A question about using "fscanf_s" to read data in the txt file.

Dear all,

In my app, my program has to read a set of data (each per line) in the txt file. I used the following statement in order to do this, i.e. while (fscanf_s(fp_t, "%ld\n", &number)), where the type of "number" is unsigned int.

While, if there is a number, e.g.9512312222222222222222222, in the txt file, no matter which format I specified in fscanf_s (%u or %ld), the data read from the file is 4087407502.

However, if my program, such "9512312222222222222222222" in the file which actually exceeds the upper limit of "unsigned int" should be reported as "an error". How can I implement it?

Can you give me a hand with it?

Thanks in advance,

Bests,

Qing

You need to use a 64-bit integer. Unsigned long long should do the trick. There's a format specifier for that, as well, but I'm not 100% sure what it is. %ull or something I believe.

. you could use a 64bit int <-- best bet for ya

if you scaled this...

. without having to use a large number lib, you could check input and see if it is equal to the max value of int/long/etc and throw an error; "value may* exceed ..."

. you could use a large number lib and just do a comparison, if a > b. throw error based on that.

. you could count the number of integers for the input by reading it in as const char * ("%s" format), and starting with the most significant integer checking each value to see if it ever 'exceeds' using 'atoi()'. throw error based on that.

Ah, also, read the line as a string, then use strtox (64-bit version) [not sure there is one...] to convert it a number.
The strtox series will set an error flag if overflow occurs.
This may or may not help...

Hmm - that number is 25 digits, so I guess than 19 digits of a 64-bit integer would not be sufficient.

--
Mats

Really? I got it to hex:
7A94 E8FA F3A0 E38E
Which is less than
FFFF FFFF FFFF FFFF

So it should fit.

Quote:

---

Originally Posted by Elysia

Really? I got it to hex:
7A94 E8FA F3A0 E38E
Which is less than
FFFF FFFF FFFF FFFF

So it should fit.

---

7A94E8FAF3A0E38E = 8832940933707260814.

Huh. I guess Windows Calc has a bug.
Fortunately, the PowerCalc can calculate it and it seems it results to...
7DE4F 7A94 E8FA F3A0 E38E

Yep, too big for a 64-bit integer to handle.

Quote:

---

Originally Posted by Elysia

Really? I got it to hex:
7A94 E8FA F3A0 E38E
Which is less than
FFFF FFFF FFFF FFFF

So it should fit.

---

Hey guys,

At first, thanks for all your replies.

Secondly, I plan to adopt the way of reading the line in terms of string at first, then convert it to long int meanwhile with the checking. The data in txt file is "951231134972364976" chosen randomly.

However, when I used the following statement i.e. fscanf_s(fp_t, "&#37;s\n", temp), where temp is a string pointer, and initialized to be NULL. The debugger returns an "Access violation" error...

Quote:

---

Originally Posted by Elysia

Huh. I guess Windows Calc has a bug.

---

Love it :)

Quote:

---

Originally Posted by qingxing2005

However, when I used the following statement i.e. fscanf_s(fp_t, "%s\n", temp), where temp is a string pointer, and initialized to be NULL. The debugger returns an "Access violation" error...

---

That's because pointers aren't storage units.
http://cpwiki.sourceforge.net/A_pointer_on_pointers

Quote:

---

Originally Posted by Elysia

That's because pointers aren't storage units.
http://cpwiki.sourceforge.net/A_pointer_on_pointers

---

Hey Elysia,

I found out this statement works well, i.e. fscanf_s(fp_t, "%s\n", _temp, (size_t) 100), where _temp is a string array.

You might as well read this about fscanf.
Also, if _temp is an array of char, then use sizeof(_temp) instead.

Quote:

---

Originally Posted by gltiich

. you could use a 64bit int <-- best bet for ya

if you scaled this...

. without having to use a large number lib, you could check input and see if it is equal to the max value of int/long/etc and throw an error; "value may* exceed ..."

. you could use a large number lib and just do a comparison, if a > b. throw error based on that.

. you could count the number of integers for the input by reading it in as const char * ("%s" format), and starting with the most significant integer checking each value to see if it ever 'exceeds' using 'atoi()'. throw error based on that. <- lulz

---

Hey gltiich,

I am using "atoi" or "atol" to convert string "9512311349" to integer, while none of them raised any error (actually both of them returned value "1".) and the converted value is "2147483647".

Do not use atoi/atol; use strtol.

Quote:

---

Originally Posted by MSDN

strtol returns the value represented in the string nptr, except when the representation would cause an overflow, in which case it returns LONG_MAX or LONG_MIN

---