toolkit/components/telemetry/histogram_tools.py
author Ryan VanderMeulen <ryanvm@gmail.com>
Thu, 15 Nov 2012 21:07:38 -0500
changeset 113462 a0cd4bcb321795f59a9c05a934b6aa02cd9918b3
parent 113449 cc2469a8b41ad3d3472f3cd691e25c4d9f50558b
child 114249 16dc855a02165608489a9c1ca17b3939df022157
permissions -rw-r--r--
Backed out changeset cc2469a8b41a (bug 800557) for Windows build bustage.

# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at http://mozilla.org/MPL/2.0/.

from __future__ import with_statement

import math
import simplejson as json

def table_dispatch(kind, table, body):
    """Call body with table[kind] if it exists.  Raise an error otherwise."""
    if kind in table:
        return body(table[kind])
    else:
        raise BaseException, "don't know how to handle a histogram of kind %s" % kind

class DefinitionException(BaseException):
    pass

def check_numeric_limits(dmin, dmax, n_buckets):
    if type(dmin) != int:
        raise DefinitionException, "minimum is not a number"
    if type(dmax) != int:
        raise DefinitionException, "maximum is not a number"
    if type(n_buckets) != int:
        raise DefinitionException, "number of buckets is not a number"

def linear_buckets(dmin, dmax, n_buckets):
    check_numeric_limits(dmin, dmax, n_buckets)
    ret_array = [0] * n_buckets
    dmin = float(dmin)
    dmax = float(dmax)
    for i in range(1, n_buckets):
        linear_range = (dmin * (n_buckets - 1 - i) + dmax * (i - 1)) / (n_buckets - 2)
        ret_array[i] = int(linear_range + 0.5)
    return ret_array

def exponential_buckets(dmin, dmax, n_buckets):
    check_numeric_limits(dmin, dmax, n_buckets)
    log_max = math.log(dmax);
    bucket_index = 2;
    ret_array = [0] * n_buckets
    current = dmin
    ret_array[1] = current
    for bucket_index in range(2, n_buckets):
        log_current = math.log(current)
        log_ratio = (log_max - log_current) / (n_buckets - bucket_index)
        log_next = log_current + log_ratio
        next_value = int(math.floor(math.exp(log_next) + 0.5))
        if next_value > current:
            current = next_value
        else:
            current = current + 1
        ret_array[bucket_index] = current
    return ret_array

always_allowed_keys = ['kind', 'description', 'cpp_guard']

class Histogram:
    """A class for representing a histogram definition."""

    def __init__(self, name, definition):
        """Initialize a histogram named name with the given definition.
definition is a dict-like object that must contain at least the keys:

 - 'kind': The kind of histogram.  Must be one of 'boolean', 'flag',
   'enumerated', 'linear', or 'exponential'.
 - 'description': A textual description of the histogram.

The key 'cpp_guard' is optional; if present, it denotes a preprocessor
symbol that should guard C/C++ definitions associated with the histogram."""
        self.verify_attributes(name, definition)
        self._name = name
        self._description = definition['description']
        self._kind = definition['kind']
        self._cpp_guard = definition.get('cpp_guard')
        self.compute_bucket_parameters(definition)
        table = { 'boolean': 'BOOLEAN',
                  'flag': 'FLAG',
                  'enumerated': 'LINEAR',
                  'linear': 'LINEAR',
                  'exponential': 'EXPONENTIAL' }
        table_dispatch(self.kind(), table,
                       lambda k: self._set_nsITelemetry_kind(k))

    def name(self):
        """Return the name of the histogram."""
        return self._name

    def description(self):
        """Return the description of the histogram."""
        return self._description

    def kind(self):
        """Return the kind of the histogram.
Will be one of 'boolean', 'flag', 'enumerated', 'linear', or 'exponential'."""
        return self._kind

    def nsITelemetry_kind(self):
        """Return the nsITelemetry constant corresponding to the kind of
the histogram."""
        return self._nsITelemetry_kind

    def _set_nsITelemetry_kind(self, kind):
        self._nsITelemetry_kind = "nsITelemetry::HISTOGRAM_%s" % kind

    def low(self):
        """Return the lower bound of the histogram.  May be a string."""
        return self._low

    def high(self):
        """Return the high bound of the histogram.  May be a string."""
        return self._high

    def n_buckets(self):
        """Return the number of buckets in the histogram.  May be a string."""
        return self._n_buckets

    def cpp_guard(self):
        """Return the preprocessor symbol that should guard C/C++ definitions
associated with the histogram.  Returns None if no guarding is necessary."""
        return self._cpp_guard

    def ranges(self):
        """Return an array of lower bounds for each bucket in the histogram."""
        table = { 'boolean': linear_buckets,
                  'flag': linear_buckets,
                  'enumerated': linear_buckets,
                  'linear': linear_buckets,
                  'exponential': exponential_buckets }
        return table_dispatch(self.kind(), table,
                              lambda p: p(self.low(), self.high(), self.n_buckets()))

    def compute_bucket_parameters(self, definition):
        table = {
            'boolean': Histogram.boolean_flag_bucket_parameters,
            'flag': Histogram.boolean_flag_bucket_parameters,
            'enumerated': Histogram.enumerated_bucket_parameters,
            'linear': Histogram.linear_bucket_parameters,
            'exponential': Histogram.exponential_bucket_parameters
            }
        table_dispatch(self.kind(), table,
                       lambda p: self.set_bucket_parameters(*p(definition)))

    def verify_attributes(self, name, definition):
        global always_allowed_keys
        general_keys = always_allowed_keys + ['low', 'high', 'n_buckets']

        table = {
            'boolean': always_allowed_keys,
            'flag': always_allowed_keys,
            'enumerated': always_allowed_keys + ['n_values'],
            'linear': general_keys,
            'exponential': general_keys
            }
        table_dispatch(definition['kind'], table,
                       lambda allowed_keys: Histogram.check_keys(name, definition, allowed_keys))

    @staticmethod
    def check_keys(name, definition, allowed_keys):
        for key in definition.iterkeys():
            if key not in allowed_keys:
                raise KeyError, '%s not permitted for %s' % (key, name)

    def set_bucket_parameters(self, low, high, n_buckets):
        def try_to_coerce_to_number(v):
            try:
                return eval(v, {})
            except:
                return v
        self._low = try_to_coerce_to_number(low)
        self._high = try_to_coerce_to_number(high)
        self._n_buckets = try_to_coerce_to_number(n_buckets)

    @staticmethod
    def boolean_flag_bucket_parameters(definition):
        return (1, 2, 3)

    @staticmethod
    def linear_bucket_parameters(definition):
        return (definition.get('low', 1),
                definition['high'],
                definition['n_buckets'])

    @staticmethod
    def enumerated_bucket_parameters(definition):
        n_values = definition['n_values']
        return (1, n_values, "%s+1" % n_values)

    @staticmethod
    def exponential_bucket_parameters(definition):
        return (definition.get('low', 1),
                definition['high'],
                definition['n_buckets'])

def from_file(filename):
    """Return an iterator that provides a sequence of Histograms for
the histograms defined in filename.
    """
    with open(filename, 'r') as f:
        histograms = json.load(f, object_pairs_hook=json.OrderedDict)
        for (name, definition) in histograms.iteritems():
            yield Histogram(name, definition)