| 1 | unit UTokenizer;
|
|---|
| 2 |
|
|---|
| 3 | {$mode Delphi}{$H+}
|
|---|
| 4 |
|
|---|
| 5 | interface
|
|---|
| 6 |
|
|---|
| 7 | uses
|
|---|
| 8 | Classes, SysUtils;
|
|---|
| 9 |
|
|---|
| 10 | type
|
|---|
| 11 |
|
|---|
| 12 | TTokenType = (ttIdentifier, ttWhiteSpace, ttString, ttNumber, ttBrackets,
|
|---|
| 13 | ttSymbol);
|
|---|
| 14 |
|
|---|
| 15 | TTokenParseState = (psNone, psString, psStringEnd, psNumber,
|
|---|
| 16 | psIdentifier, psRowComment, psSpecialSymbol, psBlockComment1,
|
|---|
| 17 | psBlockComment2, psBlockComment1End, psWhiteSpace);
|
|---|
| 18 |
|
|---|
| 19 | { TVoidParser }
|
|---|
| 20 |
|
|---|
| 21 | TVoidParser = class
|
|---|
| 22 | private
|
|---|
| 23 | ParseState: TTokenParseState;
|
|---|
| 24 | public
|
|---|
| 25 | TokenValue: string;
|
|---|
| 26 | TokenType: TTokenType;
|
|---|
| 27 | Position: TPoint;
|
|---|
| 28 | TokenStartPosition: TPoint;
|
|---|
| 29 | Source: TStream;
|
|---|
| 30 | procedure Open(ASource: TStream);
|
|---|
| 31 | procedure ParseNextToken;
|
|---|
| 32 | end;
|
|---|
| 33 |
|
|---|
| 34 | implementation
|
|---|
| 35 |
|
|---|
| 36 | { TVoidParser }
|
|---|
| 37 |
|
|---|
| 38 | procedure TVoidParser.Open(ASource: TStream);
|
|---|
| 39 | begin
|
|---|
| 40 | Position := Point(0, 1);
|
|---|
| 41 | Source := ASource;
|
|---|
| 42 | ParseState := psNone;
|
|---|
| 43 | ParseNextToken;
|
|---|
| 44 | end;
|
|---|
| 45 |
|
|---|
| 46 | procedure TVoidParser.ParseNextToken;
|
|---|
| 47 | var
|
|---|
| 48 | Character: Char;
|
|---|
| 49 | begin
|
|---|
| 50 | repeat
|
|---|
| 51 | if Source.Position >= Source.Size then Break;
|
|---|
| 52 | Character := Char(TMemoryStream(Source).ReadByte);
|
|---|
| 53 |
|
|---|
| 54 | if Character = #13 then Position.X := 0
|
|---|
| 55 | else if Character = #10 then Inc(Position.Y)
|
|---|
| 56 | else Inc(Position.X);
|
|---|
| 57 |
|
|---|
| 58 | case ParseState of
|
|---|
| 59 | psNone: begin
|
|---|
| 60 | TokenStartPosition := Position;
|
|---|
| 61 | if Character in [#13, #10, ' '] then begin
|
|---|
| 62 | TokenType := ttWhiteSpace;
|
|---|
| 63 | ParseState := psWhiteSpace;
|
|---|
| 64 | end else
|
|---|
| 65 | if Character = '''' then begin
|
|---|
| 66 | TokenValue := '';
|
|---|
| 67 | ParseState := psString;
|
|---|
| 68 | end else
|
|---|
| 69 | if ((Character >= 'A') and (Character <= 'Z')) or
|
|---|
| 70 | ((Character >= 'a') and (Character <= 'z')) then begin
|
|---|
| 71 | TokenValue := Character;
|
|---|
| 72 | TokenType := ttIdentifier;
|
|---|
| 73 | ParseState := psIdentifier;
|
|---|
| 74 | end else
|
|---|
| 75 | if ((Character >= '0') and (Character <= '9')) then begin
|
|---|
| 76 | TokenType := ttNumber;;
|
|---|
| 77 | TokenValue := Character;
|
|---|
| 78 | ParseState := psNumber;
|
|---|
| 79 | end else
|
|---|
| 80 | if Character = '{' then begin
|
|---|
| 81 | ParseState := psBlockComment2;
|
|---|
| 82 | end else
|
|---|
| 83 | if Character in ['/', '(', ')', '+', '-', '*', ':', ';', ','] then begin
|
|---|
| 84 | TokenValue := Character;
|
|---|
| 85 | ParseState := psSpecialSymbol;
|
|---|
| 86 | end;
|
|---|
| 87 | end;
|
|---|
| 88 | psNumber: begin
|
|---|
| 89 | if (Character >= '0') and (Character <= '9') then
|
|---|
| 90 | TokenValue := TokenValue + Character else
|
|---|
| 91 | begin
|
|---|
| 92 | ParseState := psNone;
|
|---|
| 93 | Source.Position := Source.Position - 1;
|
|---|
| 94 | Dec(Position.X);
|
|---|
| 95 | Break;
|
|---|
| 96 | end;
|
|---|
| 97 | end;
|
|---|
| 98 | psRowComment: begin
|
|---|
| 99 | if Character in [#13, #10] then begin
|
|---|
| 100 | ParseState := psNone;
|
|---|
| 101 | Source.Position := Source.Position - 1;
|
|---|
| 102 | Dec(Position.X);
|
|---|
| 103 | end;
|
|---|
| 104 | end;
|
|---|
| 105 | psWhiteSpace: begin
|
|---|
| 106 | if Character > ' ' then begin
|
|---|
| 107 | ParseState := psNone;
|
|---|
| 108 | TokenValue := '';
|
|---|
| 109 | TokenType := ttWhiteSpace;
|
|---|
| 110 | Source.Position := Source.Position - 1;
|
|---|
| 111 | Dec(Position.X);
|
|---|
| 112 | Break;
|
|---|
| 113 | end;
|
|---|
| 114 | end;
|
|---|
| 115 | psSpecialSymbol: begin
|
|---|
| 116 | if (TokenValue = '/') and (Character = '/') then begin
|
|---|
| 117 | ParseState := psRowComment;
|
|---|
| 118 | end else
|
|---|
| 119 | if (TokenValue = '(') and (Character = '*') then begin
|
|---|
| 120 | ParseState := psBlockComment1;
|
|---|
| 121 | end else
|
|---|
| 122 | if ((TokenValue = '>') and (Character = '=')) or
|
|---|
| 123 | ((TokenValue = '<') and (Character = '=')) or
|
|---|
| 124 | ((TokenValue = '<') and (Character = '>')) or
|
|---|
| 125 | ((TokenValue = ':') and (Character = '=')) then begin
|
|---|
| 126 | TokenValue := TokenValue + Character;
|
|---|
| 127 | TokenType := ttSymbol;
|
|---|
| 128 | ParseState := psNone;
|
|---|
| 129 | Break;
|
|---|
| 130 | end else begin
|
|---|
| 131 | TokenType := ttSymbol;
|
|---|
| 132 | ParseState := psNone;
|
|---|
| 133 | Source.Position := Source.Position - 1;
|
|---|
| 134 | Dec(Position.X);
|
|---|
| 135 | Break;
|
|---|
| 136 | end;
|
|---|
| 137 | end;
|
|---|
| 138 | psBlockComment1: begin
|
|---|
| 139 | if Character = '*' then begin
|
|---|
| 140 | ParseState := psBlockComment1End;
|
|---|
| 141 | end;
|
|---|
| 142 | end;
|
|---|
| 143 | psBlockComment1End: begin
|
|---|
| 144 | if Character = ')' then begin
|
|---|
| 145 | ParseState := psNone;
|
|---|
| 146 | end else begin
|
|---|
| 147 | ParseState := psBlockComment1;
|
|---|
| 148 | Source.Position := Source.Position - 1;
|
|---|
| 149 | Dec(Position.X);
|
|---|
| 150 | end;
|
|---|
| 151 | end;
|
|---|
| 152 | psBlockComment2: begin
|
|---|
| 153 | if Character = '}' then
|
|---|
| 154 | ParseState := psNone;
|
|---|
| 155 | end;
|
|---|
| 156 | psString: begin
|
|---|
| 157 | if Character = '''' then begin
|
|---|
| 158 | ParseState := psStringEnd;
|
|---|
| 159 | end else TokenValue := TokenValue + Character;
|
|---|
| 160 | end;
|
|---|
| 161 | psStringEnd: begin
|
|---|
| 162 | if Character = '''' then begin
|
|---|
| 163 | TokenValue := TokenValue + Character;
|
|---|
| 164 | ParseState := psString;
|
|---|
| 165 | end else begin
|
|---|
| 166 | TokenType := ttString;
|
|---|
| 167 | ParseState := psNone;
|
|---|
| 168 | Source.Position := Source.Position - 1;
|
|---|
| 169 | Dec(Position.X);
|
|---|
| 170 | Break;
|
|---|
| 171 | end;
|
|---|
| 172 | end;
|
|---|
| 173 | psIdentifier: begin
|
|---|
| 174 | if ((Character >= 'A') and (Character <= 'Z')) or
|
|---|
| 175 | ((Character >= 'a') and (Character <= 'z')) or
|
|---|
| 176 | ((Character >= '0') and (Character <= '9')) then begin
|
|---|
| 177 | TokenValue := TokenValue + Character;
|
|---|
| 178 | end else begin
|
|---|
| 179 | TokenType := ttIdentifier;
|
|---|
| 180 | ParseState := psNone;
|
|---|
| 181 | Source.Position := Source.Position - 1;
|
|---|
| 182 | Dec(Position.X);
|
|---|
| 183 | Break;
|
|---|
| 184 | end;
|
|---|
| 185 | end;
|
|---|
| 186 | end;
|
|---|
| 187 | until False;
|
|---|
| 188 | end;
|
|---|
| 189 |
|
|---|
| 190 | end.
|
|---|