captchaforge 0.2.8

//! Tier-A operational configuration loaded from `.captchaforge.toml`.
//!
//! Per the CLAUDE.md two-tier configurability rule, captchaforge has
//! two distinct config layers:
//!
//! - **Tier A — operational config (this module)**: timeouts,
//!   concurrency, cache TTL, VLM endpoint/model, third-party service
//!   selection. CLI flags + TOML defaults; CLI wins. Compiled
//!   defaults → `captchaforge.toml` → CLI flags.
//! - **Tier B — community knowledge (`detect::rules`)**: vendor
//!   detection rules, never CLI-flagged. Lives in TOML data files.
//!
//! This module owns Tier A. The shape mirrors the runtime types so
//! callers materialise a config once and apply it to chain / solver /
//! cache constructors.
//!
//! # Search path
//!
//! [`Config::discover`] looks in this order, returning the first hit:
//! 1. `$CAPTCHAFORGE_CONFIG` (explicit path; if unset is skipped)
//! 2. `./.captchaforge.toml` (project-local)
//! 3. `./captchaforge.toml`
//! 4. `$XDG_CONFIG_HOME/captchaforge/config.toml`
//!    (fallback `~/.config/captchaforge/config.toml`)
//!
//! Missing-file is NOT an error from `discover` — it returns
//! [`Config::default`]. Use [`Config::load_from_path`] when you want
//! the strict "must exist" behaviour.
//!
//! # Example file
//!
//! ```toml
//! # Per-solver timeout for the chain. Default 180_000.
//! per_solver_timeout_ms = 60000
//! # Whether to capture a screenshot on full chain failure. Default true.
//! screenshot_on_failure = true
//!
//! [cache]
//! # TTL in seconds for cached solved tokens. Default 60.
//! ttl_seconds = 90
//!
//! [vlm]
//! # Override Ollama endpoint. Beats the CAPTCHAFORGE_VLM_ENDPOINT env var.
//! endpoint = "http://gpu-host:11434"
//! model    = "qwen3-vl:30b"
//!
//! [third_party]
//! # Pick one: "two_captcha" | "cap_monster" | "cap_solver" | "custom"
//! service  = "two_captcha"
//! # Required when service = "custom".
//! base_url = "https://my-gateway.example/2captcha"
//! # API key (or leave blank and use CAPTCHAFORGE_THIRDPARTY_API_KEY env var).
//! api_key  = ""
//! poll_interval_ms = 5000
//! max_polls = 30
//!
//! [solve]
//! # All SolveConfig fields are overridable individually.
//! checkbox_max_attempts = 20
//! token_max_attempts    = 24
//! ```

use serde::{Deserialize, Serialize};
use std::path::{Path, PathBuf};

use std::sync::Arc;

use crate::provider::ProviderRegistry;
use crate::solver::{
    AudioCaptchaSolver, BehavioralCaptchaSolver, CaptchaSolverChain, ChainConfig,
    MathCaptchaSolver, PowCaptchaSolver, SliderCaptchaSolver, SolveConfig, ThirdPartyCaptchaSolver,
    ThirdPartyService, TokenCache, VlmCaptchaSolver, WaitForTokenSolver,
};

const DEFAULT_CACHE_TTL_SECONDS: u64 = 60;

const ENV_CONFIG_PATH: &str = "CAPTCHAFORGE_CONFIG";

/// Top-level Tier-A config. All fields are optional; absent fields
/// fall back to compiled defaults.
#[derive(Debug, Clone, Default, Serialize, Deserialize)]
#[serde(default, deny_unknown_fields)]
pub struct Config {
    pub per_solver_timeout_ms: Option<u64>,
    pub screenshot_on_failure: Option<bool>,
    pub cache: CacheConfig,
    pub vlm: VlmConfig,
    pub third_party: ThirdPartyConfig,
    pub solve: SolveOverrides,
}

#[derive(Debug, Clone, Default, Serialize, Deserialize)]
#[serde(default, deny_unknown_fields)]
pub struct CacheConfig {
    pub ttl_seconds: Option<u64>,
}

#[derive(Debug, Clone, Default, Serialize, Deserialize)]
#[serde(default, deny_unknown_fields)]
pub struct VlmConfig {
    pub endpoint: Option<String>,
    pub model: Option<String>,
}

#[derive(Debug, Clone, Default, Serialize, Deserialize)]
#[serde(default, deny_unknown_fields)]
pub struct ThirdPartyConfig {
    /// One of `two_captcha`, `cap_monster`, `cap_solver`, `custom`.
    pub service: Option<String>,
    /// Required when `service = "custom"`. Ignored otherwise.
    pub base_url: Option<String>,
    /// API key. Leave None to fall back to the
    /// `CAPTCHAFORGE_THIRDPARTY_API_KEY` env var.
    pub api_key: Option<String>,
    pub poll_interval_ms: Option<u64>,
    pub max_polls: Option<u32>,
}

#[derive(Debug, Clone, Default, Serialize, Deserialize)]
#[serde(default, deny_unknown_fields)]
pub struct SolveOverrides {
    pub checkbox_poll_interval_ms: Option<u64>,
    pub checkbox_max_attempts: Option<u32>,
    pub token_poll_interval_ms: Option<u64>,
    pub token_max_attempts: Option<u32>,
    pub audio_button_delay_ms: Option<u64>,
    pub audio_submit_delay_ms: Option<u64>,
    pub vlm_http_timeout_ms: Option<u64>,
    pub client_http_timeout_ms: Option<u64>,
}

impl Config {
    /// Walk the search path and return the first config found, or
    /// `Config::default()` if none exists. Errors only on a malformed
    /// TOML in a found file.
    pub fn discover() -> anyhow::Result<Self> {
        for path in Self::search_paths() {
            if path.is_file() {
                return Self::load_from_path(&path);
            }
        }
        Ok(Self::default())
    }

    /// The ordered candidate paths Self::discover walks. Public so
    /// CLIs can show users where the file is expected to live.
    pub fn search_paths() -> Vec<PathBuf> {
        Self::search_paths_with_env(|n| std::env::var(n).ok())
    }

    /// Like [`Self::search_paths`] but with an injectable env lookup
    /// so unit tests don't have to mutate process env.
    pub(crate) fn search_paths_with_env<F>(env: F) -> Vec<PathBuf>
    where
        F: Fn(&str) -> Option<String>,
    {
        let mut out = Vec::new();
        if let Some(p) = env(ENV_CONFIG_PATH).filter(|s| !s.is_empty()) {
            out.push(PathBuf::from(p));
        }
        out.push(PathBuf::from(".captchaforge.toml"));
        out.push(PathBuf::from("captchaforge.toml"));
        let xdg = env("XDG_CONFIG_HOME")
            .filter(|s| !s.is_empty())
            .map(PathBuf::from)
            .or_else(|| env("HOME").map(|h| PathBuf::from(h).join(".config")));
        if let Some(base) = xdg {
            out.push(base.join("captchaforge").join("config.toml"));
        }
        out
    }

    /// Read + parse a TOML file at `path`. Errors propagate so callers
    /// can distinguish "no config" from "bad config."
    ///
    /// Emits a `tracing::warn!` if the loaded file commits a
    /// third-party API key inline — those keys belong in
    /// `CAPTCHAFORGE_THIRDPARTY_API_KEY` (env var) or a secrets
    /// store, not in a file that gets committed alongside source.
    pub fn load_from_path(path: impl AsRef<Path>) -> anyhow::Result<Self> {
        let path = path.as_ref();
        let body = std::fs::read_to_string(path)
            .map_err(|e| anyhow::anyhow!("reading {}: {e}", path.display()))?;
        let cfg = Self::from_toml_str(&body)
            .map_err(|e| anyhow::anyhow!("parsing {}: {e}", path.display()))?;
        cfg.warn_on_inline_secrets(path);
        Ok(cfg)
    }

    /// Print a `tracing::warn!` for any plaintext secret present in a
    /// loaded config file. Called by [`Self::load_from_path`].
    /// Currently flags `[third_party] api_key`; kept as a separate
    /// method so future secrets gain warnings here without touching
    /// the load-path itself.
    pub fn warn_on_inline_secrets(&self, source: &Path) {
        if self
            .third_party
            .api_key
            .as_deref()
            .is_some_and(|k| !k.is_empty())
        {
            tracing::warn!(
                config = %source.display(),
                "captchaforge: third_party.api_key is set inline — \
                 prefer the CAPTCHAFORGE_THIRDPARTY_API_KEY env var \
                 (or a secrets store) so the key doesn't get committed \
                 to git alongside the config file"
            );
        }
    }

    /// Parse from a TOML string (no I/O — useful in tests).
    pub fn from_toml_str(s: &str) -> Result<Self, toml::de::Error> {
        toml::from_str(s)
    }

    /// Materialise a [`ChainConfig`] from this config, defaulting any
    /// absent field.
    pub fn chain_config(&self) -> ChainConfig {
        let mut c = ChainConfig::default();
        if let Some(v) = self.per_solver_timeout_ms {
            c.per_solver_timeout_ms = v;
        }
        if let Some(v) = self.screenshot_on_failure {
            c.screenshot_on_failure = v;
        }
        c
    }

    /// Materialise a [`SolveConfig`] from this config, defaulting any
    /// absent field.
    pub fn solve_config(&self) -> SolveConfig {
        let mut c = SolveConfig::default();
        let s = &self.solve;
        if let Some(v) = s.checkbox_poll_interval_ms {
            c.checkbox_poll_interval_ms = v;
        }
        if let Some(v) = s.checkbox_max_attempts {
            c.checkbox_max_attempts = v;
        }
        if let Some(v) = s.token_poll_interval_ms {
            c.token_poll_interval_ms = v;
        }
        if let Some(v) = s.token_max_attempts {
            c.token_max_attempts = v;
        }
        if let Some(v) = s.audio_button_delay_ms {
            c.audio_button_delay_ms = v;
        }
        if let Some(v) = s.audio_submit_delay_ms {
            c.audio_submit_delay_ms = v;
        }
        if let Some(v) = s.vlm_http_timeout_ms {
            c.vlm_http_timeout_ms = v;
        }
        if let Some(v) = s.client_http_timeout_ms {
            c.client_http_timeout_ms = v;
        }
        c
    }

    /// Build a [`TokenCache`] honouring the configured TTL (default 60s).
    pub fn build_token_cache(&self) -> TokenCache {
        let ttl = self.cache.ttl_seconds.unwrap_or(DEFAULT_CACHE_TTL_SECONDS);
        TokenCache::with_ttl(std::time::Duration::from_secs(ttl))
    }

    /// Build a [`VlmCaptchaSolver`] with TOML overrides applied. TOML
    /// values beat env vars (which beat defaults).
    pub fn build_vlm_solver(&self) -> VlmCaptchaSolver {
        let mut s = VlmCaptchaSolver::new();
        if let Some(ep) = &self.vlm.endpoint {
            s = s.with_endpoint(ep.clone());
        }
        if let Some(m) = &self.vlm.model {
            s = s.with_model(m.clone());
        }
        s.with_config(self.solve_config())
    }

    /// Build a [`ThirdPartyCaptchaSolver`] from the config. Returns
    /// `None` when the [`ThirdPartyConfig`] is fully empty (no
    /// service selected, no key, no overrides) — a signal that the
    /// caller should leave third-party out of the chain entirely.
    pub fn build_third_party_solver(&self) -> Option<ThirdPartyCaptchaSolver> {
        let tp = &self.third_party;
        if tp.service.is_none()
            && tp.base_url.is_none()
            && tp.api_key.is_none()
            && tp.poll_interval_ms.is_none()
            && tp.max_polls.is_none()
        {
            return None;
        }

        let service = match tp.service.as_deref().unwrap_or("two_captcha") {
            "two_captcha" => ThirdPartyService::TwoCaptcha,
            "cap_monster" => ThirdPartyService::CapMonster,
            "cap_solver" => ThirdPartyService::CapSolver,
            "custom" => ThirdPartyService::Custom {
                base_url: tp.base_url.clone().unwrap_or_default(),
            },
            other => ThirdPartyService::Custom {
                base_url: other.to_string(),
            },
        };
        let mut s = match service {
            ThirdPartyService::TwoCaptcha => ThirdPartyCaptchaSolver::two_captcha(),
            ThirdPartyService::CapMonster => ThirdPartyCaptchaSolver::cap_monster(),
            ThirdPartyService::CapSolver => ThirdPartyCaptchaSolver::cap_solver(),
            ThirdPartyService::Custom { base_url } => {
                ThirdPartyCaptchaSolver::custom_endpoint(base_url)
            }
        };
        if let Some(key) = tp.api_key.as_deref().filter(|k| !k.is_empty()) {
            s = s.with_api_key(key);
        }
        if let Some(v) = tp.poll_interval_ms {
            s = s.with_poll_interval_ms(v);
        }
        if let Some(v) = tp.max_polls {
            s = s.with_max_polls(v);
        }
        Some(s)
    }

    /// Build a fully-wired [`CaptchaSolverChain`] from this config:
    /// chain config + token cache + provider registry (with bundled
    /// rules) + Behavioral + VLM (configured) + Audio + third-party
    /// (only when the `[third_party]` section is non-empty OR the
    /// API key env var is set).
    ///
    /// Single entry point used by [`crate::auto_solve`] and the CLI's
    /// `solve` subcommand so they share the same chain shape.
    pub fn build_chain(&self) -> anyhow::Result<CaptchaSolverChain> {
        let registry = Arc::new(ProviderRegistry::with_built_in_rules()?);
        let mut chain = CaptchaSolverChain::empty()
            .with_config(self.chain_config())
            .with_provider_registry(registry)
            .with_token_cache(Arc::new(self.build_token_cache()));
        chain.add_solver(WaitForTokenSolver::new().with_max_wait_ms(3_000));
        chain.add_solver(MathCaptchaSolver::new());
        chain.add_solver(PowCaptchaSolver::new());
        chain.add_solver(SliderCaptchaSolver::new());
        chain.add_solver(BehavioralCaptchaSolver::new());
        chain.add_solver(self.build_vlm_solver());
        chain.add_solver(AudioCaptchaSolver::new());
        // Third-party: prefer the explicit config; fall back to the
        // env-var path so a bare `CAPTCHAFORGE_THIRDPARTY_API_KEY`
        // export still gets the solver wired in.
        if let Some(tp) = self.build_third_party_solver() {
            chain.add_solver(tp);
        } else {
            chain.add_solver(ThirdPartyCaptchaSolver::two_captcha());
        }
        Ok(chain)
    }
}

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

    #[test]
    fn empty_toml_round_trips_to_defaults() {
        let c = Config::from_toml_str("").unwrap();
        assert!(c.per_solver_timeout_ms.is_none());
        assert!(c.cache.ttl_seconds.is_none());
        assert!(c.vlm.endpoint.is_none());
        assert!(c.third_party.service.is_none());
    }

    #[test]
    fn typo_in_top_level_key_is_a_parse_error() {
        // deny_unknown_fields catches typos like `per_sovler_timeout_ms`
        // so users don't lose hours wondering why their override is
        // ignored.
        let r = Config::from_toml_str("per_sovler_timeout_ms = 100");
        assert!(r.is_err(), "deny_unknown_fields should reject typos");
    }

    #[test]
    fn full_toml_parses_and_materialises() {
        let toml = r#"
per_solver_timeout_ms = 60000
screenshot_on_failure = false

[cache]
ttl_seconds = 120

[vlm]
endpoint = "http://gpu:11434"
model    = "qwen3-vl:7b"

[third_party]
service  = "cap_monster"
api_key  = "xyz"
poll_interval_ms = 7000
max_polls = 12

[solve]
checkbox_max_attempts = 25
vlm_http_timeout_ms   = 30000
"#;
        let c = Config::from_toml_str(toml).unwrap();
        assert_eq!(c.per_solver_timeout_ms, Some(60000));
        assert_eq!(c.screenshot_on_failure, Some(false));
        assert_eq!(c.cache.ttl_seconds, Some(120));
        assert_eq!(c.vlm.endpoint.as_deref(), Some("http://gpu:11434"));
        assert_eq!(c.vlm.model.as_deref(), Some("qwen3-vl:7b"));
        assert_eq!(c.third_party.service.as_deref(), Some("cap_monster"));
        assert_eq!(c.third_party.api_key.as_deref(), Some("xyz"));
        assert_eq!(c.solve.checkbox_max_attempts, Some(25));
        assert_eq!(c.solve.vlm_http_timeout_ms, Some(30000));

        let cc = c.chain_config();
        assert_eq!(cc.per_solver_timeout_ms, 60000);
        assert!(!cc.screenshot_on_failure);

        let sc = c.solve_config();
        assert_eq!(sc.checkbox_max_attempts, 25);
        assert_eq!(sc.vlm_http_timeout_ms, 30000);
        // Untouched fields keep defaults.
        assert_eq!(
            sc.token_max_attempts,
            SolveConfig::default().token_max_attempts
        );
    }

    #[test]
    fn build_vlm_solver_applies_endpoint_and_model() {
        let c = Config::from_toml_str(
            r#"
[vlm]
endpoint = "http://x:11434"
model    = "m:latest"
"#,
        )
        .unwrap();
        let s = c.build_vlm_solver();
        assert_eq!(s.endpoint, "http://x:11434");
        assert_eq!(s.model, "m:latest");
    }

    #[test]
    fn build_third_party_solver_is_none_when_section_empty() {
        let c = Config::default();
        assert!(c.build_third_party_solver().is_none());
    }

    #[test]
    fn build_third_party_solver_dispatches_service_variants() {
        for (toml, expected_base) in [
            (
                r#"[third_party]
service = "two_captcha"
"#,
                "https://2captcha.com",
            ),
            (
                r#"[third_party]
service = "cap_monster"
"#,
                "https://api.capmonster.cloud",
            ),
            (
                r#"[third_party]
service = "cap_solver"
"#,
                "https://api.capsolver.com",
            ),
            (
                r#"[third_party]
service = "custom"
base_url = "https://my.gw"
"#,
                "https://my.gw",
            ),
        ] {
            let c = Config::from_toml_str(toml).unwrap();
            let s = c
                .build_third_party_solver()
                .expect("config has third_party set; should build a solver");
            assert_eq!(s.service.base_url(), expected_base);
        }
    }

    #[test]
    fn build_third_party_solver_applies_api_key_and_polling() {
        let c = Config::from_toml_str(
            r#"
[third_party]
service = "two_captcha"
api_key = "k123"
poll_interval_ms = 9000
max_polls = 7
"#,
        )
        .unwrap();
        let s = c.build_third_party_solver().unwrap();
        assert_eq!(s.api_key.as_deref(), Some("k123"));
        assert_eq!(s.poll_interval_ms, 9000);
        assert_eq!(s.max_polls, 7);
        assert!(s.has_api_key());
    }

    #[test]
    fn build_third_party_solver_skips_empty_api_key() {
        let c = Config::from_toml_str(
            r#"
[third_party]
service = "two_captcha"
api_key = ""
"#,
        )
        .unwrap();
        let s = c.build_third_party_solver().unwrap();
        // Empty api_key in the TOML doesn't override the env-var
        // fallback path (which we mock as "no key set" here).
        assert_eq!(s.api_key.as_deref().unwrap_or(""), "");
    }

    #[test]
    fn build_token_cache_honours_configured_ttl() {
        let c = Config::from_toml_str(
            r#"[cache]
ttl_seconds = 5
"#,
        )
        .unwrap();
        let cache = c.build_token_cache();
        assert_eq!(cache.ttl(), std::time::Duration::from_secs(5));
    }

    #[test]
    fn build_token_cache_defaults_to_60_seconds() {
        let c = Config::default();
        let cache = c.build_token_cache();
        assert_eq!(cache.ttl(), std::time::Duration::from_secs(60));
    }

    #[test]
    fn search_paths_includes_explicit_env_first() {
        let paths = Config::search_paths_with_env(|name| match name {
            "CAPTCHAFORGE_CONFIG" => Some("/tmp/explicit.toml".into()),
            "HOME" => Some("/home/user".into()),
            _ => None,
        });
        assert!(paths
            .first()
            .unwrap()
            .to_string_lossy()
            .contains("explicit"));
    }

    #[test]
    fn search_paths_skips_empty_explicit_env() {
        let paths = Config::search_paths_with_env(|name| match name {
            "CAPTCHAFORGE_CONFIG" => Some(String::new()),
            "HOME" => Some("/home/user".into()),
            _ => None,
        });
        // First path should be the project-local file, not the empty env.
        assert_eq!(paths.first().unwrap(), &PathBuf::from(".captchaforge.toml"));
    }

    #[test]
    fn search_paths_appends_xdg_config_home() {
        let paths = Config::search_paths_with_env(|name| match name {
            "XDG_CONFIG_HOME" => Some("/x".into()),
            _ => None,
        });
        let last = paths.last().unwrap();
        assert!(
            last.to_string_lossy()
                .contains("/x/captchaforge/config.toml"),
            "expected XDG path; got {:?}",
            last
        );
    }

    #[test]
    fn search_paths_falls_back_to_home_dotconfig() {
        let paths = Config::search_paths_with_env(|name| match name {
            "HOME" => Some("/home/u".into()),
            _ => None,
        });
        let last = paths.last().unwrap();
        assert!(
            last.to_string_lossy()
                .contains("/home/u/.config/captchaforge/config.toml"),
            "expected HOME-based fallback; got {:?}",
            last
        );
    }

    #[test]
    fn load_from_path_round_trips_disk_file() {
        let dir = tempfile::tempdir().unwrap();
        let path = dir.path().join("captchaforge.toml");
        std::fs::write(&path, "per_solver_timeout_ms = 12345\n").unwrap();
        let c = Config::load_from_path(&path).unwrap();
        assert_eq!(c.per_solver_timeout_ms, Some(12345));
    }

    #[test]
    fn load_from_path_errors_on_missing_file() {
        assert!(Config::load_from_path("/no/such/captchaforge.toml").is_err());
    }

    #[test]
    fn build_chain_default_config_has_eight_solvers() {
        let c = Config::default();
        let chain = c.build_chain().unwrap();
        // Eight solvers: wait_for_token + math + pow + slider +
        // behavioral + vlm + audio + third-party.
        assert_eq!(chain.solvers.len(), 8);
    }

    #[test]
    fn build_chain_honors_per_solver_timeout_from_toml() {
        let c = Config::from_toml_str("per_solver_timeout_ms = 7777").unwrap();
        let chain = c.build_chain().unwrap();
        assert_eq!(chain.config.per_solver_timeout_ms, 7777);
    }

    #[test]
    fn warn_on_inline_secrets_is_method_callable_idempotent() {
        // The warning is observable only via tracing; here we just
        // assert the method exists, accepts a path, and doesn't panic
        // on either an empty or populated api_key.
        let empty = Config::default();
        empty.warn_on_inline_secrets(std::path::Path::new("test.toml"));

        let with_key = Config::from_toml_str(
            r#"[third_party]
api_key = "secret"
"#,
        )
        .unwrap();
        with_key.warn_on_inline_secrets(std::path::Path::new("test.toml"));
    }

    #[test]
    fn build_chain_uses_configured_third_party_when_section_present() {
        let c = Config::from_toml_str(
            r#"
[third_party]
service = "cap_solver"
api_key = "k"
"#,
        )
        .unwrap();
        let chain = c.build_chain().unwrap();
        assert_eq!(chain.solvers.len(), 8);
        assert_eq!(chain.solvers[7].name(), "ThirdPartyCaptchaSolver");
    }
}