Oleksandr Smirnov
Oleksandr Smirnov
olexsmir@gmail.com refactor: use an actual parser instead of reflection..., 11 days ago
olexsmir@gmail.com refactor: use an actual parser instead of reflection..., 11 days ago
| 1 | package json2go |
| 2 | |
| 3 | import ( |
| 4 | "fmt" |
| 5 | "strconv" |
| 6 | ) |
| 7 | |
| 8 | type Parser struct { |
| 9 | lexer *Lexer |
| 10 | cur Token |
| 11 | peek Token |
| 12 | } |
| 13 | |
| 14 | func NewParser(l *Lexer) *Parser { |
| 15 | p := &Parser{lexer: l} |
| 16 | p.advance() // populate .peek |
| 17 | p.advance() // populate .cur |
| 18 | return p |
| 19 | } |
| 20 | |
| 21 | // Parse starts parsing and returns the root Value AST. |
| 22 | // Expects well-formed JSON with a single top-level value. |
| 23 | // Returns an error if the JSON is malformed or has unexpected tokens after the main value. |
| 24 | func (p *Parser) Parse() (Value, error) { |
| 25 | p.skipNoise() |
| 26 | v, err := p.parseValue() |
| 27 | if err != nil { |
| 28 | return Value{}, err |
| 29 | } |
| 30 | p.skipNoise() |
| 31 | if !p.got(EOF) { |
| 32 | return Value{}, fmt.Errorf("unexpected token after value: %q", p.cur.Literal) |
| 33 | } |
| 34 | return v, nil |
| 35 | } |
| 36 | |
| 37 | func (p *Parser) parseValue() (Value, error) { |
| 38 | p.skipNoise() |
| 39 | switch p.cur.Type { |
| 40 | case LBRACE: |
| 41 | return p.parseObject() |
| 42 | case LBRACKET: |
| 43 | return p.parseArray() |
| 44 | case STRING: |
| 45 | v := Value{Kind: StringValue, Str: p.cur.Literal} |
| 46 | p.advance() |
| 47 | return v, nil |
| 48 | case NUMBER: |
| 49 | n, err := strconv.ParseInt(p.cur.Literal, 10, 64) |
| 50 | if err != nil { |
| 51 | f, ferr := strconv.ParseFloat(p.cur.Literal, 64) |
| 52 | if ferr != nil { |
| 53 | return Value{}, fmt.Errorf("invalid number: %w", err) |
| 54 | } |
| 55 | p.advance() |
| 56 | return Value{Kind: DecimalValue, Float: f}, nil |
| 57 | } |
| 58 | p.advance() |
| 59 | return Value{Kind: NumberValue, Int: n}, nil |
| 60 | case DECIMAL: |
| 61 | f, err := strconv.ParseFloat(p.cur.Literal, 64) |
| 62 | if err != nil { |
| 63 | return Value{}, fmt.Errorf("invalid decimal: %w", err) |
| 64 | } |
| 65 | p.advance() |
| 66 | return Value{Kind: DecimalValue, Float: f}, nil |
| 67 | case BOOL: |
| 68 | v := Value{Kind: BoolValue, Bool: p.cur.Literal == "true"} |
| 69 | p.advance() |
| 70 | return v, nil |
| 71 | case NULL: |
| 72 | p.advance() |
| 73 | return Value{Kind: NullValue}, nil |
| 74 | default: |
| 75 | return Value{}, fmt.Errorf("unexpected token %q (%q)", p.cur.Type, p.cur.Literal) |
| 76 | } |
| 77 | } |
| 78 | |
| 79 | func (p *Parser) parseObject() (Value, error) { |
| 80 | p.advance() |
| 81 | var fields []Field |
| 82 | for { |
| 83 | p.skipNoise() |
| 84 | if p.got(RBRACE) { |
| 85 | p.advance() |
| 86 | return Value{Kind: ObjectValue, Object: fields}, nil |
| 87 | } |
| 88 | if p.got(EOF) { |
| 89 | return Value{}, fmt.Errorf("unterminated object") |
| 90 | } |
| 91 | |
| 92 | keyTok, err := p.expect(STRING) |
| 93 | if err != nil { |
| 94 | return Value{}, err |
| 95 | } |
| 96 | |
| 97 | p.skipNoise() |
| 98 | if _, cerr := p.expect(COLON); cerr != nil { |
| 99 | return Value{}, cerr |
| 100 | } |
| 101 | |
| 102 | val, err := p.parseValue() |
| 103 | if err != nil { |
| 104 | return Value{}, err |
| 105 | } |
| 106 | fields = append(fields, Field{keyTok.Literal, val}) |
| 107 | |
| 108 | p.skipNoise() |
| 109 | if p.got(COMMA) { |
| 110 | p.advance() |
| 111 | } else if !p.got(RBRACE) { |
| 112 | _, err := p.expect(RBRACE) |
| 113 | return Value{}, err |
| 114 | } |
| 115 | } |
| 116 | } |
| 117 | |
| 118 | func (p *Parser) parseArray() (Value, error) { |
| 119 | p.advance() |
| 120 | var items []Value |
| 121 | for { |
| 122 | p.skipNoise() |
| 123 | if p.got(RBRACKET) { |
| 124 | p.advance() |
| 125 | return Value{Kind: ArrayValue, Array: items}, nil |
| 126 | } |
| 127 | if p.got(EOF) { |
| 128 | return Value{}, fmt.Errorf("unterminated array") |
| 129 | } |
| 130 | |
| 131 | val, err := p.parseValue() |
| 132 | if err != nil { |
| 133 | return Value{}, err |
| 134 | } |
| 135 | items = append(items, val) |
| 136 | |
| 137 | p.skipNoise() |
| 138 | if p.got(COMMA) { |
| 139 | p.advance() |
| 140 | } else if !p.got(RBRACKET) { |
| 141 | _, err := p.expect(RBRACKET) |
| 142 | return Value{}, err |
| 143 | } |
| 144 | } |
| 145 | } |
| 146 | |
| 147 | func (p *Parser) got(kind TokenType) bool { return p.cur.Type == kind } |
| 148 | func (p *Parser) advance() Token { |
| 149 | prev := p.cur |
| 150 | p.cur = p.peek |
| 151 | p.peek = p.lexer.Next() |
| 152 | return prev |
| 153 | } |
| 154 | |
| 155 | func (p *Parser) expect(kind TokenType) (Token, error) { |
| 156 | if p.got(kind) { |
| 157 | return p.advance(), nil |
| 158 | } |
| 159 | return p.cur, fmt.Errorf("expected %s, got %s", kind, p.cur.Type) |
| 160 | } |
| 161 | |
| 162 | func (p *Parser) skipNoise() { |
| 163 | for p.cur.Type == NEWLINE || p.cur.Type == INDENT || |
| 164 | p.cur.Type == COMMENTLINE || p.cur.Type == COMMENTBLOCK { |
| 165 | p.advance() |
| 166 | } |
| 167 | } |