crq.mod

一个Modula-2语言分析器

(* CR   Main Module of Coco/R
   ==   =====================

   This is a compiler generator that produces a scanner and a parser
   from an attributed grammar, and optionally a complete small compiler.

   Original code in Oberon by Hanspeter Moessenboeck, ETH Zurich
   Ported at ETH to Apple Modula, and thence to JPI-2 Modula.

   JPI version of 27 January 1991 was then modified to make more
   portable by Pat Terry, January - October 1992

   This is the WinTel version
   This version outputs error messages in "standard" form for use with editors

Usage:
          COCOR [-options] GrammarName[.atg] [$options]

Input:
  attributed grammar   input grammar
  scanner.frm          frame file
  parser.frm           frame file
  compiler.frm         frame file (optional)

(Frame files must be in the sme directory as the grammar, or may be
found on a path specified by DOS environment variable CRFRAMES).

Output:
  <GrammarName>S.def + mod  generated scanner
  <GrammarName>P.def + mod  generated parser
  <GrammarName>.err         error numbers and corresponding error messages
  <GrammarName>.lst         source listing with error messages and trace output

Optionally

  <GrammarName>G.def + mod  generated symbolic names
  <GrammarName>.mod         generated compiler main module

Implementation restrictions
  1  too many nodes in graph (>1500)                 CRT.NewNode
  2  too many symbols (>500)                         CRT.NewSym, MovePragmas
  3  too many sets (>256 ANY-syms or SYNC syms)      CRT.NewSet,
  4  too many character classes (>250)               CRT.NewClass
  5  too many conditions in generated code (>100)    CRX.NewCondSet
  6  too many token names in "NAMES" (>100)          CRT.NewName
  7  too many states in automata (>500)              CRA.NewState

Trace output
(To activate a trace switch, write "${letter}" in the input grammar, or
invoke Coco with a second command line parameter)

  A  Prints states of automaton

  C  Generates complete compiler module

  D  Suppresses Def Mod generation

  F  Prints start symbols and followers of nonterminals.

  G  Prints the top-down graph.

  I  Trace of start symbol set computation.

  L  Forces a listing (otherwise a listing is only printed if errors are found).
  
  M  Suppresses FORWARD declarations in parser (for multipass compilers).

  N  Uses default names for symbol value constants.  This generates an
     extra module <grammar name>G, and corresponding import statements
     using constant names instead of numbers for symbols in parser and
     scanner.
     The constants are used unqualified and hence all needed constants
     have to be imported; so a complete import list for these constants
     is generated.
     There is no decision whether a constant is actually needed.

     The default conventions are (only terminals or pragmas can have names):
     single character   -->  <ASCII name (lowercase)>Sym
          eg. "+"       -->  plusSym
     character string   -->  <string>Sym
          eg. "PROGRAM" -->  PROGRAMSym
     scanner token      -->  <token name>Sym
          eg. ident     -->  identSym

  O  Trace of follow set computation (not yet implemented).

  P  Generates parser only

  S  Prints the symbol list.

  T  Suppresses generation of def and mod files (grammar tests only).

  X  Prints a cross reference list.

==========================================================================*)

MODULE CR;

  FROM CRS IMPORT lst, src, errors, directory, Error, CharAt;
  FROM CRP IMPORT Parse;
  IMPORT CRC, CRT, CRA, CRP, CRS, CRX, FileIO, Storage;
  IMPORT SYSTEM (* for TSIZE only *);

  CONST
    ATGExt = ".atg";
    LSTExt = ".lst";
    Version = "1.48q";
    ReleaseDate = "9 July 1999";

  TYPE
    INT32 = FileIO.INT32;

  VAR
    Options,
    GrammarName,
    ATGFileName,
    lstFileName: ARRAY [0 .. 63] OF CHAR;
    ll1:         BOOLEAN; (* TRUE, if grammar is LL(1) *)
    IDE,
    ok:          BOOLEAN; (* TRUE, if grammar tests ok so far *)

  MODULE ListHandler;
  (* ------------------- Source Listing and Error handler -------------- *)
  IMPORT FileIO, Storage, SYSTEM;
  IMPORT lst, CharAt, ATGFileName, IDE, errors, INT32;
  EXPORT StoreError, PrintListing;

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

  CONST
    tab = 11C;

  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
      Storage.ALLOCATE(nextErr, SYSTEM.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 # FileIO.CR) & (ch # FileIO.LF) & (ch # FileIO.EOF) DO
        line[i] := ch; INC(i); ch := CharAt(pos); INC(pos);
      END;
      eof := (i = 0) & (ch = FileIO.EOF); line[i] := 0C;
      IF ch = FileIO.CR THEN (* check for MsDos *)
        ch := CharAt(pos);
        IF ch = FileIO.LF THEN INC(pos); Extra := 0 END
      END
    END GetLine;

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

    PROCEDURE Msg (s: ARRAY OF CHAR);
      BEGIN
        FileIO.WriteString(lst, s)
      END Msg;

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

    BEGIN
      IF ~ IDE THEN Pointer END;
      CASE nr OF
         0: Msg("EOF expected")
      |  1: Msg("ident expected")
      |  2: Msg("string expected")
      |  3: Msg("badstring expected")
      |  4: Msg("number expected")
      |  5: Msg("'COMPILER' expected")
      |  6: Msg("'PRODUCTIONS' expected")
      |  7: Msg("'=' expected")
      |  8: Msg("'.' expected")
      |  9: Msg("'END' expected")
      | 10: Msg("'CHARACTERS' expected")
      | 11: Msg("'TOKENS' expected")
      | 12: Msg("'NAMES' expected")
      | 13: Msg("'PRAGMAS' expected")
      | 14: Msg("'COMMENTS' expected")
      | 15: Msg("'FROM' expected")
      | 16: Msg("'TO' expected")
      | 17: Msg("'NESTED' expected")
      | 18: Msg("'IGNORE' expected")
      | 19: Msg("'CASE' expected")
      | 20: Msg("'+' expected")
      | 21: Msg("'-' expected")
      | 22: Msg("'..' expected")
      | 23: Msg("'ANY' expected")
      | 24: Msg("'CHR' expected")
      | 25: Msg("'(' expected")
      | 26: Msg("')' expected")
      | 27: Msg("'|' expected")
      | 28: Msg("'WEAK' expected")
      | 29: Msg("'[' expected")
      | 30: Msg("']' expected")
      | 31: Msg("'{' expected")
      | 32: Msg("'}' expected")
      | 33: Msg("'SYNC' expected")
      | 34: Msg("'CONTEXT' expected")
      | 35: Msg("'<' expected")
      | 36: Msg("'>' expected")
      | 37: Msg("'<.' expected")
      | 38: Msg("'.>' expected")
      | 39: Msg("'(.' expected")
      | 40: Msg("'.)' expected")
      | 41: Msg("not expected")
      | 42: Msg("invalid TokenFactor")
      | 43: Msg("invalid Factor")
      | 44: Msg("invalid Factor")
      | 45: Msg("invalid Term")
      | 46: Msg("invalid Symbol")
      | 47: Msg("invalid SimSet")
      | 48: Msg("invalid NameDecl")
      | 49: Msg("this symbol not expected in TokenDecl")
      | 50: Msg("invalid TokenDecl")
      | 51: Msg("invalid Attribs")
      | 52: Msg("invalid Declaration")
      | 53: Msg("invalid Declaration")
      | 54: Msg("invalid Declaration")
      | 55: Msg("this symbol not expected in CR")
      | 56: Msg("invalid CR")
      
      | 102: Msg("string literal may not extend over line end");
      | 103: Msg("a literal must not have attributes");
      | 104: Msg("this symbol kind not allowed in production");
      | 105: Msg("attribute mismatch between declaration and use");
      | 106: Msg("undefined string in production");
      | 107: Msg("name declared twice");
      | 108: Msg("this type not allowed on left side of production");
      | 109: Msg("earlier semantic action was not terminated");
      | 111: Msg("no production found for grammar name");
      | 112: Msg("grammar symbol must not have attributes");
      | 113: Msg("a literal must not be declared with a structure")
      | 114: Msg("semantic action not allowed here")
      | 115: Msg("undefined name")
      | 116: Msg("attributes not allowed in token declaration")
      | 117: Msg("name does not match grammar name")
      | 118: Msg("unacceptable constant value")
      | 119: Msg("may not ignore CHR(0)")
      | 120: Msg("token may not be empty")
      | 121: Msg("token must not start with an iteration")
      | 122: Msg("only characters allowed in comment declaration")
      | 123: Msg("only terminals may be weak")
      | 124: Msg("literal tokens may not contain white space")
      | 125: Msg("comment delimiter must be 1 or 2 characters long")
      | 126: Msg("character set contains more than one character")
      | 127: Msg("could not make deterministic automaton")
      | 128: Msg("semantic action text too long - please split it")
      | 129: Msg("literal tokens may not be empty")
      ELSE Msg("Error: "); FileIO.WriteInt(lst, nr, 1);
      END;
      FileIO.WriteLn(lst)
    END PrintErr;

  PROCEDURE PrintListing;
  (* Print a source listing with error messages *)
    VAR
      nextErr:   Err;