Migrate ReasoningIT to BDD (#81)

## What is the goal of this PR?

To identify and de-duplicate integration tests that test the fundamentals of Reasoner, categorise them appropriately and migrate them to BDD scenarios.

This PR focuses on ReasoningIT in grakn.

This PR forms part of the work towards typedb/typedb#5721

## What are the changes implemented in this PR?

Added tests that were migrated from the relevant IT files in grakn. Also moved some tests around, so that everything is in the right place.

behaviour/graql/language/BUILD

@@ -24,5 +24,6 @@ exports_files([
     "get.feature",
     "insert.feature",
     "match.feature",
+    "rule-validation.feature",
     "undefine.feature",
 ])

behaviour/graql/language/define.feature

@@ -1253,6 +1253,8 @@ Feature: Graql Define Query
   # RULES #
   #########
 
+  # Note: These tests verify only the ability to create rules, and are not concerned with their application.
+
   Scenario: a rule can infer both an attribute and its ownership
     Given graql define
       """
@@ -1281,7 +1283,7 @@ Feature: Graql Define Query
       | RUL |
 
 
-  Scenario: a rule can infer a relation
+  Scenario: when defining a rule using `sub rule`, the rule is successfully created
     Given graql define
       """
       define

behaviour/graql/language/insert.feature

@@ -1427,6 +1427,43 @@ Feature: Graql Insert Query
       | MIC | TAR |
 
 
+  Scenario: match-insert can take an attribute's value and copy it to an attribute of a different type
+    Given graql define
+      """
+      define
+      height sub attribute, value long;
+      person has height;
+      """
+    Given graql insert
+      """
+      insert
+      $x isa person, has name "Susie", has age 16, has ref 0;
+      $y isa person, has name "Donald", has age 25, has ref 1;
+      $z isa person, has name "Ralph", has age 18, has ref 2;
+      """
+    Given the integrity is validated
+    Given graql insert
+      """
+      match
+        $x isa person, has age 16;
+      insert
+        $x has height 16;
+      """
+    Given the integrity is validated
+    When get answers of graql query
+      """
+      match
+        $x has height $z;
+      get $x;
+      """
+    And concept identifiers are
+      |     | check | value |
+      | SUS | key   | ref:0 |
+    Then uniquely identify answer concepts
+      | x   |
+      | SUS |
+
+
   Scenario: if match-insert matches nothing, then nothing is inserted
     Given graql define
       """

behaviour/graql/language/match.feature

@@ -746,6 +746,75 @@ Feature: Graql Match Clause
       | REF0 | REF1 |
 
 
+  Scenario: relations between distinct concepts are not retrieved when matching concepts that relate to themselves
+    Given graql insert
+      """
+      insert
+      $x isa person, has ref 1;
+      $y isa person, has ref 2;
+      (friend: $x, friend: $y) isa friendship, has ref 0;
+      """
+    Given the integrity is validated
+    When get answers of graql query
+      """
+      match (friend: $x, friend: $x) isa friendship; get;
+      """
+    Then answer size is: 0
+
+
+  Scenario: matching a chain of relations only returns answers if there is a chain of the required length
+    Given graql define
+      """
+      define
+
+      gift-delivery sub relation,
+        relates sender,
+        relates recipient;
+
+      person plays sender,
+        plays recipient;
+      """
+    Given the integrity is validated
+    Given graql insert
+      """
+      insert
+      $x1 isa person, has name "Soroush", has ref 0;
+      $x2a isa person, has name "Martha", has ref 1;
+      $x2b isa person, has name "Patricia", has ref 2;
+      $x2c isa person, has name "Lily", has ref 3;
+
+      (sender: $x1, recipient: $x2a) isa gift-delivery;
+      (sender: $x1, recipient: $x2b) isa gift-delivery;
+      (sender: $x1, recipient: $x2c) isa gift-delivery;
+      (sender: $x2a, recipient: $x2b) isa gift-delivery;
+      """
+    Given the integrity is validated
+    When get answers of graql query
+      """
+      match
+        (sender: $a, recipient: $b) isa gift-delivery;
+        (sender: $b, recipient: $c) isa gift-delivery;
+      get;
+      """
+    When concept identifiers are
+      |     | check | value |
+      | SOR | key   | ref:0 |
+      | MAR | key   | ref:1 |
+      | PAT | key   | ref:2 |
+    Then uniquely identify answer concepts
+      | a   | b   | c   |
+      | SOR | MAR | PAT |
+    When get answers of graql query
+      """
+      match
+        (sender: $a, recipient: $b) isa gift-delivery;
+        (sender: $b, recipient: $c) isa gift-delivery;
+        (sender: $c, recipient: $d) isa gift-delivery;
+      get;
+      """
+    Then answer size is: 0
+
+
   Scenario: an error is thrown when matching an entity type as if it were a role
     Then graql get throws
       """
@@ -1308,14 +1377,43 @@ Feature: Graql Match Clause
       | PER |
 
 
-  Scenario: value comparison of unbound variables throws an error
+  @ignore
+  # TODO: re-enable when variables used in multiple value predicates are resolvable (grakn#5845)
+  Scenario: an attribute variable used in both `==` and `>=` predicates is correctly resolved
+    Given graql insert
+      """
+      insert
+      $x isa person, has name "Susie", has age 16, has ref 0;
+      $y isa person, has name "Donald", has age 25, has ref 1;
+      $z isa person, has name "Ralph", has age 18, has ref 2;
+      """
+    Given the integrity is validated
+    When get answers of graql query
+      """
+      match
+        $x has age == $z;
+        $z >= 17;
+        $z isa age;
+      get $x;
+      """
+    And concept identifiers are
+      |     | check | value |
+      | DON | key   | ref:1 |
+      | RAL | key   | ref:2 |
+    Then uniquely identify answer concepts
+      | x   |
+      | DON |
+      | RAL |
+
+
+  Scenario: concept comparison of unbound variables throws an error
     Then graql get throws
       """
       match $x != $y; get;
       """
 
 
-  Scenario: concept comparison of unbound variables throws an error
+  Scenario: value comparison of unbound variables throws an error
     Then graql get throws
       """
       match $x !== $y; get;

behaviour/graql/reasoner/rule-validation.feature → behaviour/graql/language/rule-validation.feature

@@ -15,7 +15,7 @@
 # along with this program.  If not, see <https://www.gnu.org/licenses/>.
 #
 
-Feature: Graql Reasoner Rule Validation
+Feature: Graql Rule Validation
 
   Background: Initialise a session and transaction for each scenario
     Given connection has been opened

behaviour/graql/reasoner/BUILD

@@ -18,6 +18,14 @@
 package(default_visibility = ["//visibility:public"])
 
 exports_files([
-    "resolution.feature",
-    "rule-validation.feature",
+    "attribute-attachment.feature",
+    "concept-inequality.feature",
+    "negation.feature",
+    "relation-inference.feature",
+    "roleplayer-attachment.feature",
+    "resolution-test-framework.feature",
+    "type-generation.feature",
+    "type-hierarchy.feature",
+    "value-predicate.feature",
+    "variable-roles.feature",
 ])

behaviour/graql/reasoner/resolution/concept-inequality.feature → behaviour/graql/reasoner/concept-inequality.feature

@@ -31,6 +31,9 @@ Feature: Concept Inequality Resolution
       """
       define
 
+      person sub entity,
+          has name;
+
       ball sub entity,
           has name,
           plays ball1,
@@ -395,3 +398,128 @@ Feature: Concept Inequality Resolution
 #    Then all answers are correct in reasoned keyspace
     Then answer size in reasoned keyspace is: 36
 #    Then materialised and reasoned keyspaces are the same size
+
+
+  @ignore
+  # TODO: re-enable once grakn#5821 is fixed (in some answers, $typeof_ax is 'base-attribute' which is incorrect)
+  # TODO: re-enable all steps once implicit attribute variables are resolvable
+  # TODO: migrate to concept-inequality.feature
+  Scenario: when restricting concept types of a pair of inferred attributes with `!=`, the answers have distinct types
+    Given for each session, graql define
+      """
+      define
+      soft-drink sub entity,
+        has name,
+        has retailer;
+      base-attribute sub attribute, value string, abstract;
+      string-attribute sub base-attribute;
+      name sub base-attribute;
+      retailer sub base-attribute;
+      person has string-attribute;
+
+      tesco-sells-all-soft-drinks sub rule,
+      when {
+        $x isa soft-drink;
+      },
+      then {
+        $x has retailer 'Tesco';
+      };
+      """
+    Given for each session, graql insert
+      """
+      insert
+      $x isa person, has string-attribute "Tesco";
+      $y isa soft-drink, has name "Tesco";
+      """
+    When materialised keyspace is completed
+    Then for graql query
+      """
+      match
+        $x has base-attribute $ax;
+        $y has base-attribute $ay;
+        $ax isa! $typeof_ax;
+        $ay isa! $typeof_ay;
+        $typeof_ax != $typeof_ay;
+      get;
+      """
+    Then all answers are correct in reasoned keyspace
+    # x   | ax  | y   | ay  |
+    # PER | STA | SOF | NAM |
+    # PER | STA | SOF | RET |
+    # SOF | NAM | PER | STA |
+    # SOF | RET | PER | STA |
+    # SOF | NAM | SOF | STA |
+    # SOF | STA | SOF | NAM |
+    Then answer size in reasoned keyspace is: 6
+    Then materialised and reasoned keyspaces are the same size
+
+  @ignore
+  # TODO: re-enable once grakn#5821 is fixed
+  # TODO: re-enable all steps once implicit attribute variables are resolvable
+  # TODO: migrate to concept-inequality.feature
+  Scenario: inferred attribute matches can be simultaneously restricted by both concept type and attribute value
+    Given for each session, graql define
+      """
+      define
+      soft-drink sub entity,
+        has name,
+        has retailer;
+      base-attribute sub attribute, value string, abstract;
+      string-attribute sub base-attribute;
+      retailer sub base-attribute;
+      person has string-attribute;
+
+      transfer-string-attribute-to-other-people sub rule,
+      when {
+        $x isa person, has string-attribute $r1;
+        $y isa person;
+      },
+      then {
+        $y has string-attribute $r1;
+      };
+
+      tesco-sells-all-soft-drinks sub rule,
+      when {
+        $x isa soft-drink;
+      },
+      then {
+        $x has retailer 'Tesco';
+      };
+
+      if-ocado-exists-it-sells-all-soft-drinks sub rule,
+      when {
+        $x isa retailer;
+        $x == 'Ocado';
+        $y isa soft-drink;
+      },
+      then {
+        $y has retailer $x;
+      };
+      """
+    Given for each session, graql insert
+      """
+      insert
+      $w isa person, has string-attribute "Ocado";
+      $x isa person, has string-attribute "Tesco";
+      $y isa soft-drink, has name "Sprite";
+      $z "Ocado" isa retailer;
+      """
+    When materialised keyspace is completed
+    Then for graql query
+      """
+      match
+        $x has base-attribute $value;
+        $y has base-attribute $unwantedValue;
+        $value !== $unwantedValue;
+        $unwantedValue "Ocado";
+        $value isa! $type;
+        $unwantedValue isa! $type;
+        $type != $unwantedType;
+        $unwantedType type string-attribute;
+      get $x, $value, $type;
+      """
+    Then all answers are correct in reasoned keyspace
+    # x      | value | type     |
+    # Sprite | Tesco | retailer |
+    Then answer size in reasoned keyspace is: 1
+#    Then materialised and reasoned keyspaces are the same size