Format Checking in C - c

Question

I'm trying to make a program in c to read in from text a large database of ice rink activities. Does anyone know how to check for something that is not in the format
ie the text document will have something like this
sample
---------------------------------------------
date startT endT END
_______________________________________________
Ice Rink 1
1/13/2014 1:50 3:50 PM Public Skating
1/13/2014 1:50 3:50 PM Game
ice rink 2
1/13/2014 1:50 3:50 PM OPEN
I can already successfully read in one line of the event, date time and description
but how do I skip or detect the lines that don't match my scan in style of
fscanf(ifp,"%d/%d/%d\t%d:%d%s\t%d:%d%s\t\t %20c",
&e1[i].month,&e1[i].day,&e1[i].year,&e1[i].startH,&e1[i].startM,e1[i].MER1,&e1[i].endH,&e1[i].endM,e1[i].MER2,e1[i].event);
In short: how do detect cases that don't match this exactly?
Thanks in advance

Answer 1

As others have already said, you can check the return value of fscanf to find out whether a line is in the given format. That isn't the ideal approach, however. First, your data is organised line-wise, but fscanf treats the newline character like any other whitespace. You could read the line with fgets first and then apply sscanf on the line, but you'd still have one big monolithic format specifier that is easy to lose track of.
I'd like to propose another approach. Yor data lines seem to be organised in fields, which are separated from each other with tab characters. You could read the lines with fgets, then split them with strtok and finally scan the separate fields with sscanf. If you write custom wrapper functions to your sscanf statements, you can run a sanity check on the data when it's read.
/*
* Return true if str has format "hh:min AM/PM"
*/
int scan_time(const char *str, int *hh, int *mm)
{
char buf[4] = {0};
int n;
char c;
n = sscanf(str, "%d:%d%4s %c", hh, mm, buf, &c);
if (n == 4) return 0; /* trailing extra chars */
if (n < 2) return 0; /* missing minutes */
if (n == 3) {
int key = (buf[0] << 16) + (buf[1] << 8) + buf[2];
#define KEY(a, b) ((a << 16) + (b << 8))
switch (key) {
case KEY('a', 'm'):
case KEY('A', 'M'):
break;
case KEY('p', 'm'):
case KEY('P', 'M'):
*hh += 12;
break;
default:
return 0; /* invalid am/pm spec */
}
}
if (*hh < 0 || *hh >= 24) return 0; /* invalid hours */
if (*mm < 0 || *mm >= 60) return 0; /* invalid minutes */
return 1;
}
/*
* Return true, if str has format "mm/dd/year"
*/
int scan_date(const char *str, int *yy, int *mm, int *dd)
{
static const int mdays[] = {
0, 31, 29, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31
};
int n;
char c;
n = sscanf(str, "%d/%d/%d %c", mm, dd, yy, &c);
if (n == 4) return 0; /* trailing extra chars */
if (n < 2) return 0; /* missing day */
if (n == 2) *yy = 2014; /* set default value */
if (*yy < 100) *yy += 2000; /* allow 1/1/14 */
if (*mm < 1 || *mm > 12) return 0; /* invalid month */
if (*dd < 1 || *dd > mdays[*mm]) return 0;
if (*mm == 2 && *dd == 29 % *yy % 4) return 0;
/* invalid day */
return 1;
}
/*
* Return true if line is "date \t time \t time \t text"
*/
int scan_line(char *str, struct Event *ev)
{
char *token;
token = strtok(str, "\t");
if (token == NULL) return 0;
if (!scan_date(token, &ev->year, &ev->month, &ev->day)) return 0;
token = strtok(NULL, "\t");
if (token == NULL) return 0;
if (!scan_time(token, &ev->startH, &ev->startM)) return 0;
token = strtok(NULL, "\t");
if (token == NULL) return 0;
if (!scan_time(token, &ev->endH, &ev->endM)) return 0;
token = strtok(NULL, "\t");
if (token == NULL) return 0;
strncpy(ev->event, token, 40);
return 1;
}
/*
* Remove trailing newline
*/
void chomp(char *str)
{
int l = strlen(str);
if (l && str[l - 1] == '\n') str[l - 1] = '\0';
}
/*
* Scan file with events
*/
int scan_file(const char *fn)
{
FILE *f = fopen(fn, "r");
if (f == NULL) return -1;
for (;;) {
struct Event ev;
char line[200];
if (fgets(line, 200, f) == NULL) break;
chomp(line);
if (scan_line(line, &ev)) {
printf("%s on %d/%d/%d\n",
ev.event, ev.month, ev.day, ev.year);
}
}
return 0;
}
Here, the scan_xxx functions scan a piece of data, check the format, assign the data and run a basic check on the data, so that yo'll never get an event on the 32nd of January or at 35:00h.
This makes the scanning functions more complicated than a single call to sscanf, but there are some benefits. First, the checks are done when reading the format. That means you don't have to check your data in the client code, because you can rely on sensible values. That also means that you don't have to duplicate code: Note how the checks for the time are coded only once, namely in scan_time, although the are applied twice per line, for the start and end times.
Treating the data field-wise in encapsulated functions allows you to change the format. For example, you could allow "1pm" as valid shortcut for "1:00 pm". You'd just have to re-scanf your time field with a second format string when the first format fails. You can also do that with your long single-line format, but since you have two time fields, that wouldn't be so easy.
Also note how the code above accepts 14 as shortcut for 2014 and interprets a missing year as 2014. All this might seem a bit too complicated for a simple data scanning tool, but you can re-use your functions in similar projects. Also, writing these tidy functions is more fun than wrangling longish scanf formats.

Answer 2

You can check the return of fscanf: "On success, the function returns the number of items of the argument list successfully filled. This count can match the expected number of items or be less (even zero) due to a matching failure, a reading error, or the reach of the end-of-file." If you know how many items you want to match, you can check how many were matched successfully. Now realize this, a subsequent one will pick up where the first one stopped. Meaning the next fscanf start where the other one stopped, either at the completion of a full fscanf, of the first time it encountered something not within the format.
Just brain storming but what you can do to get around this is use some form of fgets to get the line until a '\n' appears and do nothing with that line.

Answer 3

Note that whatever you do, because you have more than "type" of input line, you aren't going to be fscanfing the data straight into variables in your program. You have to know whether you have a rink name or an activity entry, before you can decide what to do with the line.
So you will first read a whole line in, then process it (and dump empty lines as you go)
You can use sscanf to see if the line is of an acceptable format. You will want to test it against the format of the activity entry first, because you will conclude that if the first element doesn't match (the first digit of a time) then you must have a rink-name. Then see if you can scan the result into a suitable rink name (you might want to check something about these).
If the sscanf for the activity entry fails on anything other than the first entry, you can tell your user which one it was and thus what it is that is wrong (IE if sscanf returns 3, then you know that the date didn't scan in properly).