Export to CSV

This page walks through exporting a graph to CSV files using gqldb-exporter. The exporter writes nodes and edges to separate CSV files in a chosen output directory.

Usage Guides

Generate Configuration File

Bash
./gqldb-exporter -sample csv

A file named export.sample.csv.yml will be created in the current directory. Rename it before editing so a re-run of -sample csv doesn't clobber your changes:

Bash
mv export.sample.csv.yml export.csv.yml

Modify Configuration File

Edit export.csv.yml to point at your source graph and choose an output directory. See the Export Configurations for every field; CSV-specific settings are below.

config snippet
settings:
  format: csv
  output_path: "./exported"
  batch_size: 5000
  write_header: true        # CSV-only: include a header row in each output file
  export_nodes: true
  export_edges: true
  include_metadata: true
  nodes_file: "nodes"       # output filename prefix for node files
  edges_file: "edges"       # output filename prefix for edge files
  # node_labels: ["Person", "Company"]
  # edge_labels: ["KNOWS"]
  log_level: info

CSV-specific knobs:

• write_header — when true, every output file begins with a row of column names. The header uses the typed form <colName>:<type> (e.g., _id:_id, age:int32), so the resulting files can be re-imported with head: true without further configuration.
• nodes_file / edges_file — filename prefixes (without extension). One file is produced per label; see Output Layout below.

Execute Export

Bash
./gqldb-exporter -c export.csv.yml

Output Layout

Within output_path, the exporter writes one CSV file per label:

Directory layout
exported/
├── nodes.Person.csv
├── nodes.Company.csv
├── edges.KNOWS.csv
└── edges.WORKS_AT.csv

The pattern is <nodes_file>.<Label>.csv for nodes and <edges_file>.<Label>.csv for edges. If you set nodes_file: "n" and edges_file: "e", the files become n.Person.csv, e.KNOWS.csv, etc.

Header Format

When write_header: true, the first row of each file uses the <colName>:<type> form. For nodes:

nodes.Person.csv
_id:_id,name:string,age:int32
P001,Alice,30
P002,Bob,25

For edges, the source and destination IDs are emitted as _from:_from and _to:_to:

edges.KNOWS.csv
_from:_from,_to:_to,since:int32
P001,P002,2024
P002,P001,2025

This typed header is what the importer's CSV import expects when head: true, so an exported CSV directory can be re-imported without writing a separate property configuration.