> ## Documentation Index
> Fetch the complete documentation index at: https://docs.scrapeunblocker.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Parsed data extraction

> Get structured JSON instead of raw HTML. Extraction uses Schema.org, __NEXT_DATA__, or AI-generated rules.

For most scraping workloads you don't want raw HTML - you want clean JSON with the fields that matter. Pass `parsed_data=true` to `/getPageSource` and ScrapeUnblocker extracts structured data using the best available method for that page.

## Request

<CodeGroup>
  ```bash cURL theme={null}
  curl -X POST "https://api.scrapeunblocker.com/getPageSource?url=https://www.amazon.com/dp/B08N5WRWNW&parsed_data=true" \
    -H "x-scrapeunblocker-key: YOUR_API_KEY"
  ```

  ```python Python theme={null}
  import requests

  r = requests.post(
      "https://api.scrapeunblocker.com/getPageSource",
      params={
          "url": "https://www.amazon.com/dp/B08N5WRWNW",
          "parsed_data": True,
      },
      headers={"x-scrapeunblocker-key": "YOUR_API_KEY"},
      timeout=120,
  )
  payload = r.json()
  ```

  ```javascript Node.js theme={null}
  const res = await fetch(
    "https://api.scrapeunblocker.com/getPageSource?url=https://www.amazon.com/dp/B08N5WRWNW&parsed_data=true",
    {
      method: "POST",
      headers: { "x-scrapeunblocker-key": "YOUR_API_KEY" },
    }
  );
  const payload = await res.json();
  ```
</CodeGroup>

## Response shape

```json theme={null}
{
  "data": {
    "page_type": "product",
    "source": "schema_org",
    "data": {
      "title": "Echo Dot (4th Gen)",
      "price": "49.99",
      "currency": "USD",
      "brand": "Amazon",
      "availability": "InStock",
      "rating": 4.7,
      "review_count": 123456
    }
  }
}
```

### `page_type`

Detected category of the page. Common values:

* `product` - e-commerce product detail page
* `listing` - search results or category page
* `article` - news, blog, or editorial content
* `job` - job posting
* `real_estate` - property listing
* `social_profile` - a creator/brand profile on TikTok or Instagram
* `social_post` - a single TikTok video or Instagram post/reel
* `unknown` - extractor could not classify the page

### `source`

Which extraction strategy produced the data:

| Source       | What it means                                                                                           |
| ------------ | ------------------------------------------------------------------------------------------------------- |
| `schema_org` | The page exposed JSON-LD or microdata using [schema.org](https://schema.org) vocabulary. Most reliable. |
| `next_data`  | Extracted from a Next.js `__NEXT_DATA__` `<script>` block. Common on modern e-commerce.                 |
| `nuxt_data`  | Extracted from a Nuxt `__NUXT__` block.                                                                 |
| `og_meta`    | Fell back to OpenGraph / Twitter Card meta tags. Limited fields but always normalized.                  |
| `ai_rule`    | Custom selector rule generated by AI for this domain. Used when no structured data is available.        |

### `data`

The extracted fields. Schema depends on `page_type`. Field names are normalized across sources - a `product` always has `title` and `price` regardless of whether `source` is `schema_org` or `ai_rule`.

## Social pages: TikTok and Instagram

TikTok and Instagram build their pages in the browser from a JSON payload
embedded in the HTML, so there is very little in the markup to read. Both are
handled by dedicated extractors that read that payload directly, which means the
engagement numbers come back as **exact integers** rather than the rounded,
prose-formatted values the page eventually paints (`36K likes`).

**TikTok** - a video URL returns `page_type: "social_post"`:

```json theme={null}
{
  "page_type": "social_post",
  "data": {
    "video_id": "7665302162330766606",
    "description": "The next giant leap isn't a footprint.",
    "plays": 338217, "likes": 39511, "comments": 904,
    "shares": 2514, "saves": 4208, "reposts": 0,
    "posted_at": "2026-07-22T10:48:34+00:00",
    "duration_s": 56, "cover_url": "https://...", "hashtags": ["moon"],
    "music": { "id": "76653", "title": "original sound", "author": "NASA" },
    "author": {
      "username": "nasa", "nickname": "NASA", "verified": true,
      "followers": 421511, "likes": 1618595, "videos": 18
    }
  }
}
```

A profile URL returns `page_type: "social_profile"` with `username`, `nickname`,
`bio`, `verified`, `avatar_url`, `followers`, `following`, `likes` and `videos`.

**Instagram** - a post or reel URL returns `page_type: "social_post"` with
`shortcode`, `caption`, `likes`, `comments`, `posted_at`, `media_type`
(`image` / `video` / `carousel`), `thumbnail_url`, `video_url`, `hashtags` and an
`author` block. A profile URL returns `page_type: "social_profile"` with
`username`, `full_name`, `biography`, `followers`, `following`, `verified`,
`external_urls` and a `recent_posts` array.

<Note>
  Instagram does not include per-post engagement in the payload it serves on a
  **profile** page, so `recent_posts[].likes` and `.comments` are `null` there
  rather than guessed at. Each entry carries a post `url` - fetch that for the
  exact counts. TikTok profiles are the same: the profile carries account
  totals, and per-video numbers come from the video URL.

  Verified against live payloads: a profile timeline node carries only
  `code`, `caption`, `display_uri`, `media_type` and the author - Instagram
  serves engagement for exactly one media per page, the one the page is about.
  Both shapes also carry `counts_hidden`: on a post it is `true` when the
  creator has turned like counts off, which tells a `null` you can act on apart
  from one you cannot. On a profile grid it is `null` along with the counts,
  because Instagram sends neither. `posts_count` is `null` on a logged-out
  profile fetch for the same reason - the field is in the payload, sent empty.
</Note>

## When parsed data is the right choice

<Check>**Use it when** you're scraping a known page type at scale - products, articles, listings, jobs. Saves you from writing per-site parsers.</Check>
<Warning>**Skip it when** you need a field the extractor doesn't expose, or when you need raw HTML for downstream tooling. Fetch the HTML and parse it yourself instead.</Warning>

## Combining with `get_cookies`

You can set both `parsed_data=true` and `get_cookies=true` on the same request. The response gains a `cookies` field and a `proxy` field alongside `data`:

```json theme={null}
{
  "data": { "page_type": "product", "source": "schema_org", "data": { ... } },
  "cookies": [ { "name": "session", "value": "...", "domain": "..." } ],
  "proxy": "us"
}
```

See [cookies and sessions](/guides/cookies-and-sessions).