I've made a program that opens files and searches for a word
I want it to only work on TEXT Files
Is there a way provided by C to check if a file is BINARY, and if so, I want to exit the program before any operations take place
Thanks
4 Answers
No, there isn't, because it's impossible to tell for sure. If you expect a specific encoding, you can check yourself whether the file contents are valid in this encoding, e.g. if you expect ASCII, all bytes must be <= 0x7f. If you expect UTF-8, it's a bit more complicated, see a description of it.
In any case, there's no guarantee that a "binary" file would not by accident look like a valid file in any given text encoding. In fact, the term "binary file" doesn't make too much sense, as all files contain binary data.
1 Comment
Shaun F
Right now this seems to work, I don't expect to be working on special symbols. I'm trying to produce a program that works like GREP on linux. Thanks!
If we assume that by text you mean ASCII and not UTF-8, you can do this by reading each character and using isascii() and isspace() to check if it is a valid character:
void is_text(char *filename) {
FILE *f = fopen(filename, "r");
if (!f) {
perror("fopen failed");
return;
}
int c;
while ((c=fgetc(f) != EOF) {
if ((!isascii(c) || iscntrl(c)) && !isspace(c)) {
printf("is binary\n");
fclose(f);
return;
}
}
printf("is text\n");
fclose(f);
}
If the file contains UTF-8 characters, it becomes more complicated as you have to look at multiple bytes at once and see if they are valid UTF-8 byte sequences. There's also the question of which Unicode code points are considered text.
2 Comments
Shaun F
I am trying to emulate the functioning of GREP in Linux, I don't think it works on unicode characters so I don't think it will be a problem. Thanks for your help!
Andrushenko Alexander
There is typo in the answer: instead of
c=fgetc(c) it must be c=fgetc(f).It's not the file per se which is binary or text; it is just about how you interpret the content of the file when opening it.
You may interpret a file containing solely text as binary, thereby avoiding that a /r/n might get translated to a /n only; And you may open a file containing raw data like, for example, a bitmap using a text mode, thereby probably corrupting the content in that a 0x0D 0x0A gets converted to a 0x0D only.
So you cannot check the file per se, but you may open the file in binary mode and see if the content contains anything which you do not interpret as text.
Comments
perhaps: system(file "path/filename");
isprint().