Output & Logging

Extracting data from GSQL can be done via standard JSON responses, external files, or system logs.

1. The PRINT Statement

The primary way to return data from a query. Results are aggregated into a single JSON object delivered at the end of the query.

gsqlterminal
Copy
PRINT @@total_count, StartNodes AS seed_set;

Vertex Expression Sets

You can customize the output of a vertex set by specifying only the attributes you want, or by computing values on the fly.

gsqlterminal
Copy
# Only print id and country, and alias the country attribute
PRINT MyNodes[MyNodes.id, MyNodes.country AS location];

2. Writing to Files (FILE)

For large datasets where JSON is inefficient, use the FILE object to write CSV or custom formatted text.

gsqlterminal
Copy
CREATE QUERY export_data(STRING path) {
  FILE f1 (path);
  
  # Print headers
  PRINT "ID,Score" TO_CSV f1;
  
  # Print data in parallel (order not guaranteed)
  V = SELECT s FROM Start:s
      ACCUM f1.println(s.id, s.@score);
}

Key Keywords:

• TO_CSV: Automatically formats output as comma-separated values.
• .println(): Writes a custom line to the file. Can be used inside ACCUM for high-speed parallel writes.

3. Subquery Returns

Use RETURN to pass data from a subquery to a parent query. The return type must match the RETURNS clause in the subquery definition.

gsqlterminal
Copy
CREATE QUERY get_score(VERTEX<Person> v) RETURNS (FLOAT) {
  # ... calculation ...
  RETURN @@final_score;
}

4. Debugging with LOG

The LOG statement writes directly to the GPE (Graph Processing Engine) logs. It is essential for debugging parallel logic.

gsqlterminal
Copy
# Syntax: LOG(condition, args...)
LOG(true, "Processing vertex", s.id, "at hop", @@hop_count);

[!TIP] To view logs, run gadmin log gpe in the shell and look for entries prefixed with UDF_.