feat: structured-data extraction (JSON-LD, Microdata, OpenGraph, RDFa)

[gaborbernat]

Gap

turbohtml extracts no structured data. Scrapers want JSON-LD, Microdata, OpenGraph and Twitter meta tags, and RDFa, and reach for extruct, metadata_parser, or hand-rolled walks. This is the highest-value capability turbohtml lacks for the scraping audience.

Proposed API

data = document.structured_data()
# {"json_ld": [...], "microdata": [...], "opengraph": {...}, "microformats": [...], "rdfa": [...]}

Pull each format from the tree:

• JSON-LD: every <script type="application/ld+json">, parsed with stdlib json.
• Microdata: itemscope / itemprop / itemtype walked into nested dicts.
• OpenGraph and Twitter: <meta property="og:*"> and <meta name="twitter:*">.
• RDFa and microformats2 in a later phase.

The tree walk runs in C; JSON parsing stays in the Python facade. Per-format helpers (document.json_ld(), document.opengraph()) sit beside the combined call.

Notes

Unlocks an extruct migration section. JSON-LD and OpenGraph cover most real use, Microdata next, RDFa and microformats2 last. Top of the new-feature list. Severity: high.

[gaborbernat]

Working on this.