Refactor reverse proxy configuration to remove unnecessary URL trunca…

…tion

.vscode/reverse_proxy.yaml

@@ -1,11 +1,9 @@
 default_host:
   - location: /v1/chat/completions
     upstream:
-      scheme_and_authority: https://models.inference.ai.azure.com
-      replacement: /chat/completions
-other_host:
+      url_base: https://models.inference.ai.azure.com/chat/completions
+127.0.0.1:
   - location: / # 默认为 /
     upstream:
-      scheme_and_authority: https://www.baidu.com
-      replacement: / # 默认为 /
+      url_base: https://www.baidu.com/ # 当location以/结尾时，url_base也必须以/结尾
       version: H1 # 可以填H1、H2、AUTO，默认为AUTO

README.md

@@ -95,7 +95,6 @@ Options:
           便捷反向代理配置
           例如：--append-upstream-url=https://cdnjs.cloudflare.com
           则访问 https://your_domain/https://cdnjs.cloudflare.com 会被代理到 https://cdnjs.cloudflare.com
-          注意！这个url的PATH需要为空
   -h, --help
           Print help
 ```
@@ -122,9 +121,8 @@ curl  https://ip.im/info -U "username:password" -x https://localhost:7788  --pro
 YOUR_DOMAIN:
   - location: / # 默认为 /
     upstream:
-      scheme_and_authority: https://www.baidu.com # 末尾不包含 /
-      replacement: / # 默认为 /
-      version: AUTO # 可以填H1、H2、AUTO，默认为AUTO
+      url_base: https://www.baidu.com/ # 当location以/结尾时，url_base也必须以/结尾
+      version: H1 # 可以填H1、H2、AUTO，默认为AUTO
 ```
 
 > 如果 `YOUR_DOMAIN` 填 `default_host` 则对所有的域名生效
@@ -133,41 +131,45 @@ YOUR_DOMAIN:
 
 在github原始url前加上`https://YOUR_DOMAIN`，以便在国内访问raw.githubusercontent.com、github.com和gist.githubusercontent.com
 
-
-增加 `--enable-github-proxy`，或手动置顶下面的反向代理配置文件：
+启动参数中增加 `--enable-github-proxy`，相当于以下配置：
 
 ```yaml
 default_host:
   - location: /https://gist.githubusercontent.com
     upstream:
-      scheme_and_authority: https://gist.githubusercontent.com
-      replacement:
+      url_base: https://gist.githubusercontent.com
   - location: /https://gist.github.com
     upstream:
-      scheme_and_authority: https://gist.github.com
-      replacement:
+      url_base: https://gist.github.com
   - location: /https://github.com
     upstream:
-      scheme_and_authority: https://github.com
-      replacement:
+      url_base: https://github.com
   - location: /https://objects.githubusercontent.com
     upstream:
-      scheme_and_authority: https://objects.githubusercontent.com
-      replacement: 
+      url_base: https://objects.githubusercontent.com
   - location: /https://raw.githubusercontent.com
     upstream:
-      scheme_and_authority: https://raw.githubusercontent.com
-      replacement:
+      url_base: https://raw.githubusercontent.com
+```
+
+#### 例子2： 反向代理https://cdnjs.cloudflare.com
+
+启动参数中增加 `--append-upstream-url=https://cdnjs.cloudflare.com`，相当于以下配置：
+
+```yaml
+default_host:
+  - location: /https://cdnjs.cloudflare.com
+    upstream:
+      url_base: https://cdnjs.cloudflare.com
 ```
 
-#### 例子2: 改写Github Models的url为openai api的url格式
+#### 例子3: 改写Github Models的url为openai api的url格式
 
 ```yaml
 default_host:
   - location: /v1/chat/completions
     upstream:
-      scheme_and_authority: https://models.inference.ai.azure.com
-      replacement: /chat/completions
+      url_base: https://models.inference.ai.azure.com/chat/completions
 ```
 
 ## 可观测

rust_http_proxy/src/config.rs

@@ -15,7 +15,7 @@ use crate::tls_helper::tls_config;
 use crate::{DynError, IDLE_TIMEOUT, REFRESH_INTERVAL};
 
 pub(crate) const DEFAULT_HOST: &str = "default_host";
-const GITHUB_SCHEME_AND_AUTHORITY: [&str; 5] = [
+const GITHUB_URL_BASE: [&str; 5] = [
     "https://github.com",
     "https://gist.githubusercontent.com",
     "https://gist.github.com",
@@ -187,7 +187,7 @@ fn parse_reverse_proxy_config(
         None => HashMap::new(),
     };
     if enable_github_proxy {
-        GITHUB_SCHEME_AND_AUTHORITY.iter().for_each(|domain| {
+        GITHUB_URL_BASE.iter().for_each(|domain| {
             append_upstream_url.push((*domain).to_owned());
         });
     }
@@ -204,18 +204,17 @@ fn parse_reverse_proxy_config(
                             return;
                         }
                         let upstream_url_tmp = upstream_url.to_string();
-                        let scheme_and_authority =
+                        let upstream_url_base =
                             truncate_string(upstream_url_tmp.as_str(), upstream_url.path().len());
                         let path = match upstream_url.path() {
                             "/" => "",
                             other => other,
                         };
 
                         vec.push(LocationConfig {
-                            location: "/".to_string() + scheme_and_authority + path,
+                            location: "/".to_string() + upstream_url_base + path,
                             upstream: crate::reverse::Upstream {
-                                scheme_and_authority: (*scheme_and_authority).to_owned(),
-                                replacement: path.to_string(),
+                                url_base: (*upstream_url_base).to_owned() + path,
                                 version: crate::reverse::Version::Auto,
                             },
                         });
@@ -230,12 +229,55 @@ fn parse_reverse_proxy_config(
     locations
         .iter_mut()
         .for_each(|(_, reverse_proxy_configs)| reverse_proxy_configs.sort());
+    for ele in &locations {
+        for location_config in ele.1 {
+            if location_config.location.is_empty() {
+                return Err("location is empty, location should start with '/'".into());
+            }
+            if location_config.location.ends_with('/')
+                && !location_config.upstream.url_base.ends_with('/')
+            {
+                return Err(format!(
+                    "location ends with '/', but upstream_url_base not: {}",
+                    location_config.upstream.url_base
+                )
+                .into());
+            }
+            match location_config.upstream.url_base.parse::<Uri>() {
+                Ok(upstream_url_base) => {
+                    if upstream_url_base.scheme().is_none() {
+                        return Err(format!(
+                            "wrong upstream_url_base: {} --- scheme is empty",
+                            location_config.upstream.url_base
+                        )
+                        .into());
+                    }
+                    if upstream_url_base.authority().is_none() {
+                        return Err(format!(
+                            "wrong upstream_url_base: {} --- authority is empty",
+                            location_config.upstream.url_base
+                        )
+                        .into());
+                    }
+                    if upstream_url_base.query().is_some() {
+                        return Err(format!(
+                            "wrong upstream_url_base: {} --- query is not empty",
+                            location_config.upstream.url_base
+                        )
+                        .into());
+                    }
+                }
+                Err(e) => {
+                    return Err(format!("parse upstream upstream_url_base error:{}", e).into())
+                }
+            }
+        }
+    }
     let mut redirect_bachpaths = Vec::<RedirectBackpaths>::new();
     for (host, locations) in &locations {
         for location in locations {
             redirect_bachpaths.push(RedirectBackpaths {
-                redirect_url: location.upstream.scheme_and_authority.clone()
-                    + location.upstream.replacement.as_str(),
+                redirect_url: location.upstream.url_base.clone(),
                 host: host.clone(),
                 location: location.location.clone(),
             });
@@ -275,47 +317,6 @@ pub(crate) fn load_config() -> Result<Config, DynError> {
     }
     info!("hostname seems to be {}", param.hostname);
     let config = Config::try_from(param)?;
-    for ele in &config.reverse_proxy_config.locations {
-        for location_config in ele.1 {
-            match location_config.upstream.scheme_and_authority.parse::<Uri>() {
-                Ok(scheme_and_authority) => {
-                    if scheme_and_authority.scheme().is_none() {
-                        return Err(format!(
-                            "wrong scheme_and_authority: {} --- scheme is empty",
-                            location_config.upstream.scheme_and_authority
-                        )
-                        .into());
-                    }
-                    if scheme_and_authority.authority().is_none() {
-                        return Err(format!(
-                            "wrong scheme_and_authority: {} --- authority is empty",
-                            location_config.upstream.scheme_and_authority
-                        )
-                        .into());
-                    }
-                    if scheme_and_authority.path() != "/"
-                        || location_config.upstream.scheme_and_authority.ends_with("/")
-                    {
-                        return Err(format!(
-                            "wrong scheme_and_authority: {} --- path is not empty",
-                            location_config.upstream.scheme_and_authority
-                        )
-                        .into());
-                    }
-                    if scheme_and_authority.query().is_some() {
-                        return Err(format!(
-                            "wrong scheme_and_authority: {} --- query is not empty",
-                            location_config.upstream.scheme_and_authority
-                        )
-                        .into());
-                    }
-                }
-                Err(e) => {
-                    return Err(format!("parse upstream scheme_and_authority error:{}", e).into())
-                }
-            }
-        }
-    }
     log_config(&config);
     info!("auto close connection after idle for {:?}", IDLE_TIMEOUT);
     Ok(config)
@@ -344,13 +345,12 @@ fn log_config(config: &Config) {
         .for_each(|reverse_proxy_config| {
             for ele in reverse_proxy_config.1 {
                 info!(
-                    "    {:<70} -> {}{}**",
+                    "    {:<70} -> {}**",
                     format!(
                         "http(s)://{}:port{}**",
                         reverse_proxy_config.0, ele.location
                     ),
-                    ele.upstream.scheme_and_authority,
-                    ele.upstream.replacement
+                    ele.upstream.url_base,
                 );
             }
         });

rust_http_proxy/src/proxy.rs

@@ -138,8 +138,7 @@ impl ProxyHandler {
                             client: client_socket_addr.ip().to_canonical().to_string(),
                             origin: origin_scheme_host_port.to_string()
                                 + location_config.location.as_str(),
-                            upstream: location_config.upstream.scheme_and_authority.clone()
-                                + location_config.upstream.replacement.as_str(),
+                            upstream: location_config.upstream.url_base.clone(),
                         }))
                         .inc();
                     self.metrics
@@ -483,19 +482,11 @@ fn build_upstream_req(
         Some(path_and_query) => path_and_query.as_str(),
         None => "",
     };
-    let url = format!(
-        "{}{}",
-        location_config.upstream.scheme_and_authority.clone(),
-        location_config.upstream.replacement.clone()
-            + &path_and_query[location_config.location.len()..]
-    );
+    let url = location_config.upstream.url_base.clone()
+        + &path_and_query[location_config.location.len()..];
 
     let mut builder = Request::builder().method(method).uri(url).version(
-        if !location_config
-            .upstream
-            .scheme_and_authority
-            .starts_with("https:")
-        {
+        if !location_config.upstream.url_base.starts_with("https:") {
             match location_config.upstream.version {
                 reverse::Version::H1 => Version::HTTP_11,
                 reverse::Version::H2 => Version::HTTP_2,
@@ -658,10 +649,7 @@ fn ensure_absolute(
         .parse::<Uri>()
         .map_err(|e| io::Error::new(ErrorKind::InvalidData, e))?;
     if redirect_url.scheme_str().is_none() {
-        Ok(format!(
-            "{}{}",
-            context.upstream.scheme_and_authority, location
-        ))
+        Ok(format!("{}{}", context.upstream.url_base, location))
     } else {
         Ok(location.to_string())
     }

rust_http_proxy/src/reverse.rs

@@ -21,9 +21,7 @@ impl std::cmp::Ord for LocationConfig {
 
 #[derive(Serialize, Deserialize, Eq, PartialEq, PartialOrd)]
 pub(crate) struct Upstream {
-    pub(crate) scheme_and_authority: String, // https://google.com
-    #[serde(default = "root")]
-    pub(crate) replacement: String, // /
+    pub(crate) url_base: String, // https://google.com
     #[serde(default = "default_version")]
     pub(crate) version: Version,
 }