Module: Datadog::Tracing::Contrib::Elasticsearch::Quantize

Defined in:
lib/datadog/tracing/contrib/elasticsearch/quantize.rb

Overview

Quantize contains ES-specific resource quantization tools.

Constant Summary collapse

PLACEHOLDER = 
'?'
ID_PLACEHOLDER = 
'\1?'
EXCLUDE_KEYS = 
[].freeze
SHOW_KEYS = 
[:_index, :_type, :_id].freeze
DEFAULT_OPTIONS = 
{
  exclude: EXCLUDE_KEYS,
  show: SHOW_KEYS,
  placeholder: PLACEHOLDER
}.freeze

Class Method Summary collapse

Class Method Details

.format_body(body, options = {}) ⇒ Object 



28
29
30
31
32
 
# File 'lib/datadog/tracing/contrib/elasticsearch/quantize.rb', line 28

def format_body(body, options = {})
  format_body!(body, options)
rescue StandardError
  options[:placeholder] || PLACEHOLDER
end

.format_body!(body, options = {}) ⇒ Object 



34
35
36
37
38
39
40
41
42
43
44
45
46
 
# File 'lib/datadog/tracing/contrib/elasticsearch/quantize.rb', line 34

def format_body!(body, options = {})
  options = merge_options(DEFAULT_OPTIONS, options)

  # Determine if bulk query or not, based on content
  statements = body.end_with?("\n") ? body.split("\n") : [body]

  # Parse each statement and quantize them.
  statements.collect do |string|
    reserialize_json(string, options[:placeholder]) do |obj|
      Contrib::Utils::Quantization::Hash.format(obj, options)
    end
  end.join("\n")
end

.format_url(url) ⇒ Object 



23
24
25
26
 
# File 'lib/datadog/tracing/contrib/elasticsearch/quantize.rb', line 23

def format_url(url)
  sanitize_fragment_with_id(url)
    .gsub(/(?:\d+)/, PLACEHOLDER)
end

.merge_options(original, additional) ⇒ Object 



48
49
50
51
52
53
54
55
56
57
58
59
60
61
 
# File 'lib/datadog/tracing/contrib/elasticsearch/quantize.rb', line 48

def merge_options(original, additional)
  {}.tap do |options|
    # Show
    # If either is :all, value becomes :all
    options[:show] = if original[:show] == :all || additional[:show] == :all
                       :all
                     else
                       (original[:show] || []).dup.concat(additional[:show] || []).uniq
                     end

    # Exclude
    options[:exclude] = (original[:exclude] || []).dup.concat(additional[:exclude] || []).uniq
  end
end

.reserialize_json(string, fail_value = PLACEHOLDER) ⇒ Object

Parses a JSON object from a string, passes its value to the block provided, and dumps its result back to JSON. If JSON parsing fails, it prints fail_value.



66
67
68
69
70
71
72
73
74
75
 
# File 'lib/datadog/tracing/contrib/elasticsearch/quantize.rb', line 66

def reserialize_json(string, fail_value = PLACEHOLDER)
  return string unless block_given?

  begin
    JSON.dump(yield(JSON.parse(string)))
  rescue JSON::ParserError
    # If it can't parse/dump, don't raise an error.
    fail_value
  end
end

.sanitize_fragment_with_id(url) ⇒ Object

Sanitizes URL fragment by changing it to ? whenever a number is detected This is meant as simple heuristic that attempts to detect if particular fragment represents document Id. This is meant to reduce the cardinality in most frequent cases.



80
81
82
 
# File 'lib/datadog/tracing/contrib/elasticsearch/quantize.rb', line 80

def sanitize_fragment_with_id(url)
  url.gsub(%r{^(/?[^/]*/[^/]*/)(?:[^?/\d]*\d+[^?/]*)}, ID_PLACEHOLDER)
end