Myujth

I've just come across an issue where I was jumping between valgrind in Linux and other testing in Windows cmd.

I'm reading a certain line from a file like this:

fgets(buf, MAX_LINE_LEN, f_input);

Of course, buf is the size MAX_LINE_LEN + 1, but I digress.

This is the output of

printf("String length: %u; Contents: ", strlen(buf));
for (usint i = 0; i < strlen(buf); i++)
 printf("%x ", buf[i]);
puts(";");

in Windows:

String length: 14; Contents: 41 6e 64 72 65 6a 20 50 6c 61 76 6b 61 a ;
String length: 22; Contents: 41 6e 6e 61 20 4d 61 72 69 61 20 43 69 63 6d 61 6e 63 6f 76 61 a ;
String length: 25; Contents: 4d 61 72 69 61 20 52 61 7a 75 73 6f 76 61 20 4d 61 72 74 61 6b 6f 76 61 a ;
String length: 24; Contents: 4d 69 6c 61 6e 20 52 61 73 74 69 73 6c 61 76 20 50 6f 6b 6f 6a 6e 79 a ;
String length: 21; Contents: 4d 69 6c 65 6e 61 20 53 65 64 6d 69 6b 72 61 73 6b 6f 76 61 a ;
String length: 15; Contents: 56 69 6e 63 65 6e 74 20 53 69 6b 75 6c 61 a ;
String length: 17; Contents: 56 69 6e 63 65 6e 74 20 76 61 6e 20 47 6f 67 68 a ;

and in Linux:

String length: 15; Contents: 41 6e 64 72 65 6a 20 50 6c 61 76 6b 61 d a ;
String length: 23; Contents: 41 6e 6e 61 20 4d 61 72 69 61 20 43 69 63 6d 61 6e 63 6f 76 61 d a ;
String length: 26; Contents: 4d 61 72 69 61 20 52 61 7a 75 73 6f 76 61 20 4d 61 72 74 61 6b 6f 76 61 d a ;
String length: 25; Contents: 4d 69 6c 61 6e 20 52 61 73 74 69 73 6c 61 76 20 50 6f 6b 6f 6a 6e 79 d a ;
String length: 22; Contents: 4d 69 6c 65 6e 61 20 53 65 64 6d 69 6b 72 61 73 6b 6f 76 61 d a ;
String length: 16; Contents: 56 69 6e 63 65 6e 74 20 53 69 6b 75 6c 61 d a ;
String length: 18; Contents: 56 69 6e 63 65 6e 74 20 76 61 6e 20 47 6f 67 68 d a ;

As you can see in Linux, there is another character before the NL, a Carriage Return. If anyone can explain this and save me the pain of adding ifdef statements for a Linux and Windows code, I'd appreciate it. I understand, that linux appends a Carriage Return after each line, but is this really the intended behaviour when it then gets read by fgets?

As you can see in Linux, there is another character before the NL, a Carriage Return.

That is because your files use CR+LF newlines, i.e. each newline is actually two characters: "rn".

If you open files without the "b" flag in Windows, its C library will convert each n you write to rn, and each rn you read to n.

Use the "b" fopen() flag in Windows to see the actual file contents.

When you read a line using fgets(buf, sizeof buf, handle), you can use buf[strcspn(buf, "rn")] = ''; to remove the newline.

MS and Linux has a different expectation of a text file line ending:"rn" vs "n".

To cope, recommend after fgets() use strcspn() to lop off the potential end of line sequence, be it "n", "rn" or missing.

fgets(buf, MAX_LINE_LEN, f_input);
buf[strcspn(buf, "nr")] = '';

Some compilers on Windows will use "n" as the end-of-line sequence and others use "rn". So I attribute the variation to compilers and their manufacturers more so than the OS. Also some old MAC text files end with 'r' and will foul fgets() on Linux.

Further: reading a file that has "rn" as a text file that expects "n" as the end-of-line sequence has a problem when reading a full buffer as "......r" and the line remainder as "n" on the next fgets(). Additional processing is needed to cope as is the case whenever the buffer is insufficient for a line of input.

Text files of one variation are often copied to the other platforms, so this is a not-so-rare occurrence.

Due to editing, some text files will have a mixture of line-ending-sequences.

Pedantic code will read the file as binary and process variant line endings itself without fgets(). Good luck.

In C you an open a file stream in text or binary mode. In binary mode, no translation takes place, and the input and output are the bytes in the file. In text mode, the C "newline" character is translated into what is common on the platform in question. One UNIX-like systems, this is a 0A byte, and on DOS-like systems this is a 0D byte followed by a 0A byte. There are other cases on other operating systems, listed here:

https://en.wikipedia.org/wiki/Newline

So that you don't have to cope with every different text format in every program, these all get translated into an n character as far as the C program sees in the default case (text mode). The input/output layer does the necessary translations for you.

When you use fopen() to open a file stream in C for reading or writing, you provide a "file mode" parameter - you're probably using it here as "r" to read a file, or "w" to write one. If you want to newline translation done you can specify that the stream is opened in binary mode, with "rb" for reading or "wb" for writing.