From f5d8f56330ea5455ae2e420811885336917f675d Mon Sep 17 00:00:00 2001
From: Christoph Urlacher <christoph.urlacher@protonmail.com>
Date: Sat, 7 Jun 2025 20:34:02 +0200
Subject: [PATCH] Lib: Implement starting grid scraper

---
 src/lib/fetch.ts         | 14 ++++++++++++
 src/lib/schema.ts        |  8 +++++++
 src/lib/server/scrape.ts | 49 +++++++++++++++++++++++++++++++++++++++-
 3 files changed, 70 insertions(+), 1 deletion(-)

diff --git a/src/lib/fetch.ts b/src/lib/fetch.ts
index 8f32779..3e43828 100644
--- a/src/lib/fetch.ts
+++ b/src/lib/fetch.ts
@@ -13,6 +13,7 @@ import type {
   RaceResult,
   ScrapedDriverStanding,
   ScrapedRaceResult,
+  ScrapedStartingGrid,
   ScrapedTeamStanding,
   SeasonPick,
   SeasonPickedUser,
@@ -333,6 +334,19 @@ export const fetch_scraped_teamstandings = async (
   return scraped_teamstandings;
 };
 
+/**
+ * Fetch all [ScrapedStartingGrids] from the database, ordered descendingly by race step.
+ */
+export const fetch_scraped_startinggrids = async (
+  fetch: (_: any) => Promise<Response>,
+): Promise<ScrapedStartingGrid[]> => {
+  const scraped_startinggrids: ScrapedStartingGrid[] = await pb
+    .collection("scraped_startinggrids")
+    .getFullList({ fetch: fetch, sort: "-race_step,+position" });
+
+  return scraped_startinggrids;
+};
+
 /**
  * Fetch all [ScrapedRaceResults] from the database, ordered descendingly by race step.
  */
diff --git a/src/lib/schema.ts b/src/lib/schema.ts
index e088425..c83ec81 100644
--- a/src/lib/schema.ts
+++ b/src/lib/schema.ts
@@ -159,6 +159,14 @@ export interface RacePickPointsTotal {
 
 // Scraped Data
 
+export interface ScrapedStartingGrid {
+  id: string;
+  race_step: number; // This maps to races
+  driver_code: string; // This maps to drivers
+  position: number;
+  time: string;
+}
+
 export interface ScrapedRaceResult {
   id: string;
   race_step: number; // This maps to races
diff --git a/src/lib/server/scrape.ts b/src/lib/server/scrape.ts
index 94b1d11..1f7194f 100644
--- a/src/lib/server/scrape.ts
+++ b/src/lib/server/scrape.ts
@@ -1,4 +1,9 @@
-import type { ScrapedDriverStanding, ScrapedRaceResult, ScrapedTeamStanding } from "$lib/schema";
+import type {
+  ScrapedDriverStanding,
+  ScrapedStartingGrid,
+  ScrapedRaceResult,
+  ScrapedTeamStanding,
+} from "$lib/schema";
 import * as cheerio from "cheerio";
 
 // TODO: Validate the generated stuff
@@ -26,6 +31,48 @@ export const scrape_race_links = async (): Promise<string[]> => {
   return race_links;
 };
 
+/**
+ * Returns a list of [ScrapedStartingGrids] for all races contained in [race_links],
+ * based on official f1.com data.
+ */
+export const scrape_starting_grids = async (
+  race_links: string[],
+): Promise<ScrapedStartingGrid[]> => {
+  // Update the race_links to point to the qualifications
+  const starting_grid_links: string[] = race_links.map((link: string) =>
+    link.replace("/race-result", "/starting-grid"),
+  );
+
+  const starting_grids: ScrapedStartingGrid[] = [];
+  await Promise.all(
+    starting_grid_links.map(async (link: string, index: number) => {
+      console.log(`Fetching qualifying results from ${base_url}/${link}...`);
+      const starting_grids_response = await fetch(`${base_url}/${link}`);
+      const starting_grids_text = await starting_grids_response.text();
+
+      const $ = cheerio.load(starting_grids_text);
+
+      // Obtain the positions for this starting grid for each driver
+      $("tbody > tr", "div.f1-inner-wrapper table.f1-table").each((driver_index, element) => {
+        const $$ = cheerio.load(element);
+
+        let result: ScrapedStartingGrid = {
+          id: "",
+          race_step: index + 1,
+          driver_code: $$("td:nth-child(3) > p > span:last-child").text(),
+          position: driver_index + 1, // parseInt($$("td:nth-child(1) > p").text()),
+          time: $$("td:nth-child(5) > p").text(),
+        };
+
+        starting_grids.push(result);
+      });
+    }),
+  );
+  console.log(`Scraped ${starting_grids.length} starting grids...`);
+
+  return starting_grids;
+};
+
 /**
  * Returns a list of [ScrapedRaceResults] for all races contained in [race_links],
  * based on official f1.com data.