◐ Shell
clean mode source ↗

Issue 3587: Bad parsing of compiling regex with re.MULTILINE

import re
regex = r"[\w]+"

# Normal behaviour:
>>> re.findall(regex, "hello world", re.M)
['hello', 'world']
>>> re.compile(regex).findall("hello world")
['hello', 'world']

# Bug behaviour:
>>> re.compile(regex).findall("hello world", re.M)
['rld']
The re.M flag is an attribute of the compiled pattern, and as such it
must be passed to compile(), not to findall(). 

These all work:

>>> re.compile(r"[a-z]+").findall("hello world")
['hello', 'world']
>>> re.compile(r"[a-z]+", re.M).findall("hello world")
['hello', 'world']
>>> re.compile(r"(?m)[a-z]+").findall("hello world")
['hello', 'world']

The second argument to the findall() method of compile objects is the
start position to match from (see
http://docs.python.org/lib/re-objects.html). This explains the behaviour
you are witnessing:

>>> re.M
8
>>> re.compile(r"[a-z]+").findall("hello world", 8)
['rld']