summaryrefslogtreecommitdiff
path: root/app/scrapers/flat-nishiogikubo.ts
diff options
context:
space:
mode:
authoryyamashita <yyamashita@mosquit.one>2026-05-07 10:16:43 +0900
committeryyamashita <yyamashita@mosquit.one>2026-05-07 10:16:43 +0900
commit0cd5fb770ca9bd3f304d9556a4b33a4ad4f45e7e (patch)
treeaa316c21d7195689d87669338373d83b0b6ac3fb /app/scrapers/flat-nishiogikubo.ts
parent538fd636e25595d88a958344d285c0e7cf44e530 (diff)
Playwright scraping for FLAT/Pitbar; web UI display-only
- Install Playwright + Chromium; add shared browser singleton (playwright.server.ts) - Rewrite flat-nishiogikubo scraper: Wix calendar via headless browser, month navigation via date picker, extracts .WPczEB/.ExCBIq selectors - Rewrite pitbar-nishiogikubo scraper: freecalend.com via headless browser, parses cal-{member}-{year}-{month}-{day} cell IDs - scraper-runner: close shared browser after each run with closeBrowser() - Remove all scrape trigger buttons from web UI (events index, venues page); remove /api/scrape and /api/scrape-status from routes.ts Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Diffstat (limited to 'app/scrapers/flat-nishiogikubo.ts')
-rw-r--r--app/scrapers/flat-nishiogikubo.ts142
1 files changed, 130 insertions, 12 deletions
diff --git a/app/scrapers/flat-nishiogikubo.ts b/app/scrapers/flat-nishiogikubo.ts
index 03cc70c..da6752f 100644
--- a/app/scrapers/flat-nishiogikubo.ts
+++ b/app/scrapers/flat-nishiogikubo.ts
@@ -1,17 +1,20 @@
/**
* FLAT 西荻窪 — https://www.flat.rinky.info/schedule
*
- * ⚠️ Wix サイトのためクライアントサイド JS レンダリング。
- * 静的 fetch ではイベントデータを取得できない。
+ * Wix イベントカレンダー。JS レンダリングが必要なため Playwright を使用。
*
- * 代替案:
- * - Playwright/Puppeteer でヘッドレスブラウザを使用
- * - Wix Events API (要サイトオーナーによる API キー発行)
+ * DOM 構造:
+ * [data-hook="calendar-cell-<UTC ISO>"] ← 各日付セル
+ * .WPczEB → 開始時刻
+ * .ExCBIq → イベントタイトル
+ * aria-label が "イベントなし" のセルはスキップ
*
- * 現在は空配列を返す(エラーにはしない)。
+ * 月ナビ: calendar-date-picker-button を開いて datepicker-right-arrow で翌月へ。
*/
+import type { Page } from "playwright";
import type { Scraper, VenueMeta } from "./base";
import type { EventInput } from "~/lib/db.server";
+import { getBrowser } from "~/lib/playwright.server";
export const venue: VenueMeta = {
id: "flat-nishiogikubo",
@@ -20,14 +23,129 @@ export const venue: VenueMeta = {
area: "西荻窪",
};
+const SCHEDULE_URL = "https://www.flat.rinky.info/schedule";
+
+async function extractMonthEvents(page: Page): Promise<EventInput[]> {
+ const events: EventInput[] = [];
+ const cells = await page.locator('[data-hook^="calendar-cell-"]').all();
+
+ for (const cell of cells) {
+ const ariaLabel = (await cell.getAttribute("aria-label")) ?? "";
+ if (ariaLabel.includes("イベントなし")) continue;
+
+ const dataHook = (await cell.getAttribute("data-hook")) ?? "";
+ const isoStr = dataHook.replace("calendar-cell-", "");
+ if (!isoStr) continue;
+
+ // UTC timestamp → JST date (UTC+9)
+ const utcMs = new Date(isoStr).getTime();
+ if (isNaN(utcMs)) continue;
+ const jstDate = new Date(utcMs + 9 * 3_600_000).toISOString().slice(0, 10);
+
+ const timeLocs = cell.locator(".WPczEB");
+ const titleLocs = cell.locator(".ExCBIq");
+ const titleCount = await titleLocs.count();
+ const timeCount = await timeLocs.count();
+
+ for (let i = 0; i < titleCount; i++) {
+ const title = (await titleLocs.nth(i).innerText()).trim();
+ if (!title) continue;
+ const time = i < timeCount
+ ? (await timeLocs.nth(i).innerText()).trim()
+ : null;
+
+ events.push({
+ venue_id: venue.id,
+ title,
+ date: jstDate,
+ start_time: time || null,
+ source_url: SCHEDULE_URL,
+ });
+ }
+ }
+
+ return events;
+}
+
+async function navigateToMonth(page: Page, targetYYYYMM: string): Promise<void> {
+ const [targetYear, targetMonth] = targetYYYYMM.split("-").map(Number);
+
+ // Open the date picker
+ await page.click('[data-hook="calendar-date-picker-button"]');
+ await page.waitForTimeout(500);
+
+ // Click next-month arrow until we reach the target month
+ for (let attempt = 0; attempt < 6; attempt++) {
+ const monthText = await page.locator('[data-hook="datepicker-month-dropdown-button"]').innerText();
+ const yearText = await page.locator('[data-hook="datepicker-year-dropdown-button"]').innerText();
+
+ const currentYear = parseInt(yearText);
+ const months: Record<string, number> = {
+ "1月": 1, "2月": 2, "3月": 3, "4月": 4, "5月": 5, "6月": 6,
+ "7月": 7, "8月": 8, "9月": 9, "10月": 10, "11月": 11, "12月": 12,
+ };
+ const currentMonth = months[monthText.trim()] ?? 0;
+
+ if (currentYear === targetYear && currentMonth === targetMonth) break;
+
+ const diff = (targetYear * 12 + targetMonth) - (currentYear * 12 + currentMonth);
+ if (diff > 0) {
+ await page.click('[data-hook="datepicker-right-arrow"]');
+ } else {
+ await page.click('[data-hook="datepicker-left-arrow"]');
+ }
+ await page.waitForTimeout(300);
+ }
+
+ // Click any date in the mini-calendar that belongs to the target month
+ const allDays = await page.locator('[role="dialog"] button, [data-hook="datepicker-right-arrow"] ~ * button').all();
+ // Simpler: find a button with aria-label matching target year/month
+ const targetPrefix = `${targetYear}年${targetMonth}月`;
+ const dayBtns = await page.locator(`button[aria-label*="${targetPrefix}"]`).all();
+ if (dayBtns.length > 0) {
+ await dayBtns[0].click();
+ } else {
+ // Fallback: press Escape to close picker
+ await page.keyboard.press("Escape");
+ }
+ await page.waitForTimeout(2000);
+}
+
export const scraper: Scraper = {
venue,
async scrape(): Promise<EventInput[]> {
- // Wix renders events with JavaScript; static fetch returns an empty calendar.
- // TODO: Replace with a headless browser implementation (e.g. Playwright).
- throw new Error(
- "FLAT 西荻窪 は Wix サイトのため JS レンダリングが必要です。" +
- "ヘッドレスブラウザ(Playwright 等)への移行が必要です。"
- );
+ const browser = await getBrowser();
+ const page = await browser.newPage();
+
+ try {
+ await page.goto(SCHEDULE_URL, {
+ waitUntil: "domcontentloaded",
+ timeout: 30_000,
+ });
+ await page.waitForTimeout(5_000);
+
+ const events: EventInput[] = [];
+
+ // Current month events
+ events.push(...(await extractMonthEvents(page)));
+
+ // Navigate to next month for 35-day window coverage
+ const now = new Date();
+ const nextMonth = new Date(now.getFullYear(), now.getMonth() + 1, 1);
+ const nextYYYYMM = `${nextMonth.getFullYear()}-${String(nextMonth.getMonth() + 1).padStart(2, "0")}`;
+ await navigateToMonth(page, nextYYYYMM);
+ events.push(...(await extractMonthEvents(page)));
+
+ // Deduplicate by date + title
+ const seen = new Set<string>();
+ return events.filter((e) => {
+ const key = `${e.date}|${e.title}`;
+ if (seen.has(key)) return false;
+ seen.add(key);
+ return true;
+ });
+ } finally {
+ await page.close();
+ }
},
};