Read a page’s head metadata¶

A page’s <head> holds the facts a link preview or an index wants: the title, the description, the canonical URL, and the OpenGraph card. Read them off a parsed document with select_one() and attr(), and let structured_data() gather the OpenGraph block.

Read the head tags¶

select_one() finds each element by a CSS selector, and attr() reads an attribute (returning None when it is absent, so a missing tag never raises):

import turbohtml

doc = turbohtml.parse(
    "<html><head><title>Widget — Shop</title>"
    '<meta name="description" content="Buy the widget, ships today.">'
    '<link rel="canonical" href="https://shop.example/widget">'
    "</head><body><h1>Widget</h1></body></html>"
)
print(doc.select_one("title").text)
print(doc.select_one('meta[name="description"]').attr("content"))
print(doc.select_one('link[rel="canonical"]').attr("href"))

Widget — Shop
Buy the widget, ships today.
https://shop.example/widget

Read the OpenGraph card¶

The OpenGraph and Twitter-card tags come off structured_data(), which reads them into a dict in one walk rather than a selector per property, the metadata_parser and opengraph use case:

doc = turbohtml.parse(
    '<head><meta property="og:title" content="Widget">'
    '<meta property="og:image" content="https://shop.example/w.png">'
    '<meta name="twitter:card" content="summary"></head>'
)
card = doc.structured_data().opengraph
print(card["og:title"], card["og:image"])
print(card["twitter:card"])

Widget https://shop.example/w.png
summary

Read the title, the description, the canonical link, and the card in one parse, no per-field library. For the full set of embedded formats – JSON-LD, Microdata, RDFa – see Extract structured data.