JP Article Extractor — Japanese Article Metadata Extraction

Extract structured metadata from Japanese article URLs (news sites, blogs, note, Zenn, and more). Built specifically for Japanese-language content, including normalization of Japanese date formats.

日本語記事（ニュース / ブログ / note / Zenn など）の URL リストから、メタデータを構造化 JSON で抽出します。和文日付表記の正規化に対応した、日本語コンテンツ特化の Actor です。

What it does

Give it a list of URLs, and for each page it returns:

Field	Description
title	Article title (og:title → JSON-LD → <title> fallback chain)
description	Meta description
og_image	Social share image URL
published_at	Publication date, normalized to ISO format — including Japanese formats like 2026年7月3日 → 2026-07-03
excerpt	First ~200 characters of body text

Failed URLs return {"url": "...", "error": "..."} without stopping the batch.

Why Japanese-specific?

Generic article extractors often fail on Japanese sites:

• Japanese date formats (2026年7月3日, 2026/7/3 etc.) are normalized to ISO YYYY-MM-DD
• Character encoding detection handles UTF-8 and Shift_JIS (still common on Japanese sites) automatically
• Multi-stage fallback (Open Graph → JSON-LD → meta tags → title tag) tuned against Japanese news sites, blogs, note and Zenn

Input

{
  "urls": [
    "https://example.com/article-1",
    "https://example.com/article-2"
  ]
}

Output (dataset, one record per URL)

{
  "url": "https://note.com/x/n/1",
  "title": "記事タイトル",
  "description": "ディスクリプション",
  "og_image": "https://example.com/og.png",
  "published_at": "2026-07-01",
  "excerpt": "本文冒頭200字..."
}

Use cases

• Media monitoring dashboards for Japanese news and blogs
• Enriching link lists with titles, dates and thumbnails (newsletters, curation sites)
• Feeding Japanese article metadata into LLM / RAG pipelines
• Content research on note / Zenn / Japanese tech blogs

Works well with

Japanese news sites, personal and corporate blogs (WordPress, はてなブログ), note.com, Zenn, and most pages that expose Open Graph, JSON-LD, or standard meta tags. It is a lightweight alternative to a full Japanese web scraper when you only need article metadata — no headless browser, no proxies required for most public pages.

Pricing example

Pay-per-event pricing, charged only for what you extract:

• 100 articles ≈ $0.50
• 1,000 articles ≈ $5.00

(Plus a small Actor start fee and Apify platform usage — kept low by the 1 GB default memory.)

FAQ

Does it work with sites other than news/blogs? Any page with standard metadata (OG tags, JSON-LD, meta description) works — e.g. corporate press releases or product announcement pages in Japanese.

How is this different from generic article extractors? Japanese date normalization (2026年7月3日 → 2026-07-03), Shift_JIS support, and fallback order tuned for Japanese sites (note, Zenn, national news media). Generic extractors often return empty dates or garbled text on these.

Can I use it from the API / integrate it into a pipeline? Yes — call it via the Apify API or scheduled runs, and read results from the dataset as JSON/CSV. Works as a building block for media monitoring, newsletters, and LLM/RAG pipelines.

日本語で使えますか？ はい。入力は URL のリストだけです。日本語記事のタイトル・説明・OGP画像・公開日（ISO 形式に正規化）・冒頭抜粋を JSON で返します。

Notes

• Lightweight and fast: standard-library HTTP fetching, no headless browser
• Runs at 1 GB memory by default to keep your platform usage cost low
• Pages that block simple HTTP clients may return an error record