assembler.cpp

c语言的简化编译器

#include "stdafx.h"

#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include "assembler.h"
#include "asmutils.h"
#include "memory.h"
#include "error.h"
#include "opcodes.h"
#include "pxdScriptHeaders.h"


/* ****************** PROTOTYPES ****************** */
void AssembleFunction();
void AssembleProgram();


/* ****************** STRUCTS ******************* */

typedef struct LABELADDRESS {
   char *label;
   unsigned int offset;
 
   struct LABELADDRESS *next;
} LABELADDRESS, LABELPLUG;	



/* ******************* GLOBALS ******************** */
unsigned int curCodeArraySize = 0; // Utility variable for the growing byte code array

// Where to add the next byte code.
unsigned int codeposition = 0;     

// Our final output file with the binary code.
static FILE *emitFILE;

// The input text file with the symbolic byte code
FILE *sourceFILE;

// This is the (single) growing array of byte codes
unsigned char *bytecode = NULL;

// A linked list of unfilled "offset holes" in the code segment.
LABELPLUG *labelPlugs = NULL;

// A linked list of the labels and their corresponding offsets in the code segment.
unsigned int numLabels = 0;
LABELADDRESS *asmLabels = NULL;

// A linked list of the constants in the programs and functions
unsigned int nextconst = 0; 
unsigned int numConsts = 0;	
CONSTANT *asmConstants = NULL;

// The name of the function or program we are currently assembling.
// This is used to prefix all label names so we get truely unique names
// (remember label names are otherwise only unique within the function or
//  program in which they are used)
char *currentName = NULL;



/* *************** UTILITY FUNCTIONS ************** */

// Check that we have 'nrBytes' bytes of free space in the code segment
void checkSpace(int nrBytes) {
  unsigned char *temp = NULL;
  
  // Grow code segment if there isn't already room  
  if (codeposition+nrBytes >= curCodeArraySize) {	
    temp = (unsigned char *)Malloc(2*curCodeArraySize);
    memcpy(temp, bytecode, curCodeArraySize);
    bytecode = temp;
    curCodeArraySize *= 2;	
  }	
}


/* Add a plug to the list of label plugs at the current code position */
void addPlug(char *label) {
  LABELPLUG *plug = NULL;
        
  // Generate unique name for this label. 
  char *uniqueName = strconcat(currentName, label);
  
  plug = NEW(LABELPLUG);
  plug->label = uniqueName;        // which label?
  plug->offset = codeposition;     // where in the code should its address be placed? 
    
  // link it in 
  plug->next = labelPlugs;
  labelPlugs = plug;
    
  // Make sure there is space:
  checkSpace(4);
       
  // increase code position to make room for address later when we fill the plugs 
  codeposition += sizeof(unsigned int);  
}


/* Search the list of labels and returns the address */
unsigned int getAddressOfLabel(char *uniqueLabel) {
  LABELADDRESS *labels = asmLabels;
  
  while (labels != NULL) {
    if (strcmp(labels->label, uniqueLabel) == 0) 
      return labels->offset;
    
    labels = labels->next; 	
  }	
  
  // Error - label was not found! 
  printf("Internal assembler error - label '%s' is referenced but not defined\n", uniqueLabel);
  return 0;
}  


/* Run through the list of plugs, look up their labels addresses and insert into byte code buffer */
void fillPlugs() {
  unsigned int address = 0;
  LABELPLUG *plugs = labelPlugs;
  
  // Loop over all plugs
  while (plugs != NULL) {
    
    // look up address of labels (by now it MUST be defined) 
    address = getAddressOfLabel(plugs->label);
    
    // fill address into the code segment 
    *((unsigned int*)&bytecode[plugs->offset]) = address;

    plugs = plugs->next; 
  }
}



// Add the byte code 'bc' to the code segment.
void addCode(char bc) {
 // Make sure there is room 
 checkSpace(sizeof(char));
 
 // add code 
 bytecode[codeposition] = bc;
 codeposition++;
}


/* save a 32bit integer in the code segment */
void addIndex(unsigned int index) {
 // Make sure there is room 
 checkSpace(sizeof(unsigned int));

 // add 32bit index to code segment 
 *((unsigned int*)&bytecode[codeposition]) = index;
 
 // increase code pointer 
 codeposition += sizeof(unsigned int); 
}


/* save a 16bit integer in the code segment */
void addInt16(short unsigned int index) {
 // Make sure there is room 
 checkSpace(sizeof(short unsigned int));

 // add 16bit integer to code segment 
 *((short unsigned int*)&bytecode[codeposition]) = index;
 
 // increase code pointer 
 codeposition += sizeof(short unsigned int); 
}


// Is a given token a label? It is if its name ends with a ':'
int isLabel(char *token) {
 if (token == NULL) return 0;
 return (token[strlen(token)-1] == ':');	
}


/* We must prefix the names of the labels because they are only
   unique in their own scopes and we do global assembling of labels
   and constants */
void setLabelPosition(char *name) {
  
  LABELADDRESS *newLabel = NULL;
  LABELADDRESS **pList = NULL;

  // generate unique name for label
  char *uniqueName = strconcat(currentName, name);
 
  // Add the label 	
  newLabel = NEW(LABELADDRESS);	
  newLabel->label = uniqueName;
  newLabel->offset = codeposition;	
  newLabel->next = NULL;
 
  // Link it into the global list of labels.
  pList = &asmLabels;
  while (*pList) {
   pList = &(*pList)->next;
  }
  *pList = newLabel;  
}



/*  Get the number of a certain int constant in the constant pool */
unsigned int getIntConstNr(int val) {
 CONSTANT *temp = asmConstants;
 CONSTANT *newConst = NULL;	
 CONSTANT **pList;

 // first try to find label in list 
 while (temp != NULL) {
  if (temp->type == INT_CONST && temp->val.intval == val) 
    return temp->nr;
 
  temp = temp->next; 	
 }	
  
 	
 // Constant has not been defined yet, so we create a new one. 	
 newConst = NEW(CONSTANT);	
 newConst->nr = nextconst++;
 newConst->type = INT_CONST;
 newConst->val.intval = val;
 newConst->next = NULL; 
 
 // Link it in 
 pList = &asmConstants;
 while (*pList) {
  pList = &(*pList)->next;
 }  
 *pList = newConst;  
  
 return newConst->nr;	
}


unsigned int getStrConstNr(char *val) {
 CONSTANT *temp = asmConstants;
 CONSTANT *newConst = NULL;	
 CONSTANT **pList;

 // first try to find label in list 
 while (temp != NULL) {
  if (temp->type == ZSTRING_CONST && strcmp(temp->val.str, val) == 0) 
    return temp->nr;
    
  temp = temp->next; 	
 }	
  	
 // Constant has not been defined yet, so we create a new one. 	
 newConst = NEW(CONSTANT);	
 newConst->nr = nextconst++;
 newConst->type = ZSTRING_CONST;
 newConst->val.str = val;
 newConst->next = NULL;
 
 // Link it in
 pList = &asmConstants;
 while (*pList) {
  pList = &(*pList)->next;
 }  
 *pList = newConst;  

 return newConst->nr;	
}




/* The most boring function in the assembler */
void asmCODE(char *token) {
  int intparam;
  char *param;
       
  if (strcmp(token, "nop") == 0) {
    addCode(nopC);
  } else
  if (strcmp(token, "imul") == 0) {
    addCode(imulC);
  } else
  if (strcmp(token, "ineg") == 0) {
    addCode(inegC);
  } else
  if (strcmp(token, "irem") == 0) {
    addCode(iremC);
  } else
  if (strcmp(token, "isub") == 0) {
    addCode(isubC);
  } else
  if (strcmp(token, "idiv") == 0) {
    addCode(idivC);
  } else
  if (strcmp(token, "iadd") == 0) {
    addCode(iaddC);
  } else
  if (strcmp(token, "aadd") == 0) {
    addCode(aaddC);
  } else 
  if (strcmp(token, "goto") == 0) {
    addCode(gotoC);
    param = getToken(sourceFILE);
    addPlug(param); /* label */
  } else
  if (strcmp(token, "ifeq") == 0) {
    addCode(ifeqC);
    param = getToken(sourceFILE);
    addPlug(param); /* label */
  } else
  if (strcmp(token, "ifne") == 0) {
    addCode(ifneC);
    param = getToken(sourceFILE);
    addPlug(param); /* label */
  } else
  if (strcmp(token, "if_acmpeq") == 0) {
    addCode(if_acmpeqC);
    param = getToken(sourceFILE);
    addPlug(param); /* label */
  } else
  if (strcmp(token, "if_acmpne") == 0) {
    addCode(if_acmpneC);
    param = getToken(sourceFILE);
    addPlug(param); /* label */
  } else
  if (strcmp(token, "if_icmpeq") == 0) {
    addCode(if_icmpeqC);
    param = getToken(sourceFILE);
    addPlug(param); /* label */
  } else
  if (strcmp(token, "if_icmpgt") == 0) {
    addCode(if_icmpgtC);
    param = getToken(sourceFILE);
    addPlug(param); /* label */
  } else
  if (strcmp(token, "if_icmplt") == 0) {
    addCode(if_icmpltC);
    param = getToken(sourceFILE);
    addPlug(param); /* label */
  } else
  if (strcmp(token, "if_icmple") == 0) {
    addCode(if_icmpleC);
    param = getToken(sourceFILE);
    addPlug(param); /* label */
  } else
  if (strcmp(token, "if_icmpge") == 0) {
    addCode(if_icmpgeC);
    param = getToken(sourceFILE);
    addPlug(param); /* label */
  } else
  if (strcmp(token, "if_icmpne") == 0) {
    addCode(if_icmpneC);
    param = getToken(sourceFILE);
    addPlug(param); /* label */
  } else
  if (strcmp(token, "ireturn") == 0) {
    addCode(ireturnC);
  } else
  if (strcmp(token, "areturn") == 0) {
    addCode(areturnC);
  } else
  if (strcmp(token, "return") == 0) {
    addCode(returnC);
  } else
  if (strcmp(token, "aload") == 0) {
    addCode(aloadC);
    intparam = atoi(getToken(sourceFILE));
    addInt16(intparam);
  } else
  if (strcmp(token, "astore") == 0) {
    addCode(astoreC);
    intparam = atoi(getToken(sourceFILE));
    addInt16(intparam);
  } else