Schema Testing

Ready-to-adapt JSON-LD templates for the most common Schema.org types, plus a Python validator that extracts structured data from any URL or file, checks that it parses, and flags missing required and recommended properties.

What it does

Templates (templates/): clean, production-style JSON-LD examples using stable @ids, absolute URLs, ISO dates and ISO currencies, with GCC examples (AED, SAR, Arabic names, en-AE / ar-AE):

File	Type	Typical page
organization.json	Organization	Homepage / About
local-business.json	LocalBusiness	Branch or location page
product.json	Product + Offer	Product detail page
faq-page.json	FAQPage	Any page with a visible FAQ
breadcrumb-list.json	BreadcrumbList	All deep pages (Arabic example)
article.json	Article	Blog posts and guides
website-searchaction.json	WebSite + SearchAction	Homepage

Validator (validate_schema.py):

• Accepts a URL, a local HTML file or a raw .json / .jsonld file (several at once)
• Extracts every <script type="application/ld+json"> block
• Reports JSON syntax errors with line and column
• Walks @graph arrays and nested entities (for example the Offer inside a Product)
• Flags missing required properties (errors) and recommended properties (warnings) per type, with common subtypes such as BlogPosting → Article and Dentist → LocalBusiness
• Catches common value mistakes: non-ISO dates, currencies like "Dirham" instead of AED, prices with symbols ("AED 1,299"), relative URLs, breadcrumb positions out of order, and a SearchAction without {search_term_string}
• Exits with code 1 when errors are found, so it can run in CI

Why it matters

Structured data helps search engines and AI systems understand what a page is about and who is behind it. It also makes pages eligible for rich results such as product prices, breadcrumbs and article details. But:

• One misplaced comma makes the whole block invalid, and it fails silently.
• CMS plugins and developers often ship markup with missing fields, wrong currencies or relative URLs.
• Google's Rich Results Test checks one URL at a time. This script can check many pages or files in one run and fits into a deployment pipeline.

Use this validator for fast bulk checks, and Google's Rich Results Test plus the Schema.org validator for final confirmation.

Folder structure

schema-testing/
├── README.md
├── LICENSE
├── requirements.txt
├── validate_schema.py            # Extract + validate JSON-LD
├── examples/
│   └── sample-page.html          # Page with deliberate mistakes, for trying the validator
└── templates/
    ├── organization.json
    ├── local-business.json
    ├── product.json
    ├── faq-page.json
    ├── breadcrumb-list.json
    ├── article.json
    └── website-searchaction.json

How to use it

Requires Python 3.10+.

git clone https://github.com/haninbigleap-max/schema-testing.git
cd schema-testing
pip install -r requirements.txt

Validate a live page:

python validate_schema.py https://example.com/

Validate local files (HTML or JSON), including several at once:

python validate_schema.py examples/sample-page.html
python validate_schema.py templates/*.json --quiet

Save the findings as CSV:

python validate_schema.py https://example.com/ https://example.com/en-ae/blog/ --csv schema_report.csv

Options: --quiet hides passing checks, --csv FILE writes all findings, --timeout 20 sets the URL fetch timeout.

Using the templates

1. Copy the template that matches your page type.
2. Replace every example.com value with real data that is visible on the page.
3. Keep @id values stable and consistent across the site. For example, the organization is always https://example.com/#organization.
4. Delete properties you can't fill truthfully. Never add invented ratings or reviews.
5. Validate with validate_schema.py, then the Rich Results Test, before deploying.

Adding your own rules

Edit the RULES dictionary at the top of validate_schema.py. Each type has required, recommended and optional one_of lists. Map extra subtypes in SUBTYPES.

Example output

$ python validate_schema.py examples/sample-page.html

== examples/sample-page.html
  ! [WARNING] block 1 Product          $: missing recommended property 'image'
  ! [WARNING] block 1 Product          $: missing recommended property 'brand'
  ✗ [ERROR  ] block 1 Offer            $.offers: priceCurrency should be an ISO 4217 code (e.g. AED, SAR): 'Dirham'
  ✗ [ERROR  ] block 1 Offer            $.offers: price should be a plain number without symbols: 'AED 1,299'
  ✗ [ERROR  ] block 2 BreadcrumbList   $: ListItem positions should be 1..n in order, got [1, 3]
  ✓ [OK     ] block 2 ListItem         $.itemListElement[0]: all checked properties present
  ! [WARNING] block 2 ListItem         $.itemListElement[1]: 'item' should be an absolute URL: '/en-ae/air-purifiers/'
  ✗ [ERROR  ] block 3 -                $: JSON syntax error at line 5, column 3: Expecting property name enclosed in double quotes

Summary: 4 error(s), 8 warning(s) across 1 source(s).

(Output trimmed. The full run also lists the other missing recommended properties.)

All bundled templates pass:

$ python validate_schema.py templates/*.json --quiet

Summary: 0 error(s), 0 warning(s) across 7 source(s).

Limitations

• The rules are a practical subset of Schema.org and Google's documentation, not a full schema validator. Unknown types are listed but not checked.
• Pages that inject JSON-LD with JavaScript need to be rendered first. Save the rendered HTML from your browser or a headless crawler, then validate the file.
• Passing this validator doesn't guarantee rich results. Google decides eligibility per page and per feature.

License

MIT