M2compiler-V3/compiler/src/scanner.frm

IMPLEMENTATION MODULE -->modulename;

(* Scanner generated by Coco/R - using the FileIO library supplied with this project. *)

IMPORT FileIO, Storage;
FROM Storage IMPORT ALLOCATE;  (* gm2 needs this for NEW substitution *)

CONST
  noSYMB  = -->unknownsym; (*error token code*)
  (* not only for errors but also for not finished states of scanner analysis *)
  eof     = CHR(26) (* MS-DOS Keyboard eof char *);
  EOF     = CHR(0);
  EOL     = CHR(13);
  CR      = CHR(13);
  LF      = CHR(10);
  Long0   = 0;
  Long1   = 1;
  BlkSize = 16384;
TYPE
  BufBlock   = ARRAY [0 .. BlkSize-1] OF CHAR;
  Buffer     = ARRAY [0 .. 31] OF POINTER TO BufBlock;
  StartTable = ARRAY [0 .. 255] OF INTEGER;
  GetCH      = PROCEDURE (INT32): CHAR;
VAR
  lastCh,
  ch:        CHAR;       (*current input character*)
  curLine:   INTEGER;    (*current input line (may be higher than line)*)
  lineStart: INT32;      (*start position of current line*)
  apx:       INT32;      (*length of appendix (CONTEXT phrase)*)
  oldEols:   INTEGER;    (*number of EOLs in a comment*)
  bp, bp0:   INT32;      (*current position in buf
                           (bp0: position of current token)*)
  inputLen:  INT32;      (*source file size*)
  buf:       Buffer;     (*source buffer for low-level access*)
  start:     StartTable; (*start state for every character*)
  CurrentCh: GetCH;
  (* Two-phase parse probe (milestone 1): exact save/restore of the
     scanner state around a token-level shadow skip.  The pushed state
     covers the low-level buffer cursor plus every published position
     variable, so PopPos resumes the parse bit-exactly.  `errors` is
     included: re-scanning a bad character must not double-count. *)
  posDepth: CARDINAL;
  saveLastCh: ARRAY [0 .. 7] OF CHAR;
  saveCh: ARRAY [0 .. 7] OF CHAR;
  saveCurLine: ARRAY [0 .. 7] OF INTEGER;
  saveLineStart: ARRAY [0 .. 7] OF INT32;
  saveApx: ARRAY [0 .. 7] OF INT32;
  saveOldEols: ARRAY [0 .. 7] OF INTEGER;
  saveBp: ARRAY [0 .. 7] OF INT32;
  saveBp0: ARRAY [0 .. 7] OF INT32;
  saveLine: ARRAY [0 .. 7] OF INTEGER;
  saveCol: ARRAY [0 .. 7] OF INTEGER;
  saveLen: ARRAY [0 .. 7] OF CARDINAL;
  savePos: ARRAY [0 .. 7] OF INT32;
  saveNextLine: ARRAY [0 .. 7] OF INTEGER;
  saveNextCol: ARRAY [0 .. 7] OF INTEGER;
  saveNextLen: ARRAY [0 .. 7] OF CARDINAL;
  saveNextPos: ARRAY [0 .. 7] OF INT32;
  saveErrors: ARRAY [0 .. 7] OF INTEGER;

PROCEDURE Err (nr, line, col: INTEGER; pos: INT32);
  BEGIN
    INC(errors)
  END Err;

PROCEDURE NextCh;
(* Return global variable ch *)
  BEGIN
    lastCh := ch; INC(bp); ch := CurrentCh(bp);
    IF (ch = EOL) OR (ch = LF) AND (lastCh # EOL) THEN
      INC(curLine); lineStart := bp
    END
  END NextCh;

PROCEDURE Comment (): BOOLEAN;
  VAR
    level, startLine: INTEGER;
    oldLineStart: INT32;
  BEGIN
    level := 1; startLine := curLine; oldLineStart := lineStart;
    -->commentRETURN FALSE;
  END Comment;

PROCEDURE Get (VAR sym: CARDINAL);
  VAR
    state: CARDINAL;

  PROCEDURE Equal (s: ARRAY OF CHAR): BOOLEAN;
    VAR
      i: CARDINAL;
      q: INT32;
    BEGIN
      IF nextLen # LENGTH(s) THEN RETURN FALSE END;
      i := 1; q := bp0; INC(q);
      WHILE i < nextLen DO
        IF CurrentCh(q) # s[i] THEN RETURN FALSE END;
        INC(i); INC(q)
      END;
      RETURN TRUE
    END Equal;

  PROCEDURE CheckLiteral;
    BEGIN
      -->literals
    END CheckLiteral;

  BEGIN (*Get*)
    -->GetSy1
    pos := nextPos;   nextPos := bp;
    col := nextCol;   nextCol := VAL(INTEGER, bp - lineStart);
    line := nextLine; nextLine := curLine;
    len := nextLen;   nextLen := 0;
    apx := 0; state := start[ORD(ch)]; bp0 := bp;
    LOOP
      NextCh; INC(nextLen);
      CASE state OF
      -->GetSy2
      ELSE sym := noSYMB; RETURN (*NextCh already done*)
      END
    END
  END Get;

PROCEDURE GetString (pos: INT32; len: CARDINAL; VAR s: ARRAY OF CHAR);
  VAR
    i: CARDINAL;
    p: INT32;
  BEGIN
    IF len > HIGH(s) THEN len := HIGH(s) END;
    p := pos; i := 0;
    WHILE i < len DO
      s[i] := CharAt(p); INC(i); INC(p)
    END;
    s[len] := CHR(0);
  END GetString;

PROCEDURE GetName (pos: INT32; len: CARDINAL; VAR s: ARRAY OF CHAR);
  VAR
    i: CARDINAL;
    p: INT32;
  BEGIN
    IF len > HIGH(s) THEN len := HIGH(s) END;
    p := pos; i := 0;
    WHILE i < len DO
      s[i] := CurrentCh(p); INC(i); INC(p)
    END;
    s[len] := CHR(0);
  END GetName;

PROCEDURE CharAt (pos: INT32): CHAR;
  VAR
    ch: CHAR;
  BEGIN
    IF pos >= inputLen THEN RETURN EOF END;
    ch := buf[ORD(pos DIV BlkSize)]^[ORD(pos MOD BlkSize)];
    IF ch # eof THEN RETURN ch ELSE RETURN EOF END
  END CharAt;

PROCEDURE CapChAt (pos: INT32): CHAR;
  VAR
    ch: CHAR;
  BEGIN
    IF pos >= inputLen THEN RETURN EOF END;
    ch := CAP(buf[ORD(pos DIV BlkSize)]^[ORD(pos MOD BlkSize)]);
    IF ch # eof THEN RETURN ch ELSE RETURN EOF END
  END CapChAt;

PROCEDURE Reset;
  VAR
    i, read: CARDINAL;
  BEGIN (*assert: src has been opened*)
    i := 0; inputLen := 0;
    REPEAT
      NEW(buf[i]);   (* typed alloc: sets the block descriptor's count *)
      read := BlkSize; FileIO.ReadBytes(src, buf[i]^, read);
      INC(i); INC(inputLen, VAL(INT32, read))
    UNTIL read # BlkSize;
    buf[i-1]^[read] := EOF;
    curLine := 1; lineStart := -2; bp := -1;
    oldEols := 0; apx := 0; errors := 0;
    posDepth := 0;
    NextCh;
  END Reset;

PROCEDURE PushPos;
(* Save the full scanner state (milestone 1 probe). *)
  BEGIN
    IF posDepth > HIGH(saveBp) THEN RETURN END;
    saveLastCh[posDepth] := lastCh; saveCh[posDepth] := ch;
    saveCurLine[posDepth] := curLine; saveLineStart[posDepth] := lineStart;
    saveApx[posDepth] := apx; saveOldEols[posDepth] := oldEols;
    saveBp[posDepth] := bp; saveBp0[posDepth] := bp0;
    saveLine[posDepth] := line; saveCol[posDepth] := col;
    saveLen[posDepth] := len; savePos[posDepth] := pos;
    saveNextLine[posDepth] := nextLine; saveNextCol[posDepth] := nextCol;
    saveNextLen[posDepth] := nextLen; saveNextPos[posDepth] := nextPos;
    saveErrors[posDepth] := errors;
    INC(posDepth)
  END PushPos;

PROCEDURE PopPos;
(* Restore the state saved by PushPos. *)
  BEGIN
    IF posDepth = 0 THEN RETURN END;
    DEC(posDepth);
    lastCh := saveLastCh[posDepth]; ch := saveCh[posDepth];
    curLine := saveCurLine[posDepth]; lineStart := saveLineStart[posDepth];
    apx := saveApx[posDepth]; oldEols := saveOldEols[posDepth];
    bp := saveBp[posDepth]; bp0 := saveBp0[posDepth];
    line := saveLine[posDepth]; col := saveCol[posDepth];
    len := saveLen[posDepth]; pos := savePos[posDepth];
    nextLine := saveNextLine[posDepth]; nextCol := saveNextCol[posDepth];
    nextLen := saveNextLen[posDepth]; nextPos := saveNextPos[posDepth];
    errors := saveErrors[posDepth]
  END PopPos;

PROCEDURE DropPos;
(* Discard the top PushPos snapshot without restoring it (the caller has
   chosen to keep the current scanner position, e.g. after a successful
   body skip). *)
  BEGIN
    IF posDepth > 0 THEN DEC(posDepth) END
  END DropPos;

PROCEDURE SeekTo (p: INT32);
(* Reposition the scanner to the token starting at buffer offset p
   (two-phase re-parse).  Replays NextCh from the start so curLine /
   lineStart / lastCh match a fresh parse exactly; the buffer is
   in-memory so this is O(p) with no file I/O.  Published pos / len /
   line / col still describe the previous token afterwards — the caller
   must Get immediately, exactly like the priming Get in Parse. *)
  VAR k: INT32;
  BEGIN
    curLine := 1; lineStart := -2; bp := -1;
    ch := EOF; lastCh := EOF;
    oldEols := 0; apx := 0; bp0 := p;
    k := 0;
    WHILE k <= p DO NextCh; INC(k) END
  END SeekTo;

BEGIN
  -->initializations
  Error := Err; lastCh := EOF;
END -->modulename.
-->definitionDEFINITION MODULE -->modulename;

(* Scanner generated by Coco/R - using the FileIO library supplied with this project. *)

IMPORT FileIO;

TYPE
  INT32 = FileIO.INT32 (* need 32 bit integers *);

VAR
  src, lst:    FileIO.File;(*source/list files. To be opened by the main pgm*)
  directory:   ARRAY [0 .. 255] OF CHAR (*of source file*);
  line, col:   INTEGER;      (*line and column of current symbol*)
  len:         CARDINAL;     (*length of current symbol*)
  pos:         INT32;        (*file position of current symbol*)
  nextLine:    INTEGER;      (*line of lookahead symbol*)
  nextCol:     INTEGER;      (*column of lookahead symbol*)
  nextLen:     CARDINAL;     (*length of lookahead symbol*)
  nextPos:     INT32;        (*file position of lookahead symbol*)
  errors:      INTEGER;      (*number of detected errors*)
  Error:       PROCEDURE ((*nr*)INTEGER, (*line*)INTEGER, (*col*)INTEGER,
                          (*pos*)INT32);

PROCEDURE Get (VAR sym: CARDINAL);
(* Gets next symbol from source file *)

PROCEDURE GetString (pos: INT32; len: CARDINAL; VAR name: ARRAY OF CHAR);
(* Retrieves exact string of max length len from position pos in source file *)

PROCEDURE GetName (pos: INT32; len: CARDINAL; VAR name: ARRAY OF CHAR);
(* Retrieves name of symbol of length len at position pos in source file *)

PROCEDURE CharAt (pos: INT32): CHAR;
(* Returns exact character at position pos in source file *)

PROCEDURE Reset;
(* Reads and stores source file internally *)

PROCEDURE PushPos;
(* Save the full scanner state for a shadow skip (two-phase probe) *)

PROCEDURE PopPos;
(* Restore the state saved by PushPos *)

PROCEDURE DropPos;
(* Discard the top PushPos snapshot without restoring *)

PROCEDURE SeekTo (p: INT32);
(* Reposition the scanner to buffer offset p (two-phase re-parse);
   caller must Get immediately afterwards *)

END -->modulename.