golox/scanner/scanner_test.go

package scanner

import (
	"golox/errors"
	"golox/token"
	"testing"
)

func TestScanTokens(t *testing.T) {
	tests := []struct {
		name   string
		source string
		tokens []token.TokenType
	}{
		{
			name:   "Single character tokens",
			source: "(){}.,-+;*",
			tokens: []token.TokenType{
				token.LEFT_PAREN, token.RIGHT_PAREN, token.LEFT_BRACE, token.RIGHT_BRACE,
				token.DOT, token.COMMA, token.MINUS, token.PLUS, token.SEMICOLON, token.STAR,
			},
		},
		{
			name:   "Operators",
			source: "! != = == < <= > >=",
			tokens: []token.TokenType{
				token.BANG, token.BANG_EQUAL, token.EQUAL, token.EQUAL_EQUAL,
				token.LESS, token.LESS_EQUAL, token.GREATER, token.GREATER_EQUAL,
			},
		},
		{
			name:   "Comments",
			source: "// this is a comment\n+",
			tokens: []token.TokenType{
				token.PLUS,
			},
		},
		{
			name:   "Whitespace",
			source: " \r\t\n",
			tokens: []token.TokenType{},
		},
		{
			name:   "String literals",
			source: `"hello world"`,
			tokens: []token.TokenType{
				token.STRING,
			},
		},
		{
			name:   "Number literals",
			source: "123 45.67",
			tokens: []token.TokenType{
				token.NUMBER, token.NUMBER,
			},
		},
		{
			name:   "Identifiers and keywords",
			source: "and class else false for fun if nil or print return super this true var while",
			tokens: []token.TokenType{
				token.AND, token.CLASS, token.ELSE, token.FALSE, token.FOR, token.FUN, token.IF,
				token.NIL, token.OR, token.PRINT, token.RETURN, token.SUPER, token.THIS, token.TRUE,
				token.VAR, token.WHILE,
			},
		},
		{
			name:   "Unterminated string",
			source: `"unterminated string`,
			tokens: []token.TokenType{},
		},
		{
			name:   "Unexpected character",
			source: "@",
			tokens: []token.TokenType{},
		},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			scanner := New(tt.source, errors.NewMockErrorLogger())
			tokens := scanner.ScanTokens()

			if len(tokens) != len(tt.tokens)+1 { // +1 for EOF token
				t.Fatalf("expected %d tokens, got %d", len(tt.tokens)+1, len(tokens))
			}

			for i, tokenType := range tt.tokens {
				if tokens[i].Type != tokenType {
					t.Errorf("expected token %v, got %v", tokenType, tokens[i].Type)
				}
			}

			if tokens[len(tokens)-1].Type != token.EOF {
				t.Errorf("expected EOF token, got %v", tokens[len(tokens)-1].Type)
			}
		})
	}
}

func TestIsAtEnd(t *testing.T) {
	tests := []struct {
		name     string
		source   string
		expected bool
	}{
		{"Not at end", "abc", false},
		{"At end", "", true},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			scanner := New(tt.source, errors.NewMockErrorLogger())
			if got := scanner.isAtEnd(); got != tt.expected {
				t.Errorf("expected %v, got %v", tt.expected, got)
			}
		})
	}
}

func TestMatch(t *testing.T) {
	tests := []struct {
		name     string
		source   string
		expected bool
		char     byte
	}{
		{"Match character", "=", true, '='},
		{"No match character", "!", false, '='},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			scanner := New(tt.source, errors.NewMockErrorLogger())
			if got := scanner.match(tt.char); got != tt.expected {
				t.Errorf("expected %v, got %v", tt.expected, got)
			}
		})
	}
}

func TestPeek(t *testing.T) {
	tests := []struct {
		name     string
		source   string
		expected byte
	}{
		{"Peek character", "abc", 'a'},
		{"Peek at end", "", '\000'},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			scanner := New(tt.source, errors.NewMockErrorLogger())
			if got := scanner.peek(); got != tt.expected {
				t.Errorf("expected %v, got %v", tt.expected, got)
			}
		})
	}
}

func TestPeekNext(t *testing.T) {
	tests := []struct {
		name     string
		source   string
		expected byte
	}{
		{"Peek next character", "abc", 'b'},
		{"Peek next at end", "a", '\000'},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			scanner := New(tt.source, errors.NewMockErrorLogger())
			if got := scanner.peekNext(); got != tt.expected {
				t.Errorf("expected %v, got %v", tt.expected, got)
			}
		})
	}
}

func TestAdvance(t *testing.T) {
	tests := []struct {
		name     string
		source   string
		expected byte
	}{
		{"Advance character", "abc", 'a'},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			scanner := New(tt.source, errors.NewMockErrorLogger())
			if got := scanner.advance(); got != tt.expected {
				t.Errorf("expected %v, got %v", tt.expected, got)
			}
		})
	}
}

func TestIsAlpha(t *testing.T) {
	tests := []struct {
		name     string
		char     byte
		expected bool
	}{
		{"Is alpha", 'a', true},
		{"Is not alpha", '1', false},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			if got := isAlpha(tt.char); got != tt.expected {
				t.Errorf("expected %v, got %v", tt.expected, got)
			}
		})
	}
}

func TestIsDigit(t *testing.T) {
	tests := []struct {
		name     string
		char     byte
		expected bool
	}{
		{"Is digit", '1', true},
		{"Is not digit", 'a', false},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			if got := isDigit(tt.char); got != tt.expected {
				t.Errorf("expected %v, got %v", tt.expected, got)
			}
		})
	}
}

func TestString(t *testing.T) {
	tests := []struct {
		name     string
		source   string
		expected string
	}{
		{"Valid string", `"hello"`, "hello"},
		{"Unterminated string", `"hello`, ""},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			scanner := New(tt.source, errors.NewMockErrorLogger())
			scanner.advance() // Move to the first character of the string
			scanner.string()
			if tt.expected == "" {
				if len(scanner.tokens) != 0 {
					t.Errorf("expected no tokens, got %d", len(scanner.tokens))
				}
			} else {
				if len(scanner.tokens) != 1 {
					t.Errorf("expected 1 token, got %d", len(scanner.tokens))
				} else if scanner.tokens[0].Literal != tt.expected {
					t.Errorf("expected %v, got %v", tt.expected, scanner.tokens[0].Literal)
				}
			}
		})
	}
}

func TestNumber(t *testing.T) {
	tests := []struct {
		name     string
		source   string
		expected float64
	}{
		{"Integer number", "123", 123},
		{"Floating point number", "45.67", 45.67},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			scanner := New(tt.source, errors.NewMockErrorLogger())
			scanner.number()
			if tt.expected == 0 {
				if len(scanner.tokens) != 0 {
					t.Errorf("expected no tokens, got %d", len(scanner.tokens))
				}
			} else {
				if len(scanner.tokens) != 1 {
					t.Errorf("expected 1 token, got %d", len(scanner.tokens))
				} else if scanner.tokens[0].Literal != tt.expected {
					t.Errorf("expected %v, got %v", tt.expected, scanner.tokens[0].Literal)
				}
			}
		})
	}
}

func TestScanToken(t *testing.T) {
	tests := []struct {
		name     string
		source   string
		expected token.TokenType
		lexeme   string
	}{
		{"Left paren", "(", token.LEFT_PAREN, "("},
		{"Right paren", ")", token.RIGHT_PAREN, ")"},
		{"Left brace", "{", token.LEFT_BRACE, "{"},
		{"Right brace", "}", token.RIGHT_BRACE, "}"},
		{"Comma", ",", token.COMMA, ","},
		{"Dot", ".", token.DOT, "."},
		{"Minus", "-", token.MINUS, "-"},
		{"Plus", "+", token.PLUS, "+"},
		{"Semicolon", ";", token.SEMICOLON, ";"},
		{"Star", "*", token.STAR, "*"},
		{"Bang", "!", token.BANG, "!"},
		{"Bang equal", "!=", token.BANG_EQUAL, "!="},
		{"Equal", "=", token.EQUAL, "="},
		{"Equal equal", "==", token.EQUAL_EQUAL, "=="},
		{"Less", "<", token.LESS, "<"},
		{"Less equal", "<=", token.LESS_EQUAL, "<="},
		{"Greater", ">", token.GREATER, ">"},
		{"Greater equal", ">=", token.GREATER_EQUAL, ">="},
		{"Slash", "/", token.SLASH, "/"},
		{"Comment", "// comment\n", token.EOF, ""},
		{"Whitespace", " \r\t\n", token.EOF, ""},
		{"String", `"hello"`, token.STRING, `"hello"`},
		{"Number", "123", token.NUMBER, "123"},
		{"Identifier", "var", token.VAR, "var"},
		{"Unexpected character", "@", token.EOF, ""},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			scanner := New(tt.source, errors.NewMockErrorLogger())
			scanner.scanToken()
			if len(scanner.tokens) > 0 {
				if scanner.tokens[0].Type != tt.expected {
					t.Errorf("expected %v, got %v", tt.expected, scanner.tokens[0].Type)
				}
				if scanner.tokens[0].Lexeme != tt.lexeme {
					t.Errorf("expected %v, got %v", tt.lexeme, scanner.tokens[0].Lexeme)
				}
			} else if tt.expected != token.EOF {
				t.Errorf("expected %v, got no tokens", tt.expected)
			}
		})
	}
}

func TestIdentifier(t *testing.T) {
	tests := []struct {
		name     string
		source   string
		expected token.TokenType
	}{
		{"Keyword and", "and", token.AND},
		{"Keyword class", "class", token.CLASS},
		{"Keyword else", "else", token.ELSE},
		{"Keyword false", "false", token.FALSE},
		{"Keyword for", "for", token.FOR},
		{"Keyword fun", "fun", token.FUN},
		{"Keyword if", "if", token.IF},
		{"Keyword nil", "nil", token.NIL},
		{"Keyword or", "or", token.OR},
		{"Keyword print", "print", token.PRINT},
		{"Keyword return", "return", token.RETURN},
		{"Keyword super", "super", token.SUPER},
		{"Keyword this", "this", token.THIS},
		{"Keyword true", "true", token.TRUE},
		{"Keyword var", "var", token.VAR},
		{"Keyword while", "while", token.WHILE},
		{"Identifier", "myVar", token.IDENTIFIER},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			scanner := New(tt.source, errors.NewMockErrorLogger())
			scanner.identifier()
			if len(scanner.tokens) != 1 {
				t.Fatalf("expected 1 token, got %d", len(scanner.tokens))
			}
			if scanner.tokens[0].Type != tt.expected {
				t.Errorf("expected %v, got %v", tt.expected, scanner.tokens[0].Type)
			}
		})
	}
}