|
| 1 | +use crate::{IGNORED_EXTENSIONS, USER_AGENTS}; |
| 2 | +use actix_web::dev::ServiceRequest; |
| 3 | +use actix_web::http::header::HeaderMap; |
| 4 | +use actix_web::http::uri::PathAndQuery; |
| 5 | +use actix_web::http::{header, Method}; |
| 6 | + |
| 7 | +#[derive(Debug)] |
| 8 | +struct PrerenderMiddleware { |
| 9 | + prerender_service_url: String, |
| 10 | +} |
| 11 | + |
| 12 | +impl PrerenderMiddleware { |
| 13 | + pub fn prepare_build_api_url(&self, req: &ServiceRequest) -> String { |
| 14 | + let req_uri = req.uri(); |
| 15 | + let req_headers = req.headers(); |
| 16 | + |
| 17 | + // TODO: this.host? |
| 18 | + let host = req |
| 19 | + .uri() |
| 20 | + .host() |
| 21 | + .or_else(|| { |
| 22 | + req_headers |
| 23 | + .get("X-Forwarded-Host") |
| 24 | + .and_then(|hdr| hdr.to_str().ok()) |
| 25 | + }) |
| 26 | + .or_else(|| { |
| 27 | + req_headers |
| 28 | + .get(header::HOST) |
| 29 | + .and_then(|hdr| hdr.to_str().ok()) |
| 30 | + }) |
| 31 | + .unwrap(); |
| 32 | + |
| 33 | + let scheme = req.uri().scheme_str().unwrap_or("http"); |
| 34 | + let url_path_query = req_uri.path_and_query().map(PathAndQuery::as_str).unwrap(); |
| 35 | + |
| 36 | + format!( |
| 37 | + "{}{}://{}{}", |
| 38 | + &*self.prerender_service_url, scheme, host, url_path_query |
| 39 | + ) |
| 40 | + } |
| 41 | +} |
| 42 | + |
| 43 | +#[derive(Debug)] |
| 44 | +struct PrerenderMiddlewareBuilder {} |
| 45 | + |
| 46 | +impl PrerenderMiddlewareBuilder { |
| 47 | + pub fn use_prerender_io() -> PrerenderMiddleware { |
| 48 | + PrerenderMiddleware { |
| 49 | + prerender_service_url: prerender_url().to_string(), |
| 50 | + } |
| 51 | + } |
| 52 | + |
| 53 | + pub fn use_custom_prerender_url(prerender_service_url: &impl ToString) -> PrerenderMiddleware { |
| 54 | + PrerenderMiddleware { |
| 55 | + prerender_service_url: prerender_service_url.to_string(), |
| 56 | + } |
| 57 | + } |
| 58 | +} |
| 59 | + |
| 60 | +impl PrerenderMiddleware { |
| 61 | + pub fn builder() -> PrerenderMiddlewareBuilder { |
| 62 | + PrerenderMiddlewareBuilder {} |
| 63 | + } |
| 64 | +} |
| 65 | + |
| 66 | +/// Decides if should prerender the page or not. |
| 67 | +/// |
| 68 | +/// Will NOT prerender on the following cases: |
| 69 | +/// * HTTP is not GET or HEAD |
| 70 | +/// * User agent is NOT crawler |
| 71 | +/// * Is requesting a resource on `IGNORED_EXTENSIONS` |
| 72 | +pub fn should_prerender(req: &ServiceRequest) -> bool { |
| 73 | + let request_headers = req.headers(); |
| 74 | + let mut is_crawler = false; |
| 75 | + |
| 76 | + if ![Method::GET, Method::HEAD].contains(req.method()) { |
| 77 | + return false; |
| 78 | + } |
| 79 | + |
| 80 | + let req_ua_lowercase = if let Some(user_agent) = request_headers.get(header::USER_AGENT) { |
| 81 | + let user_agent = user_agent.to_str(); |
| 82 | + if let Ok(ua) = user_agent { |
| 83 | + ua.to_lowercase() |
| 84 | + } else { |
| 85 | + return false; |
| 86 | + } |
| 87 | + } else { |
| 88 | + return false; |
| 89 | + }; |
| 90 | + |
| 91 | + if USER_AGENTS |
| 92 | + .iter() |
| 93 | + .any(|crawler_ua| req_ua_lowercase.contains(&*crawler_ua.to_lowercase())) |
| 94 | + { |
| 95 | + is_crawler = true; |
| 96 | + } |
| 97 | + |
| 98 | + // check for ignored extensions |
| 99 | + let is_ignored_extension_url = req.uri().path_and_query().map_or_else( |
| 100 | + || false, |
| 101 | + |path_query| { |
| 102 | + IGNORED_EXTENSIONS |
| 103 | + .iter() |
| 104 | + .any(|ext| path_query.as_str().contains(ext)) |
| 105 | + }, |
| 106 | + ); |
| 107 | + if is_ignored_extension_url { |
| 108 | + return false; |
| 109 | + } |
| 110 | + |
| 111 | + is_crawler |
| 112 | +} |
| 113 | + |
| 114 | +pub fn get_prerendered_response(req: ServiceRequest) { |
| 115 | + let mut prerender_request_headers = HeaderMap::new(); |
| 116 | + let forward_headers = true; |
| 117 | + |
| 118 | + if forward_headers { |
| 119 | + prerender_request_headers = req.headers().clone(); |
| 120 | + prerender_request_headers.remove(header::HOST); |
| 121 | + } |
| 122 | + |
| 123 | + prerender_request_headers.append(header::ACCEPT_ENCODING, "gzip".parse().unwrap()); |
| 124 | + |
| 125 | + // TODO: accept `X-Prerender-Token` |
| 126 | + // prerender_request_headers.insert("X-Prerender-Token", pre_render_token); |
| 127 | +} |
| 128 | + |
| 129 | +pub fn prerender_url() -> &'static str { |
| 130 | + "https://service.prerender.io/" |
| 131 | +} |
| 132 | + |
| 133 | +#[cfg(test)] |
| 134 | +mod tests { |
| 135 | + use crate::middleware::should_prerender; |
| 136 | + use actix_web::http::{header, Method}; |
| 137 | + use actix_web::test::TestRequest; |
| 138 | + |
| 139 | + fn init_logger() { |
| 140 | + let _ = env_logger::builder().is_test(true).try_init(); |
| 141 | + } |
| 142 | + |
| 143 | + #[test] |
| 144 | + fn test_human_valid_resource() { |
| 145 | + let req = TestRequest::get() |
| 146 | + .insert_header(( |
| 147 | + header::USER_AGENT, |
| 148 | + "Mozilla/5.0 (X11; Linux x86_64; rv:62.0) Gecko/20100101 Firefox/62.0", |
| 149 | + )) |
| 150 | + .uri("http://yourserver.com/clothes/tshirts?query=xl") |
| 151 | + .to_srv_request(); |
| 152 | + |
| 153 | + assert!(!should_prerender(&req)); |
| 154 | + } |
| 155 | + |
| 156 | + #[test] |
| 157 | + fn test_crawler_valid_resource() { |
| 158 | + let req = TestRequest::get() |
| 159 | + .insert_header(( |
| 160 | + header::USER_AGENT, |
| 161 | + "Mozilla/5.0 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)", |
| 162 | + )) |
| 163 | + .uri("http://yourserver.com/clothes/tshirts?query=xl") |
| 164 | + .to_srv_request(); |
| 165 | + |
| 166 | + assert!(should_prerender(&req)); |
| 167 | + } |
| 168 | + |
| 169 | + #[test] |
| 170 | + fn test_crawler_ignored_resource() { |
| 171 | + let req = TestRequest::get() |
| 172 | + .insert_header(( |
| 173 | + header::USER_AGENT, |
| 174 | + "LinkedInBot/1.0 (compatible; Mozilla/5.0; Jakarta Commons-HttpClient/3.1 +http://www.linkedin.com)", |
| 175 | + )) |
| 176 | + .uri("http://yourserver.com/clothes/tshirts/blue.jpg") |
| 177 | + .to_srv_request(); |
| 178 | + |
| 179 | + let render = should_prerender(&req); |
| 180 | + assert!(!render); |
| 181 | + } |
| 182 | + |
| 183 | + #[test] |
| 184 | + fn test_crawler_wrong_http_method() { |
| 185 | + let req = TestRequest::post() |
| 186 | + .insert_header(( |
| 187 | + header::USER_AGENT, |
| 188 | + "LinkedInBot/1.0 (compatible; Mozilla/5.0; Jakarta Commons-HttpClient/3.1 +http://www.linkedin.com)", |
| 189 | + )) |
| 190 | + .uri("http://yourserver.com/clothes/tshirts/red-dotted") |
| 191 | + .to_srv_request(); |
| 192 | + |
| 193 | + let render = should_prerender(&req); |
| 194 | + assert!(!render); |
| 195 | + } |
| 196 | +} |
0 commit comments