How can I write to a file with the same formatting as print?

Question

TL;DR

While trying to write a string to a file the following error occurred:

Code

logfile.write(cli_args.last_name)

Output

UnicodeEncodeError: 'ascii' codec can't encode characters in position 8-9: ordinal not in range(128)

But this works:

Code

print(cli_args.last_name)

Output

Pérez

Why?

FULL CONTEXT

I made a script which receives data from a Linux CLI, processes it and finally creates a Zendesk ticket with the provided data. It is kind of a CLI API, since before my script there is a bigger system which has a web interface with forms, where users fill the values of the fields and are then replaced into the CLI script. For example:

myscript.py --first_name '_first_name_' --last_name '_last_name_'

The script was working with no issues, until yesterday when the web was updated. I think they changed something related to charsets or encoding.

I do some simple logging with F-strings by opening a file and writing some informative messages in case anything fails, so I can go back to check where it happened. Also the CLI attributes are read using the argparse module. Example:

logfile.write(f"\tChecking for opened tickets for user '{cli_args.first_name} {cli_args.last_name}'\n")

After the website update I am getting an error like this:

UnicodeEncodeError: 'ascii' codec can't encode characters in position 8-9: ordinal not in range(128)

Doing some troubleshooting I found it is because some users input names with accent marks like Carlos Pérez.

I need the script to work again and also prepare it for inputs like that, so I looked for answers by checking the HTTP headers in the input forms of the web console and found out it uses a Content-Type: text/html; charset=UTF-8; my first try was to encode the str passed in the CLI argument to utf-8 and decode it again using the same codec, but didn't succeed.

On my second try, I checked the Python docs str.encode() and bytes.decode(). So I tried this:

logfile.write(
    "\tChecking for opened tickets for user "
    f"'{cli_args.first_name.encode(encoding='utf-8', errors='ignore').decode('utf-8')} "
    f"{cli_args.last_name.encode(encoding='utf-8', errors='ignore').decode('utf-8')}'"
)

It worked but removed the accent marked letter so Carlos Pérez became Carlos Prez which is of no use to me in this case, I need the full input.

As a desperate move I tried printing the same F-string I was trying to write to the logfile, which to my surprise it worked. It printed to the console Carlos Pérez without any kind of encoding/decoding process.

How does print work? and Why trying to write to the file didn't work? But most importantly How can I write to a file with the same formatting as print?

Edit 1 @MarkTolonen

Tried the following:

logfile = open("/usr/share/pandora_server/util/plugin/plugin_mcm/sandbox/755bug.txt", mode="a", encoding="utf8")
logfile.write(cli_args.body)
logfile.close()

Output:

Traceback (most recent call last): File "/usr/share/pandora_server/util/plugin/plugin_mcm/sandbox/ticket_query_app.py", line 414, in main() File "/usr/share/pandora_server/util/plugin/plugin_mcm/sandbox/ticket_query_app.py", line 81, in main logfile.write(cli_args.body) UnicodeEncodeError: 'utf-8' codec can't encode characters in position 8-9: surrogates not allowed

Edit 2

I managed to get the text that is causing the issue:

if __name__ == "__main__":
    string = (
        "Buenos d\udcc3\udcadas,\r\n\r\n"
        "Mediante  monitoreo autom\udcc3\udca1tico se ha detectado un evento fuera de lo normal:\r\n\r\n"
        "Descripci\udcc3\udcb3n del evento: _snmp_f13_\r\n"
        "Causas sugeridas del evento: _snmp_f14_\r\n"
        "Posible afectaci\udcc3\udcb3n del evento: _snmp_f15_\r\n"
        "Validaciones de bajo impacto: _snmp_f16_\r\n"
        "Fecha y hora del evento: 2021-07-14 17:47:51\r\n\r\n"
        "Saludos."
    )

    # Output: Text with the unicodes translated
    print(string)

    # Output: "UnicodeEncodeError: 'utf-8' codec can't encode characters in position 8-9: surrogates not allowed"
    with open(file="test.log", mode="w", encoding="utf8") as logfile:
        logfile.write(string)

Answer 1

It looks like something upstream is misconfigured. Your string appears to have been produced by a decode operation with the wrong encoding, with errors='surrogateescape' error handling. From the data shown, it looks like the decoding operation tried to decode UTF-8-encoded text as ASCII.

errors='surrogateescape' is a way for an encoding to handle invalid bytes during a decode operation. The error handler replaces the invalid bytes with partial surrogates in the range U+DC80..U+DCFF when converting to a Unicode string, and the process can be reversed to get the original byte string back by performing an encode with errors='surrogateescape' and the same encoding.

The partial surrogates in your string match the pattern of what a decode(encoding='ascii', errors='surrogateescape') call would produce when given data actually encoded in UTF-8 - the surrogates are all in the range surrogateescape uses, and the bytes they correspond to form valid UTF-8. In the code below, I recover the original bytes, then decode them correctly as UTF-8. Once the Unicode string is valid, it can be written to the log file with encoding='utf8'.

string = (
    "Buenos d\udcc3\udcadas,\r\n\r\n"
    "Mediante  monitoreo autom\udcc3\udca1tico se ha detectado un evento fuera de lo normal:\r\n\r\n"
    "Descripci\udcc3\udcb3n del evento: _snmp_f13_\r\n"
    "Causas sugeridas del evento: _snmp_f14_\r\n"
    "Posible afectaci\udcc3\udcb3n del evento: _snmp_f15_\r\n"
    "Validaciones de bajo impacto: _snmp_f16_\r\n"
    "Fecha y hora del evento: 2021-07-14 17:47:51\r\n\r\n"
    "Saludos."
)

fixed = string.encode('ascii',errors='surrogateescape').decode('utf8')
print(fixed)

with open(file="test.log", mode="w", encoding="utf8") as logfile:
    logfile.write(fixed)

You can read more about surrogate escapes in PEP 383.

Answer 2

The answer is the encoding parameter to open. Observe:

Last login: Wed Jul 14 15:05:24 2021 from 50.126.68.34
[timrprobocom@jared-ingersoll ~]$ python3
Python 3.6.9 (default, Jan 26 2021, 15:33:00)
[GCC 8.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> f = open('x.txt','a')
>>> g = open('y.txt','a',encoding='utf-8')
>>> s = "spades \u2660 spades"
>>> f.write(s)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
UnicodeEncodeError: 'ascii' codec can't encode character '\u2660' in position 7: ordinal not in range(128)
>>> g.write(s)
15
>>>
[timrprobocom@jared-ingersoll ~]$ hexdump -C y.txt
00000000  73 70 61 64 65 73 20 e2  99 a0 20 73 70 61 64 65  |spades ... spade|
*
00000011