r+ mode in files

[deanhystad]

I understand your confusion. r+ is confusing. When you write, were do you write?

The logical answer is that writing starts at the file pointer. When you open a file with r+, the file pointer starts at 0. When you read a file, the file pointer moves to the end. This explain why writing happens at the start of the file when you open a file with r+ and do a write. It also explains why you write at the end of a file when you open a file, do a read(), then do a write.

What isn't so well explained is what happens if you open a file, read one line, and do a write.

When I run this:

with open("test.txt", "w") as file:
    file.write("One\nTwo\nThree\n")

with open("test.txt", "r+") as file:
    file.readline()
    print(file.tell())
    file.write("Four\n")

The file pointer is 5:

Output:
5

But it creates this file:

Output:
One
Two
Three
Four

This looks like the file pointer was 13,, not 5, for the write()

And look at what happens when I set the file pointer before doing a write.

with open("test.txt", "w") as file:
    file.write("One\nTwo\nThree\n")

with open("test.txt", "r+") as file:
    file.readline()
    file.seek(file.tell())  # Move file pointer to the reported file pointer location.
    file.write("Four\n")

It creates this file:

Output:
One
Four
hree

Here the write() starts at the file position.

Bizarre!

I cannot find anything documenting the expected behavior in Python, so I looked at C++ documetnation. I found this:

https://cplusplus.com/reference/cstdio/fopen/

For files open for update (those which include a "+" sign), on which both input and output operations are allowed, the stream shall be flushed (fflush) or repositioned (fseek, fsetpos, rewind) before a reading operation that follows a writing operation. The stream shall be repositioned (fseek, fsetpos, rewind) before a writing operation that follows a reading operation (whenever that operation did not reach the end-of-file).

This doesn't explicitly say where it will be repositioned, but implies it is repositioned to the end of the file.

I think that the takeaway is that you should never write anywhere except at the end of a file, and that r+ is an odd mode.