captchaforge 0.2.11

//! [`CaptchaProvider`] — bundles a captcha's detector with its
//! recommended solver routing in one self-contained unit.
//!
//! Adding a new captcha type to captchaforge previously meant editing
//! at least four places: the `DetectedCaptcha` enum, the
//! `parse_detection_result` mapper, the `DetectorRegistry`, and the
//! `detected_to_type` mapper in `solver/util.rs` — easy to forget one,
//! easy to disagree across the four. With [`CaptchaProvider`], a new
//! captcha is one struct with three trait impls (`Detector`,
//! `CaptchaProvider`) in one file.
//!
//! The provider does NOT own the [`Detector`] — it borrows it via
//! [`CaptchaProvider::detector`]. This means existing zero-size
//! detector structs (`TurnstileDetector` etc.) implement both traits
//! themselves; no wrapper allocation, no extra registry sync to keep.
//!
//! Backward compatibility: [`DetectorRegistry`] still works. A new
//! [`ProviderRegistry`] surface is added; consumers can use either,
//! and `ProviderRegistry::detectors()` returns the same set the legacy
//! registry would have produced.
use crate::captcha_detect::{
    AudioCaptchaDetector, CanvasCaptchaDetector, ChallengePageDetector, DetectedCaptcha, Detector,
    HCaptchaDetector, ImageCaptchaDetector, MultiStepCaptchaDetector, PowCaptchaDetector,
    RecaptchaDetector, ShadowDomDetector, SliderCaptchaDetector, TurnstileDetector,
};
use crate::solver::{CaptchaType, SolveMethod};

/// One captcha vendor / type, fully described.
///
/// New captchas implement this trait alongside [`Detector`] in a single
/// file under `src/detect/`. The bundled metadata lets the orchestrator
/// dispatch detection AND make an informed first guess about which
/// solver to try first — without depending on the [`PatternStore`]
/// having seen the domain before.
///
/// `recommended_solver_methods` is a HINT, not a constraint: the
/// solver chain may override the order based on per-domain learned
/// success rates from the pattern store.
pub trait CaptchaProvider: Send + Sync {
    /// Stable identifier — same value as [`Detector::name`] by
    /// convention so logs and reports correlate.
    fn name(&self) -> &'static str;

    /// The [`DetectedCaptcha`] variant this provider produces.
    fn detected_kind(&self) -> DetectedCaptcha;

    /// The [`CaptchaType`] used by the solver chain to look up
    /// per-type behaviour.
    fn captcha_type(&self) -> CaptchaType;

    /// Solver methods, ordered by recommended preference for this
    /// captcha. The chain consults this for cold-start routing; once
    /// the pattern store has data for the domain, learned ordering
    /// takes priority.
    ///
    /// Method-based routing collides when multiple solvers share a
    /// `SolveMethod` (Math, Pow, Slider, Behavioral all return
    /// `BehavioralBypass`) — the chain has to fall back to the
    /// first solver whose `supports()` returns true, which is
    /// fragile. New code SHOULD implement
    /// [`Self::recommended_solver_names`] instead; that path is
    /// name-based and avoids collisions. This method stays for
    /// backwards compatibility and is the fallback when
    /// `recommended_solver_names()` returns empty.
    fn recommended_solver_methods(&self) -> &'static [SolveMethod];

    /// Solver `name()`s, ordered by recommended preference. When
    /// non-empty, the chain prefers this list over
    /// [`Self::recommended_solver_methods`] — every name is matched
    /// against `solver.name()` exactly. Empty default means the
    /// chain falls back to method-based routing (the older path).
    ///
    /// Use this for vendors with dedicated solvers
    /// (`TurnstileInteractiveSolver`, `DataDomeSolver`, etc.) so the
    /// vendor's specific solver wins over the generic one even when
    /// both implement the same `SolveMethod`.
    fn recommended_solver_names(&self) -> &'static [&'static str] {
        &[]
    }

    /// Borrow the page-side detector this provider is bound to.
    fn detector(&self) -> &dyn Detector;
}

/// Registry of all known captcha providers. Replaces the standalone
/// [`DetectorRegistry`] for new code; existing code can still use the
/// detector-only registry.
///
/// The default-constructed registry contains every built-in provider
/// in priority order (matching the legacy detector registry exactly).
/// Custom providers can be appended via [`ProviderRegistry::push`].
pub struct ProviderRegistry {
    providers: Vec<Box<dyn CaptchaProvider>>,
}

impl ProviderRegistry {
    /// Construct a registry containing the built-in providers.
    pub fn new() -> Self {
        let mut providers: Vec<Box<dyn CaptchaProvider>> = vec![
            Box::new(ChallengePageDetector),
            Box::new(TurnstileDetector),
            Box::new(RecaptchaDetector),
            Box::new(HCaptchaDetector),
            Box::new(ImageCaptchaDetector),
            Box::new(AudioCaptchaDetector),
            Box::new(PowCaptchaDetector),
            Box::new(CanvasCaptchaDetector),
            Box::new(SliderCaptchaDetector),
            Box::new(MultiStepCaptchaDetector),
            Box::new(ShadowDomDetector),
        ];
        providers.sort_by_key(|p| p.detector().priority());
        Self { providers }
    }

    /// Built-in providers PLUS every rule from the bundled
    /// `community.toml` rule pack — DataDome, Akamai, PerimeterX/HUMAN,
    /// Arkose, Geetest v3+v4, AWS WAF, Friendly Captcha, MTCaptcha,
    /// WordPress math captchas, and any other rules added to the pack.
    ///
    /// The single-call entry point for production deployments that
    /// want maximum coverage out of the box.
    ///
    /// Returns `Err` if the bundled TOML somehow fails to parse —
    /// which only happens if the in-tree `community.toml` has been
    /// corrupted, since it is `include_str!`-ed at compile time.
    pub fn with_built_in_rules() -> anyhow::Result<Self> {
        use crate::detect::rules::{built_in_rules, RuleDetector};

        let mut reg = Self::new();
        let rules = built_in_rules()?;
        for rule in rules.providers {
            reg.providers.push(Box::new(RuleDetector::from(rule)));
        }
        reg.providers.sort_by_key(|p| p.detector().priority());
        Ok(reg)
    }

    /// Append a custom provider.
    pub fn push<P: CaptchaProvider + 'static>(&mut self, provider: P) {
        self.providers.push(Box::new(provider));
        self.providers.sort_by_key(|p| p.detector().priority());
    }

    /// Borrow the providers in priority order.
    pub fn providers(&self) -> &[Box<dyn CaptchaProvider>] {
        &self.providers
    }

    /// Look up a provider by its [`DetectedCaptcha`] kind. Returns
    /// `None` if no registered provider claims the kind. Used by the
    /// solver chain to find the recommended solver list for a freshly
    /// detected captcha without re-running detection.
    pub fn find_by_kind(&self, kind: &DetectedCaptcha) -> Option<&dyn CaptchaProvider> {
        self.providers
            .iter()
            .find(|p| p.detected_kind() == *kind)
            .map(|b| b.as_ref())
    }
}

impl Default for ProviderRegistry {
    fn default() -> Self {
        Self::new()
    }
}

// ─── Built-in CaptchaProvider impls ─────────────────────────────────────────
//
// Each existing detector picks up a CaptchaProvider impl here so the new
// trait is functional without touching every per-detector file. New
// captchas added in future PRs SHOULD put their CaptchaProvider impl in
// the same file as their Detector impl — the impls living here is a
// migration convenience, not a long-term pattern.

macro_rules! provider_impl {
    (
        $detector:ident,
        kind = $kind:ident,
        type = $captcha_type:expr,
        methods = [$($method:ident),+ $(,)?],
    ) => {
        impl CaptchaProvider for $detector {
            fn name(&self) -> &'static str {
                <Self as Detector>::name(self)
            }
            fn detected_kind(&self) -> DetectedCaptcha {
                DetectedCaptcha::$kind
            }
            fn captcha_type(&self) -> CaptchaType {
                $captcha_type
            }
            fn recommended_solver_methods(&self) -> &'static [SolveMethod] {
                &[$(SolveMethod::$method),+]
            }
            fn detector(&self) -> &dyn Detector {
                self
            }
        }
    };
}

provider_impl!(
    ChallengePageDetector,
    kind = Turnstile,
    type = CaptchaType::CloudflareTurnstile,
    methods = [BehavioralBypass, ThirdPartyService],
);
provider_impl!(
    TurnstileDetector,
    kind = Turnstile,
    type = CaptchaType::CloudflareTurnstile,
    methods = [BehavioralBypass, ThirdPartyService],
);
provider_impl!(
    RecaptchaDetector,
    kind = RecaptchaV2,
    type = CaptchaType::RecaptchaV2,
    methods = [BehavioralBypass, AudioBypass, VisionLLM, ThirdPartyService],
);
provider_impl!(
    HCaptchaDetector,
    kind = HCaptcha,
    type = CaptchaType::HCaptcha,
    methods = [VisionLLM, AudioBypass, ThirdPartyService],
);
provider_impl!(
    ImageCaptchaDetector,
    kind = ImageCaptcha,
    type = CaptchaType::ImageGrid,
    methods = [VisionLLM, ThirdPartyService],
);
provider_impl!(
    AudioCaptchaDetector,
    kind = AudioCaptcha,
    type = CaptchaType::AudioCaptcha,
    methods = [AudioBypass, ThirdPartyService],
);
provider_impl!(
    PowCaptchaDetector,
    kind = PowCaptcha,
    type = CaptchaType::PowCaptcha,
    methods = [BehavioralBypass],
);
provider_impl!(
    SliderCaptchaDetector,
    kind = SliderCaptcha,
    type = CaptchaType::Slider,
    methods = [BehavioralBypass, VisionLLM, ThirdPartyService],
);
provider_impl!(
    CanvasCaptchaDetector,
    kind = CanvasCaptcha,
    type = CaptchaType::CanvasCaptcha,
    methods = [VisionLLM, ThirdPartyService],
);
provider_impl!(
    ShadowDomDetector,
    kind = ShadowDomCaptcha,
    type = CaptchaType::ShadowDomCaptcha,
    methods = [VisionLLM, BehavioralBypass],
);
provider_impl!(
    MultiStepCaptchaDetector,
    kind = MultiStepCaptcha,
    type = CaptchaType::MultiStepCaptcha,
    methods = [BehavioralBypass, VisionLLM],
);

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn registry_contains_every_builtin_provider() {
        let reg = ProviderRegistry::new();
        let names: Vec<_> = reg.providers().iter().map(|p| p.name()).collect();
        for expected in [
            "challenge_page",
            "turnstile",
            "recaptcha",
            "hcaptcha",
            "image_captcha",
            "audio_captcha",
            "pow_captcha",
            "canvas_captcha",
            "slider_captcha",
            "multi_step_captcha",
            "shadow_dom",
        ] {
            assert!(names.contains(&expected), "missing provider: {expected}");
        }
    }

    #[test]
    fn registry_is_priority_ordered() {
        let reg = ProviderRegistry::new();
        let prios: Vec<i32> = reg
            .providers()
            .iter()
            .map(|p| p.detector().priority())
            .collect();
        let mut sorted = prios.clone();
        sorted.sort_unstable();
        assert_eq!(prios, sorted, "registry must be priority-ordered");
    }

    #[test]
    fn find_by_kind_locates_known_kinds() {
        let reg = ProviderRegistry::new();
        let p = reg
            .find_by_kind(&DetectedCaptcha::HCaptcha)
            .expect("hcaptcha provider");
        assert_eq!(p.name(), "hcaptcha");
        assert_eq!(p.captcha_type(), CaptchaType::HCaptcha);
    }

    #[test]
    fn find_by_kind_returns_none_for_unhandled_kind() {
        let reg = ProviderRegistry::new();
        assert!(reg.find_by_kind(&DetectedCaptcha::None).is_none());
    }

    #[test]
    fn turnstile_recommends_behavioral_first() {
        let reg = ProviderRegistry::new();
        let p = reg.find_by_kind(&DetectedCaptcha::Turnstile).unwrap();
        assert_eq!(
            p.recommended_solver_methods().first(),
            Some(&SolveMethod::BehavioralBypass),
            "Turnstile should try BehavioralBypass first; was: {:?}",
            p.recommended_solver_methods(),
        );
    }

    #[test]
    fn image_captcha_recommends_vision_llm_first() {
        let reg = ProviderRegistry::new();
        let p = reg.find_by_kind(&DetectedCaptcha::ImageCaptcha).unwrap();
        assert_eq!(
            p.recommended_solver_methods().first(),
            Some(&SolveMethod::VisionLLM),
            "ImageCaptcha should try VisionLLM first",
        );
    }

    #[test]
    fn with_built_in_rules_includes_built_in_providers() {
        let reg = ProviderRegistry::with_built_in_rules().expect("bundled rules parse");
        let names: Vec<_> = reg.providers().iter().map(|p| p.name()).collect();
        // Built-ins must still be present.
        for expected in ["turnstile", "hcaptcha", "recaptcha"] {
            assert!(names.contains(&expected), "lost built-in: {expected}");
        }
    }

    #[test]
    fn with_built_in_rules_includes_community_rules() {
        let reg = ProviderRegistry::with_built_in_rules().unwrap();
        let names: Vec<_> = reg.providers().iter().map(|p| p.name()).collect();
        for expected in [
            "datadome",
            "akamai_bot_manager",
            "perimeterx_human",
            "arkose_funcaptcha",
            "geetest_v3",
            "geetest_v4",
            "aws_waf_captcha",
            "friendly_captcha",
            "mtcaptcha",
            "wp_math_captcha",
            // Round-12 vendors
            "imperva_incapsula",
            "kasada",
            "f5_distributed_cloud",
            "yandex_smartcaptcha",
            "tencent_captcha",
            "keycaptcha",
        ] {
            assert!(
                names.contains(&expected),
                "missing community rule: {expected}"
            );
        }
    }

    #[test]
    fn with_built_in_rules_keeps_priority_order_across_built_in_and_rules() {
        let reg = ProviderRegistry::with_built_in_rules().unwrap();
        let prios: Vec<i32> = reg
            .providers()
            .iter()
            .map(|p| p.detector().priority())
            .collect();
        let mut sorted = prios.clone();
        sorted.sort_unstable();
        assert_eq!(prios, sorted, "merged registry must be priority-sorted");
    }

    #[test]
    fn community_rules_route_to_custom_captcha_type() {
        let reg = ProviderRegistry::with_built_in_rules().unwrap();
        let datadome = reg
            .providers()
            .iter()
            .find(|p| p.name() == "datadome")
            .expect("datadome registered");
        // TOML rules surface as Custom kind / type so downstream
        // routing can key on the vendor identifier.
        assert!(
            matches!(datadome.detected_kind(), DetectedCaptcha::Custom(ref s) if s == "datadome"),
            "datadome.detected_kind should be Custom(datadome)",
        );
        assert!(
            matches!(datadome.captcha_type(), CaptchaType::Custom(ref s) if s == "datadome"),
            "datadome.captcha_type should be Custom(datadome)",
        );
    }

    #[test]
    fn custom_provider_can_be_pushed_and_sorted() {
        struct DummyDetector;
        #[async_trait::async_trait]
        impl Detector for DummyDetector {
            fn name(&self) -> &'static str {
                "dummy"
            }
            fn priority(&self) -> i32 {
                1
            }
            async fn detect(
                &self,
                _page: &chromiumoxide::Page,
            ) -> anyhow::Result<Option<crate::captcha_detect::CaptchaInfo>> {
                Ok(None)
            }
        }
        impl CaptchaProvider for DummyDetector {
            fn name(&self) -> &'static str {
                "dummy"
            }
            fn detected_kind(&self) -> DetectedCaptcha {
                DetectedCaptcha::None
            }
            fn captcha_type(&self) -> CaptchaType {
                CaptchaType::Custom("dummy".into())
            }
            fn recommended_solver_methods(&self) -> &'static [SolveMethod] {
                &[SolveMethod::BehavioralBypass]
            }
            fn detector(&self) -> &dyn Detector {
                self
            }
        }

        let mut reg = ProviderRegistry::new();
        reg.push(DummyDetector);
        // priority(1) is below all built-ins (lowest is challenge_page=5),
        // so the custom provider should sort to the front.
        assert_eq!(reg.providers().first().unwrap().name(), "dummy");
    }
}