Updating a matrix with special cases - c

Question

I am trying to write a function which gets a matrix 9x9 and updates it accordingly to user's input with the following rules:
Valid number is between 1 and 9 (zero is invalid).
I have to use scanf until I get EOF.
Input has digits and symbols. valid input is a pair of two digits following with a symbol or EOF or space. string with more than two digits is invalid. for example (123% isn't valid but 12% is valid).
Example:
Input: 10 33%55^21 $123%
Output:
0 0 0 0 0 0 0 0 0
1 0 0 0 0 0 0 0 0
0 0 1 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 1 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
Explanation: 10 and 123 are invalid. 33, 55 and 21 are valid so we will put 1 in 22, 44 and 10.
What I tried to do:
void updateMarix(int matrix[][9]) {
int digits = 0, one_previous, two_previous;
char input;
while (scanf("%c", &input) != EOF) {
if(isValidDigit(input)) {
digits++;
if(digits == 1) {
two_previous = input - '0' - 1;
continue;
} else if(digits == 2){
one_previous = input - '0' -1;
continue;
}
} else if(digits == 2) {
matrix[two_previous][one_previous]++;
}
digits = 0; // reset
}
}
most tests are end with success, but some of them are fail. I think that is because I don't handle with the last input (if for example it ends with 22 it won't update it because with my implementation, the update is in the next iteration when other symbol got as input).
Is there a better implementation for this? My code became messy and not clean.
*Edit: It should ignore invalid input and a3b doesn't counts, a03b also doesn't counts but a13b does counts as 13 meaning we should increase the number in matrix[0][2].
Edit 2: #JonathanLeffler menationed FSM so I tried to create one:
Although it doesn't handles the case of 1234 (invalid number) or 123 (also invalid). The most similar thing was to create an arrow from second number to symbol (but it isn't quite true because in 1234%12 only 12 is valid.

Answer 1

I think your FSM needs 4 states plus the end state:
Zero digits read (D0).
One digit read (D1).
Two digits read (D2).
Digits are invalid but no more error reporting needed (DI).
There are 4 different inputs, too:
Digit 1-9.
Digit 0.
Other.
EOF.
I've used a switch on state and if/else code in each state, but it leads to somewhat verbose code. OTOH, I believe it handles inputs correctly.
/*
** FSM
** States: 0 digits (D0), 1 digit (D1), 2 digits (D2), digits invalid (DI)
** Inputs: digit 1-9 (D), digit 0 (0), other (O), EOF.
** Action: S - save, E - error, I - ignore, P - print
** Body of FSM encodes "action;state"
**
** State D0 D1 D2 DI
** Input
** D S;D1 S;D2 E;D2 I;DI
** O I;D0 E;D0 P;D0 I;D0
** 0 E;D2 E;D2 E;D2 I;DI
** EOF I;end E;end P;end I;end
*/
#include <assert.h>
#include <ctype.h>
#include <stdio.h>
enum State { D0, D1, D2, DI };
enum Input { Digit, Zero, Other, End };
static int debug = 0;
static enum Input input(int *rv)
{
int c = getchar();
if (debug)
printf("Input: %c\n", (c == EOF) ? 'X' : c);
*rv = c;
if (c == EOF)
return End;
if (isdigit(c))
{
*rv = c - '0';
return (c == '0') ? Zero : Digit;
}
return Other;
}
static void updateMatrix(int matrix[9][9])
{
char pair[2] = { 0, 0 };
enum State state = D0;
int c;
enum Input value;
while ((value = input(&c)) != End)
{
switch (state)
{
case D0:
if (value == Digit)
{
pair[0] = c;
state = D1;
}
else if (value == Zero)
{
fprintf(stderr, "Received zero digit - invalid\n");
state = DI;
}
else
{
assert(value == Other);
}
break;
case D1:
if (value == Digit)
{
pair[1] = c;
state = D2;
}
else if (value == Zero)
{
fprintf(stderr, "Received zero digit - invalid\n");
state = DI;
}
else
{
assert(value == Other);
fprintf(stderr, "Received one digit where two expected\n");
state = D0;
}
break;
case D2:
if (value == Digit)
{
fprintf(stderr, "Received more than two digits where two were expected\n");
state = DI;
}
else if (value == Zero)
{
fprintf(stderr, "Received zero digit - invalid\n");
state = DI;
}
else
{
assert(value == Other);
printf("Valid number %d%d\n", pair[0], pair[1]);
matrix[pair[0]-1][pair[1]-1] = 1;
state = D0;
}
break;
case DI:
if (value == Other)
state = D0;
break;
}
}
if (state == D2)
{
printf("Valid number %d%d\n", pair[0], pair[1]);
matrix[pair[0]-1][pair[1]-1] = 1;
}
else if (state == D1)
fprintf(stderr, "Received one digit where two expected\n");
}
static void dump_matrix(const char *tag, int matrix[9][9])
{
printf("%s:\n", tag);
for (int i = 0; i < 9; i++)
{
for (int j = 0; j < 9; j++)
printf("%4d", matrix[i][j]);
putchar('\n');
}
}
int main(void)
{
int matrix[9][9] = { 0 };
updateMatrix(matrix);
dump_matrix("After input", matrix);
return 0;
}
On your test input, it produces the output:
Received zero digit - invalid
Valid number 33
Valid number 55
Valid number 21
Received more than two digits where two were expected
After input:
0 0 0 0 0 0 0 0 0
1 0 0 0 0 0 0 0 0
0 0 1 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 1 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
On the mostly-invalid input file:
123345132
bbbb12cccc1dddd011dd
it produces the output:
Received more than two digits where two were expected
Valid number 12
Received one digit where two expected
Received zero digit - invalid
After input:
0 1 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
You can argue (easily) that the error messages could be more informative (identifying the erroneous character, and possibly the prior valid digits), but it only produces one error message for each invalid sequence, which is beneficial.

Answer 2

You could use a combination of fgets(), sscanf() and strpbrk() for this.
The input line is read into a character array str and a pointer ptr pointing to the part of the string in str being processed is maintained.
First, set up a loop to read input line by line. fgets() will return NULL on EOF.
for(; fgets(str, sizeof(str), stdin); )
{
...
...
...
}
fgets() will read in the trailing newline as well. You could remove it like
str[strlen(str)-1]='\0';
Now inside the above loop, use another loop to process the input line in str like
for(ptr=str; (ptr=strpbrk(ptr, "0123456789"))!=NULL; ptr+=len)
{
sscanf(ptr, "%d%n", &n, &len);
if(n>10 && n<100)
{
//accepted
printf("\n%d", n);
arr[n/10][n%10]=1;
}
//else discarded
}
strpbrk()'s prototype is
char *strpbrk(const char *s1, const char *s2);
and it returns a pointer to the first character in s1 which is a character in the string s2. If there is no match, NULL is returned.
So we are looking to see the first digit part in str that remains to be processed with strpbrk(ptr, "0123456789").
This number part is read into n via sscanf(). If this number is in the range you need, you may accept it.
The %n format specifier is used to find out the number of characters which has been scanned with the sscanf() inorder to find the value by which ptr must be updated. See this post.
The digit in the ones place will be n%10 and that in the tens place will be n/10 as the number you need is a 2-digit number.
You may set your array representing the matrix like
arr[n/10][n%10]=1;
So the whole thing could look something like
char *ptr, str[50];
for(; fgets(str, sizeof(str), stdin); )
{
for(ptr=str, str[strlen(str)-1]=0; (ptr=strpbrk(ptr, "0123456789"))!=NULL; ptr+=len)
{
sscanf(ptr, "%d%n", &n, &len);
if(n>10 && n<100)
{
printf("\n%d", n);
arr[n/10][n%10]=1;
}
}
}
And for your input 10 33%55^21 $123%, the output would be
33
55
21
as 10 and 123 will be discarded.