1 files changed, 99 insertions, 0 deletions
diff --git a/app/scrapers/warp-kichijoji.ts b/app/scrapers/warp-kichijoji.ts
new file mode 100644
index 0000000..8a828ea
--- /dev/null
+++ b/app/scrapers/warp-kichijoji.ts
@@ -0,0 +1,99 @@
+/**
+ * 吉祥寺 WARP — http://warp.rinky.info/schedules
+ *
+ * WordPress カスタムテーマ。年月は:
+ *   <h3>2026<br /><span>05</span></h3>
+ *
+ * イベント構造:
+ *   <article id="box-DD-ID" class="schedules-box">
+ *     <section class="date-box[-sun|-sat]">DD<span class="dayofweek">...</span></section>
+ *     <h4>タイトル</h4>
+ *     <section class="notes-wrapper">
+ *       <p>OPEN / START<br/><span class="strong">HH:MM / HH:MM</span></p>
+ *       <p>ADV / DOOR<br/><span class="strong">¥XXXX / ¥XXXX</span></p>
+ *     </section>
+ *     <section class="flyer"><img data-src="..." /></section>
+ *   </article>
+ */
+import * as cheerio from "cheerio";
+import type { Scraper, VenueMeta } from "./base";
+import type { EventInput } from "~/lib/db.server";
+
+export const venue: VenueMeta = {
+  id: "warp-kichijoji",
+  name: "吉祥寺 WARP",
+  url: "http://warp.rinky.info",
+  area: "吉祥寺",
+};
+
+export const scraper: Scraper = {
+  venue,
+  async scrape(): Promise<EventInput[]> {
+    const res = await fetch("http://warp.rinky.info/schedules");
+    if (!res.ok) throw new Error(`HTTP ${res.status}`);
+    const $ = cheerio.load(await res.text());
+    const events: EventInput[] = [];
+
+    // Extract year + month from <h3>2026<br/><span>05</span></h3>
+    const h3Text = $("h3").first().text().trim(); // e.g. "2026\n05"
+    const yearMonthMatch = h3Text.match(/(\d{4})\D*(\d{2})/);
+    if (!yearMonthMatch) return events;
+    const year = yearMonthMatch[1];
+    const month = yearMonthMatch[2];
+
+    $("article.schedules-box").each((_, el) => {
+      const $el = $(el);
+
+      // Day from article id: "box-03-23546" → "03"
+      const id = $el.attr("id") ?? "";
+      const dayMatch = id.match(/^box-(\d{2})-/);
+      if (!dayMatch) return;
+      const day = dayMatch[1];
+      const date = `${year}-${month}-${day}`;
+
+      const title = $el.find("h4").first().text().replace(/<br\s*\/?>/gi, " ").trim();
+      if (!title) return;
+
+      // First notes-wrapper <p> contains OPEN/START times
+      const $notes = $el.find("section.notes-wrapper p");
+      const timeStrong = $notes.eq(0).find("span.strong").text().trim();
+      // e.g. "18:30 / 19:00"
+      const [openTime, startTime] = timeStrong.split("/").map((s) => s.trim());
+
+      // Second <p> contains ADV/DOOR price
+      const priceStrong = $notes.eq(1).find("span.strong").text().trim();
+      // e.g. "¥3,000 / ¥3,500"
+      const price = priceStrong !== "TBA / TBA" && priceStrong ? priceStrong : null;
+
+      // Image: prefer data-src (lazy), fall back to noscript img src
+      const $flyer = $el.find("section.flyer img").first();
+      const rawImg =
+        $flyer.attr("data-src") ??
+        $el.find("section.flyer noscript img").first().attr("src") ??
+        null;
+      // Strip ShortPixel CDN prefix if present
+      const imageUrl = rawImg
+        ? rawImg.replace(/^https?:\/\/sp-ao\.shortpixel\.ai\/client\/[^/]+\//, "")
+        : null;
+
+      events.push({
+        venue_id: venue.id,
+        title,
+        artist: null,
+        date,
+        open_time: isTime(openTime) ? openTime : null,
+        start_time: isTime(startTime) ? startTime : null,
+        price,
+        ticket_url: $el.find("a[href*='livepocket'], a[href*='eplus'], a[href*='pia']").first().attr("href") ?? null,
+        image_url: imageUrl,
+        source_url: null,
+      });
+    });
+
+    return events;
+  },
+};
+
+function isTime(s: string | undefined): boolean {
+  return !!s && /^\d{2}:\d{2}$/.test(s.trim());
+}