re.sub regex won't match

[devnull]

https://regex101.com/r/vouYnV/1

Any idea why my regex won't find & remove this hex value embedded in carriarge return / line feeds :

Code: Select all

>>> re.sub(r'^\\r\\n[0-9,a-f]+\\r\\n','','\r\nfe5\r\nabcdefghxxxxx')
'\r\nfe5\r\nabcdefghxxxxx'

on regex101 it matches but I can't get it to work in uPython !

what it should return is:

Code: Select all

abcdefghxxxxx

This is used to removed 'chunked' data from http 1.1 requests.

[jickster]

Somewhere I read micropython has issues with r-strings; create the string as a non r-string.

[pfalcon]

MicroPython doesn't have issues with r-strings (unless someone provides a testcase).

As often happens, this case has 2 layers of confusion.

First of all, the regexp is not correct for the purpose stated (can be easily verified in CPython).

Secondly ure (sic!) module doesn't implement any superfluous escapes which can be otherwise done in Python itself. So, r"\n" isn't going to match "\n". But of course "\n" will match "\n".

Module re-pcre from micropython-lib provides (almost) complete regex syntax.

[devnull]

Sorry pfalcon I don't understand your reply and also I don't use cpython.

EDITED

This without using 'r' strings, this works in uPython:

Code: Select all

>>> re.sub('^\r\n[0-9,a-f]+\r\n','','\r\nfe5\r\nabcdefghxxxxx')
'abcdefghxxxxx'
>>> 

[efahl]

Paul is saying that r'\n' and '\n' are completely different strings, which can be verified in any version of python.

r'\n' -> two-character string containing '\' and 'n'.
'\n' -> string containing a single newline character.

This is not a bug in r-strings, it is in fact the reason they exist.

[dhylands]

The bug is in the re module which if passed a pattern containing the two characters \ followed by r (i.e. r'\r') then it should match the single character carriage return (i.e. '\r').

i.e. in CPython:

Code: Select all

>>> re.sub(r'a\rb', '', 'aa\rbb')
'ab'

whereas MicroPython (on my pyboard) gives:

Code: Select all

>>> re.sub(r'a\rb', '', 'aa\rbb')
'aa\rbb'

[pfalcon]

and also I don't use cpython.

Well, you should. MicroPython Unix port and reference implementation of the Python language, CPython, should be at the fingertips of every MicroPython programmer, or they complicate their lives unnecessarily - literally, looking at the vast (Micro)Python landscape thru narrow tube of a serial connection.

[pfalcon]

The bug is in the re module

As mentioned above, there's no (builtin) "re" module in MicroPython, only "ure". And "ure" works per the specification (that being absolutely smallest and not duplicating functionality already provided by MicroPython, like quoting).

However, there's a bug in some ports, where native MicroPython modules, starting with "u", are accessible by different name (Unix port for example doesn't have such a bug, so testing with it would show problem right away). Yes, it initially was conceived as a convenience, but long since has become mis-convenience. The rule of thumb is simple - someone developing for MicroPython should always use "u" modules, exactly how they're named in the docs: http://docs.micropython.org/en/latest/l ... -libraries

The only reason to use non-"u" module is writing code which is compatible with CPython. Which can be called "advanced topic" given confessions like "I don't use cpython".


Oh, and the docs for ure module are updated to be more complete: https://pycopy.readthedocs.io/en/latest ... y/ure.html

[devnull]

@pfalcon - I think you have your wires crossed my friend !

I have never implied any bug (other posters may have), I was simply asking for help with a regex query.

I'm sorry if asking for this kind of help is not acceptable to you ?!

[pfalcon]

devnull, I'm not sure what you're talking about. There're 2 my replies above, one clearly addressed to you, another to dhylands. Feel free to skip anything not addressed to you (or anything at all).