Class: Dbd::Fact

Inherits:
Object
  • Object
show all
Defined in:
lib/dbd/fact.rb,
lib/dbd/fact/id.rb,
lib/dbd/fact/subject.rb,
lib/dbd/fact/collection.rb

Overview

Basic Fact of knowledge.

The database is built as an ordered sequence of facts, a "fact stream".

This is somewhat similar to a "triple" in the RDF (Resource Description Framework) concept, but with different and extended functionality.

Each basic fact has:

  • a unique and invariant id (a uuid)

    To allow referencing back to it (e.g. to invalidate it later in a fact stream).

  • a time_stamp (time with nanosecond granularity)

    To allow verifying that the order in a fact stream is correct.

    A time_stamp does not need to represent the exact time of the creation of the fact, but it has to increase in strictly monotic order in a fact stream.

  • a provenance_subject (a uuid)

    The subject of the ProvenanceResource (a set of ProvenanceFacts with the same subject) about this fact. Each Fact, points back to a ProvenanceResource (the ProvenanceResource must have been fully defined, earlier in a fact stream).

  • a subject (a uuid)

    "About which Resource is this fact?".

    Similar to the subject of an RDF triple, except that this subject is not a URI, but an abstract uuid (that is world-wide unique and invariant).

    Links to "real-world" URI's and URL's can be added later as separate facts (this also allows linking multiple "real-world" URI's to a single Resource).

  • a predicate (a string)

    "Which property of the resource are we describing?".

    Currently this is a string, but I suggest modeling this similar to predicate in RDF. Probably more detailed modeling using RDF predicate will follow.

  • an object (a string)

    "What is the value of the property of the resource we are describing?".

    Currently this is a string, but I suggest modeling this similar to object in RDF. Probably more detailed modeling using RDF object will follow.

Direct Known Subclasses

ProvenanceFact

Defined Under Namespace

Modules: Collection, ID, Subject

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(options) ⇒ Fact

Builds a new Fact.

Parameters:

  • options (Hash{Symbol => Object})

Options Hash (options):

  • :predicate (#to_s) —

    Required : the predicate for this Fact

  • :object (#to_s) —

    Required : the object for this Fact (required)

  • :provenance_subject (String (uuid)) — default: nil —

    Optional: the subject of the provenance(resource|fact)

  • :subject (String (uuid)) — default: nil —

    Optional: the subject for this Fact

  • :time_stamp (TimeStamp) — default: nil —

    Optional: the time_stamp for this Fact

  • :id (String (uuid)) —

    Optional : set the id

Raises:



130
131
132
133
134
135
136
137
138
139
140
 
# File 'lib/dbd/fact.rb', line 130

def initialize(options)
  @id = options[:id] || self.class.new_id
  @time_stamp = options[:time_stamp]
  validate_time_stamp_class(@time_stamp)
  @provenance_subject = options[:provenance_subject]
  @subject = options[:subject]
  @predicate = options[:predicate]
  @object = options[:object]
  raise PredicateError, "predicate cannot be nil" if predicate.nil?
  raise ObjectError, "object cannot be nil" if object.nil?
end

Class Method Details

.attributes ⇒ Array

Returns The 6 attributes of a Fact.

Returns:

  • (Array) —

    The 6 attributes of a Fact.



62
63
64
65
66
67
68
69
 
# File 'lib/dbd/fact.rb', line 62

def self.attributes
  [:id,
   :time_stamp,
   :provenance_subject,
   :subject,
   :predicate,
   :object]
end

.fact_from_hash(hash) ⇒ Object 



262
263
264
265
266
267
268
 
# File 'lib/dbd/fact.rb', line 262

def self.fact_from_hash(hash)
  if hash[:provenance_subject]
    Fact.new(hash)
  else
    ProvenanceFact.new(hash)
  end
end

.from_string_values(string_values) ⇒ Fact, ProvenanceFact

Constructs a Fact or ProvenanceFact from a string values array (e.g. pulled from a CSV row).

Parameters:

  • values (Array) —

    Required : the array with values, organized as in attributes

Returns:



161
162
163
164
 
# File 'lib/dbd/fact.rb', line 161

def self.from_string_values(string_values)
  string_hash = hash_from_values(string_values)
  fact_from_hash(values_hash(string_hash))
end

.hash_from_values(values) ⇒ Object

FIXME This has to move to a Fact::Factory



250
251
252
253
254
 
# File 'lib/dbd/fact.rb', line 250

def self.hash_from_values(values)
  # Do not keep "empty" values (e.g. the provenance_subject for a ProvenanceFact).
  attributes_values_array = [attributes, values].transpose.delete_if{|a,v| v.nil? || v == ''}
  Hash[attributes_values_array]
end

.new_id ⇒ String

Returns A new id string.

Returns:

  • (String) —

    A new id string.



116
117
118
 
# File 'lib/dbd/fact.rb', line 116

def self.new_id
  ID.new_id
end

.new_subject ⇒ String

Returns A new subject string.

Returns:

  • (String) —

    A new subject string.



110
111
112
 
# File 'lib/dbd/fact.rb', line 110

def self.new_subject
  Subject.new_subject
end

.values_hash(string_hash) ⇒ Object 



256
257
258
259
260
 
# File 'lib/dbd/fact.rb', line 256

def self.values_hash(string_hash)
  string_hash.dup.tap do |h|
    h[:time_stamp] = TimeStamp.new(time: h[:time_stamp])
  end
end

Instance Method Details

#equivalent?(other) ⇒ Boolean

Equivalent facts (have all same values, except time_stamp).

For "equality" only a test on the id is used. If the id (which is a uuid) is the same, we assume that is the "same" fact. This equivalent? method is used to test is equal methods are "really" equivalent.

The time_stamp may be slightly different (because shifts of a few nanoseconds will be required to resolve collisions on merge).

Returns:

  • (Boolean) 


177
178
179
180
181
 
# File 'lib/dbd/fact.rb', line 177

def equivalent?(other)
  (self.class.attributes - [:time_stamp]).
    all?{ |attribute| self.send(attribute) == other.send(attribute) } &&
    self.time_stamp.near?(other.time_stamp)
end

#errors ⇒ Array

Checks if a fact has errors for storing in the graph.

Returns:

  • (Array) —

    an Array of error messages



210
211
212
213
214
215
216
217
218
219
 
# File 'lib/dbd/fact.rb', line 210

def errors
  # * id not validated, is set automatically upon creation
  # * time_stamp not validated, is set automatically later
  # * predicate not validated, is validated in initialize
  # * object not validated, is validated in initialize
  [].tap do |a|
    a << provenance_subject_error(provenance_subject)
    a << "Subject is missing" unless subject
  end.compact
end

#provenance_fact? ⇒ Boolean

Confirms this is not a ProvenanceFact.

Needed for validations that depend on different behavior for a provenance_fact (mainly, no provenance_subject).

Returns:

  • (Boolean) 


239
240
241
 
# File 'lib/dbd/fact.rb', line 239

def provenance_fact?
  false
end

#provenance_subject=(provenance_subject) ⇒ Object

A set_once setter for provenance_subject.

This implements a "form" of immutable behavior. The value can be set once (possibly after creation the object), but can never be changed after that.



94
95
96
 
# File 'lib/dbd/fact.rb', line 94

def provenance_subject=(provenance_subject)
  set_once(:provenance_subject, provenance_subject)
end

#provenance_subject_error(provenance_subject) ⇒ Object

Validates the presence or absence of provenance_subject.

Here, in (base) Fact, provenance_subject must be present In the derived ProvenanceFact it must not be present. This is how the difference is encoded between Fact and ProvenanceFact in the fact stream. Return [nil, String] nil or an error message

Parameters:

  • provenance_subject (#nil?) 


230
231
232
 
# File 'lib/dbd/fact.rb', line 230

def provenance_subject_error(provenance_subject)
  "Provenance subject is missing" unless provenance_subject
end

#short ⇒ String

Returns a short string representation of a Fact.

Returns:

  • (String) —

    a short string representation of a Fact



185
186
187
188
189
190
 
# File 'lib/dbd/fact.rb', line 185

def short
  "#{provenance_subject_short} : " \
  "#{subject.to_s[0...8]} : " \
  "#{predicate.to_s.ljust(24, ' ')[0...24]} : " \
  "#{object.to_s[0..60].gsub(/\n/, '_')}"
end

#string_values ⇒ Array

This is similar to the 6 entries in the to_CSV mapping

Returns:

  • (Array) —

    The 6 values of a Fact converted to a string.



151
152
153
 
# File 'lib/dbd/fact.rb', line 151

def string_values
  values.map(&:to_s)
end

#subject=(subject) ⇒ Object

A set_once setter for subject.

This implements a "form" of immutable behavior. The value can be set once (possibly after creation the object), but can never be changed after that.



104
105
106
 
# File 'lib/dbd/fact.rb', line 104

def subject=(subject)
  set_once(:subject, subject)
end

#time_stamp=(time_stamp) ⇒ Object

A set_once setter for time_stamp.

This implements a "form" of immutable behavior. The value can be set once (possibly after creation the object), but can never be changed after that.

The input class is validated (easy confusion with String or Time).



83
84
85
86
 
# File 'lib/dbd/fact.rb', line 83

def time_stamp=(time_stamp)
  validate_time_stamp_class(time_stamp)
  set_once(:time_stamp, time_stamp)
end

#update_used_provenance_subjects(h) ⇒ Object

Executes the required update in used_provenance_subjects.

For a Fact, pointing to a ProvenanceResource in it's provenance_subject, marks this provenance_subject in the "used_provenance_subjects" hash that is passed in as an argument (DCI). This will avoid further changes to the ProvenanceResource with this provenance_subject.

This is overridden in the ProvenanceFact, since only relevant for a Fact.



201
202
203
204
 
# File 'lib/dbd/fact.rb', line 201

def update_used_provenance_subjects(h)
  # using a provenance_subject sets the key
  h[provenance_subject] = true
end

#values ⇒ Array

Returns The 6 values of a Fact.

Returns:

  • (Array) —

    The 6 values of a Fact.



144
145
146
 
# File 'lib/dbd/fact.rb', line 144

def values
  self.class.attributes.map{ |attribute| self.send(attribute) }
end