LL(1) Parsing Table Program in C

The program reads grammar from a file “text.txt”.

“^” is used to represent epsilon.

The program assumes that the input grammar is LL(1).

text.txt

E->TA

A->+TA|^

T->FB

B->*FB|^

F->t|(E)

#include<stdio.h>

#include<string.h>

#define TSIZE 128

// table[i][j] stores

// the index of production that must be applied on

// ith varible if the input is

// jth nonterminal

int table[100][TSIZE];

// stores all list of terminals

// the ASCII value if use to index terminals

// terminal[i] = 1 means the character with

// ASCII value is a terminal

// stores all list of terminals

// only Upper case letters from 'A' to 'Z'

// can be nonterminals

// nonterminal[i] means ith alphabet is present as

// nonterminal is the grammar

char nonterminal[26];

// structure to hold each production

// str[] stores the production

// len is the length of production

struct product {

char str[100];

int len;

}pro[20];

// no of productions in form A->ß

int no_pro;

char first[26][TSIZE];

char follow[26][TSIZE];

// stores first of each production in form A->ß

char first_rhs[100][TSIZE];

// check if the symbol is nonterminal

int isNT(char c) {

return c >= 'A' && c <= 'Z';

// reading data from the file

FILE* fptr;

fptr = fopen("text.txt", "r");

char buffer[255];

int i;

int j;

while (fgets(buffer, sizeof(buffer), fptr)) {

printf("%s", buffer);

j = 0;

nonterminal[buffer[0] - 'A'] = 1;

for (i = 0; i < strlen(buffer) - 1; ++i) {

if (buffer[i] == '|') {

++no_pro;

pro[no_pro - 1].str[j] = '\0';

pro[no_pro - 1].len = j;

pro[no_pro].str[0] = pro[no_pro - 1].str[0];

pro[no_pro].str[1] = pro[no_pro - 1].str[1];

pro[no_pro].str[2] = pro[no_pro - 1].str[2];

j = 3;

else {

pro[no_pro].str[j] = buffer[i];

++j;

if (!isNT(buffer[i]) && buffer[i] != '-' && buffer[i] != '>') {

terminal[buffer[i]] = 1;
 }

pro[no_pro].len = j;

++no_pro;

void add_FIRST_A_to_FOLLOW_B(char A, char B) {

int i;

for (i = 0; i < TSIZE; ++i) {

if (i != '^')

follow[B - 'A'][i] = follow[B - 'A'][i] || first[A - 'A'][i];

void add_FOLLOW_A_to_FOLLOW_B(char A, char B) {

int i;

for (i = 0; i < TSIZE; ++i) {

if (i != '^')

follow[B - 'A'][i] = follow[B - 'A'][i] || follow[A - 'A'][i];

void FOLLOW() {

int t = 0;

int i, j, k, x;

while (t++ < no_pro) {

if (!nonterminal[k]) continue;

char nt = k + 'A';

for (i = 0; i < no_pro; ++i) {

for (j = 3; j < pro[i].len; ++j) {

if (nt == pro[i].str[j]) {

for (x = j + 1; x < pro[i].len; ++x) {

char sc = pro[i].str[x];

if (isNT(sc)) {

add_FIRST_A_to_FOLLOW_B(sc, nt);

if (first[sc - 'A']['^'])

continue;

else {

follow[nt - 'A'][sc] = 1;

break;

if (x == pro[i].len)

add_FOLLOW_A_to_FOLLOW_B(pro[i].str[0], nt);

}
}

void add_FIRST_A_to_FIRST_B(char A, char B) {

int i;

for (i = 0; i < TSIZE; ++i) {

if (i != '^') {

first[B - 'A'][i] = first[A - 'A'][i] || first[B - 'A'][i];

void FIRST() {

int i, j;

int t = 0;

while (t < no_pro) {

for (i = 0; i < no_pro; ++i) {

for (j = 3; j < pro[i].len; ++j) {

char sc = pro[i].str[j];

if (isNT(sc)) {

add_FIRST_A_to_FIRST_B(sc, pro[i].str[0]);

if (first[sc - 'A']['^'])

continue;

else {

first[pro[i].str[0] - 'A'][sc] = 1;

break;
 }

if (j == pro[i].len)

first[pro[i].str[0] - 'A']['^'] = 1;

++t;

void add_FIRST_A_to_FIRST_RHS__B(char A, int B) {

int i;

for (i = 0; i < TSIZE; ++i) {

if (i != '^')

first_rhs[B][i] = first[A - 'A'][i] || first_rhs[B][i];

// Calculates FIRST(ß) for each A->ß

void FIRST_RHS() {

int i, j;

int t = 0;

while (t < no_pro) {

for (i = 0; i < no_pro; ++i) {

for (j = 3; j < pro[i].len; ++j) {

char sc = pro[i].str[j];

if (isNT(sc)) {

add_FIRST_A_to_FIRST_RHS__B(sc, i);

if (first[sc - 'A']['^'])
 continue;

else {

first_rhs[i][sc] = 1;

break;

if (j == pro[i].len)

first_rhs[i]['^'] = 1;

++t;

int main() {

readFromFile();

follow[pro[0].str[0] - 'A']['$'] = 1;

FIRST();

FOLLOW();

FIRST_RHS();

int i, j, k;

// display first of each variable

printf("\n");

for (i = 0; i < no_pro; ++i) {

if (i == 0 || (pro[i - 1].str[0] != pro[i].str[0])) {

printf("FIRST OF %c: ", c);

for (j = 0; j < TSIZE; ++j) {

if (first[c - 'A'][j]) {

printf("%c ", j);

printf("\n");

// display follow of each variable

printf("\n");

for (i = 0; i < no_pro; ++i) {

if (i == 0 || (pro[i - 1].str[0] != pro[i].str[0])) {

char c = pro[i].str[0];

printf("FOLLOW OF %c: ", c);

for (j = 0; j < TSIZE; ++j) {

if (follow[c - 'A'][j]) {

printf("%c ", j);

printf("\n");

}
// display first of each variable ß

// in form A->ß

printf("\n");

for (i = 0; i < no_pro; ++i) {

printf("FIRST OF %s: ", pro[i].str);

for (j = 0; j < TSIZE; ++j) {

if (first_rhs[i][j]) {

printf("%c ", j);

printf("\n");

// the parse table contains '$'

// set terminal['$'] = 1

// to include '$' in the parse table

terminal['$'] = 1;

// the parse table do not read '^'

// as input

// so we set terminal['^'] = 0

// to remove '^' from terminals

terminal['^'] = 0;

// printing parse table

printf("\n\t**************** LL(1) PARSING TABLE *******************\n");

printf("\t--------------------------------------------------------\n");

printf("%-10s", "");

for (i = 0; i < TSIZE; ++i) {

if (terminal[i]) printf("%-10c", i);

printf("\n");

int p = 0;

for (i = 0; i < no_pro; ++i) {

if (i != 0 && (pro[i].str[0] != pro[i - 1].str[0]))

p = p + 1;

for (j = 0; j < TSIZE; ++j) {

if (first_rhs[i][j] && j != '^') {

table[p][j] = i + 1;

else if (first_rhs[i]['^']) {

for (k = 0; k < TSIZE; ++k) {

if (follow[pro[i].str[0] - 'A'][k]) {

table[p][k] = i + 1;

}
 k = 0;

for (i = 0; i < no_pro; ++i) {

if (i == 0 || (pro[i - 1].str[0] != pro[i].str[0])) {

printf("%-10c", pro[i].str[0]);

for (j = 0; j < TSIZE; ++j) {

if (table[k][j]) {

printf("%-10s", pro[table[k][j] - 1].str);

else if (terminal[j]) {

printf("%-10s", "");

++k;

printf("\n");