Scanner: oops wrong file
This commit is contained in:
parent
db10e77bc7
commit
051752c100
|
|
@ -0,0 +1,260 @@
|
|||
/* File name: platy_st.c
|
||||
* Purpose:This is the main program for Assignment #2 - Scanner
|
||||
* CST8152 - Compilers
|
||||
* Version: 1.17.1
|
||||
* Author: Svillen Ranev
|
||||
* Date: 30 January 2017
|
||||
*/
|
||||
|
||||
/* The #define _CRT_SECURE_NO_WARNINGS should be used in MS Visual Studio projects
|
||||
* to suppress the warnings about using "unsafe" functions like fopen()
|
||||
* and standard sting library functions defined in string.h.
|
||||
* The define does not have any effect in other compiler projects.
|
||||
*/
|
||||
#define _CRT_SECURE_NO_WARNINGS
|
||||
|
||||
#include <stdio.h>
|
||||
#include <stdlib.h>
|
||||
#include <string.h>
|
||||
#include <stdarg.h>
|
||||
|
||||
#include "buffer.h"
|
||||
#include "token.h"
|
||||
|
||||
/* constant definitions */
|
||||
#define INIT_CAPACITY 200 /* initial buffer capacity */
|
||||
#define INC_FACTOR 15 /* increment factor */
|
||||
|
||||
#define STR_INIT_CAPACITY 100 /* initial string literal table capacity */
|
||||
#define STR_CAPACITY_INC 50 /* initial string literal table capacity inc */
|
||||
|
||||
/*check for ANSI C compliancy */
|
||||
#define ANSI_C 0
|
||||
#if defined(__STDC__)
|
||||
#undef ANSI_C
|
||||
#define ANSI_C 1
|
||||
#endif
|
||||
|
||||
/* Global objects - variables */
|
||||
|
||||
Buffer * str_LTBL; /* this buffer implements String Literal Table */
|
||||
/* it is used as a repository for string literals */
|
||||
int scerrnum; /* run-time error number = 0 by default (ANSI) */
|
||||
|
||||
/*external objects */
|
||||
extern int line; /* source code line numbers - defined in scanner.c */
|
||||
extern int scanner_init(Buffer * sc_buf);
|
||||
extern Token malar_next_token(Buffer * sc_buf);
|
||||
/*function declarations */
|
||||
void err_printf(char *fmt, ...);
|
||||
void display (Buffer *ptrBuffer);
|
||||
long get_filesize(char *fname);
|
||||
void print_token(Token t);
|
||||
|
||||
/* main function takes a PLATYPUS source file as
|
||||
* an argument at the command line.
|
||||
* usage: scanner source_file_name"
|
||||
*/
|
||||
int main(int argc, char ** argv){
|
||||
|
||||
Buffer *sc_buf; /* pointer to input (source) buffer */
|
||||
FILE *fi; /* input file handle */
|
||||
Token t; /* token produced by the scanner */
|
||||
int loadsize = 0; /*the size of the file loaded in the buffer */
|
||||
int ansi_c = !ANSI_C; /* ANSI C compliancy flag */
|
||||
|
||||
/* Check if the compiler option is set to compile ANSI C */
|
||||
/* __DATE__, __TIME__, __LINE__, __FILE__, __STDC__ are predefined preprocessor macros*/
|
||||
if(ansi_c){
|
||||
err_printf("Date: %s Time: %s",__DATE__, __TIME__);
|
||||
err_printf("ERROR: Compiler is not ANSI C compliant!\n");
|
||||
exit(1);
|
||||
}
|
||||
|
||||
/*check for correct arrguments - source file name */
|
||||
if (argc <= 1){
|
||||
/* __DATE__, __TIME__, __LINE__, __FILE__ are predefined preprocessor macros*/
|
||||
err_printf("Date: %s Time: %s",__DATE__, __TIME__);
|
||||
err_printf("Runtime error at line %d in file %s", __LINE__, __FILE__);
|
||||
err_printf("%s%s%s",argv[0],": ","Missing source file name.");
|
||||
err_printf("%s%s%s","Usage: ", "scanner", " source_file_name");
|
||||
exit(1);
|
||||
}
|
||||
|
||||
|
||||
/* create a source code input buffer - multiplicative mode */
|
||||
sc_buf = b_create(INIT_CAPACITY,INC_FACTOR,'m');
|
||||
if (sc_buf == NULL){
|
||||
err_printf("%s%s%s",argv[0],": ","Could not create source buffer");
|
||||
exit(1);
|
||||
}
|
||||
|
||||
/*open source file */
|
||||
if ((fi = fopen(argv[1],"r")) == NULL){
|
||||
err_printf("%s%s%s%s",argv[0],": ", "Cannot open file: ",argv[1]);
|
||||
exit (1);
|
||||
}
|
||||
/* load source file into input buffer */
|
||||
printf("Reading file %s ....Please wait\n",argv[1]);
|
||||
loadsize = b_load (fi,sc_buf);
|
||||
if (loadsize == R_FAIL1)
|
||||
err_printf("%s%s%s",argv[0],": ","Error in loading buffer.");
|
||||
|
||||
/* close source file */
|
||||
fclose(fi);
|
||||
/*find the size of the file */
|
||||
if (loadsize == LOAD_FAIL){
|
||||
printf("The input file %s %s\n", argv[1],"is not completely loaded.");
|
||||
printf("Input file size: %ld\n", get_filesize(argv[1]));
|
||||
}
|
||||
/* pack and display the source buffer */
|
||||
|
||||
if(b_pack(sc_buf)){
|
||||
display(sc_buf);
|
||||
}
|
||||
|
||||
/* create string Literal Table */
|
||||
|
||||
str_LTBL = b_create(INIT_CAPACITY,INC_FACTOR,'a');
|
||||
if (str_LTBL == NULL){
|
||||
err_printf("%s%s%s",argv[0],": ","Could not create string literals buffer");
|
||||
exit(1);
|
||||
}
|
||||
|
||||
/*Testbed for the scanner */
|
||||
/* add SEOF to input program buffer*/
|
||||
b_addc(sc_buf,'\0');
|
||||
|
||||
/* Initialize scanner input buffer */
|
||||
if(scanner_init(sc_buf)){;
|
||||
err_printf("%s%s%s",argv[0],": ","Empty program buffer - scanning canceled");
|
||||
exit(1);
|
||||
}
|
||||
|
||||
printf("\nScanning source file...\n\n");
|
||||
printf("Token\t\tAttribute\n");
|
||||
printf("----------------------------------\n");
|
||||
do{
|
||||
t= malar_next_token(sc_buf);
|
||||
print_token(t);
|
||||
}while(t.code != SEOF_T);
|
||||
if(b_size(str_LTBL)) b_print(str_LTBL);
|
||||
b_free(sc_buf);
|
||||
b_free(str_LTBL);
|
||||
sc_buf = str_LTBL = NULL;
|
||||
|
||||
return (0);
|
||||
}
|
||||
|
||||
/* Error printing function with variable number of arguments
|
||||
*/
|
||||
void err_printf( char *fmt, ... ){
|
||||
|
||||
va_list ap;
|
||||
va_start(ap, fmt);
|
||||
|
||||
(void)vfprintf(stderr, fmt, ap);
|
||||
va_end(ap);
|
||||
/* Move to new line */
|
||||
if( strchr(fmt,'\n') == NULL )
|
||||
fprintf(stderr,"\n");
|
||||
}
|
||||
|
||||
/* The function displays buffer contents
|
||||
*/
|
||||
void display (Buffer *ptrBuffer){
|
||||
printf("\nPrinting buffer parameters:\n\n");
|
||||
printf("The capacity of the buffer is: %d\n",b_capacity(ptrBuffer));
|
||||
printf("The current size of the buffer is: %d\n",b_size(ptrBuffer));
|
||||
printf("\nPrinting buffer contents:\n\n");
|
||||
b_print(ptrBuffer);
|
||||
}
|
||||
|
||||
long get_filesize(char *fname){
|
||||
FILE *input;
|
||||
long flength;
|
||||
input = fopen(fname, "r");
|
||||
if (input == NULL){
|
||||
err_printf("%s%s", "Cannot open file: ", fname);
|
||||
return 0L;
|
||||
}
|
||||
fseek(input, 0L, SEEK_END);
|
||||
flength = ftell(input);
|
||||
fclose(input);
|
||||
return flength;
|
||||
}
|
||||
|
||||
/* The function prints the token returned by the scanner
|
||||
*/
|
||||
|
||||
void print_token(Token t){
|
||||
extern char * kw_table []; /* link to keyword table in */
|
||||
switch(t.code){
|
||||
case ERR_T:
|
||||
printf("ERR_T\t\t%s",t.attribute.err_lex);
|
||||
if(scerrnum){
|
||||
printf("%d",scerrnum);
|
||||
exit(scerrnum);
|
||||
}printf("\n");
|
||||
break;
|
||||
case SEOF_T:
|
||||
printf("SEOF_T\n" );
|
||||
break;
|
||||
case AVID_T:
|
||||
printf("AVID_T\t\t%s\n",t.attribute.vid_lex);
|
||||
break;
|
||||
case SVID_T:
|
||||
printf("SVID_T\t\t%s\n",t.attribute.vid_lex);
|
||||
break;
|
||||
case FPL_T:
|
||||
printf("FPL_T\t\t%f\n",t.attribute.flt_value);
|
||||
break;
|
||||
case INL_T:
|
||||
printf("IL_T\t\t%d\n",t.attribute.get_int);
|
||||
break;
|
||||
case STR_T:
|
||||
printf("STR_T\t\t%d\t ",(short)t.attribute.get_int);
|
||||
printf("%s\n", b_setmark(str_LTBL, (short)t.attribute.get_int));
|
||||
break;
|
||||
case SCC_OP_T:
|
||||
printf("SCC_OP_T\n" );
|
||||
break;
|
||||
case ASS_OP_T:
|
||||
printf("ASS_OP_T\n" );
|
||||
break;
|
||||
case ART_OP_T:
|
||||
printf("ART_OP_T\t%d\n",t.attribute.get_int);
|
||||
break;
|
||||
case REL_OP_T:
|
||||
printf("REL_OP_T\t%d\n",t.attribute.get_int);
|
||||
break;
|
||||
case LOG_OP_T:
|
||||
printf("LOG_OP_T\t%d\n",t.attribute.get_int);
|
||||
break;
|
||||
case LPR_T:
|
||||
printf("LPR_T\n" );
|
||||
break;
|
||||
case RPR_T:
|
||||
printf("RPR_T\n" );
|
||||
break;
|
||||
case LBR_T:
|
||||
printf("LBR_T\n" );
|
||||
break;
|
||||
case RBR_T:
|
||||
printf("RBR_T\n" );
|
||||
break;
|
||||
case KW_T:
|
||||
printf("KW_T\t\t%s\n",kw_table [t.attribute.get_int]);
|
||||
break;
|
||||
case COM_T:
|
||||
printf("COM_T\n");
|
||||
break;
|
||||
case EOS_T:
|
||||
printf("EOS_T\n" );
|
||||
break;
|
||||
|
||||
default:
|
||||
printf("Scanner error: invalid token code: %d\n", t.code);
|
||||
}
|
||||
}
|
||||
|
||||
34
scanner.c
34
scanner.c
|
|
@ -5,8 +5,8 @@
|
|||
* scanner_init() must be called before using the scanner.
|
||||
* The file is incomplete;
|
||||
* Provided by: Svillen Ranev
|
||||
* Version: 1.16.02
|
||||
* Date: 29 September 2016
|
||||
* Version: 1.17.1
|
||||
* Date: 30 January 2017
|
||||
*******************************************************************
|
||||
* REPLACE THIS HEADER WITH YOUR HEADER
|
||||
*******************************************************************
|
||||
|
|
@ -66,7 +66,7 @@ int scanner_init(Buffer * sc_buf) {
|
|||
/* scerrnum = 0; *//*no need - global ANSI C */
|
||||
}
|
||||
|
||||
Token mlwpar_next_token(Buffer * sc_buf)
|
||||
Token malar_next_token(Buffer * sc_buf)
|
||||
{
|
||||
Token t; /* token to return after recognition */
|
||||
unsigned char c; /* input symbol */
|
||||
|
|
@ -290,7 +290,7 @@ WHEN CALLED THE FUNCTION MUST
|
|||
IF THE lexeme IS LONGER than VID_LEN characters,
|
||||
ONLY FIRST VID_LEN-1 CHARACTERS ARE STORED
|
||||
INTO THE VARIABLE ATTRIBUTE ARRAY vid_lex[],
|
||||
AND THEN THE % CHARACTER IS APPENDED TO THE NAME.
|
||||
AND THEN THE # CHARACTER IS APPENDED TO THE NAME.
|
||||
ADD \0 AT THE END TO MAKE A C-type STRING.
|
||||
|
||||
return t;
|
||||
|
|
@ -305,22 +305,24 @@ WHICH IS THE ATTRIBUTE FOR THE TOKEN.
|
|||
THE VALUE MUST BE IN THE SAME RANGE AS the value of 4-byte float in C.
|
||||
IN CASE OF ERROR (OUT OF RANGE) THE FUNCTION MUST RETURN ERROR TOKEN
|
||||
THE ERROR TOKEN ATTRIBUTE IS lexeme. IF THE ERROR lexeme IS LONGER
|
||||
than ERR_LEN caharacters, only the first ERR_LEN character are
|
||||
stored in err_lex.
|
||||
than ERR_LEN characters, ONLY THE FIRST ERR_LEN-3 characters ARE
|
||||
STORED IN err_lex. THEN THREE DOTS ... ARE ADDED TO THE END OF THE
|
||||
err_lex C-type string.
|
||||
return t;
|
||||
}
|
||||
|
||||
ACCEPTING FUNCTION FOR THE integer literal(IL) - decimal constant (DIL) AND ZERO (0)
|
||||
ACCEPTING FUNCTION FOR THE integer literal(IL) - decimal constant (DIL)
|
||||
|
||||
Token aa_funcXX(char lexeme[]){
|
||||
|
||||
THE FUNCTION MUST CONVERT THE LEXEME REPRESENTING A DECIMAL CONSTANT AND 0
|
||||
THE FUNCTION MUST CONVERT THE LEXEME REPRESENTING A DECIMAL CONSTANT
|
||||
TO A DECIMAL INTEGER VALUE, WHICH IS THE ATTRIBUTE FOR THE TOKEN.
|
||||
THE VALUE MUST BE IN THE SAME RANGE AS the value of 2-byte integer in C.
|
||||
IN CASE OF ERROR (OUT OF RANGE) THE FUNCTION MUST RETURN ERROR TOKEN
|
||||
THE ERROR TOKEN ATTRIBUTE IS lexeme. IF THE ERROR lexeme IS LONGER
|
||||
than ERR_LEN caharacters, only the first ERR_LEN character are
|
||||
stored in err_lex.
|
||||
than ERR_LEN characters, ONLY THE FIRST ERR_LEN-3 characters ARE
|
||||
STORED IN err_lex. THEN THREE DOTS ... ARE ADDED TO THE END OF THE
|
||||
err_lex C-type string.
|
||||
return t;
|
||||
}
|
||||
|
||||
|
|
@ -338,8 +340,9 @@ THE FUNCTION atool(char * lexeme) WHICH CONVERTS AN ASCII STRING
|
|||
REPRESENTING AN OCTAL NUMBER TO INTEGER VALUE
|
||||
IN CASE OF ERROR (OUT OF RANGE) THE FUNCTION MUST RETURN ERROR TOKEN
|
||||
THE ERROR TOKEN ATTRIBUTE IS lexeme. IF THE ERROR lexeme IS LONGER
|
||||
than ERR_LEN caharacters, only the first ERR_LEN character are
|
||||
stored in err_lex.
|
||||
than ERR_LEN characters, ONLY THE FIRST ERR_LEN-3 characters ARE
|
||||
STORED IN err_lex. THEN THREE DOTS ... ARE ADDED TO THE END OF THE
|
||||
err_lex C-type string.
|
||||
|
||||
return t;
|
||||
}
|
||||
|
|
@ -350,9 +353,10 @@ Token aa_funcXX(char lexeme[]){
|
|||
|
||||
THE FUNCTION SETS THE ERROR TOKEN. lexeme[] CONTAINS THE ERROR
|
||||
THE ATTRIBUTE OF THE ERROR TOKEN IS THE lexeme ITSELF
|
||||
AND IT MUST BE STORED in err_lex. IF THE ERROR lexeme IS LONGER
|
||||
than ERR_LEN caharacters, only the first ERR_LEN character are
|
||||
stored in err_lex.
|
||||
AND IT MUST BE STORED in err_lex. IF THE ERROR lexeme IS LONGER
|
||||
than ERR_LEN characters, ONLY THE FIRST ERR_LEN-3 characters ARE
|
||||
STORED IN err_lex. THEN THREE DOTS ... ARE ADDED TO THE END OF THE
|
||||
err_lex C-type string.
|
||||
|
||||
return t;
|
||||
}
|
||||
|
|
|
|||
6
table.h
6
table.h
|
|
@ -1,8 +1,8 @@
|
|||
/* Filename: table.h
|
||||
* Transition Table and function declarations necessary for the scanner implementation
|
||||
* as required for CST8152 - Assignment #2.
|
||||
* Version: 1.16.02
|
||||
* Date: 29 September 2016
|
||||
* Version: 1.17.1
|
||||
* Date: 30 January 2017
|
||||
* Provided by: Svillen Ranev
|
||||
* The file is incomplete. You are to complete it.
|
||||
***************************************************
|
||||
|
|
@ -29,7 +29,7 @@
|
|||
* in the token-driven part of the scanner
|
||||
* '=' , ' ' , '(' , ')' , '{' , '}' , == , <> , '>' , '<' ,
|
||||
* space
|
||||
* !<comment , ',' , '"' , ';' , '-' , '+' , '*' , '/', # ,
|
||||
* !<comment , ',' , '"' , ';' , '-' , '+' , '*' , '/', << ,
|
||||
* .AND., .OR. , SEOF, 'wrong symbol',
|
||||
*/
|
||||
|
||||
|
|
|
|||
26
token.h
26
token.h
|
|
@ -1,8 +1,8 @@
|
|||
/* Filename: token.h
|
||||
* Token declarations necessary for the scanner implementation
|
||||
* CST8152, Assignment #2
|
||||
* Version: 1.16.02
|
||||
* Date: 29 September 2016
|
||||
* Version: 1.17.1
|
||||
* Date: 30 January 2017
|
||||
* Provided by: Svillen Ranev
|
||||
* The file is complete and MUST NOT be modified.
|
||||
*/
|
||||
|
|
@ -52,17 +52,17 @@ typedef enum LogicalOperators {AND,OR} Log_Op;
|
|||
/* Structure declaring the token and its attributes */
|
||||
|
||||
typedef union TokenAttribute{
|
||||
int get_int; /* integer attributes accessor */
|
||||
Arr_Op arr_op; /* arithmetic operator attribute code */
|
||||
Rel_Op rel_op; /* relational operator attribute code */
|
||||
Log_Op log_op; /* logical operator attribute code */
|
||||
int int_value; /* integer literal attribute (value) */
|
||||
int kwt_idx; /* keyword index in the keyword table */
|
||||
short str_offset; /* sring literal offset from the beginning of */
|
||||
/* the string literal buffer (str_LTBL->cb_head) */
|
||||
float flt_value; /* floating-point literal attribute (value) */
|
||||
char vid_lex[VID_LEN+1]; /* variable identifier token attribute */
|
||||
char err_lex[ERR_LEN+1]; /* error token attribite */
|
||||
int get_int; /* integer attributes accessor */
|
||||
Arr_Op arr_op; /* arithmetic operator attribute code */
|
||||
Rel_Op rel_op; /* relational operator attribute code */
|
||||
Log_Op log_op; /* logical operator attribute code */
|
||||
int int_value; /* integer literal attribute (value) */
|
||||
int kwt_idx; /* keyword index in the keyword table */
|
||||
short str_offset; /* sring literal offset from the beginning of */
|
||||
/* the string literal buffer (str_LTBL->cb_head) */
|
||||
float flt_value; /* floating-point literal attribute (value) */
|
||||
char vid_lex[VID_LEN+1]; /* variable identifier token attribute */
|
||||
char err_lex[ERR_LEN+1]; /* error token attribite */
|
||||
} TA;
|
||||
|
||||
typedef struct Token
|
||||
|
|
|
|||
Loading…
Reference in New Issue