scanner.c

来自「clips源代码」· C语言 代码 · 共 788 行 · 第 1/2 页

   /*******************************************************/
   /*      "C" Language Integrated Production System      */
   /*                                                     */
   /*             CLIPS Version 6.20  01/31/02            */
   /*                                                     */
   /*                    SCANNER MODULE                   */
   /*******************************************************/

/*************************************************************/
/* Purpose: Routines for scanning lexical tokens from an     */
/*   input source.                                           */
/*                                                           */
/* Principal Programmer(s):                                  */
/*      Gary D. Riley                                        */
/*                                                           */
/* Contributing Programmer(s):                               */
/*      Chris Culbert                                        */
/*      Brian Donnell                                        */
/*                                                           */
/* Revision History:                                         */
/*                                                           */
/*************************************************************/

#define _SCANNER_SOURCE_

#include <ctype.h>
#include <stdio.h>
#define _STDIO_INCLUDED_
#include <string.h>
#include <limits.h>

#include "setup.h"
#include "constant.h"
#include "envrnmnt.h"
#include "router.h"
#include "symbol.h"
#include "utility.h"
#include "memalloc.h"

#include "scanner.h"

#include <stdlib.h>

/***************************************/
/* LOCAL INTERNAL FUNCTION DEFINITIONS */
/***************************************/

   static void                   *ScanSymbol(void *,char *,int,unsigned short *);
   static void                   *ScanString(void *,char *);
   static void                    ScanNumber(void *,char *,struct token *);
   static void                    DeallocateScannerData(void *);

/************************************************/
/* InitializeScannerData: Allocates environment */
/*    data for scanner routines.                */
/************************************************/
globle void InitializeScannerData(
  void *theEnv)
  {
   AllocateEnvironmentData(theEnv,SCANNER_DATA,sizeof(struct scannerData),DeallocateScannerData);
  }
  
/**************************************************/
/* DeallocateScannerData: Deallocates environment */
/*    data for scanner routines.                  */
/**************************************************/
static void DeallocateScannerData(
  void *theEnv)
  {
   if (ScannerData(theEnv)->GlobalMax !=  0)
     { genfree(theEnv,ScannerData(theEnv)->GlobalString,ScannerData(theEnv)->GlobalMax); }
  }

/***********************************************************************/
/* GetToken: Reads next token from the input stream. The pointer to    */
/*   the token data structure passed as an argument is set to contain  */
/*   the type of token (e.g., symbol, string, integer, etc.), the data */
/*   value for the token (i.e., a symbol table location if it is a     */
/*   symbol or string, an integer table location if it is an integer), */
/*   and the pretty print representation.                              */
/***********************************************************************/
globle void GetToken(
 void *theEnv,
 char *logicalName,
 struct token *theToken)
 {
   int inchar;
   unsigned short type;

   /*=======================================*/
   /* Set Unknown default values for token. */
   /*=======================================*/

   theToken->type = UNKNOWN_VALUE;
   theToken->value = NULL;
   theToken->printForm = "unknown";
   ScannerData(theEnv)->GlobalPos = 0;
   ScannerData(theEnv)->GlobalMax = 0;

   /*==============================================*/
   /* Remove all white space before processing the */
   /* GetToken() request.                          */
   /*==============================================*/

   inchar = EnvGetcRouter(theEnv,logicalName);
   while ((inchar == ' ') || (inchar == '\n') || (inchar == '\f') ||
          (inchar == '\r') || (inchar == ';') || (inchar == '\t'))
     {
      /*=======================*/
      /* Remove comment lines. */
      /*=======================*/

      if (inchar == ';')
        {
         inchar = EnvGetcRouter(theEnv,logicalName);
         while ((inchar != '\n') && (inchar != '\r') && (inchar != EOF) )
           { inchar = EnvGetcRouter(theEnv,logicalName); }
        }
      inchar = EnvGetcRouter(theEnv,logicalName);
     }

   /*==========================*/
   /* Process Symbolic Tokens. */
   /*==========================*/

   if (isalpha(inchar))
     {
      theToken->type = SYMBOL;
      EnvUngetcRouter(theEnv,inchar,logicalName);
      theToken->value = (void *) ScanSymbol(theEnv,logicalName,0,&type);
      theToken->printForm = ValueToString(theToken->value);
     }

   /*===============================================*/
   /* Process Number Tokens beginning with a digit. */
   /*===============================================*/

   else if (isdigit(inchar))
     {
      EnvUngetcRouter(theEnv,inchar,logicalName);
      ScanNumber(theEnv,logicalName,theToken);
     }

   else switch (inchar)
     {
      /*========================*/
      /* Process String Tokens. */
      /*========================*/

      case '"':
         theToken->value = (void *) ScanString(theEnv,logicalName);
         theToken->type = STRING;
         theToken->printForm = StringPrintForm(theEnv,ValueToString(theToken->value));
         break;

      /*=======================================*/
      /* Process Tokens that might be numbers. */
      /*=======================================*/

      case '-':
      case '.':
      case '+':
         EnvUngetcRouter(theEnv,inchar,logicalName);
         ScanNumber(theEnv,logicalName,theToken);
         break;

      /*===================================*/
      /* Process ? and ?<variable> Tokens. */
      /*===================================*/

       case '?':
          inchar = EnvGetcRouter(theEnv,logicalName);
          if (isalpha(inchar)
#if DEFGLOBAL_CONSTRUCT
              || (inchar == '*'))
#else
              )
#endif
            {
             EnvUngetcRouter(theEnv,inchar,logicalName);
             theToken->value = (void *) ScanSymbol(theEnv,logicalName,0,&type);
             theToken->type = SF_VARIABLE;
#if DEFGLOBAL_CONSTRUCT
             if ((ValueToString(theToken->value)[0] == '*') &&
                 (((int) strlen(ValueToString(theToken->value))) > 1) &&
                 (ValueToString(theToken->value)[strlen(ValueToString(theToken->value)) - 1] == '*'))
               {
                size_t count;

                theToken->type = GBL_VARIABLE;
                theToken->printForm = AppendStrings(theEnv,"?",ValueToString(theToken->value));
                count = strlen(ScannerData(theEnv)->GlobalString);
                ScannerData(theEnv)->GlobalString[count-1] = EOS;
                theToken->value = EnvAddSymbol(theEnv,ScannerData(theEnv)->GlobalString+1);
                ScannerData(theEnv)->GlobalString[count-1] = (char) inchar;

               }
             else
#endif
             theToken->printForm = AppendStrings(theEnv,"?",ValueToString(theToken->value));
            }
          else
            {
             theToken->type = SF_WILDCARD;
             theToken->value = (void *) EnvAddSymbol(theEnv,"?");
             EnvUngetcRouter(theEnv,inchar,logicalName);
             theToken->printForm = "?";
            }
          break;

      /*=====================================*/
      /* Process $? and $?<variable> Tokens. */
      /*=====================================*/

      case '$':
         if ((inchar = EnvGetcRouter(theEnv,logicalName)) == '?')
           {
            inchar = EnvGetcRouter(theEnv,logicalName);
            if (isalpha(inchar)
#if DEFGLOBAL_CONSTRUCT
                 || (inchar == '*'))
#else
                 )
#endif
              {
               EnvUngetcRouter(theEnv,inchar,logicalName);
               theToken->value = (void *) ScanSymbol(theEnv,logicalName,0,&type);
               theToken->type = MF_VARIABLE;
#if DEFGLOBAL_CONSTRUCT
             if ((ValueToString(theToken->value)[0] == '*') &&
                 ((int) (strlen(ValueToString(theToken->value))) > 1) &&
                 (ValueToString(theToken->value)[strlen(ValueToString(theToken->value)) - 1] == '*'))
               {
                size_t count;

                theToken->type = MF_GBL_VARIABLE;
                theToken->printForm = AppendStrings(theEnv,"$?",ValueToString(theToken->value));
                count = strlen(ScannerData(theEnv)->GlobalString);
                ScannerData(theEnv)->GlobalString[count-1] = EOS;
                theToken->value = EnvAddSymbol(theEnv,ScannerData(theEnv)->GlobalString+1);
                ScannerData(theEnv)->GlobalString[count-1] = (char) inchar;
               }
             else
#endif
               theToken->printForm = AppendStrings(theEnv,"$?",ValueToString(theToken->value));
              }
            else
              {
               theToken->type = MF_WILDCARD;
               theToken->value = (void *) EnvAddSymbol(theEnv,"$?");
               theToken->printForm = "$?";
               EnvUngetcRouter(theEnv,inchar,logicalName);
              }
           }
         else
           {
            theToken->type = SYMBOL;
            ScannerData(theEnv)->GlobalString = ExpandStringWithChar(theEnv,'$',ScannerData(theEnv)->GlobalString,&ScannerData(theEnv)->GlobalPos,&ScannerData(theEnv)->GlobalMax,ScannerData(theEnv)->GlobalMax+80);
            EnvUngetcRouter(theEnv,inchar,logicalName);
            theToken->value = (void *) ScanSymbol(theEnv,logicalName,1,&type);
            theToken->printForm = ValueToString(theToken->value);
           }
         break;

      /*============================*/
      /* Symbols beginning with '<' */
      /*============================*/

      case '<':
         theToken->type = SYMBOL;
         ScannerData(theEnv)->GlobalString = ExpandStringWithChar(theEnv,'<',ScannerData(theEnv)->GlobalString,&ScannerData(theEnv)->GlobalPos,&ScannerData(theEnv)->GlobalMax,ScannerData(theEnv)->GlobalMax+80);
         theToken->value = (void *) ScanSymbol(theEnv,logicalName,1,&type);
         theToken->printForm = ValueToString(theToken->value);
         break;

      /*=============================================*/
      /* Process "(", ")", "~", "|", and "&" Tokens. */
      /*=============================================*/

      case '(':
         theToken->type = LPAREN;
         theToken->value = (void *) EnvAddSymbol(theEnv,"(");
         theToken->printForm = "(";
         break;

      case ')':
         theToken->type= RPAREN;
         theToken->value = (void *) EnvAddSymbol(theEnv,")");
         theToken->printForm = ")";
         break;

      case '~':
         theToken->type = NOT_CONSTRAINT;
         theToken->value = (void *) EnvAddSymbol(theEnv,"~");
         theToken->printForm = "~";
         break;

      case '|':
         theToken->type = OR_CONSTRAINT;
         theToken->value = (void *) EnvAddSymbol(theEnv,"|");
         theToken->printForm = "|";
         break;

      case '&':
         theToken->type =  AND_CONSTRAINT;
         theToken->value = (void *) EnvAddSymbol(theEnv,"&");
         theToken->printForm = "&";
         break;

      /*============================*/
      /* Process End-of-File Token. */
      /*============================*/

      case EOF:
      case 0:
      case 3:
         theToken->type = STOP;
         theToken->value = (void *) EnvAddSymbol(theEnv,"stop");
         theToken->printForm = "";
         break;

      /*=======================*/
      /* Process Other Tokens. */
      /*=======================*/

      default:
         if (isprint(inchar))
           {
            EnvUngetcRouter(theEnv,inchar,logicalName);
            theToken->value = (void *) ScanSymbol(theEnv,logicalName,0,&type);
            theToken->type = type;
            theToken->printForm = ValueToString(theToken->value);
           }
         else
           { theToken->printForm = "<<<unprintable character>>>"; }
         break;
     }

   /*===============================================*/
   /* Put the new token in the pretty print buffer. */
   /*===============================================*/

#if (! RUN_TIME) && (! BLOAD_ONLY)
   if (theToken->type == INSTANCE_NAME)
     {
      SavePPBuffer(theEnv,"[");
      SavePPBuffer(theEnv,theToken->printForm);
      SavePPBuffer(theEnv,"]");
     }
   else
     { SavePPBuffer(theEnv,theToken->printForm); }
#endif

   /*=========================================================*/
   /* Return the temporary memory used in scanning the token. */
   /*=========================================================*/

   if (ScannerData(theEnv)->GlobalString != NULL)
     {
      rm(theEnv,ScannerData(theEnv)->GlobalString,ScannerData(theEnv)->GlobalMax);
      ScannerData(theEnv)->GlobalString = NULL;
      ScannerData(theEnv)->GlobalMax = 0;
      ScannerData(theEnv)->GlobalPos = 0;
     }

   return;
  }

/*************************************/
/* ScanSymbol: Scans a symbol token. */
/*************************************/
static void *ScanSymbol(
  void *theEnv,
  char *logicalName,
  int count,
  unsigned short *type)
  {
   int inchar;
#if OBJECT_SYSTEM
   void *symbol;
#endif

   /*=====================================*/
   /* Scan characters and add them to the */
   /* symbol until a delimiter is found.  */
   /*=====================================*/

   inchar = EnvGetcRouter(theEnv,logicalName);
   while ( (inchar != '<') && (inchar != '"') &&
           (inchar != '(') && (inchar != ')') &&
           (inchar != '&') && (inchar != '|') && (inchar != '~') &&
           (inchar != ' ') && (inchar != ';') &&
           isprint(inchar) )
     {