p2pool/util/graph.py

   1 from __future__ import absolute_import
   2 from __future__ import division
   3
   4 import math
   5
   6 from p2pool.util import math as math2
   7
   8
   9 class DataViewDescription(object):
  10     def __init__(self, bin_count, total_width):
  11         self.bin_count = bin_count
  12         self.bin_width = total_width/bin_count
  13
  14 def _shift(x, shift, pad_item):
  15     left_pad = math2.clip(shift, (0, len(x)))
  16     right_pad = math2.clip(-shift, (0, len(x)))
  17     return [pad_item]*left_pad + x[right_pad:-left_pad if left_pad else None] + [pad_item]*right_pad
  18
  19 combine_bins = math2.add_dicts_ext(lambda (a1, b1), (a2, b2): (a1+a2, b1+b2), (0, 0))
  20
  21 nothing = object()
  22 def keep_largest(n, squash_key=nothing, key=lambda x: x, add_func=lambda a, b: a+b):
  23     def _(d):
  24         items = sorted(d.iteritems(), key=lambda (k, v): (k != squash_key, key(v)), reverse=True)
  25         while len(items) > n:
  26             k, v = items.pop()
  27             if squash_key is not nothing:
  28                 items[-1] = squash_key, add_func(items[-1][1], v)
  29         return dict(items)
  30     return _
  31
  32 def _shift_bins_so_t_is_not_past_end(bins, last_bin_end, bin_width, t):
  33     # returns new_bins, new_last_bin_end
  34     shift = max(0, int(math.ceil((t - last_bin_end)/bin_width)))
  35     return _shift(bins, shift, {}), last_bin_end + shift*bin_width
  36
  37 class DataView(object):
  38     def __init__(self, desc, ds_desc, last_bin_end, bins):
  39         assert len(bins) == desc.bin_count
  40
  41         self.desc = desc
  42         self.ds_desc = ds_desc
  43         self.last_bin_end = last_bin_end
  44         self.bins = bins
  45
  46     def _add_datum(self, t, value):
  47         if not self.ds_desc.multivalues:
  48             value = {'null': value}
  49         elif self.ds_desc.multivalue_undefined_means_0 and 'null' not in value:
  50             value = dict(value, null=0) # use null to hold sample counter
  51         self.bins, self.last_bin_end = _shift_bins_so_t_is_not_past_end(self.bins, self.last_bin_end, self.desc.bin_width, t)
  52
  53         bin = int(math.floor((self.last_bin_end - t)/self.desc.bin_width))
  54         assert bin >= 0
  55         if bin < self.desc.bin_count:
  56             self.bins[bin] = self.ds_desc.keep_largest_func(combine_bins(self.bins[bin], dict((k, (v, 1)) for k, v in value.iteritems())))
  57
  58     def get_data(self, t):
  59         bins, last_bin_end = _shift_bins_so_t_is_not_past_end(self.bins, self.last_bin_end, self.desc.bin_width, t)
  60         assert last_bin_end - self.desc.bin_width <= t <= last_bin_end
  61
  62         def _((i, bin)):
  63             left, right = last_bin_end - self.desc.bin_width*(i + 1), min(t, last_bin_end - self.desc.bin_width*i)
  64             center, width = (left+right)/2, right-left
  65             if self.ds_desc.is_gauge and self.ds_desc.multivalue_undefined_means_0:
  66                 real_count = max([0] + [count for total, count in bin.itervalues()])
  67                 if real_count == 0:
  68                     val = None
  69                 else:
  70                     val = dict((k, total/real_count) for k, (total, count) in bin.iteritems())
  71                 default = 0
  72             elif self.ds_desc.is_gauge and not self.ds_desc.multivalue_undefined_means_0:
  73                 val = dict((k, total/count) for k, (total, count) in bin.iteritems())
  74                 default = None
  75             else:
  76                 val = dict((k, total/width) for k, (total, count) in bin.iteritems())
  77                 default = 0
  78             if not self.ds_desc.multivalues:
  79                 val = None if val is None else val.get('null', default)
  80             return center, val, width, default
  81         return map(_, enumerate(bins))
  82
  83
  84 class DataStreamDescription(object):
  85     def __init__(self, dataview_descriptions, is_gauge=True, multivalues=False, multivalues_keep=20, multivalues_squash_key=None, multivalue_undefined_means_0=False, default_func=None):
  86         self.dataview_descriptions = dataview_descriptions
  87         self.is_gauge = is_gauge
  88         self.multivalues = multivalues
  89         self.keep_largest_func = keep_largest(multivalues_keep, multivalues_squash_key, key=lambda (t, c): t/c if self.is_gauge else t, add_func=lambda (a1, b1), (a2, b2): (a1+a2, b1+b2))
  90         self.multivalue_undefined_means_0 = multivalue_undefined_means_0
  91         self.default_func = default_func
  92
  93 class DataStream(object):
  94     def __init__(self, desc, dataviews):
  95         self.desc = desc
  96         self.dataviews = dataviews
  97
  98     def add_datum(self, t, value=1):
  99         for dv_name, dv in self.dataviews.iteritems():
 100             dv._add_datum(t, value)
 101
 102
 103 class HistoryDatabase(object):
 104     @classmethod
 105     def from_obj(cls, datastream_descriptions, obj={}):
 106         def convert_bin(bin):
 107             if isinstance(bin, dict):
 108                 return bin
 109             total, count = bin
 110             if not isinstance(total, dict):
 111                 total = {'null': total}
 112             return dict((k, (v, count)) for k, v in total.iteritems()) if count else {}
 113         def get_dataview(ds_name, ds_desc, dv_name, dv_desc):
 114             if ds_name in obj:
 115                 ds_data = obj[ds_name]
 116                 if dv_name in ds_data:
 117                     dv_data = ds_data[dv_name]
 118                     if dv_data['bin_width'] == dv_desc.bin_width and len(dv_data['bins']) == dv_desc.bin_count:
 119                         return DataView(dv_desc, ds_desc, dv_data['last_bin_end'], map(convert_bin, dv_data['bins']))
 120             elif ds_desc.default_func is None:
 121                 return DataView(dv_desc, ds_desc, 0, dv_desc.bin_count*[{}])
 122             else:
 123                 return ds_desc.default_func(ds_name, ds_desc, dv_name, dv_desc, obj)
 124         return cls(dict(
 125             (ds_name, DataStream(ds_desc, dict(
 126                 (dv_name, get_dataview(ds_name, ds_desc, dv_name, dv_desc))
 127                 for dv_name, dv_desc in ds_desc.dataview_descriptions.iteritems()
 128             )))
 129             for ds_name, ds_desc in datastream_descriptions.iteritems()
 130         ))
 131
 132     def __init__(self, datastreams):
 133         self.datastreams = datastreams
 134
 135     def to_obj(self):
 136         return dict((ds_name, dict((dv_name, dict(last_bin_end=dv.last_bin_end, bin_width=dv.desc.bin_width, bins=dv.bins))
 137             for dv_name, dv in ds.dataviews.iteritems())) for ds_name, ds in self.datastreams.iteritems())
 138
 139
 140 def make_multivalue_migrator(multivalue_keys, post_func=lambda bins: bins):
 141     def _(ds_name, ds_desc, dv_name, dv_desc, obj):
 142         if not obj:
 143             last_bin_end = 0
 144             bins = dv_desc.bin_count*[{}]
 145         else:
 146             inputs = dict((k, obj.get(v, {dv_name: dict(bins=[{}]*dv_desc.bin_count, last_bin_end=0)})[dv_name]) for k, v in multivalue_keys.iteritems())
 147             last_bin_end = max(inp['last_bin_end'] for inp in inputs.itervalues()) if inputs else 0
 148             assert all(len(inp['bins']) == dv_desc.bin_count for inp in inputs.itervalues())
 149             inputs = dict((k, dict(zip(['bins', 'last_bin_end'], _shift_bins_so_t_is_not_past_end(v['bins'], v['last_bin_end'], dv_desc.bin_width, last_bin_end)))) for k, v in inputs.iteritems())
 150             assert len(set(inp['last_bin_end'] for inp in inputs.itervalues())) <= 1
 151             bins = post_func([dict((k, v['bins'][i]['null']) for k, v in inputs.iteritems() if 'null' in v['bins'][i]) for i in xrange(dv_desc.bin_count)])
 152         return DataView(dv_desc, ds_desc, last_bin_end, bins)
 153     return _