# Vertical manifest v1 — `kind` selects the runtime in src/extract/vertical/.
version: 1
order: 16
name: deepwiki
# text-extract: fetch request URL as text → clean → fields.
kind: text-extract
description: DeepWiki repository documentation overview.
urlPatterns:
  - https://deepwiki.com/:owner/:repo
  - https://deepwiki.com/:owner/:repo/:page*
# Runtime requirements for the scrape host.
requirements:
  requiresBrowser: true
  requiresLLM: false
  requiresCloud: false
# Declared output facets (discovery / tooling).
capabilities:
  - wiki_overview
  - sections
  - source_files
source: builtin
# HTTP request (api-json / api-xml / text-extract). {{captures}} interpolate into urlTemplate.
request:
  urlTemplate: https://deepwiki.com/{{owner|encodeURIComponent}}/{{repo|encodeURIComponent}}
# Text cleanup applied before field rules (stripTags, collapseWhitespace, …).
clean:
  stripTags: true
  collapseWhitespace: true
# Field rules on cleaned page text (regex, tokens, jsonWalk, sectionList, …).
fields:
    owner:
      value: "{{owner}}"
    repo:
      value: "{{repo}}"
    lastIndexed:
      regex: 'Last indexed:\s*([^()]+?)\s*\('
      group: 1
      transforms: [trim]
    commit:
      regex: 'Last indexed:\s*[^)]*\(\s*([a-f0-9]+)\s*\)'
      group: 1
    sections:
      sectionList:
        afterRegex: 'Last indexed:[^)]*\)\s*'
        before: Glossary
        minLength: 3
        knownPhrases:
          - Repository Structure
          - Development Workflow
          - System Architecture
          - Core Components
          - API Reference
          - Feature Flags
          - Key Concepts
          - Key Features
          - Build System
          - Data Flow
          - Architecture
          - Configuration
          - Deployment
          - Implementation
          - Infrastructure
          - Authentication
          - Components
          - Database
          - Examples
          - Frontend
          - Backend
          - Overview
          - Packages
          - Testing
        rejectPrefixes:
          - Menu
          - Loading
          - Devin
          - Edit Wiki
          - Share
          - Index
          - DeepWiki
    activeSection:
      regex: '(?:^|\s)Menu\s+(.+?)(?=\s+Relevant source files|\s*$)'
      group: 1
      transforms: [cleanSection]
    sourceFiles:
      tokens:
        after: Relevant source files
        while: sourcePath
    githubUrl:
      value: https://github.com/{{owner}}/{{repo}}