PDF Extraction architecture

Maintainer overview of the extraction pipeline, OCR layers, and public endpoints.

Flow

1. Editor REST or WP-CLI schedules work through Action_Scheduler_Handler::schedule().
2. Extraction_Service resolves the PDF (uploads dir or remote download), calls OCR_Orchestrator::extract_text(), and saves a pdf_extraction child post.
3. OCR_Orchestrator tries providers by priority. Claude (priority 4) first; Gemini (priority 5) fallback. Each returns plain text, Markdown, and Gutenberg HTML.
4. Rewrite_Rules intercepts /{slug}/extraction and /{slug}/text early in parse_request. Markdown serving prefers prc-markdown-for-agents when active.
5. Content_Discovery adds alternate links, JSON-LD DigitalDocument, /llms.txt entries, and robots Allow rules.

Flow

Extraction pipeline

5 relationships

OCR layers

Layer	Namespace	Responsibility
Domain	OCR\Domain	Request/response types, exceptions
Application	OCR\Application	Orchestrator, validation, page analysis, merge
Infrastructure	OCR\Infrastructure	HTTP client, base64 encoder
Providers	OCR\Providers	Claude, Gemini, WP AI

Dependencies

• Required: prc-platform-core, Action Scheduler
• Optional: prc-markdown-for-agents (legacy Markdown template fallback when absent)
• Env: PRC_PLATFORM_ANTHROPIC_API_KEY (Claude), PRC_PLATFORM_GOOGLE_API_KEY (Gemini)

Key files

Path	Purpose
includes/class-bootstrap.php	Module wiring
includes/class-content-type.php	pdf_extraction CPT + meta
includes/class-extraction-service.php	Resolve PDF, OCR, save
includes/class-action-scheduler-handler.php	Async queue + complete/failed actions
includes/class-rest-api.php	Editor convert/status
includes/class-rewrite-rules.php	Public URL intercept
includes/class-content-discovery.php	Discovery metadata
includes/class-wp-cli-commands.php	Single-post CLI
includes/class-bulk-cli-command.php	Bulk process (dry-run default)

Related

• REST API
• Plugin README — hooks and CLI details