:Steps A-G

aggregate

• Description: An aggregate() sideEffect step aggregates input objects into a list<object> during a traversal. The list<object> is stored to a sideEffect.

• Note:

  • This step does not change the input objects. These objects are the input of the next step.

  • The list<object> is read-only. You can use a cap() step to extract the objects from the list<object> later at any point of the traversal.

    • We recommend that you do not use a select() step to extract the objects. This is because a select() step returns a list<object> for each element. Multiple identical lists are returned if multiple elements exist.

  • This step can be modulated by a by() step that specifies the field based on which objects are aggregated.

  • An implicit barrier() step is automatically inserted. You can use a store() step to aggregate objects in lazy mode without a barrier.

  • Differences among aggregate(), as(), and fold():

    • An aggregate() step aggregates input objects into a list and copies the list to a sideEffect. These objects are the input of the next step. The list is read-only in subsequent steps.

    • An as() step adds labels to input objects. These labeled objects are the input of the next step. You can modify the labeled objects in subsequent steps.

    • A fold() step aggregates input objects into a list. The list is the input of the next step.

g("thinkerpop").V("1;2;3;4;5;6").hasLabel("person").aggregate("x").by("name")
==> {"label":"person","age":29,"name":"marko","pk":"1"}
==> {"label":"person","age":32,"name":"josh","pk":"4"}
==> {"label":"person","age":27,"name":"vadas","pk":"2"}
==> {"label":"person","age":35,"name":"peter","pk":"6"}

g("thinkerpop").V("1;2;3;4;5;6").hasLabel("person").aggregate("x").by("name").cap("x")
==> ["josh","marko","peter","vadas"]

g("thinkerpop").V("1;2;3;4;5;6").hasLabel("person").aggregate("x").by("name").filter("age>=32").cap("x")
==> ["josh","marko","peter","vadas"]

g("thinkerpop").V("1;2;3;4;5;6").hasLabel("person").aggregate("x").by("name").outE("knows").aggregate("y").cap("x","y").dedup()
==> {"\"x\"":["josh","marko","peter","vadas"],"\"y\"":[{"label":"knows","pk":"1","sk":"2","weight":0.5},{"label":"knows","pk":"1","sk":"4","weight":1.0}]}

alias

• Description: An alias() sideEffect step defines virtual fields for input elements, including vertices and edges.

• Syntax: alias("expr1:new1;expr2;new2;field1:new3")

• Note:

  • You can define an expression as a virtual field.

  • You can define a virtual field for an original field. This way, you define an alias for the original field.

    • After an alias is defined for a field, you must use the alias to access the field in subsequent steps.

  • Field names must be unique.

  • This step is an extended step of iGraph.

and

• Description: An and() filter step returns the input objects for which all child traversals have results.

• Note: This step can take an arbitrary number of traversals as the input.

as

• Description: An as() modulator step adds labels to input objects. These objects can be accessed by subsequent steps and data structures that make use of such labels.

• Note:

  • An as() step can add multiple labels to objects at a time.

  • The labeled objects are the input of the next step. You can use a select() step to extract the labels later at any point of the traversal.

    • If you use a select() step to extract multiple labels, a map<label,objects> is returned.

    • A select() step can be modulated by a by() step that specifies the field to be obtained from the labeled objects.

  • Differences among aggregate(), as(), and fold():

    • An aggregate() step aggregates input objects into a list and copies the list to a sideEffect. These objects are the input of the next step. The list is read-only in subsequent steps.

    • An as() step adds labels to input objects. These labeled objects are the input of the next step. You can modify the labeled objects in subsequent steps.

    • A fold() step aggregates input objects into a list. The list is the input of the next step.

  • The "label":"person" field in the following result set indicates the label of an element in the graph.

    This label is different from the label added by the as() step.

barrier

• Description: A barrier() map step serves as a barrier. This means that every step prior to the barrier() step needs to be complete before the system moves to the steps after the barrier() step.

• Note:

  • By default, a barrier() step merges repeated objects.

    • Method: The bulk values of the objects are added together. The sack values of the objects are merged based on the specific merge operator.

    • Purpose: For example, the input objects contain ten Vertices A with a bulk value of 1.

      • If you run the barrier().outE() steps, the barrier() step returns a single Vertex A with a bulk value of 10, and the outE() step returns a result with a bulk value of 10 by performing one computation.

      • If you directly run an outE() step, the step returns ten results with a bulk value of 1 by performing ten computations.

    • Generally, the merge operation is imperceptible to users. This is because an object whose bulk value is greater than 1 is unfolded by multiple times based on the bulk value when the object is returned in the final result. This process does not change the bulk value.

    • You can use a bulk() step to obtain the bulk value of an element.

    • You can use a local(count()) step to count the number of each unique object.

  • A barrier(Barrier.nodedup) step serves as a barrier but does not merge repeated objects.

  • An implicit barrier() step is automatically inserted for steps such as aggregate(), cap(), count(), dedup(), distinct(), fold(), group(), groupCount(), limit(), max(), mean(), min(), order(), range(), sample(), sum(), and tail().

branch

• Description: A branch() branch step performs branching queries.

• Note:

  • This step must be used with option() steps.

  • If the condition of the first option is met, the process goes to the first branch. Otherwise, the second option is checked.

  • This step can take an arbitrary number of traversals as the input.

  • The use of this step is similar to that of a choose() step.

bulk

• Description: A bulk() map step returns the bulk value of an element.

• Note:

  • The initial bulk value of an element is 1. The bulk values of objects are added when the objects are merged by a barrier() step.

  • Generally, the merge operation is imperceptible to users. This is because an object whose bulk value is greater than 1 is unfolded by multiple times based on the bulk value when the object is returned in the final result. This process does not change the bulk value.

  • A dedup() step resets the bulk value of an element to 1.

by

• Description: A by() modulator step provides parameters such as traversals, functions, and comparators for other steps. The general pattern is step().by()…​by().

• Note: A by() step can modulate the following steps. For more information, see the description of each step.

  • aggregate(): You can use a by() step to specify the field based on which objects are aggregated.

  • cyclicPath(): You can use a by() step to specify the fields to be returned for a cyclic path.

  • dedup(): You can use a by() step to specify the field based on which repeated objects are deduplicated.

  • distinct(): You can use a by() step to specify the distinction rule.

  • group(): You can use two by() steps to specify the grouping rule and the fields to be stored in each group.

  • groupCount(): You can use a by() step to specify the grouping rule.

  • order(): You can use a by() step to specify the sort rule, such as the ascending order, descending order, or random order.

  • path(): You can use a by() step to specify the fields to be returned for a path.

  • project(): You can use a by() step to specify the values that are mapped.

  • sample(): You can use a by() step to specify the weight field for sampling. A larger weight indicates a higher probability that an object is sampled.

  • select(): You can use a by() step to specify the field to be obtained from the labeled objects.

  • simplePath(): You can use a by() step to specify the fields to be returned for a simple path. A path that is not cyclic is considered a simple path.

cap

• Description: A cap() map step returns the value of a sideEffect.

• Note:

  • This step can read multiple sideEffects and return a map<key,sideEffect>.

  • An implicit barrier() step is automatically inserted.

choose

• Description: A choose() branch step performs branching queries.

• Note:

  • If the condition of the first option is met, the process goes to the first branch. Otherwise, the second option is checked.

  • This step can take an arbitrary number of traversals as the input.

  • The use of this step is similar to that of a branch() step.

coalesce

• Description: A coalesce() branch step evaluates the input traversals in sequence and returns the first traversal that finds at least one element.

• Note: This step can take an arbitrary number of traversals as the input.

constant

• Description: A constant() map step returns a custom constant value.

• Note:

  • The constant value must be a number or a string.

  • This step is often used in branching queries.

count

• Description: A count() map step counts the number of input objects.

• Note:

  • A count(Scope.local) step can count the number of objects in a iterator-based data collection such as a list or a set.

  • An implicit barrier() step is automatically inserted.

cyclicPath

• Description: A cyclicPath() filter step returns a cyclic path.

• Note:

  • This step can be modulated by a by() step that specifies the fields to be returned for a cyclic path.

  • A simplePath() step returns a simple path.

dedup

• Description: A dedup() filter step deduplicates input objects.

• Note:

  • If the bulk value of an object is greater than 1, this step resets the bulk value to 1.

  • This step can be modulated by a by() step that specifies the field based on which objects are deduplicated.

    • A random object is retained for objects with the same field value.

  • This step can deduplicate objects based on labels added by as() steps.

  • An implicit barrier() step is automatically inserted.

distinct

• Description: A distinct() filter step hashes input objects based on a specific method and extracts a specific number of objects from each hash bucket.

• Note:

  • You can use the

    Distinct.round parameter to specify the number of extraction rounds. Default value: 1.

  • You can use the

    Distinct.amount parameter to specify the number of objects to be extracted from each bucket each round.

    Default value: 1.

  • You can use the

    Distinct.isReserved parameter to specify whether to append unextracted objects to the end of the result set. By default, unextracted objects are discarded.

  • This step can be modulated by a by() step that specifies the distinction rule.

  • An implicit barrier() step is automatically inserted.

  • This step is an extended step of iGraph.

E

• Description: An E() element step reads edges from a graph.

• Syntax: Separate multiple pkey strings with semicolons (;). Specify multiple secondary keys (skeys) for a pkey in the pk:sk1|sk2|... format.

• Note:

  • You can use a hasLabel() step to specify the label of the graph from which you want to read edges.

emit

• Description: An emit() modulator step returns the intermediate results of a loop.

• Note: This step must be used with other steps such as loop() and repeat().

fields

• Description: A field() sideEffect step retains the specified fields and removes other fields from elements, including vertices and edges.

• Syntax: Separate multiple fields with semicolons (;).

• Note:

  • You cannot remove a pkey field or an skey field.

  • This step helps reduce the data volume.

  • This step is an extended step of iGraph.

filter

• Description: A filter() filter step filters elements. Elements that meet the filter condition are retained, whereas other elements are discarded.

• Note: You can specify an expression or a child traversal as the filter condition.

fold

• Description: A fold() map step aggregates input objects into a list<object>.

• Note:

  • Differences among aggregate(), as(), and fold():

    • An aggregate() step aggregates input objects into a list and copies the list to a sideEffect. These objects are the input of the next step. The list is read-only in subsequent steps.

    • An as() step adds labels to input objects. These labeled objects are the input of the next step. You can modify the labeled objects in subsequent steps.

    • A fold() step aggregates input objects into a list. The list is the input of the next step.

  • An unfold() step has the inverse functionality to unfold a list into objects.

  • An implicit barrier() step is automatically inserted.

group

• Description: A group() map step groups input objects based on a specific rule.

• Note:

  • This step can be modulated by two by() steps.

    • You can use the first by() step to specify the grouping rule and the second by() step to specify the fields to be stored in each group.

    • You must specify the grouping rule. You can specify an expression or a child traversal as the grouping rule.

    • The fields to be stored in each group are optional. You can use an expression or a child traversal to specify the fields. By default, all fields of an object are stored.

  • An implicit barrier() step is automatically inserted.

• Common usage

groupCount

• Description: A groupCount() sideEffect step groups input objects based on a specific rule and returns the number of objects in each group.

• Note:

  • This step can be modulated by a by() step that specifies the grouping rule.

    • The grouping rule is optional. You can specify an expression or a child traversal as the grouping rule.

  • This step stores the grouping result to a sideEffect. Then, you can extract the objects by using a cap() step.

  • An implicit barrier() step is automatically inserted.