Schema & Structured Data

Schema Markup Extractor

Extract inline JSON-LD from pasted HTML, inspect types and parsing errors, and copy individual or combined schemas.

Browser processing · No account required

Page source

Up to 1 MiB HTML and 100 JSON-LD scripts. The detached HTML document and its scripts are never mounted or executed.

Pasted source only; external scripts, remote contexts and linked pages are not fetched. Microdata and RDFa are outside this extractor.

Extracted schemas

Add your input, then run the tool to see results.
WORK WITH CONFIDENCE

Understand your results

For maintainers reviewing the JSON-LD actually authored in a page's HTML source.

The browser parses supplied source inside inert template content in a detached document, locating application/ld+json scripts without mounting the HTML, loading its resources or executing code. Each inline block is parsed independently: one malformed block does not hide other results. Type labels include explicit nested and graph-node @type values. Counts distinguish script blocks from inspected objects; a block can contain several typed entities.

How to use this tool

  1. Paste the relevant HTML source, including inline JSON-LD script elements.
  2. Run Extract schemas and compare the total script count with parsed blocks and per-block errors.
  3. Copy an individual parsed block, or copy all parsed blocks as a JSON array. Review excluded failures before reusing combined output.

What the observations mean

The script count describes source elements, not unique entities or duplicate detection. Type summaries collect explicit type strings from nested objects; aliases hidden in a context are not expanded.

Worked example

Extract two inline blocks without executing the surrounding page.

<script type="application/ld+json">{"@context":"https://schema.org","@type":"Organization","name":"Example"}</script>
<script type="application/ld+json">{"@context":"https://schema.org","@type":"Article","headline":"Example guide"}</script>

Two scripts are found and parsed independently, exposing Organization and Article. Copy all combines the two objects into one array; an invalid third block would be reported and excluded from that parsed output.

Limits and processing

Browser-side inert HTML parsing only: 1 MiB HTML, 100 scripts, and the validator's 256 KiB/depth/value bounds per block. No script execution, resource/context requests, remote page fetching, Microdata/RDFa parsing, entity deduplication or complete vocabulary validation.

Common problems and practical fixes

No scripts are found
Include the JSON-LD script elements in the pasted source. A page's runtime-generated DOM may differ from its initial HTML.
One block fails while others parse
Correct that block independently. Copy all intentionally excludes unparseable or external blocks and does not silently repair them.
Unexpected object or type counts
Inspect arrays, graph containers and nested entities. Several objects can describe a single real-world entity, and this extractor does not deduplicate them.

Best practices

Compare the extraction with your source and deployed markup. Never publish copied third-party identities, reviews or business claims as your own.

Practical tips

  • View Source and the browser's rendered DOM can differ. This tool only sees the pasted sample.
  • External src scripts are reported but not fetched, even if they also contain inline text.
  • HTML parsing can recover malformed markup. Recovery does not prove that the original HTML is valid.
A LITTLE MORE CLARITY

Frequently asked questions

Practical answers for using the Schema Markup Extractor.

Does pasted HTML execute?

No. The parsed document stays detached; scripts, event handlers and images are not mounted into the page. Only extracted text is rendered safely.

What does Copy all contain?

A JSON array with each successfully parsed inline block as one entry. Existing arrays and @graph shapes are preserved. Failed/external blocks are excluded and remain visibly reported.

Can it find Microdata or JavaScript-injected schemas?

It extracts JSON-LD script text only. It does not read Microdata/RDFa, run JavaScript or visit a live page. Injected blocks must already be present in the pasted HTML.