Class: Hecks::Runtime::QueryInterpreter

Inherits:
Object
  • Object
show all
Defined in:
lib/hecks/runtime/query_interpreter.rb

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(registry) ⇒ QueryInterpreter

Returns a new instance of QueryInterpreter.



17
18
19
 
# File 'lib/hecks/runtime/query_interpreter.rb', line 17

def initialize(registry)
  @registry = registry
end

Instance Attribute Details

#registry ⇒ Object (readonly)

Returns the value of attribute registry.



15
16
17
 
# File 'lib/hecks/runtime/query_interpreter.rb', line 15

def registry
  @registry
end

Instance Method Details

#call(domain, aggregate, query_name, args) ⇒ Object 



21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
 
# File 'lib/hecks/runtime/query_interpreter.rb', line 21

def call(domain, aggregate, query_name, args)
  return entity_rows(domain, aggregate, query_name, args) if query_name.include?(".")

  declared = declared_query(aggregate, query_name)
  args = normalize_args(aggregate, declared, args)
  declared = TenantScope.apply(declared, args)
  # AFTER TenantScope, so its synthetic clause is already present
  # in `.wheres` and rides through as an ordinary LOCAL clause on
  # the OUTER query. It does not reach the hop's own inner
  # sub-query against the TARGET aggregate — a hop's target may
  # not even declare the same tenant boundary, and propagating one
  # aggregate's tenant scope onto an unrelated aggregate's own
  # query is a real design question of its own, not answered here.
  declared = ReferenceHop.apply(declared, args, registry: @registry, domain: domain, aggregate: aggregate)

  repository = @registry.repository(domain, aggregate)
  # `registry:` is threaded through so a `none_in_state` where-clause
  # can look its target aggregate up — see
  # QuerySpecification::Common::Comparison#none_in_state? for the
  # comparator itself, and for why an ordinary aggregate-level
  # Memory query needs the registry passed explicitly rather than
  # closing over an instance variable.
  if (native = Ports::Query.execute(repository, declared, args,
                                    context: { domain: domain, aggregate: aggregate, registry: @registry }))
    records = native
    # `record.state.merge(id: record.id)` — id LAST, not first. See
    # Instance#to_h's own comment: an aggregate free to declare its
    # own attribute literally named `id` has that attribute's own
    # wrapped value sitting in `record.state[:id]` already; merging
    # it OVER a `{id:}.merge(state)` used to let it silently
    # clobber the correct bare identity this row is supposed to
    # carry.
    # A QUERY ROW IS AN ANSWER, NOT A HANDLE. Mutating one edits
    # nobody's state and silently disagrees with the store.
    return Freezer.deep(records.map { |record| record.state.merge(id: record.id) })
  end

  Freezer.deep(interpret(repository.all, declared, args, domain: domain))
end

#reference_call(domain, aggregate, query_name, args) ⇒ Object

The REFERENCE answer — this interpreter's own evaluation, never an adapter's native hook. The fuzzer's query oracle replays every generated ask through both paths and treats a difference as a finding: the differential gate the retired cross-runtime harness should always have been, aimed where the divergence actually lives — between the engines inside this one runtime.

A hop clause is answered here by its OWN, deliberately naive walk (reference_where_holds?) — never Runtime::ReferenceHop's partition/fold/IN-clause. Sharing that algorithm would have made this oracle blind to exactly the code the hop feature adds: every PHASE of a shared fold would still get diffed against the native adapters, but the fold itself — the empty candidate set, a duplicate id, a dangling reference, a chain's inside-out resolution order — would only ever be compared against itself.



76
77
78
79
80
81
82
83
84
 
# File 'lib/hecks/runtime/query_interpreter.rb', line 76

def reference_call(domain, aggregate, query_name, args)
  return entity_rows(domain, aggregate, query_name, args) if query_name.include?(".")

  declared = declared_query(aggregate, query_name)
  args = normalize_args(aggregate, declared, args)
  declared = TenantScope.apply(declared, args)
  reference_interpret(@registry.repository(domain, aggregate).all, declared, args,
                      domain: domain, shape: aggregate)
end