The SELECT Statement

The SELECT statement is the heart of GSQL. It defines how to traverse the graph, filter data, and perform computations using path patterns.

1. Basic Syntax (V2/V3)

GSQL uses a path-based syntax that is compatible with GQL (Graph Query Language).

gsqlterminal
Copy
Result = SELECT target_alias
         FROM source_alias:s -(edge_type:e)-> target_type:t
         WHERE t.age > 21
         ACCUM t.@visited += 1;

2. The FROM Clause: Path Patterns

Path patterns define the template for traversal.

• Vertex Pattern: (alias:Type)
• Edge Pattern: -[alias:Type]-> (Right), <-[alias:Type]- (Left), -[alias:Type]- (Undirected).

Multi-Hop & Quantifiers

You can traverse multiple levels using quantifiers:

• (s:Person) -[:friend*1..3]-> (t:Person): Finds friends up to 3 hops away.
• *: 1 or more repetitions.
• {m,n}: m to n repetitions.

3. Computational Clauses

SELECT statements are unique in GSQL because they allow computation during traversal.

ACCUM

Executes for every edge that matches the pattern. Perfect for updating accumulators based on edge/neighbor data.

gsqlterminal
Copy
ACCUM t.@score += s.weight * e.impact

POST-ACCUM

Executes once for every vertex in the source or target set after the ACCUM phase. Ideal for final calculations or filtering based on accumulated values.

gsqlterminal
Copy
POST-ACCUM s.total_score = s.@score / s.outdegree()

4. Result Refinement

• WHERE: Filters edges and vertices during traversal.
• ORDER BY: Sorts the resulting vertex set.
• LIMIT: Restricts the number of vertices returned.
• SAMPLE: Randomly samples a subset of neighbors (essential for "hub" nodes with high degrees).

5. Conjunctive Patterns

You can combine multiple paths in a single SELECT by separating them with commas. This acts as an "AND" condition.

gsqlterminal
Copy
# Find people who are both 'Friends' and 'Coworkers'
Results = SELECT t
          FROM (s:Person) -[:Friend]-> (t:Person),
               (s:Person) -[:Works_With]-> (t:Person);