config

type AnalyzersConfig struct {
	Enabled  []string `mapstructure:"enabled"`
	Disabled []string `mapstructure:"disabled"`
	// Specialized turns on opt-in checks that are off by default: the lower-confidence
	// AI-search heuristics (AEO direct-answer-lead, GEO quotable-density) and the WordPress
	// analyzer's active security-endpoint probes. They only fire when this is set.
	Specialized bool `mapstructure:"specialized"`
	// SecurityAudit turns on the `security` analyzer's opt-in audit pass: TLS protocol and
	// certificate inspection, Set-Cookie attribute hygiene, and the deeper response-header
	// policy checks. It is passive — it reads the crawl's own responses and opens no extra
	// connections — but its findings are infrastructure rather than SEO, so it stays off
	// unless asked for.
	SecurityAudit bool `mapstructure:"security_audit"`
}

type Config struct {
	Seed      string          `mapstructure:"seed"`
	Render    string          `mapstructure:"render"`
	Crawl     CrawlConfig     `mapstructure:"crawl"`
	Output    OutputConfig    `mapstructure:"output"`
	Analyzers AnalyzersConfig `mapstructure:"analyzers"`
	Store     StoreConfig     `mapstructure:"store"`
}

type CrawlConfig struct {
	MaxDepth      int     `mapstructure:"max_depth"`
	MaxPages      int     `mapstructure:"max_pages"`
	Concurrency   int     `mapstructure:"concurrency"`
	RatePerSecond float64 `mapstructure:"rate_per_second"`
	UserAgent     string  `mapstructure:"user_agent"`
	// UserAgents is an optional pool of User-Agent strings rotated across requests; it
	// supersedes UserAgent when set. UserAgentRotation is off, round-robin, or random.
	UserAgents        []string `mapstructure:"user_agents"`
	UserAgentRotation string   `mapstructure:"user_agent_rotation"`
	// Proxy is a single proxy URL; Proxies is a pool to rotate across (Proxy, if set, is
	// prepended to the pool). ProxyRotation is off, round-robin, random, or sticky-host.
	Proxy         string   `mapstructure:"proxy"`
	Proxies       []string `mapstructure:"proxies"`
	ProxyRotation string   `mapstructure:"proxy_rotation"`
	// BasicAuth is "user:pass" sent as an HTTP Basic Authorization header on every request to
	// the crawled host — for sites gated by server-level Basic Auth, which is common on
	// staging/acceptance environments (e.g. a reverse-proxy realm challenge in front of the
	// whole site, independent of any app-level login).
	BasicAuth string        `mapstructure:"basic_auth"`
	Timeout   time.Duration `mapstructure:"timeout"`
	// MaxDuration bounds the crawl's total wall-clock time (0 = unlimited). When it elapses,
	// the crawl stops early and still produces a report from whatever was fetched so far.
	MaxDuration     time.Duration `mapstructure:"max_duration"`
	MaxBodyBytes    int64         `mapstructure:"max_body_bytes"`
	RespectRobots   bool          `mapstructure:"respect_robots"`
	AllowSubdomains bool          `mapstructure:"allow_subdomains"`
	FollowExternal  bool          `mapstructure:"follow_external"`
	FollowNofollow  bool          `mapstructure:"follow_nofollow"`
	StripQuery      bool          `mapstructure:"strip_query"`
	Verbose         bool          `mapstructure:"verbose"`
	AdaptiveDelay   bool          `mapstructure:"adaptive_delay"`
	Include         []string      `mapstructure:"include"`
	Exclude         []string      `mapstructure:"exclude"`
}

type OutputConfig struct {
	Format string `mapstructure:"format"`
	Path   string `mapstructure:"path"`
	// SitemapPath, when set, writes a standard sitemap.xml of the crawled pages to that path as
	// a side output (independent of Format). The crawled site is also always rendered as a "Site
	// map" tab in the HTML report.
	SitemapPath string `mapstructure:"sitemap_path"`
}

type StoreConfig struct {
	// Dir is the store root; empty means store.DefaultRoot() (~/.gocrawl/crawls).
	Dir string `mapstructure:"dir"`
}