---
id: "09721f1c-2a81-4d26-a2e7-ba9f176f2b74"
name: "Iframe Content Extraction"
micro_summary: "Extracts McGraw-Hill question content through the dynamically named iframe's contentWindow.document; iframe_target() leaves JS in the parent context."
status: "not_examined"
see_also: []
scope: "How to extract content from nested iframes in McGraw-Hill Connect and SIMbook pages. The iframe_target() helper doesn't change CDP execution context, so you must reach into iframes from the parent page via contentWindow.document."
warnings: "iframe name attribute is dynamically generated (e.g., Q_13252744676035422_ansb) — never hardcode it. Find via js('document.querySelector(\\\"iframe\\\").name') or search by src pattern."
---

## The Problem

McGraw-Hill Connect and SIMnet load actual question content inside nested iframes. The parent page shows only the shell (nav, buttons, progress bar). `extract_text()` and `js()` run in the parent CDP context — they never see iframe content.

## The Working Pattern

Use parent-page JS to reach into the iframe via `contentWindow.document`:

**Step 1: Find the iframe name/selector**
```python
js('document.querySelector(\"iframe\").name')
# → Q_13252744676035422_ansb (dynamic, changes per assignment)
```

**Step 2: Extract content in chunks from parent context**
```python
js('document.querySelector(\"iframe[name=\\\"<IFRAME_NAME>\\\"]\")?.contentWindow?.document?.body?.innerHTML.substring(0, 5000)')
js('document.querySelector(\"iframe[name=\\\"<IFRAME_NAME>\\\"]\")?.contentWindow?.document?.body?.innerHTML.substring(5000, 12000)')
```

**Reusable template:**
```python
js('document.querySelector(\"iframe[name=\\\"<IFRAME_NAME>\\\"]\")?.contentWindow?.document?.body?.innerHTML.substring(<START>, <END>)')
```

Chunk with increasing ranges until you have all the content. Parse the HTML for tables, dropdowns (`role=combobox`), and question text.

## What Does NOT Work

- `iframe_target()` — switches visual focus but NOT CDP execution context. `js()` and `extract_text()` still run on the parent page.
- `document.body.innerHTML` on iframe from Python — same-origin policy may block direct access, but `contentWindow.document` works because CDP executes in the same browsing context.

## Finding Iframes Dynamically

The iframe `name` attribute is dynamically generated (e.g., `Q_13252744676035422_ansb`). To find it:

```python
js('document.querySelector(\"iframe\").name')  # first iframe
# or search by src pattern:
js('document.querySelector(\"iframe[src*=\\\"accountingtool\\\"]\").name')
```