mirror of
https://github.com/zeek/zeek.git
synced 2025-10-07 17:18:20 +00:00
Merge remote-tracking branch 'origin/topic/bernhard/metrics-samples'
Closes #1003. * origin/topic/bernhard/metrics-samples: finishing touches, make test more robust, rename function in last again change names of data structures after talking with seth make last plugin nicer and samplify sqli detector add tests for sampler reservoir sampler. untested.
This commit is contained in:
commit
f76446fb4e
10 changed files with 347 additions and 29 deletions
|
@ -1,4 +1,5 @@
|
|||
@load ./average
|
||||
@load ./last
|
||||
@load ./max
|
||||
@load ./min
|
||||
@load ./sample
|
||||
|
|
54
scripts/base/frameworks/sumstats/plugins/last.bro
Normal file
54
scripts/base/frameworks/sumstats/plugins/last.bro
Normal file
|
@ -0,0 +1,54 @@
|
|||
@load base/frameworks/sumstats
|
||||
@load base/utils/queue
|
||||
|
||||
module SumStats;
|
||||
|
||||
export {
|
||||
redef enum Calculation += {
|
||||
## Keep last X observations in a queue
|
||||
LAST
|
||||
};
|
||||
|
||||
redef record Reducer += {
|
||||
## number of elements to keep.
|
||||
num_last_elements: count &default=0;
|
||||
};
|
||||
|
||||
redef record ResultVal += {
|
||||
## This is the queue where elements are maintained. Use the
|
||||
## :bro:see:`SumStats::get_elements` function to get a vector of the current element values.
|
||||
last_elements: Queue::Queue &optional;
|
||||
};
|
||||
|
||||
## Get a vector of element values from a ResultVal.
|
||||
global get_last: function(rv: ResultVal): vector of Observation;
|
||||
}
|
||||
|
||||
function get_last(rv: ResultVal): vector of Observation
|
||||
{
|
||||
local s: vector of Observation = vector();
|
||||
if ( rv?$last_elements )
|
||||
Queue::get_vector(rv$last_elements, s);
|
||||
return s;
|
||||
}
|
||||
|
||||
hook observe_hook(r: Reducer, val: double, obs: Observation, rv: ResultVal)
|
||||
{
|
||||
if ( LAST in r$apply && r$num_last_elements > 0 )
|
||||
{
|
||||
if ( ! rv?$last_elements )
|
||||
rv$last_elements = Queue::init([$max_len=r$num_last_elements]);
|
||||
Queue::put(rv$last_elements, obs);
|
||||
}
|
||||
}
|
||||
|
||||
hook compose_resultvals_hook(result: ResultVal, rv1: ResultVal, rv2: ResultVal)
|
||||
{
|
||||
# Merge $samples
|
||||
if ( rv1?$last_elements && rv2?$last_elements )
|
||||
result$last_elements = Queue::merge(rv1$last_elements, rv2$last_elements);
|
||||
else if ( rv1?$last_elements )
|
||||
result$last_elements = rv1$last_elements;
|
||||
else if ( rv2?$last_elements )
|
||||
result$last_elements = rv2$last_elements;
|
||||
}
|
|
@ -1,49 +1,120 @@
|
|||
@load base/frameworks/sumstats/main
|
||||
@load base/utils/queue
|
||||
|
||||
module SumStats;
|
||||
|
||||
export {
|
||||
redef enum Calculation += {
|
||||
## Get uniquely distributed random samples from the observation stream.
|
||||
SAMPLE
|
||||
};
|
||||
|
||||
redef record Reducer += {
|
||||
## A number of sample Observations to collect.
|
||||
samples: count &default=0;
|
||||
num_samples: count &default=0;
|
||||
};
|
||||
|
||||
redef record ResultVal += {
|
||||
## This is the queue where samples are maintained. Use the
|
||||
## :bro:see:`SumStats::get_samples` function to get a vector of the samples.
|
||||
samples: Queue::Queue &optional;
|
||||
};
|
||||
## This is the vector in which the samples are maintained.
|
||||
samples: vector of Observation &default=vector();
|
||||
|
||||
## Get a vector of sample Observation values from a ResultVal.
|
||||
global get_samples: function(rv: ResultVal): vector of Observation;
|
||||
## Number of total observed elements.
|
||||
sample_elements: count &default=0;
|
||||
};
|
||||
}
|
||||
|
||||
function get_samples(rv: ResultVal): vector of Observation
|
||||
redef record ResultVal += {
|
||||
# Internal use only. This is not meant to be publically available
|
||||
# and just a copy of num_samples from the Reducer. Needed for availability
|
||||
# in the compose hook.
|
||||
num_samples: count &default=0;
|
||||
};
|
||||
|
||||
hook init_resultval_hook(r: Reducer, rv: ResultVal)
|
||||
{
|
||||
local s: vector of Observation = vector();
|
||||
if ( rv?$samples )
|
||||
Queue::get_vector(rv$samples, s);
|
||||
return s;
|
||||
if ( SAMPLE in r$apply )
|
||||
rv$num_samples = r$num_samples;
|
||||
}
|
||||
|
||||
function sample_add_sample(obs:Observation, rv: ResultVal)
|
||||
{
|
||||
++rv$sample_elements;
|
||||
|
||||
if ( |rv$samples| < rv$num_samples )
|
||||
rv$samples[|rv$samples|] = obs;
|
||||
else
|
||||
{
|
||||
local ra = rand(rv$sample_elements);
|
||||
if ( ra < rv$num_samples )
|
||||
rv$samples[ra] = obs;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
hook observe_hook(r: Reducer, val: double, obs: Observation, rv: ResultVal)
|
||||
{
|
||||
if ( r$samples > 0 )
|
||||
if ( SAMPLE in r$apply )
|
||||
{
|
||||
if ( ! rv?$samples )
|
||||
rv$samples = Queue::init([$max_len=r$samples]);
|
||||
Queue::put(rv$samples, obs);
|
||||
sample_add_sample(obs, rv);
|
||||
}
|
||||
}
|
||||
|
||||
hook compose_resultvals_hook(result: ResultVal, rv1: ResultVal, rv2: ResultVal)
|
||||
{
|
||||
# Merge $samples
|
||||
if ( rv1?$samples && rv2?$samples )
|
||||
result$samples = Queue::merge(rv1$samples, rv2$samples);
|
||||
else if ( rv1?$samples )
|
||||
result$samples = rv1$samples;
|
||||
else if ( rv2?$samples )
|
||||
result$samples = rv2$samples;
|
||||
if ( rv1$num_samples != rv2$num_samples )
|
||||
{
|
||||
Reporter::error("Merging sample sets with differing sizes is not supported");
|
||||
return;
|
||||
}
|
||||
|
||||
local num_samples = rv1$num_samples;
|
||||
result$num_samples = num_samples;
|
||||
|
||||
if ( |rv1$samples| > num_samples || |rv2$samples| > num_samples )
|
||||
{
|
||||
Reporter::error("Sample vector with too many elements. Aborting.");
|
||||
return;
|
||||
}
|
||||
|
||||
|
||||
if ( |rv1$samples| != num_samples && |rv2$samples| < num_samples )
|
||||
{
|
||||
if ( |rv1$samples| != rv1$sample_elements || |rv2$samples| < rv2$sample_elements )
|
||||
{
|
||||
Reporter::error("Mismatch in sample element size and tracking. Aborting merge");
|
||||
return;
|
||||
}
|
||||
|
||||
for ( i in rv1$samples )
|
||||
sample_add_sample(rv1$samples[i], result);
|
||||
|
||||
for ( i in rv2$samples)
|
||||
sample_add_sample(rv2$samples[i], result);
|
||||
}
|
||||
else
|
||||
{
|
||||
local other_vector: vector of Observation;
|
||||
local othercount: count;
|
||||
|
||||
if ( rv1$sample_elements > rv2$sample_elements )
|
||||
{
|
||||
result$samples = copy(rv1$samples);
|
||||
other_vector = rv2$samples;
|
||||
othercount = rv2$sample_elements;
|
||||
}
|
||||
else
|
||||
{
|
||||
result$samples = copy(rv2$samples);
|
||||
other_vector = rv1$samples;
|
||||
othercount = rv1$sample_elements;
|
||||
}
|
||||
|
||||
local totalcount = rv1$sample_elements + rv2$sample_elements;
|
||||
result$sample_elements = totalcount;
|
||||
|
||||
for ( i in other_vector )
|
||||
{
|
||||
if ( rand(totalcount) <= othercount )
|
||||
result$samples[i] = other_vector[i];
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue