From 78bbd9a706ba102c64ade4edf333916089bb9760 Mon Sep 17 00:00:00 2001 From: dtookey Date: Fri, 15 Apr 2022 13:58:50 -0400 Subject: [PATCH] The finalization pass was a dumb idea. We'll just put the per-table/concept sanitization in their file. We need these to happen before the other scripts run because we denormalize the data for performance. --- src/sql/0-run-first/all_projects.sql | 57 +++++++++++++++++------- src/sql/0-run-first/z-sanitize_final.sql | 10 ----- 2 files changed, 42 insertions(+), 25 deletions(-) delete mode 100644 src/sql/0-run-first/z-sanitize_final.sql diff --git a/src/sql/0-run-first/all_projects.sql b/src/sql/0-run-first/all_projects.sql index a05df3b..e8f59cd 100644 --- a/src/sql/0-run-first/all_projects.sql +++ b/src/sql/0-run-first/all_projects.sql @@ -3,31 +3,58 @@ DROP TABLE IF EXISTS all_projects; #todo we have to find out how to generate this from the info on tablemetadata -CREATE TABLE all_projects AS SELECT * FROM CLT2019 +CREATE TABLE all_projects AS +SELECT * +FROM CLT2019 UNION -SELECT * FROM CLT2020 +SELECT * +FROM CLT2020 UNION -SELECT * FROM CLT2021 +SELECT * +FROM CLT2021 UNION -SELECT * FROM CLT2022 +SELECT * +FROM CLT2022 UNION -SELECT * FROM RDU2019 +SELECT * +FROM RDU2019 UNION -SELECT * FROM RDU2020 +SELECT * +FROM RDU2020 UNION -SELECT * FROM RDU2021 +SELECT * +FROM RDU2021 UNION -SELECT * FROM RDU2022 +SELECT * +FROM RDU2022 UNION -SELECT * FROM ENV2020 +SELECT * +FROM ENV2020 UNION -SELECT * FROM ENV2021 +SELECT * +FROM ENV2021 UNION -SELECT * FROM ENV2022 +SELECT * +FROM ENV2022 UNION -SELECT * FROM SRV2020 +SELECT * +FROM SRV2020 UNION -SELECT * FROM SRV2021; +SELECT * +FROM SRV2021; -CREATE UNIQUE INDEX ref_num ON all_projects(refnum); -CREATE INDEX project_flavor ON all_projects(proj_type); +CREATE UNIQUE INDEX ref_num ON all_projects (refnum); +CREATE INDEX project_flavor ON all_projects (proj_type); + + +UPDATE all_projects +SET description = TRIM(BOTH ' ' FROM description) +WHERE description RLIKE '^ ' + OR description RLIKE ' $'; + +UPDATE all_projects +SET description = 'Soil/Footings' +WHERE description IN + ( + 'Soils/Footings' + ); diff --git a/src/sql/0-run-first/z-sanitize_final.sql b/src/sql/0-run-first/z-sanitize_final.sql deleted file mode 100644 index 329039c..0000000 --- a/src/sql/0-run-first/z-sanitize_final.sql +++ /dev/null @@ -1,10 +0,0 @@ -UPDATE all_projects -set description = TRIM(both ' ' from description) -where description rlike '^ ' or description rlike ' $'; - -UPDATE all_projects -set description = 'Soil/Footings' -where description in - ( - 'Soils/Footings' - );