How to replace a string in a file without changing encoding

When you're opening the file using 'wb' and writing, you are overwriting the entire file each time through. What you want is to control the write/read head, move it where you need it, and overwrite where needed.

A few changes will need to be made. First you need a reference to the open file. with open() I didn't test, so idk if it works - I'm sure it does. You can use the plain file = open('filepath') or test with open(), but you need the reference and if you don't use with open(), you'll need to explicitly call file.close() when you're done. Second you want to open the file using rb+, not wb so that you're not overwriting the file each time. Finally, you want to control the read/write head so that you can position it correctly for your reads/writes/overwrites.

def main():
    file = open("test.txt", "rb")
    filePos = 0
    
    while True:
        # Read the file character by character
        char = file.read(1)
        # When we find the char we want, break the loop and save the read/write head position.
        # Since we're in binary, we need to decode to get it to proper format for comparison (or encode the char)
        if char.decode('ascii') == "*":
            filePos = file.tell()
            break
        # If no more characters, we're at the end of the file. Break the loop and end the program.
        elif not char:
            break
       
    
    # Resolve open/unneeded file pointers.
    file.close()
    
    # Open the file in rb+ mode for writing without overwriting.
    fileWrite = open("test.txt", 'rb+')
    
    # Move the read/write head to the location we found our char at. 
    fileWrite.seek(filePos - 1)
    
    # Overwrite our char.
    fileWrite.write(bytes("?", "ascii"))
    
    # Close the file
    fileWrite.close()
    
if __name__ == "__main__":
    main()

**Starting file contents:** This is old data with this * weird symbol that needs replacing.

**Completed file contents:** This is old data with this ? weird symbol that needs replacing.

The file is not 100% encoded text, so that's why it must be opened as binary. You can't use readlines() as the spec indicates space-padded fields in the ASCII fields, i.e., no newlines. Use 'r+b' to open for read/update of the binary file, seek() to the offset of the byte you want to replace, and write(b'?') the question mark byte.

Example:

# Create a small example file
with open("heartbeat-baseline-700001.edf", "wb") as f:
    f.write(b'ABCDEFG')

# Change a byte
with open("heartbeat-baseline-700001.edf", "r+b") as f:
    f.seek(3)     # offset of "D"
    f.write(b'?') # change to "?"

# read and display
with open("heartbeat-baseline-700001.edf", "rb") as f:
    print(f.read())

Output:

b'ABC?EFG'

Another issue is the '°' character. It's not an ASCII character which violates the spec (likely why you are replacing it), but what encoding is it using? It's probably ISO-8859-1 or Windows-1252...both of which encode as the byte b'\xb0'. but it could be UTF-8 which encodes as the two bytes b'\xc2\xb0'. Assuming the former, that byte could occur in any of the data record integer fields as well, so be careful to replace the correct byte. The UTF-8 pair could occur in the data fields as well, but less likely.

If you know the bad character is byte b'\xb0' and it occurs in the header before any data record entries, you could read the whole file, make a single replacement, and write the whole file back:

# read the whole file as binary
with open('heartbeat-baseline-700001.edf', 'rb') as f:
    data = f.read()

# replace the 1st 0xb0 byte found
data = data.replace(b'\xb0', b'?', 1)

# write the whole file back
with open('heartbeat-baseline-700001.edf', 'wb') as f:
    f.write(data)