headless_engine/network/
client.rs

1use crate::network::fingerprint::{DeviceProfile, Fingerprint};
2use anyhow::{Context, Result};
3use std::sync::Arc;
4use std::time::Duration;
5
6pub struct FetchResult {
7    pub status: u16,
8    pub final_url: String,
9    pub html: String,
10    pub is_captcha_detected: bool,
11}
12
13pub struct NetworkClient {
14    client: reqwest::Client,
15    pub profile: DeviceProfile,
16    pub fingerprint: Fingerprint,
17    #[allow(dead_code)]
18    pub cookie_jar: Arc<reqwest::cookie::Jar>,
19}
20
21impl NetworkClient {
22    #[allow(dead_code)]
23    pub fn new() -> Result<Self> {
24        Self::with_profile(DeviceProfile::ChromeWindows)
25    }
26
27    pub fn with_profile(profile: DeviceProfile) -> Result<Self> {
28        Self::with_builder_config(profile, None, Duration::from_secs(30), 10, None)
29    }
30
31    pub fn with_builder_config(
32        profile: DeviceProfile,
33        proxy_url: Option<&str>,
34        timeout: Duration,
35        max_redirects: usize,
36        custom_ua: Option<&str>,
37    ) -> Result<Self> {
38        let cookie_jar = Arc::new(reqwest::cookie::Jar::default());
39        let fingerprint = Fingerprint::for_profile(profile);
40        let mut headers = fingerprint.build_headers();
41
42        if let Some(ua) = custom_ua {
43            headers.insert(
44                reqwest::header::USER_AGENT,
45                reqwest::header::HeaderValue::from_str(ua)
46                    .context("Invalid custom User-Agent header")?,
47            );
48        }
49
50        let mut builder = reqwest::Client::builder()
51            .default_headers(headers)
52            .cookie_provider(cookie_jar.clone())
53            .timeout(timeout)
54            .redirect(reqwest::redirect::Policy::limited(max_redirects))
55            .gzip(true)
56            .brotli(true)
57            .deflate(true);
58
59        if let Some(proxy_str) = proxy_url {
60            let proxy = reqwest::Proxy::all(proxy_str)
61                .context(format!("Invalid proxy URL: {}", proxy_str))?;
62            builder = builder.proxy(proxy);
63        }
64
65        // StealthGuard: Pre-warm legitimate session and consent cookies
66        if let Ok(google_url) = "https://www.google.com".parse::<reqwest::Url>() {
67            cookie_jar.add_cookie_str("SOCS=CAESHAgBEhJnd3NfMjAyNDA5MDUtMF9SQzIaAmVuIAEaBgiA_L20Bg; Path=/; Domain=.google.com; Secure", &google_url);
68            cookie_jar.add_cookie_str("CONSENT=PENDING+987; Path=/; Domain=.google.com; Secure", &google_url);
69            cookie_jar.add_cookie_str("AEC=AZ6Zc-Wz2R61_67w88hJ; Path=/; Domain=.google.com; Secure", &google_url);
70        }
71        if let Ok(ddg_url) = "https://duckduckgo.com".parse::<reqwest::Url>() {
72            cookie_jar.add_cookie_str("5=0; Path=/; Domain=.duckduckgo.com; Secure", &ddg_url);
73            cookie_jar.add_cookie_str("l=en-us; Path=/; Domain=.duckduckgo.com; Secure", &ddg_url);
74        }
75
76        let client = builder.build().context("Failed to build HTTP client")?;
77
78        Ok(Self {
79            client,
80            profile,
81            fingerprint,
82            cookie_jar,
83        })
84    }
85
86    pub fn set_profile(&mut self, profile: DeviceProfile) -> Result<()> {
87        let new_client = Self::with_profile(profile)?;
88        self.client = new_client.client;
89        self.profile = profile;
90        self.fingerprint = new_client.fingerprint;
91        Ok(())
92    }
93
94    pub async fn fetch(&self, url: &str) -> Result<FetchResult> {
95        if url.starts_with("file://") {
96            let file_path = url.trim_start_matches("file:///").trim_start_matches("file://");
97            let clean_path = file_path.replace('/', "\\");
98            let html = std::fs::read_to_string(&clean_path)
99                .or_else(|_| std::fs::read_to_string(file_path))
100                .context(format!("Failed to read local file: {}", url))?;
101            return Ok(FetchResult {
102                status: 200,
103                final_url: url.to_string(),
104                html,
105                is_captcha_detected: false,
106            });
107        }
108
109        let response = self
110            .client
111            .get(url)
112            .send()
113            .await
114            .context(format!("Failed to send request to {}", url))?;
115
116        let status = response.status().as_u16();
117        let final_url = response.url().to_string();
118        let html = response
119            .text()
120            .await
121            .context("Failed to decode response body")?;
122
123        let is_captcha_detected = html.contains("sorry/index?continue=")
124            || html.contains("Our systems have detected unusual traffic")
125            || html.contains("id=\"captcha-form\"")
126            || (html.contains("challenges.cloudflare.com")
127                && html.contains("cf-turnstile-wrapper"))
128            || html.contains("hcaptcha-box");
129
130        Ok(FetchResult {
131            status,
132            final_url,
133            html,
134            is_captcha_detected,
135        })
136    }
137}
headless_engine/network/client.rs

headless_engine/network/
client.rs
