MMQLEngine/mmql/engine/lexer.go

package engine

import (
    "fmt"
    "io"
    "strings"
    "strconv"

    stmtd "github.com/tgckpg/mmqlengine/mmql/statements"
)

type LexerExpect struct {
    Statements bool
    Keywords bool
    Brackets bool
    Quotes bool
    Key string
}

type Lexer struct {
    SupportedStmts *map[string] func( *Lexer ) ( stmtd.IStatement, error ) 
    reader *strings.Reader
    readToken bool
}

func ( this *Lexer ) Parse( line string ) ( stmtd.IStatement, error ) {

    this.reader = strings.NewReader( line )

    ex := LexerExpect{
        Statements: true,
        Keywords: false,
        Brackets: true,
        Quotes: true,
    }

    return this.ReadStatement( &ex )
}

func ( this *Lexer ) SetReader( reader *strings.Reader ) {
    this.reader = reader
}

func ( this *Lexer ) GetPos() int {
    return this.reader.Len()
}

/**
 *  Reads
 *    123,456,789.00001
 *    1234.567
 *    1234
 *    10.5k
 */
func ( this *Lexer ) ReadDecimal() ( f float64, err error ) {
    reader := this.reader
    defer this.Rollback( reader.Len(), &err )

    var b strings.Builder

    dot := false
    comma := false
    k := false
    end := false
    for {
        r, _, _err := reader.ReadRune()
        err = _err
        if err != nil {
            if err == io.EOF {
                if comma {
                    err = fmt.Errorf( "Unexpected ',' char" )
                    return
                }
                if b.Len() == 0 {
                    err = fmt.Errorf( "Nothing left to read" )
                    return
                }
            }
            break
        }

        if end && !strings.ContainsRune( K_SPACES, r ) {
            if k {
                b.WriteRune( 'k' )
            }
            err = fmt.Errorf(
                "Cannot parse '%s' in ReadDecimal",
                this.ReadUntilNot( K_STATEMENT_CHARS, true, r ),
            )
            return
        }

        if r == ',' {
            if comma || dot {
                err = fmt.Errorf( "Unexpected ',' char in ReadDecimal" )
                return
            }
            comma = true
            continue
        }

        if strings.ContainsRune( K_SPACES, r ) {
            if comma {
                err = fmt.Errorf( "Unexpected ',' char in ReadDecimal" )
                return
            }
            if 0 < b.Len() {
                break
            }
            continue
        }

        if r == '.' {
            if !dot {
                b.WriteRune( r )
                dot = true
                continue
            }
            err = fmt.Errorf( "Unexpected '.' char in ReadDecimal" )
            return
        }

        if r == 'k' {
            k = true
            end = true
            continue
        }

        if '0' <= r && r <= '9' {
            b.WriteRune( r )
            comma = false
        } else {
            err = fmt.Errorf(
                "Cannot parse '%s' in ReadDecimal",
                this.ReadUntilNot( K_STATEMENT_CHARS, true, r ),
            )
            return
        }
    }

    f, err = strconv.ParseFloat( b.String(), 64 )
    if k {
        f *= 1000
    }
    return
}

func ( this *Lexer ) Rollback( oPos int, err *error ) {
    if *err != nil {
        this.reader.Seek( int64( this.reader.Len() - oPos ), io.SeekCurrent )
    }
}

func ( this *Lexer ) ReadKeyword( Keyword string ) ( err error ) {

    reader := this.reader
    kReader := strings.NewReader( Keyword )

    defer this.Rollback( reader.Len(), &err )

    var b strings.Builder
    for {
        r, _, _err := reader.ReadRune()
        err = _err
        if err != nil {
            if err == io.EOF {
                if 0 < b.Len() {
                    err = nil
                } else {
                    err = fmt.Errorf( "Nothing left to read" )
                    return
                }
            }
            break
        }

        if strings.ContainsRune( K_SPACES, r ) && b.Len() == 0 {
            continue
        }

        k, _, _err := kReader.ReadRune()
        if _err != nil {
            if _err != io.EOF {
                err = _err
            } else {
                err = nil
            }
            return
        }

        if k != r {
            err = fmt.Errorf(
                "Expected \"%s\", got \"%s\"",
                Keyword,
                this.ReadUntilNot( K_STATEMENT_CHARS, true, r ),
            )
            return
        }
        b.WriteRune( r )
    }
    return
}

func ( this *Lexer ) ReadAlpha() ( s string, err error ) {
    reader := this.reader

    defer this.Rollback( reader.Len(), &err )

    var b strings.Builder

    for {
        r, _, _err := reader.ReadRune()
        err = _err
        if err != nil {
            if err == io.EOF {
                if 0 < b.Len() {
                    err = nil
                } else {
                    err = fmt.Errorf( "Nothing left to read" )
                    return
                }
            }
            break
        }

        if strings.ContainsRune( K_SPACES, r ) {
            if 0 < b.Len() {
                break
            }
            continue
        }

        if ( 'A' <= r && r <= 'Z' ) || ( 'a' <= r && r <= 'z' ) {
            b.WriteRune( r )
        }
    }
    s = b.String()
    return
}

func ( this *Lexer ) ReadStatement( expecting *LexerExpect ) ( stmtd.IStatement, error ) {
    reader := this.reader

    var stmt stmtd.IStatement

    for {
        r, _, err := reader.ReadRune()
        if err != nil {
            return nil, err
        }

        if expecting.Quotes && strings.ContainsRune( K_QUOTES, r ) {
            s, err := this.ReadUntilClose( r, true, '\\' )
            if err != nil {
                return nil, err
            }

            stmt = stmtd.QuotedValue{
                RawStatement: &stmtd.RawStatement{ Value: s },
                Quote: string( r ),
            }
            return stmt, nil
        }

        bracketIndex := strings.IndexRune( K_BRACKETS, r )
        if expecting.Brackets && bracketIndex % 2 == 0 {
            cl := rune( K_BRACKETS[ bracketIndex + 1 ] )
            s, err := this.ReadUntilClose( cl, false, 'A' )
            if err != nil {
                return nil, err
            }

            var b strings.Builder
            b.WriteRune( r )
            b.WriteRune( cl )

            stmt = stmtd.BracketedValue{
                RawStatement: &stmtd.RawStatement{ Value: s },
                Brackets: b.String(),
            }
            return stmt, nil
        }

        s := expecting.Key
        if s == "" && strings.ContainsRune( K_STATEMENT_CHARS, r ) {
            s = this.ReadUntilNot( K_STATEMENT_CHARS, true, r )
        }

        if expecting.Statements && s != "" {

            if this.SupportedStmts == nil {
                return nil, fmt.Errorf( "Statement not supported: %s", s )
            }

            if f, ok := (*this.SupportedStmts)[ strings.ToUpper( s ) ]; ok {

                // If we are using key, we need to move 1 rune back
                if expecting.Key != "" {
                    err = reader.UnreadRune()
                    if err != nil {
                        return nil, err
                    }
                }

                return f( this )
            } else {
                var supported strings.Builder
                comma := false
                for k := range *this.SupportedStmts {
                    if comma {
                        supported.WriteString( ", " )
                    } else {
                        comma = true
                    }
                    supported.WriteString( k )
                }
                return nil, fmt.Errorf( "Statement not supported: %s, supported statements are: %s", s, supported.String() )
            }
        }
    }

    return nil, fmt.Errorf( "Nothing to read" )
}

func ( this *Lexer ) ReadUntilNot( charRange string, writeHead bool, head rune ) string {
    var b strings.Builder

    if writeHead {
        b.WriteRune( head )
    }

    reader := this.reader
    for {
        r, _, err := reader.ReadRune()

        if err == nil && strings.ContainsRune( charRange, r ) {
            b.WriteRune( r )
        } else {
            break
        }
    }

    return b.String()
}

func ( this *Lexer ) ReadUntilClose( closing rune, canEsc bool, escRune rune ) ( string, error ) {
    var b strings.Builder

    reader := this.reader
    for {
        r, _, err := reader.ReadRune()

        if err != nil {
            if err == io.EOF {
                return "", fmt.Errorf( "Unexpected end of data after opening %s", string( closing ) )
            }
            return "", err
        }

        if r == closing {
            break
        } else if canEsc && r == escRune {
            r2, _, err := reader.ReadRune()
            if err != nil {
                return "", err
            }

            if r2 != closing {
                b.WriteRune( r )
                b.WriteRune( r2 )
                continue
            }
        }

        b.WriteRune( r )
    }

    return b.String(), nil
}