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>

4 files changed, 132 insertions, 184 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}`;
diff --git a/app/scrapers/liquid-room.ts b/app/scrapers/liquid-room.ts
index b497759..f577ee6 100644
--- a/app/scrapers/liquid-room.ts
+++ b/app/scrapers/liquid-room.ts
@@ -1,9 +1,3 @@
-/**
- * Liquid Room (恵比寿) — https://www.liquidroom.net/schedule
- *
- * The schedule page lists events with JSON-LD or HTML data.
- * Structure: <div class="p-schedule__item"> contains date, title, etc.
- */
 import * as cheerio from "cheerio";
 import type { Scraper, VenueMeta } from "./base";
 import type { EventInput } from "~/lib/db.server";
@@ -24,64 +18,53 @@ export const scraper: Scraper = {
     const $ = cheerio.load(html);
     const events: EventInput[] = [];
 
-    $("article.p-schedule__item, .schedule-list__item, .c-event-item").each(
-      (_, el) => {
-        const $el = $(el);
+    $("article").each((_, el) => {
+      const $el = $(el);
 
-        const title =
-          $el.find(".p-schedule__title, .event-title, h3, h2").first().text().trim();
-        if (!title) return;
+      const href = $el.find("a.s_link").attr("href") ?? "";
+      // Date is encoded in the URL: e.g. /schedule/eventname_20260501
+      const dateMatch = href.match(/_(\d{4})(\d{2})(\d{2})$/);
+      if (!dateMatch) return;
+      const date = `${dateMatch[1]}-${dateMatch[2]}-${dateMatch[3]}`;
 
-        const dateStr =
-          $el.find(".p-schedule__date, .event-date, time").first().text().trim() ||
-          $el.find("time").attr("datetime") ||
-          "";
-        const date = parseJapaneseDate(dateStr);
-        if (!date) return;
+      const h2 = $el.find("h2").first().text().trim();
+      if (!h2) return;
 
-        const artist =
-          $el.find(".p-schedule__artist, .artist").first().text().trim() || null;
-        const startTime =
-          $el.find(".p-schedule__time, .open-time").first().text().trim().match(/\d{2}:\d{2}/)?.[0] ?? null;
-        const ticketUrl =
-          $el.find("a[href*='ticket'], a[href*='eplus'], a[href*='pia']").first().attr("href") ?? null;
-        const imageUrl =
-          $el.find("img").first().attr("src") ?? null;
-        const sourceUrl =
-          $el.find("a").first().attr("href") ?? null;
+      const subtitle = $el.find("p.subtitle").first().text().trim();
+      // h2 is the artist/band name; subtitle (if present) is the event title
+      const title = subtitle || h2;
+      const artist = subtitle ? h2 : null;
 
-        events.push({
-          venue_id: venue.id,
-          title,
-          artist,
-          date,
-          start_time: startTime,
-          ticket_url: ticketUrl,
-          image_url: imageUrl ? absoluteUrl(imageUrl, venue.url) : null,
-          source_url: sourceUrl ? absoluteUrl(sourceUrl, venue.url) : null,
-        });
-      }
-    );
+      const openTime =
+        $el
+          .find("dl")
+          .filter((_, dl) => $(dl).find("dt").text().includes("OPEN"))
+          .find("dd")
+          .text()
+          .trim()
+          .match(/\d{2}:\d{2}/)?.[0] ?? null;
+
+      const startTime =
+        $el
+          .find("dl")
+          .filter((_, dl) => $(dl).find("dt").text().includes("START"))
+          .find("dd")
+          .text()
+          .trim()
+          .match(/\d{2}:\d{2}/)?.[0] ?? null;
+
+      events.push({
+        venue_id: venue.id,
+        title,
+        artist,
+        date,
+        open_time: openTime,
+        start_time: startTime,
+        image_url: $el.find("div.left img").attr("src") ?? null,
+        source_url: href,
+      });
+    });
 
     return events;
   },
 };
-
-function parseJapaneseDate(raw: string): string | null {
-  // Handles "2025.06.15" "2025/06/15" "2025年06月15日" "06.15" formats
-  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;
-  if (url.startsWith("/")) return base + url;
-  return base + "/" + url;
-}
diff --git a/app/scrapers/shinjuku-loft.ts b/app/scrapers/shinjuku-loft.ts
index 8a64761..d5602e7 100644
--- a/app/scrapers/shinjuku-loft.ts
+++ b/app/scrapers/shinjuku-loft.ts
@@ -1,8 +1,3 @@
-/**
- * 新宿 LOFT — https://www.loft-prj.co.jp/schedule/loft
- *
- * The schedule page renders events inside `.eventlist` items.
- */
 import * as cheerio from "cheerio";
 import type { Scraper, VenueMeta } from "./base";
 import type { EventInput } from "~/lib/db.server";
@@ -22,59 +17,48 @@ export const scraper: Scraper = {
     const html = await res.text();
     const $ = cheerio.load(html);
     const events: EventInput[] = [];
+    const seen = new Set<string>();
 
-    $(".eventlist__item, .schedule-item, .event_list li").each((_, el) => {
-      const $el = $(el);
+    $("section.block_schedule_list a[href*='/schedule/loft/schedule/']").each(
+      (_, el) => {
+        const $el = $(el);
+        const href = $el.attr("href") ?? "";
+        if (seen.has(href)) return;
+        seen.add(href);
 
-      const title = $el.find(".eventlist__title, .event-title, h3, h2").first().text().trim();
-      if (!title) return;
+        const year = $el.find("time div.year").text().trim();
+        const month = $el.find("time div.month").text().trim();
+        const day = $el.find("time div.day").text().trim();
+        if (!year || !month || !day) return;
+        const date = `${year}-${month.padStart(2, "0")}-${day.padStart(2, "0")}`;
 
-      const rawDate =
-        $el.find(".eventlist__date, .event-date, time").first().text().trim() ||
-        $el.find("time").attr("datetime") ||
-        "";
-      const date = parseJapaneseDate(rawDate);
-      if (!date) return;
+        const title = $el.find(".c_title span").text().trim();
+        if (!title) return;
 
-      const timeText = $el.find(".eventlist__time, .time").first().text();
-      const openMatch = timeText.match(/OPEN[：: ]*(\d{2}:\d{2})/i);
-      const startMatch = timeText.match(/START[：: ]*(\d{2}:\d{2})/i);
+        const timeText = $el.find(".open").text().trim();
+        const openMatch = timeText.match(/OPEN\s*(\d{2}:\d{2})/i);
+        const startMatch = timeText.match(/START\s*(\d{2}:\d{2})/i);
 
-      const detailHref = $el.find("a[href]").first().attr("href") ?? null;
+        const artists = $el
+          .find("ul.artist_tag li")
+          .map((_, li) => $(li).text().trim())
+          .get()
+          .filter((a) => a !== "...");
+        const artist = artists.join(" / ") || null;
 
-      events.push({
-        venue_id: venue.id,
-        title,
-        artist: $el.find(".eventlist__artist, .artist").first().text().trim() || null,
-        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,
-      });
-    });
+        events.push({
+          venue_id: venue.id,
+          title,
+          artist,
+          date,
+          open_time: openMatch?.[1] ?? null,
+          start_time: startMatch?.[1] ?? null,
+          image_url: $el.find("span.bg").attr("data-bg") ?? null,
+          source_url: href,
+        });
+      }
+    );
 
     return events;
   },
 };
-
-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}`;
-}
diff --git a/app/scrapers/www-shibuya.ts b/app/scrapers/www-shibuya.ts
index 905fc61..d561332 100644
--- a/app/scrapers/www-shibuya.ts
+++ b/app/scrapers/www-shibuya.ts
@@ -1,6 +1,3 @@
-/**
- * WWW / WWW X (渋谷) — https://www-shibuya.jp/schedule/
- */
 import * as cheerio from "cheerio";
 import type { Scraper, VenueMeta } from "./base";
 import type { EventInput } from "~/lib/db.server";
@@ -21,39 +18,48 @@ export const scraper: Scraper = {
     const $ = cheerio.load(html);
     const events: EventInput[] = [];
 
-    $(".schedule-list li, .p-schedule-item, article").each((_, el) => {
+    // Month from nav: "202605May" → year=2026, month=5
+    const monthText = $("li.month").first().text().trim();
+    const monthMatch = monthText.match(/(\d{4})(\d{2})/);
+    let year = monthMatch ? parseInt(monthMatch[1]) : new Date().getFullYear();
+    let month = monthMatch ? parseInt(monthMatch[2]) : new Date().getMonth() + 1;
+    let prevDay = 0;
+
+    $("article.column").each((_, el) => {
       const $el = $(el);
 
-      const title = $el.find(".schedule-title, .title, h3, h2").first().text().trim();
-      if (!title) return;
+      const day = parseInt($el.find(".date .day").text().trim(), 10);
+      if (!day) return;
 
-      const rawDate =
-        $el.find(".schedule-date, .date, time").first().text().trim() ||
-        $el.find("time").attr("datetime") ||
-        "";
-      const date = parseJapaneseDate(rawDate);
-      if (!date) return;
+      // Detect month rollover when day numbers reset
+      if (prevDay > 0 && day < prevDay) {
+        month++;
+        if (month > 12) {
+          month = 1;
+          year++;
+        }
+      }
+      prevDay = day;
 
-      const timeText = $el.find(".schedule-time, .time").first().text();
-      const openMatch = timeText.match(/OPEN\s*(\d{2}:\d{2})/i);
-      const startMatch = timeText.match(/START\s*(\d{2}:\d{2})/i);
+      const date = `${year}-${String(month).padStart(2, "0")}-${String(day).padStart(2, "0")}`;
+
+      const title = $el.find("h3").text().trim();
+      if (!title) return;
 
-      const detailHref = $el.find("a").first().attr("href") ?? null;
+      const timeText = $el.find(".openstart").text();
+      const times = timeText.match(/\d{2}:\d{2}/g) ?? [];
+
+      const href = $el.find("a").first().attr("href") ?? null;
 
       events.push({
         venue_id: venue.id,
         title,
-        artist: $el.find(".artist").first().text().trim() || null,
+        artist: null,
         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: times[0] ?? null,
+        start_time: times[1] ?? null,
+        image_url: null,
+        source_url: href ? absoluteUrl(href, venue.url) : null,
       });
     });
 
@@ -61,18 +67,6 @@ 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}`;