diff --git a/src/shared/xparser.pas b/src/shared/xparser.pas
index 4e45fc6849cf1e6861e94c87664996b1153e6b6d..595d300f99f32291ae9af1da261b3a8907f95106 100644 (file)
--- a/src/shared/xparser.pas
+++ b/src/shared/xparser.pas
SignedNumbers, // allow signed numbers; otherwise sign will be TTDelim
DollarIsId, // allow dollar in identifiers; otherwise dollar will be TTDelim
DotIsId, // allow dot in identifiers; otherwise dot will be TTDelim
+ DashIsId, // '-' can be part of identifier (but identifier cannot start with '-')
+ HtmlColors, // #rgb or #rrggbb colors
PascalComments // allow `{}` pascal comments
);
TOptions = set of TOption;
private
type
TAnsiCharSet = set of AnsiChar;
+ const
+ CharBufSize = 8;
private
mLine, mCol: Integer;
- mCurChar, mNextChar: AnsiChar;
+ // chars for 'unget'
+ mCharBuf: packed array [0..CharBufSize-1] of AnsiChar;
+ mCharBufUsed: Integer;
+ mCharBufPos: Integer;
+ mEofHit: Boolean; // no more chars to load into mCharBuf
mOptions: TOptions;
mTokChar: AnsiChar; // for delimiters
mTokInt: Integer;
+ private
+ procedure fillCharBuf ();
+ function popFrontChar (): AnsiChar; inline; // never drains char buffer (except on "total EOF")
+ function peekCurChar (): AnsiChar; inline;
+ function peekNextChar (): AnsiChar; inline;
+ function peekChar (dest: Integer): AnsiChar; inline;
+
protected
- procedure warmup (); // called in constructor to warm up the system
- procedure loadNextChar (); virtual; abstract; // loads next char into mNextChar; #0 means 'eof'
+ function loadChar (): AnsiChar; virtual; abstract; // loads next char; #0 means 'eof'
+
+ public
+ function isIdStartChar (ch: AnsiChar): Boolean; inline;
+ function isIdMidChar (ch: AnsiChar): Boolean; inline;
public
constructor Create (aopts: TOptions=[TOption.SignedNumbers]);
procedure error (const amsg: AnsiString); noreturn;
procedure errorfmt (const afmt: AnsiString; const args: array of const); noreturn;
- function isEOF (): Boolean; inline;
-
function skipChar (): Boolean; // returns `false` on eof
function skipBlanks (): Boolean; // ...and comments; returns `false` on eof
function skipToken1 (): Boolean;
{$ENDIF}
+ function isEOF (): Boolean; inline;
+ function isId (): Boolean; inline;
+ function isInt (): Boolean; inline;
+ function isStr (): Boolean; inline;
+ function isDelim (): Boolean; inline;
function isIdOrStr (): Boolean; inline;
function expectId (): AnsiString;
function expectStr (allowEmpty: Boolean=false): AnsiString;
function expectInt (): Integer;
- function expectStrOrId (allowEmpty: Boolean=false): AnsiString;
+ function expectIdOrStr (allowEmpty: Boolean=false): AnsiString;
procedure expectTT (ttype: Integer);
function eatTT (ttype: Integer): Boolean;
property col: Integer read mCol;
property line: Integer read mLine;
- property curChar: AnsiChar read mCurChar;
- property nextChar: AnsiChar read mNextChar;
+ property curChar: AnsiChar read peekCurChar;
+ property nextChar: AnsiChar read peekNextChar;
// token start
property tokCol: Integer read mTokCol;
mBufPos: Integer;
protected
- procedure loadNextChar (); override; // loads next char into mNextChar; #0 means 'eof'
+ function loadChar (): AnsiChar; override; // loads next char; #0 means 'eof'
public
constructor Create (const fname: AnsiString; aopts: TOptions=[TOption.SignedNumbers]);
mPos: Integer;
protected
- procedure loadNextChar (); override; // loads next char into mNextChar; #0 means 'eof'
+ function loadChar (): AnsiChar; override; // loads next char; #0 means 'eof'
public
constructor Create (const astr: AnsiString; aopts: TOptions=[TOption.SignedNumbers]);
begin
mLine := 1;
mCol := 1;
- mCurChar := #0;
- mNextChar := #0;
+ mCharBufUsed := 0;
+ mCharBufPos := 0;
+ mEofHit := false;
mTokType := TTNone;
mTokStr := '';
mTokChar := #0;
mTokInt := 0;
mOptions := aopts;
- warmup();
skipToken();
+ // fuck you, BOM!
+ {
+ if (mBufLen >= 3) and (mBuffer[0] = #$EF) and (mBuffer[1] = #$BB) and (mBuffer[2] = #$BF) then
+ begin
+ for f := 3 to mBufLen-1 do mBuffer[f-3] := mBuffer[f];
+ Dec(mBufLen, 3);
+ end;
+ }
end;
end;
-function TTextParser.isEOF (): Boolean; inline; begin result := (mCurChar = #0); end;
-
+function TTextParser.isIdStartChar (ch: AnsiChar): Boolean; inline;
+begin
+ result :=
+ (ch = '_') or
+ ((ch >= 'A') and (ch <= 'Z')) or
+ ((ch >= 'a') and (ch <= 'z')) or
+ (ch >= #128) or
+ ((ch = '$') and (TOption.DollarIsId in mOptions)) or
+ ((ch = '.') and (TOption.DotIsId in mOptions));
+end;
-procedure TTextParser.warmup ();
+function TTextParser.isIdMidChar (ch: AnsiChar): Boolean; inline;
begin
- mNextChar := ' ';
- loadNextChar();
- mCurChar := mNextChar;
- if (mNextChar <> #0) then loadNextChar();
+ result :=
+ ((ch >= '0') and (ch <= '9')) or
+ ((ch = '-') and (TOption.DashIsId in mOptions)) or
+ isIdStartChar(ch);
end;
-function TTextParser.skipChar (): Boolean;
+procedure TTextParser.fillCharBuf ();
+var
+ ch: AnsiChar;
begin
- if (mCurChar = #0) then begin result := false; exit; end;
- if (mCurChar = #10) then begin mCol := 1; Inc(mLine); end else Inc(mCol);
- mCurChar := mNextChar;
- if (mCurChar = #0) then begin result := false; exit; end;
- loadNextChar();
- // skip CR in CR/LF
- if (mCurChar = #13) then
+ if (mEofHit) then begin mCharBuf[mCharBufPos] := #0; exit; end;
+ while (not mEofHit) and (mCharBufUsed < CharBufSize) do
begin
- if (mNextChar = #10) then loadNextChar();
- mCurChar := #10;
+ ch := loadChar();
+ mCharBuf[(mCharBufPos+mCharBufUsed) mod CharBufSize] := ch;
+ if (ch = #0) then begin mEofHit := true; break; end;
+ Inc(mCharBufUsed);
end;
+end;
+
+
+// never drains char buffer (except on "total EOF")
+function TTextParser.popFrontChar (): AnsiChar; inline;
+begin
+ if (mEofHit) and (mCharBufUsed = 0) then begin result := #0; exit; end;
+ assert(mCharBufUsed > 0);
+ result := mCharBuf[mCharBufPos];
+ mCharBufPos := (mCharBufPos+1) mod CharBufSize;
+ Dec(mCharBufUsed);
+ if (not mEofHit) and (mCharBufUsed = 0) then fillCharBuf();
+end;
+
+function TTextParser.peekCurChar (): AnsiChar; inline;
+begin
+ if (mCharBufUsed = 0) and (not mEofHit) then fillCharBuf();
+ result := mCharBuf[mCharBufPos]; // it is safe, 'cause `fillCharBuf()` will put #0 on "total EOF"
+end;
+
+function TTextParser.peekNextChar (): AnsiChar; inline;
+begin
+ if (mCharBufUsed < 2) and (not mEofHit) then fillCharBuf();
+ if (mCharBufUsed < 2) then result := #0 else result := mCharBuf[(mCharBufPos+1) mod CharBufSize];
+end;
+
+function TTextParser.peekChar (dest: Integer): AnsiChar; inline;
+begin
+ if (dest < 0) or (dest >= CharBufSize) then error('internal text parser error');
+ if (mCharBufUsed < dest+1) then fillCharBuf();
+ if (mCharBufUsed < dest+1) then result := #0 else result := mCharBuf[(mCharBufPos+dest) mod CharBufSize];
+end;
+
+
+function TTextParser.skipChar (): Boolean;
+var
+ ch: AnsiChar;
+begin
+ ch := popFrontChar();
+ if (ch = #0) then begin result := false; exit; end;
result := true;
+ // CR?
+ case ch of
+ #10:
+ begin
+ mCol := 1;
+ Inc(mLine);
+ end;
+ #13:
+ begin
+ mCol := 1;
+ Inc(mLine);
+ if (mCharBufUsed > 0) and (mCharBuf[0] = #10) then
+ begin
+ if (popFrontChar() = #0) then result := false;
+ end;
+ end;
+ else
+ Inc(mCol);
+ end;
end;
var
level: Integer;
begin
- while not isEOF do
+ //writeln('line=', mLine, '; col=', mCol, '; char0=', Integer(peekChar(0)));
+ if (mLine = 1) and (mCol = 1) and
+ (peekChar(0) = #$EF) and
+ (peekChar(1) = #$BB) and
+ (peekChar(2) = #$BF) then
+ begin
+ skipChar();
+ skipChar();
+ skipChar();
+ end;
+
+ while (curChar <> #0) do
begin
if (curChar = '/') then
begin
// single-line comment
if (nextChar = '/') then
begin
- while not isEOF and (curChar <> #10) do skipChar();
+ //writeln('spos=(', mLine, ',', mCol, ')');
+ while (curChar <> #0) and (curChar <> #10) and (curChar <> #13) do skipChar();
skipChar(); // skip EOL
+ //writeln('{', curChar, '}');
+ //writeln('epos=(', mLine, ',', mCol, ')');
continue;
end;
// multline comment
// skip comment start
skipChar();
skipChar();
- while not isEOF do
+ while (curChar <> #0) do
begin
if (curChar = '*') and (nextChar = '/') then
begin
skipChar();
skipChar();
level := 1;
- while not isEOF do
+ while (curChar <> #0) do
begin
if (curChar = '+') and (nextChar = '/') then
begin
// pascal comment; skip comment start
skipChar();
skipChar();
- while not isEOF do
+ while (curChar <> #0) do
begin
if (curChar = '*') and (nextChar = ')') then
begin
begin
// pascal comment; skip comment start
skipChar();
- while not isEOF do
+ while (curChar <> #0) do
begin
if (curChar = '}') then
begin
if (curChar > ' ') then break;
skipChar(); // skip blank
end;
- result := not isEOF;
+ result := (curChar <> #0);
end;
end;
// default base
if (base < 0) then base := 10;
- if (digitInBase(curChar, base) < 0) then raise Exception.Create('invalid number');
+ if (digitInBase(curChar, base) < 0) then error('invalid number');
mTokType := TTInt;
mTokInt := 0; // just in case
- while not isEOF do
+ while (curChar <> #0) do
begin
+ if (curChar = '_') then
+ begin
+ skipChar();
+ if (curChar = #0) then break;
+ end;
n := digitInBase(curChar, base);
if (n < 0) then break;
n := mTokInt*10+n;
- if (n < 0) or (n < mTokInt) then raise Exception.Create('integer overflow');
+ if (n < 0) or (n < mTokInt) then error('integer overflow');
mTokInt := n;
skipChar();
end;
// check for valid number end
- if not isEOF then
+ if (curChar <> #0) then
begin
- if (curChar = '.') then raise Exception.Create('floating numbers aren''t supported yet');
- if (curChar = '_') or ((curChar >= 'A') and (curChar <= 'Z')) or ((curChar >= 'a') and (curChar <= 'z')) or (curChar >= #128) then
- begin
- raise Exception.Create('invalid number');
- end;
+ if (curChar = '.') then error('floating numbers aren''t supported yet');
+ if (isIdMidChar(curChar)) then error('invalid number');
end;
if neg then mTokInt := -mTokInt;
end;
mTokStr := ''; // just in case
qch := curChar;
skipChar(); // skip starting quote
- while not isEOF do
+ while (curChar <> #0) do
begin
// escape
if (qch = '"') and (curChar = '\') then
begin
- if (nextChar = #0) then raise Exception.Create('unterminated string escape');
+ if (nextChar = #0) then error('unterminated string escape');
ch := nextChar;
// skip backslash and escape type
skipChar();
'x', 'X': // hex escape
begin
n := digitInBase(curChar, 16);
- if (n < 0) then raise Exception.Create('invalid hexstr escape');
+ if (n < 0) then error('invalid hexstr escape');
skipChar();
if (digitInBase(curChar, 16) > 0) then
begin
begin
mTokType := TTId;
mTokStr := ''; // just in case
- while (curChar = '_') or ((curChar >= '0') and (curChar <= '9')) or
- ((curChar >= 'A') and (curChar <= 'Z')) or
- ((curChar >= 'a') and (curChar <= 'z')) or
- (curChar >= #128) or
- ((TOption.DollarIsId in mOptions) and (curChar = '$')) or
- ((TOption.DotIsId in mOptions) and (curChar = '.') and (nextChar <> '.')) do
+ while (isIdMidChar(curChar)) do
begin
+ if (curChar = '.') and (nextChar = '.') then break; // dotdot is a token by itself
mTokStr += curChar;
skipChar();
end;
end;
+var
+ xpos: Integer;
begin
- mTokType := TTEOF;
+ mTokType := TTNone;
mTokStr := '';
mTokChar := #0;
mTokInt := 0;
if not skipBlanks() then
begin
result := false;
+ mTokType := TTEOF;
mTokLine := mLine;
mTokCol := mCol;
exit;
if (curChar >= '0') and (curChar <= '9') then begin parseInt(); exit; end;
// string?
- if (curChar = '"') or (curChar = '''') then begin parseString(); exit; end;
+ if (curChar = '"') or (curChar = '''') or (curChar = '`') then begin parseString(); exit; end;
+
+ // html color?
+ if (curChar = '#') and (TOption.HtmlColors in mOptions) then
+ begin
+ if (digitInBase(peekChar(1), 16) >= 0) and (digitInBase(peekChar(2), 16) >= 0) and (digitInBase(peekChar(3), 16) >= 0) then
+ begin
+ if (digitInBase(peekChar(4), 16) >= 0) and (digitInBase(peekChar(5), 16) >= 0) and (digitInBase(peekChar(6), 16) >= 0) then xpos := 7 else xpos := 4;
+ if (not isIdMidChar(peekChar(xpos))) then
+ begin
+ mTokType := TTId;
+ mTokStr := '';
+ while (xpos > 0) do
+ begin
+ mTokStr += curChar;
+ skipChar();
+ Dec(xpos);
+ end;
+ exit;
+ end;
+ end;
+ end;
// identifier?
- if (curChar = '_') or ((curChar >= 'A') and (curChar <= 'Z')) or ((curChar >= 'a') and (curChar <= 'z')) or (curChar >= #128) then begin parseId(); exit; end;
- if (TOption.DollarIsId in mOptions) and (curChar = '$') then begin parseId(); exit; end;
- if (TOption.DotIsId in mOptions) and (curChar = '.') and (nextChar <> '.') then begin parseId(); exit; end;
+ if (isIdStartChar(curChar)) then
+ begin
+ if (curChar = '.') and (nextChar = '.') then
+ begin
+ // nothing to do here, as dotdot is a token by itself
+ end
+ else
+ begin
+ parseId();
+ exit;
+ end;
+ end;
// known delimiters?
mTokChar := curChar;
end;
-function TTextParser.isIdOrStr (): Boolean; inline;
-begin
- result := (mTokType = TTId) or (mTokType = TTStr);
-end;
+function TTextParser.isEOF (): Boolean; inline; begin result := (mTokType = TTEOF); end;
+function TTextParser.isId (): Boolean; inline; begin result := (mTokType = TTId); end;
+function TTextParser.isInt (): Boolean; inline; begin result := (mTokType = TTInt); end;
+function TTextParser.isStr (): Boolean; inline; begin result := (mTokType = TTStr); end;
+function TTextParser.isDelim (): Boolean; inline; begin result := (mTokType = TTDelim); end;
+function TTextParser.isIdOrStr (): Boolean; inline; begin result := (mTokType = TTId) or (mTokType = TTStr); end;
function TTextParser.expectId (): AnsiString;
begin
- if (mTokType <> TTId) then raise Exception.Create('identifier expected');
+ if (mTokType <> TTId) then error('identifier expected');
result := mTokStr;
skipToken();
end;
@@ -672,11 +814,11 @@ procedure TTextParser.expectId (const aid: AnsiString; caseSens: Boolean=true);
begin
if caseSens then
begin
- if (mTokType <> TTId) or (mTokStr <> aid) then raise Exception.Create('identifier '''+aid+''' expected');
+ if (mTokType <> TTId) or (mTokStr <> aid) then error('identifier '''+aid+''' expected');
end
else
begin
- if (mTokType <> TTId) or (not strEquCI1251(mTokStr, aid)) then raise Exception.Create('identifier '''+aid+''' expected');
+ if (mTokType <> TTId) or (not strEquCI1251(mTokStr, aid)) then error('identifier '''+aid+''' expected');
end;
skipToken();
end;
function TTextParser.expectStr (allowEmpty: Boolean=false): AnsiString;
begin
- if (mTokType <> TTStr) then raise Exception.Create('string expected');
- if (not allowEmpty) and (Length(mTokStr) = 0) then raise Exception.Create('non-empty string expected');
+ if (mTokType <> TTStr) then error('string expected');
+ if (not allowEmpty) and (Length(mTokStr) = 0) then error('non-empty string expected');
result := mTokStr;
skipToken();
end;
-function TTextParser.expectStrOrId (allowEmpty: Boolean=false): AnsiString;
+function TTextParser.expectIdOrStr (allowEmpty: Boolean=false): AnsiString;
begin
case mTokType of
TTStr:
- if (not allowEmpty) and (Length(mTokStr) = 0) then raise Exception.Create('non-empty string expected');
+ if (not allowEmpty) and (Length(mTokStr) = 0) then error('non-empty string expected');
TTId:
begin end;
else
- raise Exception.Create('string or identifier expected');
+ error('string or identifier expected');
end;
result := mTokStr;
skipToken();
function TTextParser.expectInt (): Integer;
begin
- if (mTokType <> TTInt) then raise Exception.Create('string expected');
+ if (mTokType <> TTInt) then error('string expected');
result := mTokInt;
skipToken();
end;
procedure TTextParser.expectTT (ttype: Integer);
begin
- if (mTokType <> ttype) then raise Exception.Create('unexpected token');
+ if (mTokType <> ttype) then error('unexpected token');
skipToken();
end;
procedure TTextParser.expectDelim (const ch: AnsiChar);
begin
- if (mTokType <> TTDelim) or (mTokChar <> ch) then raise Exception.CreateFmt('delimiter ''%s'' expected', [ch]);
+ if (mTokType <> TTDelim) or (mTokChar <> ch) then errorfmt('delimiter ''%s'' expected', [ch]);
skipToken();
end;
function TTextParser.expectDelims (const ch: TAnsiCharSet): AnsiChar;
begin
- if (mTokType <> TTDelim) then raise Exception.Create('delimiter expected');
- if not (mTokChar in ch) then raise Exception.Create('delimiter expected');
+ if (mTokType <> TTDelim) then error('delimiter expected');
+ if not (mTokChar in ch) then error('delimiter expected');
result := mTokChar;
skipToken();
end;
GetMem(mBuffer, BufSize);
mBufPos := 0;
mBufLen := mFile.Read(mBuffer^, BufSize);
- if (mBufLen < 0) then raise Exception.Create('TFileTextParser: read error');
+ if (mBufLen < 0) then error('TFileTextParser: read error');
inherited Create(aopts);
end;
constructor TFileTextParser.Create (st: TStream; astOwned: Boolean=true; aopts: TOptions=[TOption.SignedNumbers]);
begin
- if (st = nil) then raise Exception.Create('cannot create parser for nil stream');
+ if (st = nil) then error('cannot create parser for nil stream');
mFile := st;
mStreamOwned := astOwned;
GetMem(mBuffer, BufSize);
mBufPos := 0;
mBufLen := mFile.Read(mBuffer^, BufSize);
- if (mBufLen < 0) then raise Exception.Create('TFileTextParser: read error');
+ if (mBufLen < 0) then error('TFileTextParser: read error');
inherited Create(aopts);
end;
mBuffer := nil;
mBufPos := 0;
mBufLen := 0;
- if mStreamOwned then mFile.Free();
- mFile := nil;
+ if (mStreamOwned) then FreeAndNil(mFile) else mFile := nil;
inherited;
end;
-procedure TFileTextParser.loadNextChar ();
+function TFileTextParser.loadChar (): AnsiChar;
begin
- if (mBufLen = 0) then begin mNextChar := #0; exit; end;
+ if (mBufLen = 0) then begin result := #0; exit; end;
if (mBufPos >= mBufLen) then
begin
mBufLen := mFile.Read(mBuffer^, BufSize);
- if (mBufLen < 0) then raise Exception.Create('TFileTextParser: read error');
- if (mBufLen = 0) then begin mNextChar := #0; exit; end;
+ if (mBufLen < 0) then error('TFileTextParser: read error');
+ if (mBufLen = 0) then begin result := #0; exit; end;
mBufPos := 0;
end;
assert(mBufPos < mBufLen);
- mNextChar := mBuffer[mBufPos];
+ result := mBuffer[mBufPos];
Inc(mBufPos);
- if (mNextChar = #0) then mNextChar := ' ';
+ if (result = #0) then result := ' ';
end;
end;
-procedure TStrTextParser.loadNextChar ();
+function TStrTextParser.loadChar (): AnsiChar;
begin
- mNextChar := #0;
+ result := #0;
if (mPos > Length(mStr)) then exit;
- mNextChar := mStr[mPos]; Inc(mPos);
- if (mNextChar = #0) then mNextChar := ' ';
+ result := mStr[mPos];
+ Inc(mPos);
+ if (result = #0) then result := ' ';
end;