chore(scripts/sql): save useful SQL queries to file

This patch copies over some SQL queries I had saved in Beekeeper Studio
to local files tracked by version control. They may be useful to have in
the future. Many of these are already codified in migration files.

scripts/sql/README.md

@@ -0,0 +1,3 @@
+# `sql`
+
+This directory contains some SQL queries that could be useful in the future and were complex enough to track in version control.

scripts/sql/duplicate-looks.sql

@@ -0,0 +1,9 @@
+  SELECT ARRAY_AGG("id") "id", MIN("id") "idmin"
+	FROM "Look"
+	GROUP BY "showId", "number"
+	HAVING COUNT(*) > 1;
+
+ 	SELECT ARRAY_AGG("id") "id", "showId", "number", COUNT(*)
+	FROM "Look"
+	GROUP BY "showId", "number"
+	HAVING COUNT(*) > 1;

scripts/sql/looks-with-images.sql

@@ -0,0 +1,5 @@
+-- Looks with more than one image.
+SELECT "Look"."id", COUNT("Image"."id") as "images_count"
+FROM "Look" JOIN "Image" ON "Image"."lookId" = "Look"."id"
+GROUP BY "Look"."id"
+HAVING COUNT("Image"."id") > 1;

scripts/sql/remove-special-chars-from-slugs.sql

@@ -0,0 +1,68 @@
+CREATE EXTENSION IF NOT EXISTS unaccent;
+
+-- Normalize the brand slugs.
+CREATE OR REPLACE TEMP VIEW "BrandWithSlug" AS
+SELECT
+  REGEXP_REPLACE(NORMALIZE("slug", NFKD), '[\x0300-\x036f]', '', 'g') as "normalized_slug",
+  REGEXP_REPLACE(NORMALIZE("slug", NFKD), E'[\\x0300-\\x036f\\x2019\'"]', '', 'g') as "unquoted_normalized_slug",
+  UNACCENT("slug") as "unaccent_slug",
+  REGEXP_REPLACE(UNACCENT("slug"), E'[\'"]', '') as "unquoted_slug",
+  *
+FROM "Brand";
+SELECT * FROM "BrandWithSlug";
+
+-- See if there are any duplicate brands.
+CREATE OR REPLACE TEMP VIEW "DuplicateBrands" AS
+SELECT "unquoted_normalized_slug", ARRAY_AGG("BrandWithSlug"."id") AS "id", MAX("id") "idmax"
+FROM "BrandWithSlug"
+GROUP BY "unquoted_normalized_slug"
+HAVING COUNT(*) > 1;
+SELECT * FROM "DuplicateBrands";
+
+-- See if there are any duplicate shows.
+SELECT * FROM "Show" INNER JOIN "DuplicateBrands" ON "Show"."brandId" = ANY("DuplicateBrands"."id");
+
+-- Combine duplicates. For every show:
+-- (a) move all reviews to the last created show (i.e. the larger "ID");
+-- (b) add all look images to the last created show;
+-- (c) punt on all other relations as they aren't populated yet;
+-- (d) delete the duplicate show;
+-- (e) delete the duplicate brand.
+
+-- Note this only works given the assumption that the most recently created brand
+-- is also associated with the most recently created show, looks, images, etc.
+--
+-- This assumption is generally correct as I import brands, shows, looks, and images
+-- all at the same time. Should this change in the future, this migration will no longer
+-- work (but then again, it won't have to because it will have already been applied).
+
+CREATE OR REPLACE TEMP VIEW "DuplicateShows" AS
+SELECT ARRAY_AGG("Show"."id") "id", MAX("Show"."id") "idmax", "DuplicateBrands"."idmax" "brandId"
+FROM "Show"
+INNER JOIN "DuplicateBrands" ON "Show"."brandId" = ANY("DuplicateBrands"."id")
+GROUP BY "DuplicateBrands"."idmax", "seasonId", "sex", "level"
+HAVING COUNT(*) > 1;
+SELECT * FROM "DuplicateShows";
+
+CREATE OR REPLACE TEMP VIEW "DuplicateLooks" AS
+SELECT ARRAY_AGG("Look"."id") "id", MAX("Look"."id") "idmax", "DuplicateShows"."idmax" "showId"
+FROM "Look"
+INNER JOIN "DuplicateShows" ON "Look"."showId" = ANY("DuplicateShows"."id")
+GROUP BY "DuplicateShows"."idmax", "Look"."number"
+HAVING COUNT(*) > 1;
+SELECT * FROM "DuplicateLooks";
+
+-- (a) move all reviews to the last created show (i.e. the larger "ID");
+UPDATE "Review" SET "showId" = "DuplicateShows"."idmax" FROM "DuplicateShows"
+WHERE "Review"."showId" = ANY("DuplicateShows"."id") AND "Review"."showId" != "DuplicateShows"."idmax";
+
+-- (b) add all look images to the last created show;
+UPDATE "Image" SET "lookId" = "DuplicateLooks"."idmax" FROM "DuplicateLooks"
+WHERE "Image"."lookId" = ANY("DuplicateLooks"."id") AND "Image"."lookId" != "DuplicateLooks"."idmax";
+
+-- Delete duplicates (i.e. the smaller "ID" as we're keeping the larger "ID").
+DELETE FROM "Brand" USING "DuplicateBrands"
+WHERE "Brand"."id" = ANY("DuplicateBrands"."id") AND "Brand"."id" != "DuplicateBrands"."idmax";
+
+-- Update the brand slugs (to remove special characters).
+UPDATE "Brand" SET "slug" = "unquoted_normalized_slug" FROM "BrandWithSlug" WHERE "BrandWithSlug"."id" = "Brand"."id";

scripts/sql/reviewed-by-person.sql

@@ -0,0 +1,4 @@
+SELECT "Show"."name", "Review"."publicationId" FROM "Show"
+INNER JOIN "Review" ON "Review"."showId" = "Show"."id"
+INNER JOIN "User" on "User"."id" = "Review"."authorId"
+WHERE "User"."name" = 'Luisa Zargani'

scripts/sql/shows-in-wwd.sql

@@ -0,0 +1 @@
+SELECT "Show"."name" FROM "Review" INNER JOIN "Show" ON "Show"."id" = "Review"."showId" WHERE "Review"."publicationId" = 2 ORDER BY "Show"."name" ASC

scripts/sql/vogue-shows-with-locations.sql

@@ -0,0 +1,14 @@
+-- Gets the shows that were originally imported from Vogue with locations, and
+-- had the same locations in WWD and thus the records merged as expected.
+--
+-- This query only returned 3 rows, meaning that I should probably simply omit
+-- the location from the show name when merging from WWD (as there were more records
+-- that were alike besides the location) and then manually merge those three.
+--
+-- Hyke FALL 2023 WOMAN RTW TOKYO, Hyke SPRING 2022 WOMAN RTW TOKYO, Hyke SPRING 2023 WOMAN RTW TOKYO
+SELECT * FROM (
+  SELECT "Show"."name", "Show"."location", COUNT("Review"."id") as "review_count"
+  FROM "Show"
+  LEFT OUTER JOIN "Review" ON "Review"."showId" = "Show"."id"
+  GROUP BY "Show"."name", "Show"."location"
+) "Shows" WHERE "review_count" > 1 AND "location" != 'PARIS' AND "location" IS NOT NULL