python - recursively traverse multidimensional dictionary and export to csv -

i've have complex multidimensional dictionary want export of the key value pairs csv file running log file. i've tried various exporting cvs functions , hacked away @ of code example in stackoverflow on traversing multidimensional dictionaries have failed arrive @ solution. problem unique in has key values want export.

here dictionary:

cpu_stats = {'time_stamp': {'hour': 22, 'month': 5, 'second': 43, 'year': 2014, 'day': 29, 'minute': 31}, 'cpus': [[{'metric_type': 'cpu_index', 'value': 1}, {'metric_type': 'cpu_temperature', 'value': 39}, {'metric_type': 'cpu_fan_speed', 'value': 12000}]]} 

i need format values in time_stamp yyyy-mm-dd hh:mm:ss , store first cell of row. need values in 'cpus' cpu_index, cpu_temperature, , cpu_fan_speed in same row time stamp.

the csv file should this:

time_stamp, cpu_index, cpu_temperature, cpu_fan_speed 2014-05-29, 1, 38, 12000 

one example i've been hacking away on is:

def walk_dict(seq, level=0): """recursively traverse multidimensional dictionary , print keys , values. """  items = seq.items() items.sort() v in items:     if isinstance(v[1], dict):         # print key before make recursive call         print "%s%s" % ("  " * level, v[0])         nextlevel = level + 1         walk_dict(v[1], nextlevel)     else:         print "%s%s %s" % ("  " * level, v[0], v[1]) 

i following output

walk_dict(cpu_stats)  cpus [[{'metric_type': 'cpu_index', 'value': 1}, {'metric_type': 'cpu_temperature', 'value': 38}, {'metric_type': 'cpu_fan_speed', 'value': 12000}]] time_stamp   day 29   hour 22   minute 17   month 5   second 19   year 2014 

i have been hacking away @ function hoping can store date information variables can formatted single string. unfortuatly has recursive calls loose local variables on subsequent calls. using global futile.

def parsedictionary(obj, nested_level=0, output=sys.stdout):  spacing = '   ' if type(obj) == dict:     print >> output, '%s{' % ((nested_level) * spacing)     k, v in obj.items():         if hasattr(v, '__iter__'):             # 1st level, prints time , cpus             print >> output, '%s:' % (k)             parsedictionary(v, nested_level + 1, output)         else:             # here work             if k == "hour":                 hour = v             elif k == "month":                 month = v             elif k == "second":                 second = v             elif k == "year":                 year = v             elif k == "day":                 day = v             elif k == "minute":                 minute = v             print >> output, '%s %s' % (k, v)     print >> output, '%s}' % (nested_level * spacing) elif type(obj) == list:     print >> output, '%s[' % ((nested_level) * spacing)     v in obj:         if hasattr(v, '__iter__'):             parsedictionary(v, nested_level + 1, output)         else:             print >> output, '%s%s' % ((nested_level + 1) * spacing, v)     print >> output, '%s]' % ((nested_level) * spacing) else:     print >> output, '%s%s' % (nested_level * spacing, obj)   if __name__ == "__main__":     global year     global month     global day     global hour     global minute     global second      cpu_stats = {'time_stamp': {'hour': 22, 'month': 5, 'second': 43, 'year': 2014, 'day': 29, 'minute': 31}, 'cpus': [[{'metric_type': 'cpu_index', 'value': 1}, {'metric_type': 'cpu_temperature', 'value': 39}, {'metric_type': 'cpu_fan_speed', 'value': 12000}]]}     parsedictionary(cpu_stats)     print '%s-%s-%s %s:%s:%s' % (year, month, day, hour, minute, second) 

output:

{ time_stamp:    { hour 22 month 5 second 27 year 2014 day 29 minute 57 cpus:    [       [          { metric_type cpu_index value 1          { metric_type cpu_temperature value 39          { metric_type cpu_fan_speed value 12000       ]    ] traceback (most recent call last):   file "./cpu.py", line 135, in <module>     print '%s-%s-%s %s:%s:%s' % (year, month, day, hour, minute, second) nameerror: global name 'year' not defined 

thanks, appreciate in pointing me in right direction i'm @ loss.

i agree @desired login, assuming have no control of incoming data , had work showed in questions... traverse so:

cpu_stats = {'time_stamp': {'hour': 22, 'month': 5, 'second': 43, 'year': 2014, 'day': 29, 'minute': 31},               'cpus': [ [{'metric_type': 'cpu_index', 'value': 1}, {'metric_type': 'cpu_temperature', 'value': 39}, {'metric_type': 'cpu_fan_speed', 'value': 12000} ] ]              }  timestamp = '' stats in cpu_stats.keys():     if stats == 'time_stamp':         timestamp = '{year}-{month}-{day}'.format(**cpu_stats[stats])     if stats == 'cpus':         cpu in cpu_stats[stats]:             cpu_index = ''             cpu_temperature = ''             cpu_fan_speed = ''             metric in cpu:                 if metric['metric_type'] == 'cpu_index':                     cpu_index = str(metric['value'])                 elif metric['metric_type'] == 'cpu_temperature':                     cpu_temperature = str(metric['value'])                 elif metric['metric_type'] == 'cpu_fan_speed':                     cpu_fan_speed = str(metric['value'])             print ','.join([timestamp, cpu_index, cpu_temperature, cpu_fan_speed])