*: basic support for builtin function STR_TO_DATE

ast/functions.go

@@ -103,6 +103,7 @@ const (
 	MonthName        = "monthname"
 	Now              = "now"
 	Second           = "second"
+	StrToDate        = "str_to_date"
 	Sysdate          = "sysdate"
 	Time             = "time"
 	UTCDate          = "utc_date"

evaluator/builtin.go

@@ -77,6 +77,7 @@ var Funcs = map[string]Func{
 	ast.MonthName:        {builtinMonthName, 1, 1},
 	ast.Now:              {builtinNow, 0, 1},
 	ast.Second:           {builtinSecond, 1, 1},
+	ast.StrToDate:        {builtinStrToDate, 2, 2},
 	ast.Sysdate:          {builtinSysDate, 0, 1},
 	ast.Time:             {builtinTime, 1, 1},
 	ast.UTCDate:          {builtinUTCDate, 0, 0},

evaluator/builtin_time.go

@@ -21,8 +21,10 @@ import (
 	"fmt"
 	"math"
 	"regexp"
+	"strconv"
 	"strings"
 	"time"
+	"unicode"
 
 	"github.com/juju/errors"
 	"github.com/pingcap/tidb/ast"
@@ -673,6 +675,326 @@ func builtinFromUnixTime(args []types.Datum, _ context.Context) (d types.Datum,
 	return builtinDateFormat([]types.Datum{d, args[1]}, nil)
 }
 
+// strToDate converts date string according to format, return true on success,
+// the value will be stored in argument t.
+func strToDate(t *time.Time, date string, format string) bool {
+	date = skipWhiteSpace(date)
+	format = skipWhiteSpace(format)
+
+	token, formatRemain, succ := getFormatToken(format)
+	if !succ {
+		return false
+	}
+
+	if token == "" {
+		return date == ""
+	}
+
+	dateRemain, succ := matchDateWithToken(t, date, token)
+	if !succ {
+		return false
+	}
+
+	return strToDate(t, dateRemain, formatRemain)
+}
+
+func getFormatToken(format string) (token string, remain string, succ bool) {
+	if len(format) == 0 {
+		return "", "", true
+	}
+
+	// Just one character.
+	if len(format) == 1 {
+		if format[0] == '%' {
+			return "", "", false
+		}
+		return format, "", true
+	}
+
+	// More than one character.
+	if format[0] == '%' {
+		return format[:2], format[2:], true
+	}
+
+	return format[:1], format[1:], true
+}
+
+func skipWhiteSpace(input string) string {
+	for i, c := range input {
+		if !unicode.IsSpace(c) {
+			return input[i:]
+		}
+	}
+	return ""
+}
+
+var weekdayAbbrev = map[string]time.Weekday{
+	"Sun": time.Sunday,
+	"Mon": time.Monday,
+	"Tue": time.Tuesday,
+	"Wed": time.Wednesday,
+	"Thu": time.Tuesday,
+	"Fri": time.Friday,
+	"Sat": time.Saturday,
+}
+
+var monthAbbrev = map[string]time.Month{
+	"Jan": time.January,
+	"Feb": time.February,
+	"Mar": time.March,
+	"Apr": time.April,
+	"May": time.May,
+	"Jun": time.June,
+	"Jul": time.July,
+	"Aug": time.August,
+	"Sep": time.September,
+	"Oct": time.October,
+	"Nov": time.November,
+	"Dec": time.December,
+}
+
+type dateFormatParser func(t *time.Time, date string) (remain string, succ bool)
+
+var dateFormatParserTable = map[string]dateFormatParser{
+	"%a": abbreviatedWeekday,
+	"%b": abbreviatedMonth,
+	"%c": monthNumeric,
+	"%D": dayOfMonthWithSuffix,
+	"%Y": yearNumericFourDigits,
+	"%m": monthNumericTwoDigits,
+	"%d": dayOfMonthNumericTwoDigits,
+	"%H": hour24TwoDigits,
+	"%i": minutesNumeric,
+	"%s": secondsNumeric,
+}
+
+func matchDateWithToken(t *time.Time, date string, token string) (remain string, succ bool) {
+	if parse, ok := dateFormatParserTable[token]; ok {
+		return parse(t, date)
+	}
+
+	if strings.HasPrefix(date, token) {
+		return date[len(token):], true
+	}
+	return date, false
+}
+
+func parseTwoDigits(input string) (int, bool) {
+	if len(input) < 2 {
+		return 0, false
+	}
+
+	v, err := strconv.ParseUint(input[:2], 10, 64)
+	if err != nil {
+		return int(v), false
+	}
+	return int(v), true
+}
+
+func hour24TwoDigits(t *time.Time, input string) (string, bool) {
+	v, succ := parseTwoDigits(input)
+	if !succ || v >= 24 {
+		return input, false
+	}
+	timeSetHour(t, v)
+	return input[2:], true
+}
+
+func secondsNumeric(t *time.Time, input string) (string, bool) {
+	v, succ := parseTwoDigits(input)
+	if !succ || v >= 60 {
+		return input, false
+	}
+	timeSetSecond(t, v)
+	return input[2:], true
+}
+
+func minutesNumeric(t *time.Time, input string) (string, bool) {
+	v, succ := parseTwoDigits(input)
+	if !succ || v >= 60 {
+		return input, false
+	}
+	timeSetMinute(t, v)
+	return input[2:], true
+}
+
+func dayOfMonthNumericTwoDigits(t *time.Time, input string) (string, bool) {
+	v, succ := parseTwoDigits(input)
+	if !succ || v >= 32 {
+		return input, false
+	}
+	timeSetDay(t, v)
+	return input[2:], true
+}
+
+func yearNumericFourDigits(t *time.Time, input string) (string, bool) {
+	if len(input) < 4 {
+		return input, false
+	}
+
+	v, err := strconv.ParseUint(input[:4], 10, 64)
+	if err != nil {
+		return input, false
+	}
+	timeSetYear(t, int(v))
+	return input[4:], true
+}
+
+func monthNumericTwoDigits(t *time.Time, input string) (string, bool) {
+	v, succ := parseTwoDigits(input)
+	if !succ || v > 12 {
+		return input, false
+	}
+
+	timeSetMonth(t, time.Month(v))
+	return input[2:], true
+}
+
+func abbreviatedWeekday(t *time.Time, input string) (string, bool) {
+	if len(input) >= 3 {
+		dayName := input[:3]
+		if _, ok := weekdayAbbrev[dayName]; ok {
+			// TODO: we need refact mysql time to support this
+			return input, false
+		}
+	}
+	return input, false
+}
+
+func abbreviatedMonth(t *time.Time, input string) (string, bool) {
+	if len(input) >= 3 {
+		monthName := input[:3]
+		if month, ok := monthAbbrev[monthName]; ok {
+			timeSetMonth(t, month)
+			return input[len(monthName):], true
+		}
+	}
+	return input, false
+}
+
+func monthNumeric(t *time.Time, input string) (string, bool) {
+	// TODO: this code is ugly!
+	for i := 12; i >= 0; i-- {
+		str := strconv.FormatInt(int64(i), 10)
+		if strings.HasPrefix(input, str) {
+			timeSetMonth(t, time.Month(i))
+			return input[len(str):], true
+		}
+	}
+
+	return input, false
+}
+
+// 0th 1st 2nd 3rd ...
+func dayOfMonthWithSuffix(t *time.Time, input string) (string, bool) {
+	month, remain := parseOrdinalNumbers(input)
+	if month >= 0 {
+		timeSetMonth(t, time.Month(month))
+		return remain, true
+	}
+	return input, false
+}
+
+func parseOrdinalNumbers(input string) (value int, remain string) {
+	for i, c := range input {
+		if !unicode.IsDigit(c) {
+			v, err := strconv.ParseUint(input[:i], 10, 64)
+			if err != nil {
+				return -1, input
+			}
+			value = int(v)
+			break
+		}
+	}
+	switch {
+	case strings.HasPrefix(remain, "st"):
+		if value == 1 {
+			remain = remain[2:]
+			return
+		}
+	case strings.HasPrefix(remain, "nd"):
+		if value == 2 {
+			remain = remain[2:]
+			return
+		}
+	case strings.HasPrefix(remain, "th"):
+		remain = remain[2:]
+		return
+	}
+	return -1, input
+}
+
+func timeSetYear(t *time.Time, year int) {
+	_, month, day := t.Date()
+	hour, min, sec := t.Clock()
+	nsec := t.Nanosecond()
+	loc := t.Location()
+	*t = time.Date(year, month, day, hour, min, sec, nsec, loc)
+}
+
+func timeSetMonth(t *time.Time, month time.Month) {
+	year, _, day := t.Date()
+	hour, min, sec := t.Clock()
+	nsec := t.Nanosecond()
+	loc := t.Location()
+	*t = time.Date(year, month, day, hour, min, sec, nsec, loc)
+}
+
+func timeSetDay(t *time.Time, day int) {
+	year, month, _ := t.Date()
+	hour, min, sec := t.Clock()
+	nsec := t.Nanosecond()
+	loc := t.Location()
+	*t = time.Date(year, month, day, hour, min, sec, nsec, loc)
+}
+
+func timeSetHour(t *time.Time, hour int) {
+	year, month, day := t.Date()
+	_, min, sec := t.Clock()
+	nsec := t.Nanosecond()
+	loc := t.Location()
+	*t = time.Date(year, month, day, hour, min, sec, nsec, loc)
+}
+
+func timeSetMinute(t *time.Time, min int) {
+	year, month, day := t.Date()
+	hour, _, sec := t.Clock()
+	nsec := t.Nanosecond()
+	loc := t.Location()
+	*t = time.Date(year, month, day, hour, min, sec, nsec, loc)
+}
+
+func timeSetSecond(t *time.Time, sec int) {
+	year, month, day := t.Date()
+	hour, min, _ := t.Clock()
+	nsec := t.Nanosecond()
+	loc := t.Location()
+	*t = time.Date(year, month, day, hour, min, sec, nsec, loc)
+}
+
+// See https://dev.mysql.com/doc/refman/5.5/en/date-and-time-functions.html#function_str-to-date
+func builtinStrToDate(args []types.Datum, _ context.Context) (types.Datum, error) {
+	date := args[0].GetString()
+	format := args[1].GetString()
+	var (
+		d      types.Datum
+		goTime time.Time
+	)
+	goTime = types.ZeroTime
+	if !strToDate(&goTime, date, format) {
+		d.SetNull()
+		return d, nil
+	}
+
+	t := types.Time{
+		Time: goTime,
+		Type: mysql.TypeDatetime,
+		Fsp:  types.UnspecifiedFsp,
+	}
+	d.SetMysqlTime(t)
+	return d, nil
+}
+
 func builtinSysDate(args []types.Datum, ctx context.Context) (types.Datum, error) {
 	// SYSDATE is not the same as NOW if NOW is used in a stored function or trigger.
 	// But here we can just think they are the same because we don't support stored function

evaluator/builtin_time_test.go

@@ -579,3 +579,31 @@ func (s *testEvaluatorSuite) TestDateArith(c *C) {
 		}
 	}
 }
+
+func (s *testEvaluatorSuite) TestStrToDate(c *C) {
+	tests := []struct {
+		Date    string
+		Format  string
+		Success bool
+		Expect  time.Time
+	}{
+		{"20161122165022", "%Y%m%d%H%i%s", true, time.Date(2016, 11, 22, 16, 50, 22, 0, time.Local)},
+		{"2016 11 22 16 50 22", "%Y%m%d%H%i%s", true, time.Date(2016, 11, 22, 16, 50, 22, 0, time.Local)},
+		{"16-50-22 2016 11 22", "%H-%i-%s%Y%m%d", true, time.Date(2016, 11, 22, 16, 50, 22, 0, time.Local)},
+		{"16-50 2016 11 22", "%H-%i-%s%Y%m%d", false, time.Time{}},
+	}
+
+	for _, test := range tests {
+		date := types.NewStringDatum(test.Date)
+		format := types.NewStringDatum(test.Format)
+		result, err := builtinStrToDate([]types.Datum{date, format}, nil)
+		if !test.Success {
+			c.Assert(err, IsNil)
+			c.Assert(result.IsNull(), IsTrue)
+			continue
+		}
+		c.Assert(result.Kind(), Equals, types.KindMysqlTime)
+		value := result.GetMysqlTime()
+		c.Assert(value.Time, Equals, test.Expect)
+	}
+}