BlueArchiveToolkit/infrastructure/src/curl_transfer.rs

use std::env;
use std::path::{Path, PathBuf};
use std::process::{Command, Output};

const PROXY_ENV_PRIORITY: [&str; 6] = [
    "HTTPS_PROXY",
    "https_proxy",
    "ALL_PROXY",
    "all_proxy",
    "HTTP_PROXY",
    "http_proxy",
];
const NO_PROXY_ENV_PRIORITY: [&str; 2] = ["NO_PROXY", "no_proxy"];

/// Proxy selection mode for official HTTP transfers executed through `curl`.
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum CurlProxyMode {
    /// Detect proxy settings from the current process environment.
    Auto,
    /// Force direct connections and ignore proxy environment variables.
    Disabled,
    /// Force a specific proxy URL.
    Url(String),
}

/// Proxy configuration for official HTTP transfers executed through `curl`.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct CurlProxyConfig {
    mode: CurlProxyMode,
}

impl CurlProxyConfig {
    /// Creates automatic proxy detection.
    pub fn auto() -> Self {
        Self {
            mode: CurlProxyMode::Auto,
        }
    }

    /// Creates a direct-connection configuration.
    pub fn disabled() -> Self {
        Self {
            mode: CurlProxyMode::Disabled,
        }
    }

    /// Creates a forced proxy configuration.
    pub fn url(url: impl Into<String>) -> Self {
        Self {
            mode: CurlProxyMode::Url(url.into()),
        }
    }

    /// Returns the proxy mode.
    pub fn mode(&self) -> &CurlProxyMode {
        &self.mode
    }
}

impl Default for CurlProxyConfig {
    fn default() -> Self {
        Self::auto()
    }
}

/// Resolved proxy settings for one `curl` command.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct ResolvedCurlProxy {
    /// Whether a proxy URL will be passed to `curl`.
    pub enabled: bool,
    /// Whether direct mode was explicitly requested.
    pub disabled: bool,
    /// Proxy URL, when one is selected.
    pub url: Option<String>,
    /// Where this proxy decision came from.
    pub source: String,
    /// `NO_PROXY`/`no_proxy` value preserved for `curl`, when present.
    pub no_proxy: Option<String>,
}

impl ResolvedCurlProxy {
    /// Returns a human-readable summary with credentials redacted.
    pub fn human_summary(&self) -> String {
        if self.disabled {
            return format!("代理已关闭；来源={}", self.source);
        }
        if let Some(url) = self.url.as_ref() {
            let no_proxy = self
                .no_proxy
                .as_ref()
                .map(|value| format!("；no_proxy={value}"))
                .unwrap_or_default();
            return format!(
                "使用代理 {}；来源={}{}",
                redact_proxy_url(url),
                self.source,
                no_proxy
            );
        }
        format!("未检测到代理；来源={}", self.source)
    }
}

/// Resolves the proxy setting that should be applied to `curl`.
pub fn resolve_curl_proxy(config: &CurlProxyConfig) -> ResolvedCurlProxy {
    let pairs = proxy_env_pairs();
    resolve_curl_proxy_from_pairs(config, &pairs)
}

/// Redacts user info from a proxy URL for diagnostics and logs.
///
/// `curl` accepts proxy strings without an explicit scheme (defaulting to
/// `http://`), so credentials such as `user:secret@host:port` must be redacted
/// even when no `://` separator is present.
pub fn redact_proxy_url(url: &str) -> String {
    if let Some((scheme, rest)) = url.split_once("://") {
        let Some(at_index) = rest.find('@') else {
            return url.to_string();
        };
        return format!("{scheme}://<redacted>@{}", &rest[at_index + 1..]);
    }
    let Some(at_index) = url.find('@') else {
        return url.to_string();
    };
    format!("<redacted>@{}", &url[at_index + 1..])
}

fn proxy_env_pairs() -> Vec<(String, String)> {
    PROXY_ENV_PRIORITY
        .iter()
        .chain(NO_PROXY_ENV_PRIORITY.iter())
        .filter_map(|name| {
            env::var(name)
                .ok()
                .map(|value| ((*name).to_string(), value))
        })
        .collect()
}

fn resolve_curl_proxy_from_pairs(
    config: &CurlProxyConfig,
    pairs: &[(String, String)],
) -> ResolvedCurlProxy {
    match config.mode() {
        CurlProxyMode::Auto => {
            let no_proxy = first_env_value(pairs, &NO_PROXY_ENV_PRIORITY).map(|(_, value)| value);
            if let Some((name, url)) = first_env_value(pairs, &PROXY_ENV_PRIORITY) {
                ResolvedCurlProxy {
                    enabled: true,
                    disabled: false,
                    url: Some(url),
                    source: name,
                    no_proxy,
                }
            } else {
                ResolvedCurlProxy {
                    enabled: false,
                    disabled: false,
                    url: None,
                    source: "auto".to_string(),
                    no_proxy,
                }
            }
        }
        CurlProxyMode::Disabled => ResolvedCurlProxy {
            enabled: false,
            disabled: true,
            url: None,
            source: "--no-proxy".to_string(),
            no_proxy: None,
        },
        CurlProxyMode::Url(url) => ResolvedCurlProxy {
            enabled: true,
            disabled: false,
            url: Some(url.clone()),
            source: "--proxy".to_string(),
            no_proxy: first_env_value(pairs, &NO_PROXY_ENV_PRIORITY).map(|(_, value)| value),
        },
    }
}

fn first_env_value(pairs: &[(String, String)], priority: &[&str]) -> Option<(String, String)> {
    priority.iter().find_map(|name| {
        pairs
            .iter()
            .find(|(candidate, value)| candidate == name && !value.trim().is_empty())
            .map(|(candidate, value)| (candidate.clone(), value.clone()))
    })
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub(crate) enum CurlFailureKind {
    HttpForbidden,
    HttpNotFound,
    HttpClientError,
    HttpTooManyRequests,
    HttpServerError,
    HttpUnknown,
    Dns,
    Connect,
    Timeout,
    Tls,
    Interrupted,
    Network,
    /// 代理自身故障：无法解析/连接代理，或代理返回 407 认证失败等。
    Proxy,
    ProcessFailed,
    Unknown,
}

impl CurlFailureKind {
    pub(crate) fn as_str(&self) -> &'static str {
        match self {
            Self::HttpForbidden => "http_forbidden",
            Self::HttpNotFound => "http_not_found",
            Self::HttpClientError => "http_client_error",
            Self::HttpTooManyRequests => "http_too_many_requests",
            Self::HttpServerError => "http_server_error",
            Self::HttpUnknown => "http_unknown",
            Self::Dns => "dns",
            Self::Connect => "connect",
            Self::Timeout => "timeout",
            Self::Tls => "tls",
            Self::Interrupted => "interrupted",
            Self::Network => "network",
            Self::Proxy => "proxy",
            Self::ProcessFailed => "process_failed",
            Self::Unknown => "unknown",
        }
    }

    pub(crate) fn is_retryable(&self) -> bool {
        match self {
            // 代理故障（认证失败、代理主机解析/连接失败）多为配置错误，per-URL 层
            // 不重试以免空转；daemon 的 error-retry 循环仍会在下一轮整体重试，覆盖
            // 代理短暂抖动。
            Self::HttpForbidden
            | Self::HttpNotFound
            | Self::HttpClientError
            | Self::Proxy
            | Self::ProcessFailed => false,
            Self::HttpTooManyRequests
            | Self::HttpServerError
            | Self::HttpUnknown
            | Self::Dns
            | Self::Connect
            | Self::Timeout
            | Self::Tls
            | Self::Interrupted
            | Self::Network
            | Self::Unknown => true,
        }
    }
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub(crate) struct CurlFailure {
    pub(crate) url: String,
    pub(crate) destination: Option<PathBuf>,
    pub(crate) exit_code: Option<i32>,
    pub(crate) stderr: String,
    pub(crate) http_status: Option<u16>,
    pub(crate) kind: CurlFailureKind,
}

impl CurlFailure {
    fn from_output(url: &str, destination: Option<&Path>, output: &Output) -> Self {
        let stderr = String::from_utf8_lossy(&output.stderr).trim().to_string();
        let exit_code = output.status.code();
        let http_status = parse_http_status(&stderr);
        let kind = classify_failure(exit_code, http_status, &stderr);
        Self {
            url: url.to_string(),
            destination: destination.map(Path::to_path_buf),
            exit_code,
            stderr,
            http_status,
            kind,
        }
    }

    fn process_failed(
        url: &str,
        destination: Option<&Path>,
        curl_command: &Path,
        error: std::io::Error,
    ) -> Self {
        Self {
            url: url.to_string(),
            destination: destination.map(Path::to_path_buf),
            exit_code: None,
            stderr: format!("启动 curl 命令失败 {}：{error}", curl_command.display()),
            http_status: None,
            kind: CurlFailureKind::ProcessFailed,
        }
    }

    pub(crate) fn retryable(&self) -> bool {
        self.kind.is_retryable()
    }

    fn format_compact(&self) -> String {
        let mut parts = vec![
            format!("kind={}", self.kind.as_str()),
            format!("retryable={}", self.retryable()),
            format!("url={}", self.url),
        ];
        if let Some(status) = self.http_status {
            parts.push(format!("http_status={status}"));
        }
        if let Some(code) = self.exit_code {
            parts.push(format!("curl_exit={code}"));
        }
        if let Some(destination) = self.destination.as_ref() {
            parts.push(format!("destination={}", destination.display()));
        }
        if !self.stderr.is_empty() {
            parts.push(format!("stderr={}", self.stderr));
        }
        parts.join(" ")
    }
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub(crate) struct CurlAttemptFailure {
    pub(crate) attempt: usize,
    pub(crate) attempts: usize,
    pub(crate) failure: CurlFailure,
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub(crate) struct CurlRetryError {
    pub(crate) attempts: Vec<CurlAttemptFailure>,
}

impl CurlRetryError {
    pub(crate) fn final_failure(&self) -> Option<&CurlFailure> {
        self.attempts.last().map(|attempt| &attempt.failure)
    }

    pub(crate) fn attempt_count(&self) -> usize {
        self.attempts.len()
    }

    pub(crate) fn final_kind_label(&self) -> Option<&'static str> {
        self.final_failure().map(|failure| failure.kind.as_str())
    }

    pub(crate) fn final_http_status(&self) -> Option<u16> {
        self.final_failure().and_then(|failure| failure.http_status)
    }

    pub(crate) fn final_retryable(&self) -> Option<bool> {
        self.final_failure().map(CurlFailure::retryable)
    }

    pub(crate) fn summary(&self) -> String {
        let Some(final_failure) = self.final_failure() else {
            return "curl 未执行就已失败".to_string();
        };
        let policy = if final_failure.retryable() {
            "重试次数已耗尽"
        } else {
            "不可重试，已提前停止"
        };
        let attempts = self
            .attempts
            .iter()
            .map(|attempt| {
                format!(
                    "第 {}/{} 次尝试失败：{}",
                    attempt.attempt,
                    attempt.attempts,
                    attempt.failure.format_compact()
                )
            })
            .collect::<Vec<_>>()
            .join("；");
        format!("{policy}；{attempts}")
    }
}

impl std::fmt::Display for CurlRetryError {
    fn fmt(&self, formatter: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        formatter.write_str(&self.summary())
    }
}

impl std::error::Error for CurlRetryError {}

pub(crate) fn run_curl_with_retry_with_proxy(
    curl_command: &Path,
    url: &str,
    destination: Option<&Path>,
    attempts: usize,
    proxy_config: &CurlProxyConfig,
    mut build_command: impl FnMut() -> Command,
) -> Result<Output, CurlRetryError> {
    let attempts = attempts.max(1);
    let mut failures = Vec::new();
    let proxy = resolve_curl_proxy(proxy_config);

    for attempt in 1..=attempts {
        let mut command = build_command();
        apply_curl_proxy(&mut command, &proxy);
        let failure = match command.output() {
            Ok(output) if output.status.success() => return Ok(output),
            Ok(output) => CurlFailure::from_output(url, destination, &output),
            Err(error) => CurlFailure::process_failed(url, destination, curl_command, error),
        };
        let retryable = failure.retryable();
        failures.push(CurlAttemptFailure {
            attempt,
            attempts,
            failure,
        });
        if !retryable {
            break;
        }
    }

    Err(CurlRetryError { attempts: failures })
}

fn apply_curl_proxy(command: &mut Command, proxy: &ResolvedCurlProxy) {
    if proxy.disabled {
        remove_proxy_env(command);
        command.arg("--noproxy").arg("*");
        return;
    }

    if let Some(url) = proxy.url.as_ref() {
        // 通过环境变量而非 --proxy 参数把代理传给 curl：代理 URL 可能带凭据，
        // 放进 argv 会出现在 curl 子进程的 /proc/<pid>/cmdline（0444，世界可读）。
        // 先清除所有继承的代理变量，再设 ALL_PROXY/all_proxy（覆盖 http 与 https，
        // 大小写两种写法都设，避免 curl 只读其中一种）。no_proxy 列表不含凭据，仍走参数。
        remove_proxy_env(command);
        command.env("ALL_PROXY", url);
        command.env("all_proxy", url);
        if let Some(no_proxy) = proxy.no_proxy.as_ref() {
            command.arg("--noproxy").arg(no_proxy);
        }
    }
}

fn remove_proxy_env(command: &mut Command) {
    for name in PROXY_ENV_PRIORITY
        .iter()
        .chain(NO_PROXY_ENV_PRIORITY.iter())
    {
        command.env_remove(name);
    }
}

/// 从 curl stderr 判断是否为代理自身故障（而非源站故障）。
///
/// 覆盖：无法解析代理（curl exit 5）、CONNECT 隧道被代理拒绝或返回 4xx（如 407
/// 认证失败）等。curl 会把这些原因写进 stderr，比仅凭 exit code 更可靠地区分
/// 代理故障与源站故障。
fn is_proxy_failure(stderr: &str) -> bool {
    let lower = stderr.to_ascii_lowercase();
    lower.contains("resolve proxy")
        || lower.contains("connect tunnel failed")
        || lower.contains("from proxy after connect")
        || (lower.contains("proxy") && lower.contains("407"))
}

fn classify_failure(
    exit_code: Option<i32>,
    http_status: Option<u16>,
    stderr: &str,
) -> CurlFailureKind {
    // 代理故障优先识别：这些错误来自代理链路而非官方源站，单独分类并 fail-fast。
    if is_proxy_failure(stderr) {
        return CurlFailureKind::Proxy;
    }

    if exit_code == Some(22) {
        return match http_status {
            Some(403) => CurlFailureKind::HttpForbidden,
            Some(404) => CurlFailureKind::HttpNotFound,
            Some(408 | 425 | 429) => CurlFailureKind::HttpTooManyRequests,
            Some(400..=499) => CurlFailureKind::HttpClientError,
            Some(500..=599) => CurlFailureKind::HttpServerError,
            Some(_) => CurlFailureKind::HttpUnknown,
            None => CurlFailureKind::HttpUnknown,
        };
    }

    match exit_code {
        Some(5 | 6) => CurlFailureKind::Dns,
        Some(7) => CurlFailureKind::Connect,
        Some(18) => CurlFailureKind::Interrupted,
        Some(28) => CurlFailureKind::Timeout,
        Some(35) => CurlFailureKind::Tls,
        Some(52 | 56) => CurlFailureKind::Network,
        Some(_) => CurlFailureKind::Unknown,
        None => CurlFailureKind::ProcessFailed,
    }
}

fn parse_http_status(stderr: &str) -> Option<u16> {
    let bytes = stderr.as_bytes();
    let mut index = 0usize;
    while index < bytes.len() {
        if !bytes[index].is_ascii_digit() {
            index += 1;
            continue;
        }
        let start = index;
        while index < bytes.len() && bytes[index].is_ascii_digit() {
            index += 1;
        }
        if index - start == 3 {
            let status = stderr[start..index].parse::<u16>().ok()?;
            if (100..=599).contains(&status) {
                return Some(status);
            }
        }
    }
    None
}

#[cfg(test)]
mod tests {
    use super::*;
    use std::os::unix::process::ExitStatusExt;

    #[test]
    fn classifies_http_status_codes_from_curl_stderr() {
        let status = std::process::ExitStatus::from_raw(22 << 8);
        let output = Output {
            status,
            stdout: Vec::new(),
            stderr: b"curl: (22) The requested URL returned error: 404".to_vec(),
        };
        let failure = CurlFailure::from_output("https://example.test/a", None, &output);

        assert_eq!(failure.http_status, Some(404));
        assert_eq!(failure.kind, CurlFailureKind::HttpNotFound);
        assert!(!failure.retryable());
    }

    #[test]
    fn treats_5xx_as_retryable() {
        let status = std::process::ExitStatus::from_raw(22 << 8);
        let output = Output {
            status,
            stdout: Vec::new(),
            stderr: b"curl: (22) The requested URL returned error: 503".to_vec(),
        };
        let failure = CurlFailure::from_output("https://example.test/a", None, &output);

        assert_eq!(failure.http_status, Some(503));
        assert_eq!(failure.kind, CurlFailureKind::HttpServerError);
        assert!(failure.retryable());
    }

    #[test]
    fn classifies_proxy_auth_and_resolution_failures() {
        // 代理 407 认证失败：CONNECT 隧道失败，识别为 Proxy 且不重试。
        let auth = Output {
            status: std::process::ExitStatus::from_raw(56 << 8),
            stdout: Vec::new(),
            stderr: b"curl: (56) CONNECT tunnel failed, response 407".to_vec(),
        };
        let failure = CurlFailure::from_output("https://example.test/a", None, &auth);
        assert_eq!(failure.kind, CurlFailureKind::Proxy);
        assert!(!failure.retryable());

        // 无法解析代理主机（curl exit 5）：同样归为 Proxy 而非普通 DNS。
        let resolve = Output {
            status: std::process::ExitStatus::from_raw(5 << 8),
            stdout: Vec::new(),
            stderr: b"curl: (5) Could not resolve proxy: proxy.invalid".to_vec(),
        };
        let failure = CurlFailure::from_output("https://example.test/a", None, &resolve);
        assert_eq!(failure.kind, CurlFailureKind::Proxy);
        assert!(!failure.retryable());

        // 源站普通连接失败（无代理标记）仍归为 Connect 且可重试。
        let origin = Output {
            status: std::process::ExitStatus::from_raw(7 << 8),
            stdout: Vec::new(),
            stderr: b"curl: (7) Failed to connect to example.test port 443".to_vec(),
        };
        let failure = CurlFailure::from_output("https://example.test/a", None, &origin);
        assert_eq!(failure.kind, CurlFailureKind::Connect);
        assert!(failure.retryable());
    }

    #[test]
    fn auto_proxy_prefers_https_and_preserves_no_proxy() {
        let pairs = vec![
            (
                "HTTP_PROXY".to_string(),
                "http://proxy-http:8080".to_string(),
            ),
            (
                "HTTPS_PROXY".to_string(),
                "http://proxy-https:7890".to_string(),
            ),
            ("NO_PROXY".to_string(), "localhost,127.0.0.1".to_string()),
        ];

        let resolved = resolve_curl_proxy_from_pairs(&CurlProxyConfig::auto(), &pairs);

        assert!(resolved.enabled);
        assert!(!resolved.disabled);
        assert_eq!(resolved.url.as_deref(), Some("http://proxy-https:7890"));
        assert_eq!(resolved.source, "HTTPS_PROXY");
        assert_eq!(resolved.no_proxy.as_deref(), Some("localhost,127.0.0.1"));
    }

    #[test]
    fn disabled_proxy_ignores_environment() {
        let pairs = vec![(
            "HTTPS_PROXY".to_string(),
            "http://proxy-https:7890".to_string(),
        )];

        let resolved = resolve_curl_proxy_from_pairs(&CurlProxyConfig::disabled(), &pairs);

        assert!(!resolved.enabled);
        assert!(resolved.disabled);
        assert!(resolved.url.is_none());
        assert_eq!(resolved.source, "--no-proxy");
    }

    #[test]
    fn explicit_proxy_wins_and_redacts_credentials() {
        let pairs = vec![(
            "HTTPS_PROXY".to_string(),
            "http://environment:7890".to_string(),
        )];

        let resolved = resolve_curl_proxy_from_pairs(
            &CurlProxyConfig::url("http://user:secret@127.0.0.1:7890"),
            &pairs,
        );

        assert!(resolved.enabled);
        assert_eq!(
            resolved.url.as_deref(),
            Some("http://user:secret@127.0.0.1:7890")
        );
        assert_eq!(resolved.source, "--proxy");
        assert_eq!(
            redact_proxy_url(resolved.url.as_deref().unwrap()),
            "http://<redacted>@127.0.0.1:7890"
        );
    }

    #[test]
    fn redact_proxy_url_handles_schemeless_credentials() {
        assert_eq!(
            redact_proxy_url("user:secret@127.0.0.1:7890"),
            "<redacted>@127.0.0.1:7890"
        );
        assert_eq!(
            redact_proxy_url("socks5://user:secret@127.0.0.1:1080"),
            "socks5://<redacted>@127.0.0.1:1080"
        );
        assert_eq!(redact_proxy_url("127.0.0.1:7890"), "127.0.0.1:7890");
        assert_eq!(
            redact_proxy_url("http://127.0.0.1:7890"),
            "http://127.0.0.1:7890"
        );
    }

    fn command_args(command: &Command) -> Vec<String> {
        command
            .get_args()
            .map(|arg| arg.to_string_lossy().to_string())
            .collect()
    }

    fn command_env(command: &Command, key: &str) -> Option<Option<String>> {
        command
            .get_envs()
            .find(|(name, _)| *name == key)
            .map(|(_, value)| value.map(|value| value.to_string_lossy().to_string()))
    }

    #[test]
    fn apply_curl_proxy_passes_url_via_env_not_argv() {
        let proxy = ResolvedCurlProxy {
            enabled: true,
            disabled: false,
            url: Some("http://user:secret@127.0.0.1:7897".to_string()),
            source: "--proxy".to_string(),
            no_proxy: Some("localhost,127.0.0.1".to_string()),
        };
        let mut command = Command::new("curl");
        apply_curl_proxy(&mut command, &proxy);

        // 凭据不得出现在 curl 的 argv。
        let args = command_args(&command);
        assert!(!args.iter().any(|arg| arg.contains("secret")));
        assert!(!args.iter().any(|arg| arg == "--proxy"));
        // no_proxy 列表不含凭据，仍以参数传入。
        assert!(args
            .windows(2)
            .any(|pair| pair == ["--noproxy", "localhost,127.0.0.1"]));

        // 代理 URL 经 ALL_PROXY/all_proxy 环境变量下传。
        assert_eq!(
            command_env(&command, "ALL_PROXY"),
            Some(Some("http://user:secret@127.0.0.1:7897".to_string()))
        );
        assert_eq!(
            command_env(&command, "all_proxy"),
            Some(Some("http://user:secret@127.0.0.1:7897".to_string()))
        );
        // 继承的 scheme 专用代理变量被清除，避免覆盖 ALL_PROXY。
        assert_eq!(command_env(&command, "HTTPS_PROXY"), Some(None));
        assert_eq!(command_env(&command, "https_proxy"), Some(None));
    }

    #[test]
    fn apply_curl_proxy_disabled_forces_direct() {
        let proxy = ResolvedCurlProxy {
            enabled: false,
            disabled: true,
            url: None,
            source: "--no-proxy".to_string(),
            no_proxy: None,
        };
        let mut command = Command::new("curl");
        apply_curl_proxy(&mut command, &proxy);

        let args = command_args(&command);
        assert!(args.windows(2).any(|pair| pair == ["--noproxy", "*"]));
        // 直连模式清除所有继承的代理变量，且不设置任何代理。
        assert_eq!(command_env(&command, "ALL_PROXY"), Some(None));
        assert_eq!(command_env(&command, "HTTPS_PROXY"), Some(None));
    }
}