Hi all
I have updated the regular expression to allow for arbitrary whitespace.
def dict_of_key_value_pairs(arg):
""" parse KEY=val,KEY2=val2 into {'KEY':'val',
'KEY2':'val2'}
Quotes can be used to allow commas in the value """
D = {}
import re
# Regular expression to match key=value, key='value' or key="value"
strings
# Quoted values can allow commas, unquoted will
not
regex = re.compile(r"[ \t]*(?P<key>\w+)[ \t]*=[
\t]*(?P<quote>['\"])?(?P<value>.*?)(?(quote)['\"])[,$]")
# Work-around for the fact that the non-greedy value match will not let
the
# regular expression parse the last key=value pair if value is not
quoted
if len(arg) and arg[-1] != ',':
arg += ','
matches = regex.findall(arg)
for match in matches:
# Match will be a tuple of length 3 containing (key, quote,
value)
D[match[0]] = match[2]
return D
Cheers,
Tim
On Thu, May 21, 2009 at 3:10 PM, Tim Godfrey <[email protected]> wrote:
> Hi all
>
> I'm a little new to this but I'd like to propose a change to the
> dict_of_key_value_pairs function in datatypes.py
>
> Basically my problem is this: I have a process I need to monitor defined in
> my supervisor.conf. I need to set an environment variable on it as well.
> This environment variable has commas in the string, therefore the
> dict_of_key_value_pairs parser wont accept it. I've checked svn trunk and
> the function currently looks like this:
>
> def dict_of_key_value_pairs(arg):
> """ parse KEY=val,KEY2=val2 into {'KEY':'val', 'KEY2':'val2'} """
> D = {}
> try:
> pairs = filter(None, arg.split(','))
>
> for pair in pairs:
> try:
> k, v = pair.split('=', 1)
> except ValueError:
> raise ValueError('Unknown key/value pair %s' % pair)
> D[k.strip()] = v.strip()
>
> except:
> raise ValueError("not a list of key/value pairs: " + repr(arg))
>
> return D
>
> I propose that it gets changed to this:
>
> def dict_of_key_value_pairs(arg):
> """ parse KEY=val,KEY2=val2 into {'KEY':'val',
> 'KEY2':'val2'}
>
> Quotes can be used to allow commas in the value """
> D = {}
> import re
>
> # Regular expression to match key=value, key='value' or key="value"
> strings
> # Quoted values can allow commas, unquoted will
> not
>
> regex =
> re.compile(r"(?P<key>\w+)=(?P<quote>['\"])?(?P<value>.*?)(?(quote)['\"])[,$]")
>
> # Work-around for the fact that the non-greedy value match will not let
> the
> # regular expression parse the last key=value pair if value is not
> quoted
> if len(arg) and arg[-1] != ',':
> arg += ','
>
> matches = regex.findall(arg)
>
> for match in matches:
> # Match will be a tuple of length 3 containing (key, quote,
> value)
> D[match[0]] = match[2]
>
> return D
>
> I've done thorough testing of my version of the function, and it seems to
> work fine. Let me know what you think.
>
> If you would prefer I send a patch please let me know.
>
> Cheers,
>
> Tim
>
_______________________________________________
Supervisor-users mailing list
[email protected]
http://lists.supervisord.org/mailman/listinfo/supervisor-users
