data4impact/apps/data4impact-import-scripts/scripts/patent_excel/schema.sql

43 lines
1.4 KiB
SQL

CREATE TABLE data(
pat_id text,
type_ip text,
appnum text,
appnt text,
title text,
pat_url text,
pat_ref text,
pat_auth text,
pat_num text,
pat_kind text,
note text,
appln_id text,
appln_title_patstat text,
priority_year text,
var15 text,
projectid text
);
CREATE MATERIALIZED VIEW document AS SELECT
'50|patents_____::'||MD5(lower(trim(appln_id))) AS "id",
title AS "title",
lower(regexp_replace(type_ip,'s$','')) AS "type",
priority_year AS "pubYear",
'patent repo'::text AS "repository"
FROM data
WHERE appln_id IS NOT NULL AND trim(appln_id) != '';
CREATE MATERIALIZED VIEW doc_other_identifier AS SELECT
'50|patents_____::'||MD5(lower(trim(appln_id))) AS "docId",
trim(appln_id) AS "id",
'patent'::text AS "type"
FROM data
WHERE appln_id IS NOT NULL AND trim(appln_id) != '';
CREATE MATERIALIZED VIEW doc_project AS SELECT
'50|patents_____::'||MD5(lower(trim(appln_id))) AS "docId",
'40|corda_______::'||MD5(lower(trim(projectid))) AS "projectId"
FROM data
WHERE appln_id IS NOT NULL AND trim(appln_id) != '' AND projectid IS NOT NULL AND trim(projectid) != '';