export-formats

Export Taxonomy to Multiple Formats

Export a generated taxonomy to CSV, TSV, JSON, NDJSON, SQL, or YAML. Preserve all columns and metadata.

When to use

• "Export this as JSON for my app"
• "I need a SQL seed file with INSERT statements"
• "Give me NDJSON for BigQuery"
• "Convert to YAML format"

Inputs to gather

• Taxonomy file path: Absolute path to generated CSV or JSON (auto-discover from data/<name>/ if not given).
• Format: CSV, TSV, JSON, NDJSON, SQL, YAML.
• For SQL format: Table name (inferred from filename if not given).

Procedure

1. Read the taxonomy file (CSV or JSON).
2. Normalize to in-memory structure (list of dicts, one per row).
3. Write to data/<name>/exports/ in the requested format:
   • CSV: Standard csv module; preserve headers.
   • TSV: csv with delimiter='\t'.
   • JSON: Pretty-printed array of objects; include schema hint in a _schema field if needed.
   • NDJSON: One JSON object per line (no array wrapper); good for streaming/BigQuery.
   • SQL: Generate CREATE TABLE DDL + INSERT statements. Include the same DDL as load-to-postgres skill (infer types from data); wrap in a transaction (BEGIN; ... COMMIT;).
   • YAML: Array of objects in YAML syntax (each entry as a dict with key-value pairs).
4. Log the export: write filename and format to state/exports/<timestamp>-<name>-<format>.log.
5. Report to user: file size, row count, format name, full output path.

Output / side effects

• Exported file in data/<name>/exports/<name>.<ext> where ext is csv, tsv, json, ndjson, sql, yaml.
• Export logged to state/exports/<timestamp>-<name>-<format>.log.

Safety / constraints

• No credentials in SQL: SQL seed files should not include connection strings or credentials.
• Reversible: Export does not modify the original taxonomy; purely additive.