Show log context level by indentation or prefix length

My idea is to make a context logging scheme as showed on the example below:

[   DEBUG] Parsing dialogs files
[   DEBUG] ... [DialogGroup_001]
[   DEBUG] ...... Indexing dialog xml file [c:\001_dlg.xml]
[   DEBUG] ......... dialog [LobbyA]
[   DEBUG] ............ speech nodes [3]
[   DEBUG] ............... [LobbyA_01]
[   DEBUG] ............... [LobbyA_02]
[   DEBUG] ............... [LobbyA_03]
[   DEBUG] ............ sms nodes [0]
[   DEBUG] ......... dialog [LobbyB]
[   DEBUG] ............ speech nodes [3]
[   DEBUG] ............... [LobbyB_01]
[   DEBUG] ............... [LobbyB_02]
[   DEBUG] ............... [LobbyB_03]
[   DEBUG] ............ sms nodes [0]
[   DEBUG] ... [DialogGroup_002]
[   DEBUG] ...... Indexing dialog xml file [c:\002_dlg.xml]
[   DEBUG] ......... dialog [HighGroundsA]
[   DEBUG] ............ speech nodes [3]
[   DEBUG] ..............开发者_开发问答. [HighGroundsA_01]
[   DEBUG] ............... [HighGroundsA_02]
[   DEBUG] ............... [HighGroundsA_03]
[   DEBUG] ............ sms nodes [0]

At this point, I'm using Python's logging module with custom, hand-written prefixes when logging, for example:

(...)

log.debug('')
log.debug('Parsing dialogs files')
for dlg in defDlgList:
    log.debug('... [{0}]'.format(dlg))

(...)

It's working quite ok, but there are some subtle problems, for example: when logging from inside functions - they may be called from various scopes and prefix length may vary for each call.

I'm looking for an elegant and invisible way to establish a length of a '...' prefix automatically for each log. I'd rather avoid passing prefix length as a parameter to each function or setting the length using explicit calls, for example:

(...)

logWrapper.debug('')
logWrapper.debug('Parsing dialogs files')
for dlg in defDlgList:
    logWrapper.nextLogLevelBegin()
    logWrapper.debug('[{0}]'.format(dlg))
    logWrapper.nextLogLevelEnd()

(...)

Is there a way to get the current indentation level from Python's parser or construct a scope sensitive wrapper class for logging?

Perhaps you can use inspect.getouterframes to find the indentation level:

import inspect
import logging

logger=logging.getLogger(__name__)

def debug(msg):
    frame,filename,line_number,function_name,lines,index=inspect.getouterframes(
        inspect.currentframe())[1]
    line=lines[0]
    indentation_level=line.find(line.lstrip())
    logger.debug('{i} [{m}]'.format(
        i='.'*indentation_level,
        m=msg            
        ))

def foo():    
    debug('Hi Mom')
    for i in range(1):
        debug("Now we're cookin")

if __name__=='__main__':
    logging.basicConfig(level=logging.DEBUG)
    foo()

yields

DEBUG:__main__:.... [Hi Mom]
DEBUG:__main__:........ [Now we're cookin]

Searching through the docs, I don't really see a way to get current indentation level. The best you can do, is get the current function nesting level, like this:

len(traceback.extract_stack());

Example:

import traceback;

def test():
    print len(traceback.extract_stack()); 

print len(traceback.extract_stack()); # prints 1
test(); # prints 2

Combining the previous answers with How do I add custom field to Python log format string? can achieve the same result without needing to provide a custom debug() method (since the same would need to be done for each level info(), error(), etc).

import logging
import traceback
class CustomAdapter(logging.LoggerAdapter):
    @staticmethod
    def indent():
        indentation_level = len(traceback.extract_stack())
        return indentation_level-4  # Remove logging infrastructure frames

    def process(self, msg, kwargs):
        return '{i}{m}'.format(i='\t'*self.indent(), m=msg), kwargs

logger = CustomAdapter(logging.getLogger(__name__), {})
logger.debug('A debug message')
logger.error('An error message')
logger.info('An info message')