Simple functions for converting raw graphite data to pandas time series / frame. Reasonably fast too; my test set was a 16MiB 350 metric dump, timing on my machine to parse is ~400ms.

graphite_to_pandas.py

import numpy as np
import pandas as pd
from cStringIO import StringIO

def parse_raw_graphite_metric(line):
    head, body = line.rstrip('\n').split('|', 1)
    name, since, until, freq = head.replace('\n', '').split(',')
    since = float(since) * 1e9
    until = float(until) * 1e9
    freq  = '%ss' % int(freq)
    data  = body.rstrip('\n').split(',')
    index = pd.date_range(start=since, end=until, freq=freq, tz='UTC')
    index = index[:len(data)]
    S = pd.Series(
        name  = name,
        data  = data,
        index = index,
    )
    S.replace('None', np.nan, inplace=True)
    return S.astype(float)

def parse_raw_graphite(content):
    # in testing on a 16MiB 350 metric data dump:
    # - str.split         ~ 11ms
    # - custom lazy split ~  9ms
    # - cStringIO         ~  6ms
    metrics = (
        parse_raw_graphite_metric(line)
        for line in StringIO(content)
    )
    return pd.DataFrame.from_items(
        (metric.name, metric)
        for metric in metrics
    )