Instructions to import the CSV version of the ORCiD public data produced by https://github.com/slint/orcid-data-toolkit

names.sql

-- Create a new table with the new schema (excluding constraints and indexes)
CREATE TABLE name_metadata_new
  (LIKE name_metadata EXCLUDING CONSTRAINTS EXCLUDING INDEXES);
-- Time: 0.337s

-- Copy the CSV data into the new table, running the following in a shell:
-- export DB_URI="CHANGE_ME"
-- gzip -dc orcid-names-2024.csv.gz | pv --line-mode --average-rate --timer | psql $DB_URI -c 'COPY name_metadata_new (created, updated, id, json, version_id, pid) FROM STDIN (FORMAT CSV);'
-- Time: 210.000s (3 minutes 30 seconds)

-- Rename the old indexes/constraints
ALTER INDEX pk_name_metadata RENAME TO pk_name_metadata_old;
ALTER INDEX uq_name_metadata_pid RENAME TO uq_name_metadata_pid_old;
-- Time: 0.042s

ALTER TABLE name_metadata_new ADD CONSTRAINT pk_name_metadata PRIMARY KEY (id);
-- Time: 53.613s (53 seconds)
ALTER TABLE name_metadata_new ADD CONSTRAINT uq_name_metadata_pid UNIQUE (pid);
-- Time: 112.739s (1 minute 52 seconds)

-- Rename the tables and indices/constraints in a transaction
BEGIN;
ALTER TABLE name_metadata RENAME TO name_metadata_old;
ALTER TABLE name_metadata_new RENAME TO name_metadata;
COMMIT;
-- Time: 0.013s

-- Drop the old table (optional)
DROP TABLE name_metadata_old;

-- Reindex names to OpenSearch (~6h)
-- invenio rdm rebuild-all-indices -o names
-- Time: ~1h for the command + 5h for bulk indexing