essence/
config.rs

1#[derive(Debug, Clone)]
2pub struct Settings {
3    pub server: ServerSettings,
4    pub browser: BrowserSettings,
5    pub engine: EngineSettings,
6    pub crawler: CrawlerSettings,
7    pub retry: RetrySettings,
8}
9
10#[derive(Debug, Clone)]
11pub struct ServerSettings {
12    pub port: u16,
13    pub host: String,
14    pub max_request_size_mb: usize,
15    pub log_level: String,
16}
17
18#[derive(Debug, Clone)]
19pub struct BrowserSettings {
20    pub headless: bool,
21    pub pool_size: usize,
22    pub timeout_ms: u64,
23}
24
25#[derive(Debug, Clone)]
26pub struct EngineSettings {
27    pub waterfall_enabled: bool,
28    pub waterfall_delay_ms: u64,
29    pub auto_fallback_on_block: bool,
30}
31
32#[derive(Debug, Clone)]
33pub struct CrawlerSettings {
34    pub max_concurrent_requests: usize,
35    pub max_duration_sec: u64,
36    pub rate_limit_per_sec: usize,
37}
38
39#[derive(Debug, Clone)]
40pub struct RetrySettings {
41    pub max_attempts: usize,
42    pub initial_delay_ms: u64,
43    pub max_delay_ms: u64,
44}
45
46fn env_or<T: std::str::FromStr>(key: &str, default: T) -> T {
47    std::env::var(key)
48        .ok()
49        .and_then(|v| v.parse().ok())
50        .unwrap_or(default)
51}
52
53impl Settings {
54    pub fn new() -> Result<Self, String> {
55        Ok(Self {
56            server: ServerSettings {
57                port: env_or("PORT", 8080),
58                host: std::env::var("HOST").unwrap_or_else(|_| "0.0.0.0".to_string()),
59                max_request_size_mb: env_or("MAX_REQUEST_SIZE_MB", 1),
60                log_level: std::env::var("RUST_LOG")
61                    .unwrap_or_else(|_| "essence=info".to_string()),
62            },
63            browser: BrowserSettings {
64                headless: env_or("BROWSER_HEADLESS", true),
65                pool_size: env_or("BROWSER_POOL_SIZE", 5),
66                timeout_ms: env_or("BROWSER_TIMEOUT_MS", 30000),
67            },
68            engine: EngineSettings {
69                waterfall_enabled: env_or("ENGINE_WATERFALL_ENABLED", true),
70                waterfall_delay_ms: env_or("ENGINE_WATERFALL_DELAY_MS", 5000),
71                auto_fallback_on_block: env_or("ESSENCE_ENGINE_AUTO_FALLBACK_ON_BLOCK", true),
72            },
73            crawler: CrawlerSettings {
74                max_concurrent_requests: env_or("MAX_CONCURRENT_REQUESTS", 10),
75                max_duration_sec: env_or("CRAWL_MAX_DURATION_SEC", 300),
76                rate_limit_per_sec: env_or("CRAWL_RATE_LIMIT_PER_SEC", 2),
77            },
78            retry: RetrySettings {
79                max_attempts: env_or("RETRY_MAX_ATTEMPTS", 3),
80                initial_delay_ms: env_or("RETRY_INITIAL_DELAY_MS", 500),
81                max_delay_ms: env_or("RETRY_MAX_DELAY_MS", 30000),
82            },
83        })
84    }
85}
essence/config.rs

essence/
config.rs
