M2compiler-V3/compiler/src/compiler.frm

MODULE -->Grammar;
(* This is an example of a rudimentary main module for use with COCO/R.
   It assumes the FileIO/Storage I/O libraries (as supplied with this
   project) are available.
   The auxiliary modules <Grammar>S (scanner) and <Grammar>P (parser)
   are assumed to have been constructed with COCO/R compiler generator. *)

  FROM -->Scanner IMPORT lst, src, errors, Error, CharAt;
  FROM -->Parser IMPORT Parse, Successful, GetUnit,
    InInst, InstBegin, InstAddActual, InstEnd, InstNameGet, InstActualGet,
    UnitNameGet, UnitIsGeneric, UnitIsDefModule,
    RefineSeen, RefineIsDef, RefineInst, RefineGen,
    RefineActualCount, RefineActualGet, SetTwoPhase;
  IMPORT
    Strings, Storage, SYSTEM, FileIO, SymTab, QbeGen, AST, Lower, Check;
    (* and any others needed *)

  TYPE
    INT32 = FileIO.INT32 (* 32 bit integers needed *);
    GName = SymTab.Name;
    GPath = ARRAY [0 .. 255] OF CHAR;

  MODULE ListHandler;
  (* ------------------- Source Listing and Error handler -------------- *)
    FROM FileIO IMPORT CR, LF, EOF, File, StdOut,
      WriteString, Write, WriteLn, WriteInt, Long0;
    FROM Storage IMPORT ALLOCATE;
    FROM SYSTEM IMPORT TSIZE;
    IMPORT lst, CharAt, errors, INT32;
    EXPORT StoreError, PrintListing, PrintConsoleErrors, ResetErrors;

    TYPE
      Err = POINTER TO ErrDesc;
      ErrDesc = RECORD
        nr, line, col: INTEGER;
        next: Err
      END;

    CONST
      tab = CHR(9);

    VAR
      firstErr, lastErr: Err;
      Extra: INTEGER;

    PROCEDURE StoreError (nr, line, col: INTEGER; pos: INT32);
    (* Store an error message for later printing *)
      VAR
        nextErr: Err;
      BEGIN
        ALLOCATE(nextErr, TSIZE(ErrDesc));
        nextErr^.nr := nr; nextErr^.line := line; nextErr^.col := col;
        nextErr^.next := NIL;
        IF firstErr = NIL
          THEN firstErr := nextErr
          ELSE lastErr^.next := nextErr
        END;
        lastErr := nextErr;
        INC(errors)
      END StoreError;

    PROCEDURE GetLine (VAR pos: INT32;
                       VAR line: ARRAY OF CHAR;
                       VAR eof: BOOLEAN);
    (* Read a source line. Return empty line if eof *)
      VAR
        ch: CHAR;
        i: CARDINAL;
      BEGIN
        i := 0; eof := FALSE; ch := CharAt(pos); INC(pos);
        WHILE (ch # CR) AND (ch # LF) AND (ch # EOF) DO
          (* bound-check: a line longer than the caller's buffer must
             truncate, not overflow it (stack corruption / segfault) *)
          IF i < HIGH(line) THEN line[i] := ch; INC(i) END;
          ch := CharAt(pos); INC(pos);
        END;
        eof := (i = 0) AND (ch = EOF); line[i] := CHR(0);
        IF ch = CR THEN (* check for MsDos *)
          ch := CharAt(pos);
          IF ch = LF THEN INC(pos); Extra := 0 END
        END
      END GetLine;

    PROCEDURE ErrText (nr: INTEGER; f: File);
    (* Error message text for nr on file f (listing and console share it) *)

      PROCEDURE Msg (s: ARRAY OF CHAR);
        BEGIN
          WriteString(f, s)
        END Msg;

      BEGIN
        CASE nr OF
        -->Errors
        (* add customized cases here *)
        | 200: Msg("duplicate identifier")
        | 201: Msg("undeclared identifier")
        | 202: Msg("module name mismatch")
        | 210: Msg("incompatible assignment")
        | 211: Msg("arithmetic operand must be numeric")
        | 212: Msg("boolean operand required")
        | 213: Msg("incompatible comparison")
        | 214: Msg("BOOLEAN condition required")
        | 215: Msg("not a RECORD type")
        | 216: Msg("unknown field")
        | 217: Msg("not an ARRAY type")
        | 218: Msg("array index must be integer")
        | 219: Msg("not a POINTER type")
        | 220: Msg("FOR needs integer variable and bounds")
        | 221: Msg("not a type name")
        | 222: Msg("set operand mismatch")
        | 223: Msg("cyclical type definition")
        | 224: Msg("ordinal type required")
        | 230: Msg("not supported yet")
        | 231: Msg("opaque type outside definition")
        | 232: Msg("bad RETURN")
        | 233: Msg("invalid call")
        | 234: Msg("invalid UTF-8 in U-literal")
        | 235: Msg("implementation does not match definition")
        | 236: Msg("bad variant record")
        ELSE         Msg("Error: "); WriteInt(f, nr, 0);
        END
      END ErrText;

    PROCEDURE PrintErr (line: ARRAY OF CHAR; nr, col: INTEGER);
    (* Print an error message *)

      PROCEDURE Pointer;
        VAR
          i: INTEGER;
        BEGIN
          WriteString(lst, "*****  ");
          i := 0;
          WHILE (i < col + Extra - 2)
            AND (i <= VAL(INTEGER, HIGH(line)))
            AND (line[i] # CHR(0)) DO
            IF line[i] = tab
              THEN Write(lst, tab)
              ELSE Write(lst, ' ')
            END;
            INC(i)
          END;
          WriteString(lst, "^ ")
        END Pointer;

      BEGIN
        Pointer;
        ErrText(nr, lst);
        WriteLn(lst)
      END PrintErr;

    PROCEDURE PrintListing;
    (* Print a source listing with error messages *)
      VAR
        nextErr: Err;
        eof, done: BOOLEAN;
        lnr, errC: INTEGER;
        srcPos: INT32;
        line: ARRAY [0 .. 4095] OF CHAR;   (* long lines (e.g. big enums) *)
      BEGIN
        WriteString(lst, "Listing:");
        WriteLn(lst); WriteLn(lst);
        srcPos := 0; nextErr := firstErr;
        GetLine(srcPos, line, eof); lnr := 1; errC := 0;
        WHILE NOT eof DO
          WriteInt(lst, lnr, 5); WriteString(lst, "  ");
          WriteString(lst, line); WriteLn(lst);
          done := FALSE;
          WHILE (nextErr # NIL) AND NOT done DO
            IF nextErr^.line = lnr THEN
              PrintErr(line, nextErr^.nr, nextErr^.col); INC(errC);
              nextErr := nextErr^.next
            ELSE done := TRUE
            END
          END;
          GetLine(srcPos, line, eof); INC(lnr);
        END;
        IF nextErr # NIL THEN
          WriteInt(lst, lnr, 5); WriteLn(lst);
          WHILE nextErr # NIL DO
            PrintErr(line, nextErr^.nr, nextErr^.col); INC(errC);
            nextErr := nextErr^.next
          END
        END;
        WriteLn(lst);
        WriteInt(lst, errC, 5); WriteString(lst, " error");
        IF errC # 1 THEN Write(lst, 's') END;
        WriteLn(lst); WriteLn(lst); WriteLn(lst);
      END PrintListing;

    PROCEDURE PrintConsoleErrors (fname: ARRAY OF CHAR);
    (* "file:line:col: message" errors on StdOut, for editor
       integration. Columns are 1-based token starts, matching the
       listing carets (NOT the col-1 of CRQ.FRM). *)

      PROCEDURE WriteNum (n: INTEGER);
      (* FileIO.WriteInt pads a leading blank; editors want none. *)
        VAR
          buf: ARRAY [0 .. 15] OF CHAR;
          i: CARDINAL;
        BEGIN
          IF n < 0 THEN Write(StdOut, "-"); n := -n END;
          i := 0;
          IF n = 0 THEN buf[0] := "0"; i := 1 END;
          WHILE (n > 0) AND (i <= HIGH(buf)) DO
            buf[i] := CHR(ORD("0") + VAL(CARDINAL, n MOD 10));
            n := n DIV 10; INC(i)
          END;
          WHILE i > 0 DO DEC(i); Write(StdOut, buf[i]) END
        END WriteNum;

      VAR
        e: Err;
      BEGIN
        e := firstErr;
        WHILE e # NIL DO
          WriteString(StdOut, fname);
          Write(StdOut, ":");
          WriteNum(e^.line);
          Write(StdOut, ":");
          WriteNum(e^.col);
          WriteString(StdOut, ": ");
          ErrText(e^.nr, StdOut);
          WriteLn(StdOut);
          e := e^.next
        END
      END PrintConsoleErrors;

    PROCEDURE ResetErrors;
    (* Fresh error list per source file in a session. *)
      BEGIN
        firstErr := NIL
      END ResetErrors;

    BEGIN
      firstErr := NIL; Extra := 1;
    END ListHandler;

  (* --------------------------- main module ------------------------------- *)

  PROCEDURE ChangeExtension (oldName, Ext: ARRAY OF CHAR;
                             VAR newName: ARRAY OF CHAR);
  (* Constructs newName as complete file name by appending ext to oldName
     Examples: (assume ext = "EXT")
           old.any ==> old.EXT
           old     ==> old.EXT
     This is not a file renaming facility, merely a string manipulation
     routine. *)
    VAR
      i, l: CARDINAL;
    BEGIN
      Strings.Assign(oldName, newName);
      i := LENGTH(oldName); l := i;
      WHILE (i > 0) AND (oldName[i -1] # '.')
 AND (oldName[i -1] # '\') AND (oldName[i -1] # '/') DO
        DEC(i)
      END;
      IF (i > 0) AND (oldName[i-1] = '.') THEN
        Strings.Delete(newName, i - 1, l + 1 - i)
      END;
      IF Ext[0] = '.' THEN Strings.Delete(Ext, 0, 1) END;
      Strings.Append(".", newName);
      Strings.Append(Ext, newName)
    END ChangeExtension;

  PROCEDURE BaseName (path: ARRAY OF CHAR; VAR name: ARRAY OF CHAR);
  (* Basename of a path without its extension (program image name). *)
    VAR
      i, j, k: CARDINAL;
    BEGIN
      i := 0; j := 0;
      WHILE (i <= HIGH(path)) AND (path[i] # CHR(0)) DO
        IF (path[i] = '/') OR (path[i] = '\') THEN j := i + 1 END;
        INC(i)
      END;
      k := j;
      WHILE (k < i) AND (path[k] # '.') DO INC(k) END;
      i := 0;
      WHILE j < k DO name[i] := path[j]; INC(i); INC(j) END;
      name[i] := CHR(0)
    END BaseName;

  VAR
    sourceName, listName, progName: ARRAY [0 .. 255] OF CHAR;
    files: ARRAY [0 .. 127] OF ARRAY [0 .. 255] OF CHAR;
    nFiles: CARDINAL;
    f: CARDINAL;
    bad: BOOLEAN;
    doCheck: BOOLEAN;   (* run the resolve/check pass (Check.Resolve) *)
    doTwoPhase: BOOLEAN; (* probe the two-phase body skipper (ProbeBody) *)

    (* generics (Increment 2): module name -> source path, def and impl *)
    genDefNm:  ARRAY [0 .. 31] OF ARRAY [0 .. 63] OF CHAR;
    genDefPth: ARRAY [0 .. 31] OF ARRAY [0 .. 255] OF CHAR;
    nGenDef: CARDINAL;
    genImpNm:  ARRAY [0 .. 31] OF ARRAY [0 .. 63] OF CHAR;
    genImpPth: ARRAY [0 .. 31] OF ARRAY [0 .. 255] OF CHAR;
    nGenImp: CARDINAL;
    uname, iname, gname: GName;
    gpath: GPath;
    isDef: BOOLEAN;

  PROCEDURE EqNm (a, b: ARRAY OF CHAR): BOOLEAN;
    VAR i: CARDINAL;
  BEGIN
    i := 0;
    WHILE (i <= HIGH(a)) AND (i <= HIGH(b)) DO
      IF a[i] # b[i] THEN RETURN FALSE END;
      IF a[i] = CHR(0) THEN RETURN TRUE END;
      INC(i)
    END;
    RETURN FALSE
  END EqNm;

  PROCEDURE NoteGen (name: ARRAY OF CHAR; isD: BOOLEAN;
                     path: ARRAY OF CHAR);
    VAR i: CARDINAL;
  BEGIN
    IF isD THEN
      IF nGenDef <= HIGH(genDefNm) THEN
        i := 0;
        WHILE (i <= HIGH(genDefNm[nGenDef])) AND (name[i] # CHR(0)) DO
          genDefNm[nGenDef][i] := name[i]; INC(i)
        END;
        genDefNm[nGenDef][i] := CHR(0);
        i := 0;
        WHILE (i <= HIGH(genDefPth[nGenDef])) AND (path[i] # CHR(0)) DO
          genDefPth[nGenDef][i] := path[i]; INC(i)
        END;
        genDefPth[nGenDef][i] := CHR(0);
        INC(nGenDef)
      END
    ELSE
      IF nGenImp <= HIGH(genImpNm) THEN
        i := 0;
        WHILE (i <= HIGH(genImpNm[nGenImp])) AND (name[i] # CHR(0)) DO
          genImpNm[nGenImp][i] := name[i]; INC(i)
        END;
        genImpNm[nGenImp][i] := CHR(0);
        i := 0;
        WHILE (i <= HIGH(genImpPth[nGenImp])) AND (path[i] # CHR(0)) DO
          genImpPth[nGenImp][i] := path[i]; INC(i)
        END;
        genImpPth[nGenImp][i] := CHR(0);
        INC(nGenImp)
      END
    END
  END NoteGen;

  PROCEDURE FindGenIdx (name: ARRAY OF CHAR; isD: BOOLEAN): INTEGER;
  (* Index of the generic module's file in the registry, or -1. *)
    VAR i: CARDINAL;
  BEGIN
    i := 0;
    IF isD THEN
      WHILE i < nGenDef DO
        IF EqNm(genDefNm[i], name) THEN RETURN VAL(INTEGER, i) END;
        INC(i)
      END
    ELSE
      WHILE i < nGenImp DO
        IF EqNm(genImpNm[i], name) THEN RETURN VAL(INTEGER, i) END;
        INC(i)
      END
    END;
    RETURN -1
  END FindGenIdx;

  PROCEDURE DoInst;
  (* Re-parse the generic's source with the instance name and the
     actuals bound, then lower the resulting unit. *)
    VAR j: CARDINAL;
      gi: INTEGER;
      saved: FileIO.File;
  BEGIN
    RefineInst(iname); RefineGen(gname); isDef := RefineIsDef();
    gi := FindGenIdx(gname, isDef);
    IF gi < 0 THEN
      FileIO.WriteString(FileIO.StdOut, "generic not found: ");
      FileIO.WriteString(FileIO.StdOut, gname);
      FileIO.WriteLn(FileIO.StdOut);
      RETURN
    END;
    IF isDef THEN
      j := 0;
      WHILE (j <= HIGH(gpath)) AND (genDefPth[gi][j] # CHR(0)) DO
        gpath[j] := genDefPth[gi][j]; INC(j)
      END;
      gpath[j] := CHR(0)
    ELSE
      j := 0;
      WHILE (j <= HIGH(gpath)) AND (genImpPth[gi][j] # CHR(0)) DO
        gpath[j] := genImpPth[gi][j]; INC(j)
      END;
      gpath[j] := CHR(0)
    END;
    saved := src;
    FileIO.Open(src, gpath, FALSE);
    IF NOT FileIO.Okay THEN src := saved; RETURN END;
    ResetErrors;
    InstBegin(iname);
    j := 0;
    WHILE j < RefineActualCount() DO
      InstAddActual(RefineActualGet(j)); INC(j)
    END;
    AST.Init;
    Lower.BeginUnit;
    Check.BeginUnit;
    SetTwoPhase(doTwoPhase);
    Parse;
    InstEnd;
    IF doCheck THEN Check.Resolve(GetUnit()) END;
    Lower.LowerUnitNow(GetUnit(), FALSE);
    IF (src # FileIO.StdIn) & (src # FileIO.StdOut) THEN
      FileIO.Close(src)
    END;
    src := saved
  END DoInst;

  BEGIN
    (* collect the session's source files (program module last) *)
    nFiles := 0;
    doCheck := TRUE;
    doTwoPhase := FALSE;
    LOOP
      FileIO.NextParameter(sourceName);
      IF (sourceName[0] = CHR(0)) OR (nFiles > HIGH(files)) THEN EXIT END;
      IF EqNm(sourceName, "-noresolve") THEN
        doCheck := FALSE
      ELSIF EqNm(sourceName, "-twophase") THEN
        doTwoPhase := TRUE
      ELSE
        Strings.Assign(sourceName, files[nFiles]);
        INC(nFiles)
      END
    END;
    IF nFiles = 0 THEN
      FileIO.WriteString(FileIO.StdOut, "No input file specified");
      HALT
    END;

    BaseName(files[nFiles - 1], progName);
    SymTab.Init;
    QbeGen.OpenModule(progName);
    Lower.BeginSession(progName);
    Check.BeginSession(progName);
    Error := StoreError;

    (* parse each unit into the shared session *)
    bad := FALSE; f := 0;
    WHILE f < nFiles DO
      Strings.Assign(files[f], sourceName);
      FileIO.Open(src, sourceName, FALSE);
      IF NOT FileIO.Okay THEN
        FileIO.WriteString(FileIO.StdOut, "Could not open input file");
        FileIO.WriteLn(FileIO.StdOut);
        HALT
      END;
      ChangeExtension(sourceName, ".LST", listName);
      FileIO.Open(lst, listName, TRUE);
      IF NOT FileIO.Okay THEN lst := FileIO.StdOut END;
      ResetErrors;

      FileIO.WriteString(FileIO.StdOut, "Parsing");
      FileIO.WriteLn(FileIO.StdOut);
      AST.Init;
      Lower.BeginUnit;
      Check.BeginUnit;
      SetTwoPhase(doTwoPhase);
      Parse;
      UnitNameGet(uname);
      IF doCheck THEN Check.Resolve(GetUnit()) END;
      IF RefineSeen() THEN
        DoInst
      ELSIF UnitIsGeneric() THEN
        (* record the generic's source for later instantiation; do not
           emit the abstract template *)
        NoteGen(uname, UnitIsDefModule(), sourceName)
      ELSE
        Lower.LowerUnitNow(GetUnit(), f = nFiles - 1)
      END;
      PrintListing;
      IF lst # FileIO.StdOut THEN FileIO.Close(lst) END;
      IF NOT Successful() THEN
        PrintConsoleErrors(sourceName);
        bad := TRUE
      END;
      (* Release the source handle before the next file: the host
         FileIO caps open files at BitSetSize (16), and a session may
         compile more sources than that. *)
      IF (src # FileIO.StdIn) & (src # FileIO.StdOut) THEN
        FileIO.Close(src)
      END;
      INC(f)
    END;

    Lower.EndSession;
    Check.EndSession;

    IF bad OR NOT SymTab.HaveProgram()
      THEN FileIO.WriteString(FileIO.StdOut, "Incorrect source")
      ELSE FileIO.WriteString(FileIO.StdOut, "Parsed correctly")
    END
  END -->Grammar.