        if (slashIndex == ESCAPE_INDEX_NOT_SET || (slashIndex != -1 && slashIndex < off)) {
+            nextEscapeIndex = slashIndex = IOUtils.indexOfDoubleQuote(buf, off, end);
+        }
+
+        int lineSeparatorIndex = IOUtils.indexOfLineSeparator(buf, off, end);
+        if (lineSeparatorIndex == -1) {
+            if (input != null) {
+                if (!inputEnd) {
+                    int rest = end - off;
+                    if (rest > 0) {
+                        System.arraycopy(buf, off, buf, 0, rest);
+                    }
+                    end = rest;
+
+                    while (end < buf.length) {
+                        int cnt = input.read(buf, end, buf.length - end);
+                        if (cnt != -1) {
+                            end += cnt;
+                        } else {
+                            inputEnd = true;
+                            break;
+                        }
+                    }
+                    this.nextEscapeIndex = slashIndex >= off
+                            ? slashIndex - off
+                            : IOUtils.indexOfDoubleQuote(buf, rest, end);
+                    lineSeparatorIndex = IOUtils.indexOfLineSeparator(buf, rest, end);
+                    this.off = off = 0;
+                    this.end = end;
+                }
+            }
+            if (lineSeparatorIndex == -1 && inputEnd) {
+                if (off < end) {
+                    lineSeparatorIndex = end;
+                } else {
+                    return false;
+                }
+            }
+        }
+        if ((lineSeparatorIndex != -1) && (slashIndex == -1 || slashIndex > lineSeparatorIndex)) {
+            this.lineTerminated = true;
+            this.lineStart = off;
+            this.lineEnd = lineSeparatorIndex != off && buf[lineSeparatorIndex - 1] == '\r'
+                    ? lineSeparatorIndex - 1
+                    : lineSeparatorIndex;
+            this.off = lineSeparatorIndex + 1;
+            return true;
+        }
+
+        return seekLine0(buf, off, end);
+    }
+
+    private boolean seekLine0(byte[] buf, int off, int end) throws IOException {
+        int lineNextStart = off;
         for (int k = 0; k < 3; ++k) {
             lineTerminated = false;
-
+            for (int i = off; i < end; i++) {
-            int i = off, end = this.end;
-            while (i + 8 < end && !containsQuoteOrLineSeparator(IOUtils.getLongUnaligned(buf, i))) {
-                i += 8;
-            }
-            for (; i < end; i++) {
                 byte ch = buf[i];
                 if (ch == '"') {
                     lineSize++;
@@ -160,18 +228,7 @@ final class CSVReaderUTF8<T>
                     continue;
                 }
 
-                if (ch == '\n') {
+                if (ch == '\r' || ch == '\n') {
-                    if (lineSize > 0 || (features & Feature.IgnoreEmptyLine.mask) == 0) {
-                        rowCount++;
-                    }
-                    lineTerminated = true;
-                    lineSize = 0;
-                    lineEnd = i;
-                    lineStart = lineNextStart;
-                    lineNextStart = off = i + 1;
-
-                    break;
-                } else if (ch == '\r') {
                     if (lineSize > 0 || (features & Feature.IgnoreEmptyLine.mask) == 0) {
                         rowCount++;
                     }
@@ -180,6 +237,7 @@ final class CSVReaderUTF8<T>
                     lineSize = 0;
                     lineEnd = i;
 
+                    if (ch == '\r') {
                         int n = i + 1;
                         if (n >= end) {
                             break;
@@ -187,6 +245,7 @@ final class CSVReaderUTF8<T>
                         if (buf[n] == '\n') {
                             i++;
                         }
+                    }
 
                     lineStart = lineNextStart;
                     lineNextStart = off = i + 1;
@@ -200,9 +259,11 @@ final class CSVReaderUTF8<T>
             if (!lineTerminated) {
                 if (input != null && !inputEnd) {
                     int len = end - off;
+                    int slashIndex = this.nextEscapeIndex;
                     if (off > 0) {
                         if (len > 0) {
                             System.arraycopy(buf, off, buf, 0, len);
+                            this.nextEscapeIndex = slashIndex >= off ? slashIndex - off : ESCAPE_INDEX_NOT_SET;
                         }
                         lineStart = lineNextStart = 0;
                         off = 0;
@@ -283,7 +344,7 @@ final class CSVReaderUTF8<T>
 
     public Object[] readLineValues(boolean strings) {
         try {
-            if (inputEnd) {
+            if (inputEnd && off == end) {
                 return null;
             }
 
@@ -302,21 +363,113 @@ final class CSVReaderUTF8<T>
             throw new JSONException("seekLine error", e);
         }
 
+        int escapedIndex = this.nextEscapeIndex;
+        return escapedIndex == -1 || escapedIndex >= lineEnd
+                ? readLineValue(strings)
+                : readLineValueEscaped(strings);
+    }
+
+    private Object[] readLineValue(boolean strings) {
+        List<String> columns = this.columns;
         Object[] values = null;
         List<Object> valueList = null;
         if (columns != null) {
-            if (strings) {
+            int size = columns.size();
-                values = new String[columns.size()];
+            values = strings ? new String[size] : new Object[size];
         } else {
-                values = new Object[columns.size()];
+            valueList = new ArrayList<>();
+        }
+
+        int valueStart = lineStart, lineEnd = this.lineEnd;
+        int valueSize = 0;
+        int columnIndex = 0;
+        byte[] buf = this.buf;
+        for (int i = lineStart; i < lineEnd; ++i) {
+            if (buf[i] == ',') {
+                readValue(strings, columnIndex, valueSize, buf, valueStart, values, valueList);
+                valueStart = i + 1;
+                valueSize = 0;
+                columnIndex++;
+                continue;
+            }
+
+            valueSize++;
+        }
+
+        if (valueSize > 0) {
+            readValue(strings, columnIndex, valueSize, buf, valueStart, values, valueList);
+        }
+
+        if (values == null) {
+            if (valueList != null) {
+                int size = valueList.size();
+                valueList.toArray(
+                        values = strings ? new String[size] : new Object[size]
+                );
             }
         }
 
+        if (input == null && off == end) {
+            inputEnd = true;
+        }
+
+        return values;
+    }
+
+    private void readValue(boolean strings,
+                           int columnIndex,
+                           int valueSize,
+                           byte[] buf,
+                           int valueStart,
+                           Object[] values,
+                           List<Object> valueList) {
+        Type type = types != null && columnIndex < types.length ? types[columnIndex] : null;
+
+        Object value;
+        if (type == null || type == String.class || type == Object.class || strings) {
+            byte c0, c1;
+            if (valueSize == 1 && (c0 = buf[valueStart]) >= 0) {
+                value = TypeUtils.toString((char) c0);
+            } else if (valueSize == 2
+                    && (c0 = buf[valueStart]) >= 0
+                    && (c1 = buf[valueStart + 1]) >= 0
+            ) {
+                value = TypeUtils.toString((char) c0, (char) c1);
+            } else {
+                value = new String(buf, valueStart, valueSize, charset);
+            }
+        } else {
+            try {
+                value = readValue(buf, valueStart, valueSize, type);
+            } catch (Exception e) {
+                value = error(columnIndex, e);
+            }
+        }
+
+        if (values != null) {
+            if (columnIndex < values.length) {
+                values[columnIndex] = value;
+            }
+        } else {
+            valueList.add(value);
+        }
+    }
+
+    private Object[] readLineValueEscaped(boolean strings) {
+        List<String> columns = this.columns;
+        Object[] values = null;
+        List<Object> valueList = null;
+        if (columns != null) {
+            int size = columns.size();
+            values = strings ? new String[size] : new Object[size];
+        }
+
         boolean quote = false;
-        int valueStart = lineStart;
+        int valueStart = lineStart, lineEnd = this.lineEnd;
         int valueSize = 0;
         int escapeCount = 0;
         int columnIndex = 0;
+        byte[] buf = this.buf;
         for (int i = lineStart; i < lineEnd; ++i) {
             byte ch = buf[i];
 
@@ -375,7 +528,7 @@ final class CSVReaderUTF8<T>
                         }
 
                         if (type == null || type == String.class || type == Object.class) {
-                            value = new String(bytes, 0, bytes.length, charset);
+                            value = new String(bytes, charset);
                         } else {
                             try {
                                 value = readValue(bytes, 0, bytes.length, type);
@@ -500,12 +653,10 @@ final class CSVReaderUTF8<T>
 
         if (values == null) {
             if (valueList != null) {
-                if (strings) {
+                int size = valueList.size();
-                    values = new String[valueList.size()];
+                valueList.toArray(
-                } else {
+                        values = strings ? new String[size] : new Object[size]
-                    values = new Object[valueList.size()];
+                );
-                }
-                valueList.toArray(values);
             }
         }
 
@@ -649,7 +800,7 @@ final class CSVReaderUTF8<T>
 
         for (int r = 0; r < maxRows || maxRows < 0; ++r) {
             try {
-                if (inputEnd) {
+                if (inputEnd & off == end) {
                     break;
                 }
 
@@ -670,6 +821,41 @@ final class CSVReaderUTF8<T>
 
             consumer.beforeRow(rowCount);
 
+            int escapedIndex = this.nextEscapeIndex;
+            if (escapedIndex == -1 || escapedIndex >= lineEnd) {
+                readLine(consumer);
+            } else {
+                readLineEscaped(consumer);
+            }
+
+            consumer.afterRow(rowCount);
+        }
+        consumer.end();
+    }
+
+    private void readLine(ByteArrayValueConsumer consumer) {
+        int valueStart = lineStart;
+        int valueSize = 0;
+        int columnIndex = 0;
+        byte[] buf = this.buf;
+        for (int i = lineStart; i < lineEnd; ++i) {
+            if (buf[i] == ',') {
+                consumer.accept(rowCount, columnIndex, buf, valueStart, valueSize, charset);
+                valueStart = i + 1;
+                valueSize = 0;
+                columnIndex++;
+                continue;
+            }
+
+            valueSize++;
+        }
+
+        if (valueSize > 0) {
+            consumer.accept(rowCount, columnIndex, buf, valueStart, valueSize, charset);
+        }
+    }
+
+    private void readLineEscaped(ByteArrayValueConsumer consumer) {
         boolean quote = false;
         int valueStart = lineStart;
         int valueSize = 0;
@@ -769,8 +955,5 @@ final class CSVReaderUTF8<T>
             }
             consumer.accept(rowCount, columnIndex, columnBuf, columnStart, columnSize, charset);
         }
-            consumer.afterRow(rowCount);
-        }
-        consumer.end();
     }
 }

core/src/main/java/com/alibaba/fastjson2/util/IOUtils.java

@@ -1483,6 +1483,10 @@ public class IOUtils {
         }
     }
 
+    public static boolean isTRUE(byte[] buf, int pos) {
+        return getIntUnaligned(buf, pos) == TRUE;
+    }
+
     public static boolean isALSE(byte[] buf, int pos) {
         return getIntUnaligned(buf, pos) == ALSE;
     }
@@ -1525,6 +1529,75 @@ public class IOUtils {
         );
     }
 
+    public static int digit5(byte[] bytes, int off) {
+        int d4 = digit4(bytes, off);
+        int d1 = digit2(bytes, off + 4);
+        return (d4 | d1) == -1 ? -1 : (d4 * 10 + d1);
+    }
+
+    public static int digit6(byte[] bytes, int off) {
+        int d4 = digit4(bytes, off);
+        int d2 = digit2(bytes, off + 4);
+        return (d4 | d2) == -1 ? -1 : (d4 * 100 + d2);
+    }
+
+    public static int digit7(byte[] bytes, int off) {
+        int d4 = digit4(bytes, off);
+        int d2 = digit2(bytes, off + 4);
+        int d1 = digit1(bytes, off + 6);
+        return (d4 | d2 | d1) == -1 ? -1 : (d4 * 1000 + d2 * 10 + d1);
+    }
+
+    public static int digit8(byte[] bytes, int off) {
+        return digit8(
+                getLongLE(bytes, off)
+        );
+    }
+
+    private static int digit8(long x) {
+        /*
+            Here we are doing a 4-Byte Vector operation on the Int type.
+
+            x & 0xF0 != 0xC0
+            ---------------
+            0 0b0011_0000 & 0b1111_0000 = 0b0011_0000
+            1 0b0011_0001 & 0b1111_0000 = 0b0011_0000
+            2 0b0011_0010 & 0b1111_0000 = 0b0011_0000
+            3 0b0011_0011 & 0b1111_0000 = 0b0011_0000
+            4 0b0011_0100 & 0b1111_0000 = 0b0011_0000
+            5 0b0011_0101 & 0b1111_0000 = 0b0011_0000
+            6 0b0011_0110 & 0b1111_0000 = 0b0011_0000
+            7 0b0011_0111 & 0b1111_0000 = 0b0011_0000
+            8 0b0011_1000 & 0b1111_0000 = 0b0011_0000
+            9 0b0011_1001 & 0b1111_0000 = 0b0011_0000
+
+            (((d = x & 0x0F) + 0x06) & 0xF0) != 0
+            ---------------
+            0 ((0b0011_0000) & 0b0000_1111 + 0b0110_0000) & 0b1111_0000 = 0b0110_0000
+            1 ((0b0011_0001) & 0b0000_1111 + 0b0110_0000) & 0b1111_0000 = 0b0110_0000
+            2 ((0b0011_0010) & 0b0000_1111 + 0b0110_0000) & 0b1111_0000 = 0b0110_0000
+            3 ((0b0011_0011) & 0b0000_1111 + 0b0110_0000) & 0b1111_0000 = 0b0110_0000
+            4 ((0b0011_0100) & 0b0000_1111 + 0b0110_0000) & 0b1111_0000 = 0b0110_0000
+            5 ((0b0011_0101) & 0b0000_1111 + 0b0110_0000) & 0b1111_0000 = 0b0110_0000
+            6 ((0b0011_0110) & 0b0000_1111 + 0b0110_0000) & 0b1111_0000 = 0b0110_0000
+            7 ((0b0011_0111) & 0b0000_1111 + 0b0110_0000) & 0b1111_0000 = 0b0110_0000
+            8 ((0b0011_1000) & 0b0000_1111 + 0b0110_0000) & 0b1111_0000 = 0b0110_0000
+            9 ((0b0011_1001) & 0b0000_1111 + 0b0110_0000) & 0b1111_0000 = 0b0110_0000
+         */
+        long d;
+        if ((((x & 0xF0F0F0F0F0F0F0F0L) - 0x3030303030303030L) | (((d = x & 0x0F0F0F0F0F0F0F0FL) + 0x0606060606060606L) & 0xF0F0F0F0F0F0F0F0L)) != 0) {
+            return -1;
+        }
+        return (int) ((d & 0x0FL) * 10000000
+                + ((d & 0x0F00L) >> 8) * 1000000
+                + ((d & 0x0F0000L) >> 16) * 100000
+                + ((d & 0x0F000000L) >> 24) * 10000
+                + ((d & 0x0F00000000L) >> 32) * 1000
+                + ((d & 0x0F0000000000L) >> 40) * 100
+                + ((d & 0x0F000000000000L) >> 48) * 10
+                + (d >> 56));
+    }
+
     private static int digit4(int x) {
         /*
             Here we are doing a 4-Byte Vector operation on the Int type.
@@ -1639,6 +1712,7 @@ public class IOUtils {
             throw new JSONException(e.getMessage());
         }
     }
+
     static int indexOfQuote0(byte[] value, int quote, int fromIndex, int max) {
         int i = fromIndex;
         long address = ARRAY_BYTE_BASE_OFFSET + fromIndex;
@@ -1651,6 +1725,50 @@ public class IOUtils {
         return indexOfChar0(value, quote, i, max);
     }
 
+    public static int indexOfDoubleQuote(byte[] value, int fromIndex, int max) {
+        if (INDEX_OF_CHAR_LATIN1 == null) {
+            return indexOfDoubleQuoteV(value, fromIndex, max);
+        }
+        try {
+            return (int) INDEX_OF_CHAR_LATIN1.invokeExact(value, (int) '"', fromIndex, max);
+        } catch (Throwable e) {
+            throw new JSONException(e.getMessage());
+        }
+    }
+
+    public static int indexOfDoubleQuoteV(byte[] value, int fromIndex, int max) {
+        int i = fromIndex;
+        long address = ARRAY_BYTE_BASE_OFFSET + fromIndex;
+        int upperBound = fromIndex + ((max - fromIndex) & ~7);
+        while (i < upperBound && notContains(UNSAFE.getLong(value, address), 0x2222_2222_2222_2222L)) {
+            i += 8;
+            address += 8;
+        }
+        return indexOfChar0(value, '"', i, max);
+    }
+
+    public static int indexOfLineSeparator(byte[] value, int fromIndex, int max) {
+        if (INDEX_OF_CHAR_LATIN1 == null) {
+            return indexOfLineSeparatorV(value, fromIndex, max);
+        }
+        try {
+            return (int) INDEX_OF_CHAR_LATIN1.invokeExact(value, (int) '\n', fromIndex, max);
+        } catch (Throwable e) {
+            throw new JSONException(e.getMessage());
+        }
+    }
+
+    public static int indexOfLineSeparatorV(byte[] value, int fromIndex, int max) {
+        int i = fromIndex;
+        long address = ARRAY_BYTE_BASE_OFFSET + fromIndex;
+        int upperBound = fromIndex + ((max - fromIndex) & ~7);
+        while (i < upperBound && notContains(UNSAFE.getLong(value, address), 0x0A0A0A0A0A0A0A0AL)) {
+            i += 8;
+            address += 8;
+        }
+        return indexOfChar0(value, '\n', i, max);
+    }
+
     public static int indexOfSlash(byte[] value, int fromIndex, int max) {
         if (INDEX_OF_CHAR_LATIN1 == null) {
             return indexOfSlashV(value, fromIndex, max);
@@ -1906,4 +2024,38 @@ public class IOUtils {
         }
         return true;
     }
+
+    private static boolean isDigitLatin1(int c) {
+        return c >= '0' && c <= '9';
+    }
+
+    public static int parseInt(byte[] bytes, int off, int len) {
+        int fc = bytes[off];
+        int result = isDigitLatin1(fc)
+                ? '0' - fc
+                : len != 1 && (fc == '-' || fc == '+')
+                ? 0
+                : 1;  // or any value > 0
+        int end = off + len;
+        off++;
+        int d;
+        while (off + 1 < end
+                && (d = IOUtils.digit2(bytes, off)) != -1
+                && Integer.MIN_VALUE / 100 <= result & result <= 0) {
+            result = result * 100 - d;  // overflow from d => result > 0
+            off += 2;
+        }
+        if (off < end
+                && isDigitLatin1(d = bytes[off])
+                && Integer.MIN_VALUE / 10 <= result & result <= 0) {
+            result = result * 10 + '0' - d;  // overflow from '0' - d => result > 0
+            off += 1;
+        }
+        if (off == end
+                & result <= 0
+                & (Integer.MIN_VALUE < result || fc == '-')) {
+            return fc == '-' ? result : -result;
+        }
+        throw new NumberFormatException(new String(bytes, off, len));
+    }
 }

core/src/main/java/com/alibaba/fastjson2/util/TypeUtils.java

@@ -2076,25 +2076,21 @@ public class TypeUtils {
                 return null;
             case 1: {
                 byte b0 = bytes[off];
-                if (b0 == '1' || b0 == 'Y') {
+                if (b0 == '1' | b0 == 'Y') {
                     return Boolean.TRUE;
                 }
-                if (b0 == '0' || b0 == 'N') {
+                if (b0 == '0' | b0 == 'N') {
                     return Boolean.FALSE;
                 }
                 break;
             }
             case 4:
-                if (bytes[off] == 't' && bytes[off + 1] == 'r' && bytes[off + 2] == 'u' && bytes[off + 3] == 'e') {
+                if (IOUtils.isTRUE(bytes, off)) {
                     return Boolean.TRUE;
                 }
                 break;
             case 5:
-                if (bytes[off] == 'f'
+                if (bytes[off] == 'f' && IOUtils.isALSE(bytes, off + 1)) {
-                        && bytes[off + 1] == 'a'
-                        && bytes[off + 2] == 'l'
-                        && bytes[off + 3] == 's'
-                        && bytes[off + 4] == 'e') {
                     return Boolean.FALSE;
                 }
                 break;
@@ -2143,155 +2139,47 @@ public class TypeUtils {
     }
 
     public static int parseInt(byte[] bytes, int off, int len) {
+        int digit;
         switch (len) {
             case 1: {
-                byte b0 = bytes[off];
+                digit = IOUtils.digit1(bytes, off);
-                if (b0 >= '0' && b0 <= '9') {
-                    return b0 - '0';
-                }
                 break;
             }
             case 2: {
-                byte b0 = bytes[off];
+                digit = IOUtils.digit2(bytes, off);
-                byte b1 = bytes[off + 1];
-                if (b0 >= '0' && b0 <= '9'
-                        && b1 >= '0' && b1 <= '9') {
-                    return (b0 - '0') * 10
-                            + (b1 - '0');
-                }
                 break;
             }
             case 3: {
-                byte b0 = bytes[off];
+                digit = IOUtils.digit3(bytes, off);
-                byte b1 = bytes[off + 1];
-                byte b2 = bytes[off + 2];
-                if (b0 >= '0' && b0 <= '9'
-                        && b1 >= '0' && b1 <= '9'
-                        && b2 >= '0' && b2 <= '9') {
-                    return (b0 - '0') * 100
-                            + (b1 - '0') * 10
-                            + (b2 - '0');
-                }
                 break;
             }
             case 4: {
-                byte b0 = bytes[off];
+                digit = IOUtils.digit4(bytes, off);
-                byte b1 = bytes[off + 1];
-                byte b2 = bytes[off + 2];
-                byte b3 = bytes[off + 3];
-                if (b0 >= '0' && b0 <= '9'
-                        && b1 >= '0' && b1 <= '9'
-                        && b2 >= '0' && b2 <= '9'
-                        && b3 >= '0' && b3 <= '9'
-                ) {
-                    return (b0 - '0') * 1000
-                            + (b1 - '0') * 100
-                            + (b2 - '0') * 10
-                            + (b3 - '0');
-                }
                 break;
             }
             case 5: {
-                byte b0 = bytes[off];
+                digit = IOUtils.digit5(bytes, off);
-                byte b1 = bytes[off + 1];
-                byte b2 = bytes[off + 2];
-                byte b3 = bytes[off + 3];
-                byte b4 = bytes[off + 4];
-                if (b0 >= '0' && b0 <= '9'
-                        && b1 >= '0' && b1 <= '9'
-                        && b2 >= '0' && b2 <= '9'
-                        && b3 >= '0' && b3 <= '9'
-                        && b4 >= '0' && b4 <= '9'
-                ) {
-                    return (b0 - '0') * 10_000
-                            + (b1 - '0') * 1000
-                            + (b2 - '0') * 100
-                            + (b3 - '0') * 10
-                            + (b4 - '0');
-                }
                 break;
             }
             case 6: {
-                byte b0 = bytes[off];
+                digit = IOUtils.digit6(bytes, off);
-                byte b1 = bytes[off + 1];
-                byte b2 = bytes[off + 2];
-                byte b3 = bytes[off + 3];
-                byte b4 = bytes[off + 4];
-                byte b5 = bytes[off + 5];
-                if (b0 >= '0' && b0 <= '9'
-                        && b1 >= '0' && b1 <= '9'
-                        && b2 >= '0' && b2 <= '9'
-                        && b3 >= '0' && b3 <= '9'
-                        && b4 >= '0' && b4 <= '9'
-                        && b5 >= '0' && b5 <= '9'
-                ) {
-                    return (b0 - '0') * 100_000
-                            + (b1 - '0') * 10_000
-                            + (b2 - '0') * 1_000
-                            + (b3 - '0') * 100
-                            + (b4 - '0') * 10
-                            + (b5 - '0');
-                }
                 break;
             }
             case 7: {
-                byte b0 = bytes[off];
+                digit = IOUtils.digit7(bytes, off);
-                byte b1 = bytes[off + 1];
-                byte b2 = bytes[off + 2];
-                byte b3 = bytes[off + 3];
-                byte b4 = bytes[off + 4];
-                byte b5 = bytes[off + 5];
-                byte b6 = bytes[off + 6];
-                if (b0 >= '0' && b0 <= '9'
-                        && b1 >= '0' && b1 <= '9'
-                        && b2 >= '0' && b2 <= '9'
-                        && b3 >= '0' && b3 <= '9'
-                        && b4 >= '0' && b4 <= '9'
-                        && b5 >= '0' && b5 <= '9'
-                        && b6 >= '0' && b6 <= '9'
-                ) {
-                    return (b0 - '0') * 1_000_000
-                            + (b1 - '0') * 100_000
-                            + (b2 - '0') * 10_000
-                            + (b3 - '0') * 1_000
-                            + (b4 - '0') * 100
-                            + (b5 - '0') * 10
-                            + (b6 - '0');
-                }
                 break;
             }
             case 8: {
-                byte b0 = bytes[off];
+                digit = IOUtils.digit8(bytes, off);
-                byte b1 = bytes[off + 1];
-                byte b2 = bytes[off + 2];
-                byte b3 = bytes[off + 3];
-                byte b4 = bytes[off + 4];
-                byte b5 = bytes[off + 5];
-                byte b6 = bytes[off + 6];
-                byte b7 = bytes[off + 7];
-                if (b0 >= '0' && b0 <= '9'
-                        && b1 >= '0' && b1 <= '9'
-                        && b2 >= '0' && b2 <= '9'
-                        && b3 >= '0' && b3 <= '9'
-                        && b4 >= '0' && b4 <= '9'
-                        && b5 >= '0' && b5 <= '9'
-                        && b6 >= '0' && b6 <= '9'
-                        && b7 >= '0' && b7 <= '9'
-                ) {
-                    return (b0 - '0') * 10_000_000
-                            + (b1 - '0') * 1_000_000
-                            + (b2 - '0') * 100_000
-                            + (b3 - '0') * 10_000
-                            + (b4 - '0') * 1_000
-                            + (b5 - '0') * 100
-                            + (b6 - '0') * 10
-                            + (b7 - '0');
-                }
                 break;
             }
             default:
+                digit = -1;
                 break;
         }
+        if (digit != -1) {
+            return digit;
+        }
 
         String str = new String(bytes, off, len);
         return Integer.parseInt(str);

core/src/test/java/com/alibaba/fastjson2/support/csv/CSVTest4.java

@@ -56,7 +56,8 @@ public class CSVTest4 {
 
         in = new ByteArrayInputStream(bytes);
         CSVReader<Integer[]> parser2 = CSVReader.of(in, types);
-        Object object = parser2.stream(Object[].class).findFirst().get();
+        Object object = parser2.stream(Object[].class)
+                .findFirst().get();
         assertTrue(object instanceof Object[]);
         assertFalse(object instanceof Integer[]);
         assertFalse(object instanceof String[]);

core/src/test/java/com/alibaba/fastjson2/util/IOUtilsTest.java

@@ -294,6 +294,13 @@ public class IOUtilsTest {
         }
     }
 
+    @Test
+    public void digit8() {
+        byte[] bytes = "12345678".getBytes(StandardCharsets.ISO_8859_1);
+        int d8 = IOUtils.digit8(bytes, 0);
+        System.out.println(d8);
+    }
+
     @Test
     public void printDigit() {
         System.out.println("0xF " + Integer.toBinaryString(0xF));