new post: SELECT solution FROM aoc WHERE year = 2022

_posts/2022-12-09-select-solution-from-aoc-where-year-2022.md

@@ -0,0 +1,87 @@
+---
+permalink: "/{{ year }}/{{ month }}/{{ day }}/select-solution-from-aoc-where-year-2022"
+title: "SELECT solution FROM aoc WHERE year = 2022"
+published_date: "2022-12-09 11:20:00 +0100"
+layout: post.liquid
+data:
+  route: blog
+---
+
+Every year [Advent of Code][aoc] happens.
+Every year I'm not super enthusiastic to actually solve those challenges.
+Every year I still start (and then maybe stop after some days).
+
+2022 was similar.
+I don't have a new programming language to learn and I still write enough code day-by-day that more coding is not really necessary.
+But yet sometimes it can be kind of nice to have a small challenge with a clear solution to go for
+and so I set myself some rules and went ahead.
+
+## My own rules
+
+* Solve it in SQL.
+  * I started to run that in [BigQuery], but moved to [SQLite].
+* Only the SQL the system provides. How far can I take this?
+  * It was easier on BigQuery with its large base of functionality, including string procesing, arrays, structs, ...
+  * On SQLite initially I tried to avoid any custom functionality, I later relaxed this rule: [SQLean] is okay occasionally.
+  * Once (so far) I used Python to preprocess the data. SQL is just really not good at text processing.
+* Getting the solution once is good enough.
+  * The code can be ugly, messy and inefficient.
+* I don't write tests.
+  * I do test. Just manually against the test input.
+* No persistent data.
+  * It's all in-memory tables, CTEs, temporary views and a bunch of `SELECT`.
+* Learn some arcane SQL features.
+  * Did you know SQL can recurse and select over windows? I do now!
+
+These are my rules. I break them when I feel like it.
+
+[AoC]: https://adventofcode.com/
+[bigquery]: https://cloud.google.com/bigquery/
+[sqlite]: https://sqlite.org/
+[sqlean]: https://github.com/nalgeon/sqlean
+
+## Sneak peek of day 1
+
+`SELECT solution FROM aoc WHERE year = 2022 AND day = 1;`
+
+I plan to release my solutions publicly, but I haven't yet.
+So for now you get my solution to day 1:
+
+```sql
+CREATE TABLE data(raw);
+.import input01.txt data
+WITH
+groups AS (
+  SELECT rowid, ROW_NUMBER() OVER (ORDER BY rowid) cat FROM data WHERE raw = ''
+),
+elves AS (
+  SELECT
+    COALESCE(
+      (SELECT cat FROM groups WHERE data.rowid < groups.rowid LIMIT 1),
+      (SELECT cat + 1 FROM groups ORDER BY cat DESC LIMIT 1)
+    ) AS elf,
+    raw as calories
+  FROM data WHERE raw != ''
+),
+part1 AS (
+  SELECT
+    elf,
+    SUM(calories) AS total_calories
+  FROM elves
+  GROUP BY 1
+  ORDER BY 2 DESC
+)
+
+--SELECT * FROM part1;
+SELECT SUM(total_calories) as best_3 FROM (
+    SELECT * FROM part1 LIMIT 3
+);
+```
+
+Runnable as:
+
+```
+sqlite3 < aoc01.sql
+```
+
+(Oh, another rule: It needs to run as `sqlite3 < script.sql`)