mysql: introduce icuregex package

go/mysql/collations/charset/convert.go

@@ -19,6 +19,8 @@ package charset
 import (
 	"fmt"
 	"unicode/utf8"
+
+	"vitess.io/vitess/go/hack"
 )
 
 func failedConversionError(from, to Charset, input []byte) error {
@@ -126,6 +128,78 @@ func Convert(dst []byte, dstCharset Charset, src []byte, srcCharset Charset) ([]
 	}
 }
 
+func Expand(dst []rune, src []byte, srcCharset Charset) []rune {
+	switch srcCharset := srcCharset.(type) {
+	case Charset_utf8mb3, Charset_utf8mb4:
+		if dst == nil {
+			return []rune(string(src))
+		}
+		dst = make([]rune, 0, len(src))
+		for _, cp := range string(src) {
+			dst = append(dst, cp)
+		}
+		return dst
+	case Charset_binary:
+		if dst == nil {
+			dst = make([]rune, 0, len(src))
+		}
+		for _, c := range src {
+			dst = append(dst, rune(c))
+		}
+		return dst
+	default:
+		if dst == nil {
+			dst = make([]rune, 0, len(src))
+		}
+		for len(src) > 0 {
+			cp, width := srcCharset.DecodeRune(src)
+			src = src[width:]
+			dst = append(dst, cp)
+		}
+		return dst
+	}
+}
+
+func Collapse(dst []byte, src []rune, dstCharset Charset) []byte {
+	switch dstCharset := dstCharset.(type) {
+	case Charset_utf8mb3, Charset_utf8mb4:
+		if dst == nil {
+			return hack.StringBytes(string(src))
+		}
+		return append(dst, hack.StringBytes(string(src))...)
+	case Charset_binary:
+		if dst == nil {
+			dst = make([]byte, 0, len(src))
+		}
+		for _, b := range src {
+			dst = append(dst, byte(b))
+		}
+		return dst
+	default:
+		nDst := 0
+		if dst == nil {
+			dst = make([]byte, len(src)*dstCharset.MaxWidth())
+		} else {
+			dst = dst[:cap(dst)]
+		}
+		for _, c := range src {
+			if len(dst)-nDst < 4 {
+				newDst := make([]byte, len(dst)*2)
+				copy(newDst, dst[:nDst])
+				dst = newDst
+			}
+			w := dstCharset.EncodeRune(dst[nDst:], c)
+			if w < 0 {
+				if w = dstCharset.EncodeRune(dst[nDst:], '?'); w < 0 {
+					break
+				}
+			}
+			nDst += w
+		}
+		return dst[:nDst]
+	}
+}
+
 func ConvertFromUTF8(dst []byte, dstCharset Charset, src []byte) ([]byte, error) {
 	return Convert(dst, dstCharset, src, Charset_utf8mb4{})
 }

go/mysql/collations/env.go

@@ -194,10 +194,11 @@ func makeEnv(version collver) *Environment {
 // A few interesting character set values.
 // See http://dev.mysql.com/doc/internals/en/character-set.html#packet-Protocol::CharacterSet
 const (
-	CollationUtf8ID       = 33
-	CollationUtf8mb4ID    = 255
-	CollationBinaryID     = 63
-	CollationUtf8mb4BinID = 46
+	CollationUtf8ID        = 33
+	CollationUtf8mb4ID     = 255
+	CollationBinaryID      = 63
+	CollationUtf8mb4BinID  = 46
+	CollationLatin1Swedish = 8
 )
 
 // Binary is the default Binary collation

go/mysql/constants.go

@@ -565,6 +565,31 @@ const (
 	ERJSONValueTooBig              = ErrorCode(3150)
 	ERJSONDocumentTooDeep          = ErrorCode(3157)
 
+	ERRegexpStringNotTerminated = ErrorCode(3684)
+	ERRegexpBufferOverflow      = ErrorCode(3684)
+	ERRegexpIllegalArgument     = ErrorCode(3685)
+	ERRegexpIndexOutOfBounds    = ErrorCode(3686)
+	ERRegexpInternal            = ErrorCode(3687)
+	ERRegexpRuleSyntax          = ErrorCode(3688)
+	ERRegexpBadEscapeSequence   = ErrorCode(3689)
+	ERRegexpUnimplemented       = ErrorCode(3690)
+	ERRegexpMismatchParen       = ErrorCode(3691)
+	ERRegexpBadInterval         = ErrorCode(3692)
+	ERRRegexpMaxLtMin           = ErrorCode(3693)
+	ERRegexpInvalidBackRef      = ErrorCode(3694)
+	ERRegexpLookBehindLimit     = ErrorCode(3695)
+	ERRegexpMissingCloseBracket = ErrorCode(3696)
+	ERRegexpInvalidRange        = ErrorCode(3697)
+	ERRegexpStackOverflow       = ErrorCode(3698)
+	ERRegexpTimeOut             = ErrorCode(3699)
+	ERRegexpPatternTooBig       = ErrorCode(3700)
+	ERRegexpInvalidCaptureGroup = ErrorCode(3887)
+	ERRegexpInvalidFlag         = ErrorCode(3900)
+
+	ERCharacterSetMismatch = ErrorCode(3995)
+
+	ERWrongParametersToNativeFct = ErrorCode(1583)
+
 	// max execution time exceeded
 	ERQueryTimeout = ErrorCode(3024)