// Copyright 2012, Google Inc. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.

// Copyright 2016 The kingshard Authors. All rights reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License"): you may
// not use this file except in compliance with the License. You may obtain
// a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
// WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
// License for the specific language governing permissions and limitations
// under the License.

package sqlparser

import (
	"bytes"
	"fmt"
	"strings"

	"github.com/flike/kingshard/sqltypes"
)

const EOFCHAR = 0x100

// Tokenizer is the struct used to generate SQL
// tokens for the parser.
type Tokenizer struct {
	InStream *strings.Reader
	AllowComments bool
	ForceEOF bool
	lastChar uint16
	Position int
	errorToken []byte
	LastError string
	posVarIndex int
	ParseTree Statement
}

// NewStringTokenizer creates a new Tokenizer for the
// sql string.
func NewStringTokenizer(sql string) *Tokenizer {
	return &Tokenizer{InStream: strings.NewReader(sql)}
}

var keywords = map[string]int{
	"select": SELECT,
	"insert": INSERT,
	"update": UPDATE,
	"delete": DELETE,
	"from": FROM,
	"where": WHERE,
	"group": GROUP,
	"having": HAVING,
	"order": ORDER,
	"by": BY,
	"limit": LIMIT,
	"for": FOR,

	"union": UNION,
	"all": ALL,
	"minus": MINUS,
	"except": EXCEPT,
	"intersect": INTERSECT,

	"join": JOIN,
	"straight_join": STRAIGHT_JOIN,
	"left": LEFT,
	"right": RIGHT,
	"inner": INNER,
	"outer": OUTER,
	"cross": CROSS,
	"natural": NATURAL,
	"use": USE,
	"force": FORCE,
	"on": ON,
	"into": INTO,

	"distinct": DISTINCT,
	"case": CASE,
	"when": WHEN,
	"then": THEN,
	"else": ELSE,
	"end": END,
	"as": AS,
	"and": AND,
	"or": OR,
	"not": NOT,
	"exists": EXISTS,
	"in": IN,
	"is": IS,
	"like": LIKE,
	"between": BETWEEN,
	"null": NULL,
	"asc": ASC,
	"desc": DESC,
	"values": VALUES,
	"duplicate": DUPLICATE,
	"key": KEY,
	"default": DEFAULT,
	"set": SET,
	"lock": LOCK,

	"create": CREATE,
	"alter": ALTER,
	"rename": RENAME,
	"drop": DROP,
	"table": TABLE,
	"index": INDEX,
	"view": VIEW,
	"to": TO,
	"ignore": IGNORE,
	"if": IF,
	"unique": UNIQUE,
	"using": USING,

	"begin": BEGIN,
	"rollback": ROLLBACK,
	"commit": COMMIT,

	"names": NAMES,
	"replace": REPLACE,

	//for kingshard
	"admin": ADMIN,
	"help": HELP,
	"start": START,
	"transaction": TRANSACTION,
	"collate": COLLATE,
	"offset": OFFSET,
	"truncate": TRUNCATE,
}

// Lex returns the next token form the Tokenizer.
// This function is used by go yacc.
func (tkn *Tokenizer) Lex(lval *yySymType) int {
	typ, val := tkn.Scan()
	for typ == COMMENT {
		if tkn.AllowComments {
			break
		}
		typ, val = tkn.Scan()
	}
	switch typ {
	case ID, STRING, NUMBER, VALUE_ARG, COMMENT:
		lval.bytes = val
	}
	tkn.errorToken = val
	return typ
}

// Error is called by go yacc if there's a parsing error.
func (tkn *Tokenizer) Error(err string) {
	buf := bytes.NewBuffer(make([]byte, 0, 32))
	if tkn.errorToken != nil {
		fmt.Fprintf(buf, "%s at position %v near %s", err, tkn.Position, tkn.errorToken)
	} else {
		fmt.Fprintf(buf, "%s at position %v", err, tkn.Position)
	}
	tkn.LastError = buf.String()
}

// Scan scans the tokenizer for the next token and returns
// the token type and an optional value.
func (tkn *Tokenizer) Scan() (int, []byte) {
	if tkn.ForceEOF {
		return 0, nil
	}

	if tkn.lastChar == 0 {
		tkn.next()
	}
	tkn.skipBlank()
	switch ch := tkn.lastChar; {
	case isLetter(ch):
		return tkn.scanIdentifier()
	case isDigit(ch):
		return tkn.scanNumber(false)
	case ch == ':':
		return tkn.scanBindVar()
	default:
		tkn.next()
		switch ch {
		case EOFCHAR:
			return 0, nil
		case '=', ',', ';', '(', ')', '+', '*', '%', '&', '|', '^', '~':
			return int(ch), nil
		case '?':
			tkn.posVarIndex++
			buf := new(bytes.Buffer)
			fmt.Fprintf(buf, ":v%d", tkn.posVarIndex)
			return VALUE_ARG, buf.Bytes()
		case '.':
			if isDigit(tkn.lastChar) {
				return tkn.scanNumber(true)
			} else {
				return int(ch), nil
			}
		case '/':
			switch tkn.lastChar {
			case '/':
				tkn.next()
				return tkn.scanCommentType1("//")
			case '*':
				tkn.next()
				return tkn.scanCommentType2()
			default:
				return int(ch), nil
			}
		case '-':
			if tkn.lastChar == '-' {
				tkn.next()
				return tkn.scanCommentType1("--")
			} else {
				return int(ch), nil
			}
		case '<':
			switch tkn.lastChar {
			case '>':
				tkn.next()
				return NE, nil
			case '=':
				tkn.next()
				switch tkn.lastChar {
				case '>':
					tkn.next()
					return NULL_SAFE_EQUAL, nil
				default:
					return LE, nil
				}
			default:
				return int(ch), nil
			}
		case '>':
			if tkn.lastChar == '=' {
				tkn.next()
				return GE, nil
			} else {
				return int(ch), nil
			}
		case '!':
			if tkn.lastChar == '=' {
				tkn.next()
				return NE, nil
			} else {
				return LEX_ERROR, []byte("!")
			}
		case '\'', '"':
			return tkn.scanString(ch, STRING)
		case '`':
			return tkn.scanString(ch, ID)
		default:
			return LEX_ERROR, []byte{byte(ch)}
		}
	}
}

func (tkn *Tokenizer) skipBlank() {
	ch := tkn.lastChar
	for ch == ' ' || ch == '\n' || ch == '\r' || ch == '\t' {
		tkn.next()
		ch = tkn.lastChar
	}
}

func (tkn *Tokenizer) scanIdentifier() (int, []byte) {
	buffer := bytes.NewBuffer(make([]byte, 0, 8))
	buffer.WriteByte(byte(tkn.lastChar))
	for tkn.next(); isLetter(tkn.lastChar) || isDigit(tkn.lastChar); tkn.next() {
		buffer.WriteByte(byte(tkn.lastChar))
	}
	lowered := bytes.ToLower(buffer.Bytes())
	if keywordId, found := keywords[string(lowered)]; found {
		return keywordId, lowered
	}
	return ID, buffer.Bytes()
}

func (tkn *Tokenizer) scanBindVar() (int, []byte) {
	buffer := bytes.NewBuffer(make([]byte, 0, 8))
	buffer.WriteByte(byte(tkn.lastChar))
	for tkn.next(); isLetter(tkn.lastChar) || isDigit(tkn.lastChar) || tkn.lastChar == '.'; tkn.next() {
		buffer.WriteByte(byte(tkn.lastChar))
	}
	if buffer.Len() == 1 {
		return LEX_ERROR, buffer.Bytes()
	}
	return VALUE_ARG, buffer.Bytes()
}

func (tkn *Tokenizer) scanMantissa(base int, buffer *bytes.Buffer) {
	for digitVal(tkn.lastChar) < base {
		tkn.ConsumeNext(buffer)
	}
}

func (tkn *Tokenizer) scanNumber(seenDecimalPoint bool) (int, []byte) {
	buffer := bytes.NewBuffer(make([]byte, 0, 8))
	if seenDecimalPoint {
		buffer.WriteByte('.')
		tkn.scanMantissa(10, buffer)
		goto exponent
	}

	if tkn.lastChar == '0' {
		// int or float
		tkn.ConsumeNext(buffer)
		if tkn.lastChar == 'x' || tkn.lastChar == 'X' {
			// hexadecimal int
			tkn.ConsumeNext(buffer)
			tkn.scanMantissa(16, buffer)
		} else {
			// octal int or float
			seenDecimalDigit := false
			tkn.scanMantissa(8, buffer)
			if tkn.lastChar == '8' || tkn.lastChar == '9' {
				// illegal octal int or float
				seenDecimalDigit = true
				tkn.scanMantissa(10, buffer)
			}
			if tkn.lastChar == '.' || tkn.lastChar == 'e' || tkn.lastChar == 'E' {
				goto fraction
			}
			// octal int
			if seenDecimalDigit {
				return LEX_ERROR, buffer.Bytes()
			}
		}
		goto exit
	}

	// decimal int or float
	tkn.scanMantissa(10, buffer)

fraction:
	if tkn.lastChar == '.' {
		tkn.ConsumeNext(buffer)
		tkn.scanMantissa(10, buffer)
	}

exponent:
	if tkn.lastChar == 'e' || tkn.lastChar == 'E' {
		tkn.ConsumeNext(buffer)
		if tkn.lastChar == '+' || tkn.lastChar == '-' {
			tkn.ConsumeNext(buffer)
		}
		tkn.scanMantissa(10, buffer)
	}

exit:
	return NUMBER, buffer.Bytes()
}

func (tkn *Tokenizer) scanString(delim uint16, typ int) (int, []byte) {
	buffer := bytes.NewBuffer(make([]byte, 0, 8))
	for {
		ch := tkn.lastChar
		tkn.next()
		if ch == delim {
			if tkn.lastChar == delim {
				tkn.next()
			} else {
				break
			}
		} else if ch == '\\' {
			if tkn.lastChar == EOFCHAR {
				return LEX_ERROR, buffer.Bytes()
			}
			if decodedChar := sqltypes.SqlDecodeMap[byte(tkn.lastChar)]; decodedChar == sqltypes.DONTESCAPE {
				ch = tkn.lastChar
			} else {
				ch = uint16(decodedChar)
			}
			tkn.next()
		}
		if ch == EOFCHAR {
			return LEX_ERROR, buffer.Bytes()
		}
		buffer.WriteByte(byte(ch))
	}
	return typ, buffer.Bytes()
}

func (tkn *Tokenizer) scanCommentType1(prefix string) (int, []byte) {
	buffer := bytes.NewBuffer(make([]byte, 0, 8))
	buffer.WriteString(prefix)
	for tkn.lastChar != EOFCHAR {
		if tkn.lastChar == '\n' {
			tkn.ConsumeNext(buffer)
			break
		}
		tkn.ConsumeNext(buffer)
	}
	return COMMENT, buffer.Bytes()
}

func (tkn *Tokenizer) scanCommentType2() (int, []byte) {
	buffer := bytes.NewBuffer(make([]byte, 0, 8))
	buffer.WriteString("/*")
	for {
		if tkn.lastChar == '*' {
			tkn.ConsumeNext(buffer)
			if tkn.lastChar == '/' {
				tkn.ConsumeNext(buffer)
				break
			}
			continue
		}
		if tkn.lastChar == EOFCHAR {
			return LEX_ERROR, buffer.Bytes()
		}
		tkn.ConsumeNext(buffer)
	}
	return COMMENT, buffer.Bytes()
}

func (tkn *Tokenizer) ConsumeNext(buffer *bytes.Buffer) {
	if tkn.lastChar == EOFCHAR {
		// This should never happen.
		panic("unexpected EOF")
	}
	buffer.WriteByte(byte(tkn.lastChar))
	tkn.next()
}

func (tkn *Tokenizer) next() {
	if ch, err := tkn.InStream.ReadByte(); err != nil {
		// Only EOF is possible.
		tkn.lastChar = EOFCHAR
	} else {
		tkn.lastChar = uint16(ch)
	}
	tkn.Position++
}

func isLetter(ch uint16) bool {
	return 'a' <= ch && ch <= 'z' || 'A' <= ch && ch <= 'Z' || ch == '_' || ch == '@'
}

func digitVal(ch uint16) int {
	switch {
	case '0' <= ch && ch <= '9':
		return int(ch) - '0'
	case 'a' <= ch && ch <= 'f':
		return int(ch) - 'a' + 10
	case 'A' <= ch && ch <= 'F':
		return int(ch) - 'A' + 10
	}
	return 16 // larger than any legal digit val
}

func isDigit(ch uint16) bool {
	return '0' <= ch && ch <= '9'
}