Fin/scanner.go

package config

import (
	"bufio"
	"bytes"
	"errors"
	"fmt"
	"io"
	"strconv"
)

// TokenType represents the type of token
type TokenType int

const (
	TokenError TokenType = iota
	TokenEOF
	TokenName
	TokenString
	TokenNumber
	TokenBoolean
	TokenEquals
	TokenOpenBrace
	TokenCloseBrace
	TokenComment
)

// Token represents a lexical token
type Token struct {
	Type   TokenType
	Value  []byte
	Line   int
	Column int
}

// Scanner handles the low-level parsing of the configuration format
type Scanner struct {
	reader *bufio.Reader
	line   int // Current line number
	col    int // Current column position
	buffer []byte
	token  Token // Current token
}

// NewScanner creates a new scanner with the given reader
func NewScanner(r io.Reader) *Scanner {
	return &Scanner{
		reader: bufio.NewReader(r),
		line:   1, // Start at line 1
		col:    0,
		buffer: make([]byte, 0, 128), // Pre-allocate with reasonable capacity
	}
}

// ReadByte reads a single byte from the input
func (s *Scanner) ReadByte() (byte, error) {
	b, err := s.reader.ReadByte()
	if err == nil {
		if b == '\n' {
			s.line++
			s.col = 0
		} else {
			s.col++
		}
	}
	return b, err
}

// PeekByte looks at the next byte without consuming it
func (s *Scanner) PeekByte() (byte, error) {
	b, err := s.reader.Peek(1)
	if err != nil {
		return 0, err
	}
	return b[0], nil
}

// PeekBytes looks at the next n bytes without consuming them
func (s *Scanner) PeekBytes(n int) ([]byte, error) {
	return s.reader.Peek(n)
}

// UnreadByte pushes back a byte to the reader
func (s *Scanner) UnreadByte() error {
	err := s.reader.UnreadByte()
	if err == nil && s.col > 0 {
		s.col--
	}
	return err
}

// Error creates an error with line and column information
func (s *Scanner) Error(msg string) error {
	return fmt.Errorf("line %d, column %d: %s", s.line, s.col, msg)
}

// SkipWhitespace skips whitespace characters
func (s *Scanner) SkipWhitespace() error {
	for {
		b, err := s.PeekByte()
		if err == io.EOF {
			return nil
		}
		if err != nil {
			return err
		}

		// Fast check for common whitespace bytes
		if b != ' ' && b != '\t' && b != '\n' && b != '\r' {
			return nil
		}

		_, err = s.ReadByte()
		if err != nil {
			return err
		}
	}
}

// NextToken scans and returns the next token
func (s *Scanner) NextToken() (Token, error) {
	if s.token.Type != TokenError {
		// We have a stored token
		token := s.token
		s.token = Token{Type: TokenError} // Reset
		return token, nil
	}

	// No stored token, scan a new one
	// Skip whitespace
	err := s.SkipWhitespace()
	if err == io.EOF {
		return Token{Type: TokenEOF}, nil
	}
	if err != nil {
		return Token{Type: TokenError, Value: []byte(err.Error())}, err
	}

	b, err := s.PeekByte()
	if err != nil {
		if err == io.EOF {
			return Token{Type: TokenEOF}, nil
		}
		return Token{Type: TokenError, Value: []byte(err.Error())}, err
	}

	// Record start position for error reporting
	startLine, startColumn := s.line, s.col

	// Process based on first character
	switch {
	case b == '=':
		_, _ = s.ReadByte() // consume equals
		return Token{Type: TokenEquals, Line: startLine, Column: startColumn}, nil

	case b == '{':
		_, _ = s.ReadByte() // consume open brace
		return Token{Type: TokenOpenBrace, Line: startLine, Column: startColumn}, nil

	case b == '}':
		_, _ = s.ReadByte() // consume close brace
		return Token{Type: TokenCloseBrace, Line: startLine, Column: startColumn}, nil

	case b == '-':
		// Could be a comment or a negative number
		peekBytes, err := s.PeekBytes(2)
		if err == nil && len(peekBytes) == 2 && peekBytes[1] == '-' {
			err = s.scanComment()
			if err != nil {
				return Token{Type: TokenError, Value: []byte(err.Error())}, err
			}
			return Token{Type: TokenComment, Line: startLine, Column: startColumn}, nil
		}

		// Check if it's a negative number
		if err == nil && len(peekBytes) == 2 && isDigit(peekBytes[1]) {
			return s.scanNumber(startLine, startColumn)
		}

		// Just a single dash
		_, _ = s.ReadByte() // consume dash
		return Token{Type: TokenError, Value: []byte("unexpected '-'")},
			fmt.Errorf("unexpected '-' at line %d, column %d", startLine, startColumn)

	case b == '"':
		return s.scanString(startLine, startColumn)

	case isLetter(b):
		return s.scanName(startLine, startColumn)

	case isDigit(b):
		return s.scanNumber(startLine, startColumn)

	default:
		_, _ = s.ReadByte() // consume the unexpected character
		err := fmt.Errorf("unexpected character: %c", b)
		return Token{Type: TokenError, Value: []byte(err.Error()), Line: startLine, Column: startColumn}, err
	}
}

func (s *Scanner) UnreadToken(token Token) {
	s.token = token // Store the token to be returned next
}

// scanComment processes a comment
func (s *Scanner) scanComment() error {
	// Consume the first dash
	_, err := s.ReadByte()
	if err != nil {
		return err
	}

	// Check for second dash
	b, err := s.ReadByte()
	if err != nil {
		return err
	}
	if b != '-' {
		return s.Error("invalid comment")
	}

	// Check for block comment [[
	b, err = s.PeekByte()
	if err == nil && b == '[' {
		_, _ = s.ReadByte() // consume first [
		b, err = s.PeekByte()
		if err == nil && b == '[' {
			_, _ = s.ReadByte() // consume second [
			return s.scanBlockComment()
		}
	}

	// Line comment - consume until newline or EOF
	for {
		b, err := s.ReadByte()
		if err == io.EOF {
			return nil
		}
		if err != nil {
			return err
		}
		if b == '\n' {
			return nil
		}
	}
}

// scanBlockComment processes a block comment
func (s *Scanner) scanBlockComment() error {
	for {
		b, err := s.ReadByte()
		if err != nil {
			return s.Error("unclosed block comment")
		}

		if b == ']' {
			b, err = s.PeekByte()
			if err == nil && b == ']' {
				_, _ = s.ReadByte() // consume second ]
				return nil
			}
		}
	}
}

// scanString scans a quoted string
func (s *Scanner) scanString(startLine, startColumn int) (Token, error) {
	// Reset buffer
	s.buffer = s.buffer[:0]

	// Consume opening quote
	_, err := s.ReadByte()
	if err != nil {
		return Token{Type: TokenError, Value: []byte(err.Error())}, err
	}

	for {
		b, err := s.ReadByte()
		if err != nil {
			return Token{Type: TokenError, Value: []byte("unterminated string")}, errors.New("unterminated string")
		}

		if b == '"' {
			break
		}

		// Handle escape sequences
		if b == '\\' {
			escaped, err := s.ReadByte()
			if err != nil {
				return Token{Type: TokenError, Value: []byte("unterminated escape sequence")}, errors.New("unterminated escape sequence")
			}
			switch escaped {
			case '"':
				s.buffer = append(s.buffer, '"')
			case '\\':
				s.buffer = append(s.buffer, '\\')
			case 'n':
				s.buffer = append(s.buffer, '\n')
			case 't':
				s.buffer = append(s.buffer, '\t')
			default:
				s.buffer = append(s.buffer, '\\')
				s.buffer = append(s.buffer, escaped)
			}
		} else {
			s.buffer = append(s.buffer, b)
		}
	}

	return Token{
		Type:   TokenString,
		Value:  append([]byte(nil), s.buffer...), // Make a copy of the buffer
		Line:   startLine,
		Column: startColumn,
	}, nil
}

// scanName scans an identifier
func (s *Scanner) scanName(startLine, startColumn int) (Token, error) {
	// Reset buffer
	s.buffer = s.buffer[:0]

	// Read first character
	b, err := s.ReadByte()
	if err != nil {
		return Token{Type: TokenError, Value: []byte(err.Error())}, err
	}

	if !isLetter(b) {
		return Token{Type: TokenError, Value: []byte("name must start with letter")}, s.Error("name must start with letter")
	}
	s.buffer = append(s.buffer, b)

	// Read rest of name
	for {
		b, err := s.PeekByte()
		if err == io.EOF {
			break
		}
		if err != nil {
			return Token{Type: TokenError, Value: []byte(err.Error())}, err
		}
		if !isLetter(b) && !isDigit(b) && b != '_' {
			break
		}
		s.buffer = append(s.buffer, b)
		_, _ = s.ReadByte()
	}

	// Check if it's a boolean
	if bytes.Equal(s.buffer, []byte("true")) || bytes.Equal(s.buffer, []byte("false")) {
		return Token{
			Type:   TokenBoolean,
			Value:  append([]byte(nil), s.buffer...), // Make a copy of the buffer
			Line:   startLine,
			Column: startColumn,
		}, nil
	}

	return Token{
		Type:   TokenName,
		Value:  append([]byte(nil), s.buffer...), // Make a copy of the buffer
		Line:   startLine,
		Column: startColumn,
	}, nil
}

// scanNumber scans a numeric value
func (s *Scanner) scanNumber(startLine, startColumn int) (Token, error) {
	// Reset buffer
	s.buffer = s.buffer[:0]

	// Read first character (might be a minus sign or digit)
	b, err := s.ReadByte()
	if err != nil {
		return Token{Type: TokenError, Value: []byte(err.Error())}, err
	}
	s.buffer = append(s.buffer, b)

	// Scan the rest of the number
	hasDot := false
	for {
		b, err := s.PeekByte()
		if err != nil {
			if err == io.EOF {
				break
			}
			return Token{Type: TokenError, Value: []byte(err.Error())}, err
		}

		if b == '.' && !hasDot {
			hasDot = true
			_, _ = s.ReadByte()
			s.buffer = append(s.buffer, b)
		} else if isDigit(b) {
			_, _ = s.ReadByte()
			s.buffer = append(s.buffer, b)
		} else {
			break
		}
	}

	return Token{
		Type:   TokenNumber,
		Value:  append([]byte(nil), s.buffer...), // Make a copy of the buffer
		Line:   startLine,
		Column: startColumn,
	}, nil
}

// ScanValue processes a value and returns its Go representation
func (s *Scanner) ScanValue() (any, error) {
	token, err := s.NextToken()
	if err != nil {
		return nil, err
	}

	switch token.Type {
	case TokenString:
		return string(token.Value), nil

	case TokenBoolean:
		if bytes.Equal(token.Value, []byte("true")) {
			return true, nil
		}
		return false, nil

	case TokenNumber:
		// Convert to number
		value := string(token.Value)
		if bytes.Contains(token.Value, []byte(".")) {
			// Float
			return strconv.ParseFloat(value, 64)
		}
		// Integer
		return strconv.ParseInt(value, 10, 64)

	case TokenOpenBrace:
		// Object or array
		return s.scanObjectOrArray()

	case TokenName:
		// Name identifier - could be a special value or just a string
		return string(token.Value), nil

	default:
		return nil, fmt.Errorf("unexpected token type %v at line %d, column %d", token.Type, token.Line, token.Column)
	}
}

// scanObjectOrArray processes a map or array enclosed in braces
func (s *Scanner) scanObjectOrArray() (any, error) {
	// Initialize collections
	contents := make(map[string]any)
	var arrayElements []any
	isArray := true

	for {
		err := s.SkipWhitespace()
		if err != nil {
			return nil, err
		}

		b, err := s.PeekByte()
		if err == io.EOF {
			return nil, errors.New("unclosed object/array")
		}
		if err != nil {
			return nil, err
		}

		// Check for closing brace
		if b == '}' {
			_, _ = s.ReadByte() // consume closing brace
			if isArray && len(contents) == 0 {
				return arrayElements, nil
			}
			return contents, nil
		}

		// Handle comments
		if b == '-' {
			peekBytes, err := s.PeekBytes(2)
			if err == nil && len(peekBytes) == 2 && peekBytes[1] == '-' {
				err = s.scanComment()
				if err != nil {
					return nil, err
				}
				continue
			}
		}

		// Process key-value pair or array element
		if isLetter(b) {
			// Read name
			nameToken, err := s.scanName(s.line, s.col)
			if err != nil {
				return nil, err
			}
			name := string(nameToken.Value)

			// Skip whitespace
			err = s.SkipWhitespace()
			if err != nil {
				return nil, err
			}

			// Check if it's followed by = or {
			b, err = s.PeekByte()
			if err != nil && err != io.EOF {
				return nil, err
			}

			if b == '=' {
				// It's a key-value pair
				_, _ = s.ReadByte() // consume =
				err = s.SkipWhitespace()
				if err != nil {
					return nil, err
				}

				value, err := s.ScanValue()
				if err != nil {
					return nil, err
				}

				isArray = false
				contents[name] = value
			} else if b == '{' {
				// It's a nested object/array
				_, _ = s.ReadByte() // consume {
				value, err := s.scanObjectOrArray()
				if err != nil {
					return nil, err
				}

				isArray = false
				contents[name] = value
			} else {
				// It's a simple name as an array element
				// Try to convert to appropriate type first
				var value any = name
				// Try common conversions
				if name == "true" {
					value = true
				} else if name == "false" {
					value = false
				} else if isDigit(name[0]) || (len(name) > 1 && name[0] == '-' && isDigit(name[1])) {
					// Looks like a number, try to convert
					if hasDot(name) {
						if f, err := strconv.ParseFloat(name, 64); err == nil {
							value = f
						}
					} else {
						if i, err := strconv.ParseInt(name, 10, 64); err == nil {
							value = i
						}
					}
				}

				arrayElements = append(arrayElements, value)
			}
		} else if b == '"' {
			// String value - must be an array element
			value, err := s.ScanValue()
			if err != nil {
				return nil, err
			}
			arrayElements = append(arrayElements, value)
		} else {
			// Other value type - must be an array element
			value, err := s.ScanValue()
			if err != nil {
				return nil, err
			}
			arrayElements = append(arrayElements, value)
		}
	}
}