MAGIK-935: Website URL Provider — Meta/OG/JSON-LD Extraction

[MAGIKBIT]

MAGIK-935: Website URL Provider — Meta/OG/JSON-LD Extraction

Epic: EPIC-025 — #113
Priority: P0
Estimate: 5 SP
Depends on: MAGIK-934

Description

Create WebsiteProvider that crawls a given website URL and extracts profile-relevant data from HTML meta tags, Open Graph tags, JSON-LD structured data, schema.org markup, and visible page content.

Implementation

Class: Libraries/Enrichment/WebsiteProvider.php

URL matching: Any valid HTTP/HTTPS URL not matched by social-specific providers.

Extraction layers (in priority order):

1. JSON-LD / schema.org — <script type="application/ld+json"> for Organization, LocalBusiness, Person
2. Open Graph tags — og:title, og:description, og:image, og:url, og:site_name
3. HTML meta tags — <meta name="description">, <meta name="author">, <link rel="icon">
4. Visible content heuristics — regex for emails (mailto:), phone patterns, address blocks
5. Social link discovery — <a href> matching known social platform URL patterns

Extracted fields:

Field	Source	Confidence
Company/Site name	JSON-LD > OG > <title>	0.9 / 0.8 / 0.6
Description	JSON-LD > OG > meta description	0.9 / 0.8 / 0.7
Logo/Favicon	JSON-LD logo > OG image > <link rel="icon">	0.9 / 0.7 / 0.5
Emails	JSON-LD > mailto: links > regex	0.9 / 0.8 / 0.5
Phones	JSON-LD > tel: links > regex	0.9 / 0.8 / 0.4
Address	JSON-LD PostalAddress > address block regex	0.9 / 0.3
Social links	<a> href matching fb/ig/yt/tw/li patterns	0.8

HTTP client: CodeIgniter's CURLRequest with 10s timeout, User-Agent: MagikTap-Enrichment/1.0, robots.txt check.

Files

File	Action
Libraries/Enrichment/WebsiteProvider.php	Create
Libraries/Enrichment/HtmlExtractor.php	Create (shared HTML parsing utility)
Libraries/EnrichmentService.php	Modify (register provider)
Config/Enrichment.php	Modify (add website config)

Acceptance Criteria

• canHandleUrl() matches any valid HTTP/HTTPS URL (fallback provider)
• Extracts name, description, logo from OG tags
• Extracts structured data from JSON-LD when available
• Discovers email/phone from visible content
• Discovers social media links from page anchors
• Each field includes confidence score and source evidence
• Respects robots.txt and 10s timeout
• Returns graceful error on unreachable/blocked sites
• Unit tests with fixture HTML files