summaryrefslogtreecommitdiff
path: root/app/scrapers/club-quattro.ts
diff options
context:
space:
mode:
authoryyamashita <yyamashita@mosquit.one>2026-05-07 22:06:24 +0900
committeryyamashita <yyamashita@mosquit.one>2026-05-07 22:06:24 +0900
commitb4823d4c124160dcba8e5fed1e424cf2cc12e72c (patch)
tree573de8cf82704320b3f308eaa0112e590e5f0aa5 /app/scrapers/club-quattro.ts
parentb8537eabe94b24e8530b4c1511456dc94cf8ec4c (diff)
Fix scrapers returning 0 events for 4 venues
Rewrote selectors to match actual HTML structure after inspecting each site: - LIQUID ROOM: article selector, date extracted from URL (_YYYYMMDD suffix) - WWW/WWW X: article.column selector, month from li.month nav, day rollover detection - 新宿 LOFT: section.block_schedule_list links, full date from time div.year/month/day - CLUB QUATTRO: li[data-event-date] with ISO date attribute, jp-label time parsing Result: 0件 → LIQUID ROOM 25件 / WWW 48件 / LOFT 13件 / QUATTRO 24件 Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Diffstat (limited to 'app/scrapers/club-quattro.ts')
-rw-r--r--app/scrapers/club-quattro.ts59
1 files changed, 23 insertions, 36 deletions
diff --git a/app/scrapers/club-quattro.ts b/app/scrapers/club-quattro.ts
index ae903bc..946b9a4 100644
--- a/app/scrapers/club-quattro.ts
+++ b/app/scrapers/club-quattro.ts
@@ -1,6 +1,3 @@
-/**
- * Club Quattro 渋谷 — https://www.club-quattro.com/shibuya/schedule/
- */
import * as cheerio from "cheerio";
import type { Scraper, VenueMeta } from "./base";
import type { EventInput } from "~/lib/db.server";
@@ -21,38 +18,40 @@ export const scraper: Scraper = {
const $ = cheerio.load(html);
const events: EventInput[] = [];
- $(".schedule-list__item, .c-event, li.event").each((_, el) => {
+ $("li[data-event-date]").each((_, el) => {
const $el = $(el);
- const title = $el.find(".schedule-list__title, .event-name, h3, h2").first().text().trim();
+ const date = $el.attr("data-event-date") ?? "";
+ if (!date) return;
+
+ const title = $el.find("p.txt-02").text().trim();
if (!title) return;
- const rawDate =
- $el.find(".schedule-list__date, .event-date, time").first().text().trim() ||
- $el.find("time").attr("datetime") ||
- "";
- const date = parseJapaneseDate(rawDate);
- if (!date) return;
+ const artist = $el.find("p.txt-01 span").text().trim() || null;
- const timeText = $el.find(".schedule-list__time, .time-info").first().text();
- const openMatch = timeText.match(/OPEN[:: ]*(\d{2}:\d{2})/i);
- const startMatch = timeText.match(/START[:: ]*(\d{2}:\d{2})/i);
+ let openTime: string | null = null;
+ let startTime: string | null = null;
+ $el.find("dl.detail-list .bundle").each((_, bundle) => {
+ const label = $(bundle).find("dt").text().trim();
+ if (label.includes("開場") || label.includes("開演")) {
+ const times = $(bundle).find("dd").text().trim().match(/\d{2}:\d{2}/g) ?? [];
+ openTime = times[0] ?? null;
+ startTime = times[1] ?? null;
+ }
+ });
- const detailHref = $el.find("a[href]").first().attr("href") ?? null;
+ const href = $el.find("a").first().attr("href") ?? null;
+ const imageSrc = $el.find(".front img").attr("src") ?? null;
events.push({
venue_id: venue.id,
title,
- artist: $el.find(".schedule-list__artist, .artist-name").first().text().trim() || null,
+ artist,
date,
- open_time: openMatch?.[1] ?? null,
- start_time: startMatch?.[1] ?? null,
- ticket_url:
- $el.find("a[href*='eplus'], a[href*='pia'], a[href*='ticket']").first().attr("href") ?? null,
- image_url: $el.find("img").first().attr("src")
- ? absoluteUrl($el.find("img").first().attr("src")!, venue.url)
- : null,
- source_url: detailHref ? absoluteUrl(detailHref, venue.url) : null,
+ open_time: openTime,
+ start_time: startTime,
+ image_url: imageSrc ? absoluteUrl(imageSrc, venue.url) : null,
+ source_url: href ? absoluteUrl(href, venue.url) : null,
});
});
@@ -60,18 +59,6 @@ export const scraper: Scraper = {
},
};
-function parseJapaneseDate(raw: string): string | null {
- const m =
- raw.match(/(\d{4})[./年](\d{1,2})[./月](\d{1,2})/) ||
- raw.match(/(\d{1,2})[./月](\d{1,2})/);
- if (!m) return null;
- if (m.length === 4) {
- return `${m[1]}-${m[2].padStart(2, "0")}-${m[3].padStart(2, "0")}`;
- }
- const year = new Date().getFullYear();
- return `${year}-${m[1].padStart(2, "0")}-${m[2].padStart(2, "0")}`;
-}
-
function absoluteUrl(url: string, base: string): string {
if (url.startsWith("http")) return url;
return url.startsWith("/") ? base + url : `${base}/${url}`;