Use rule file

zu1k · Sep 14, 2021 · b9dfca7 · b9dfca7
1 parent adb3ccc
commit b9dfca7
Show file tree

Hide file tree

Showing 9 changed files with 254 additions and 26 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -18,11 +18,14 @@ codegen-units = 1
 
 [dependencies]
 bytes = { version = "1", features = ["serde"] }
+clap = "2.33.3"
 env_logger = "0.9"
 fancy-regex = "0.7"
 hudsucker = "0.4"
 lazy_static = "1.4"
 log = "0.4"
+serde = { version = "1.0", features = ["derive"] }
+serde_yaml = "0.8"
 tokio = { version = "1", features = ["full"] }
 
 [dev-dependencies]

diff --git a/README.md b/README.md
@@ -36,6 +36,8 @@ Adding `http` and `https` proxies to the browser, `http://127.0.0.1:34567` if no
 
 Now I add two demo websites, removing ADs using good-MITM `rewrite` feature.
 
+You should use the demo rule file, `good-mitm.exe -f demo.yaml`
+
 See the effect by comparing the content(ads) with and without using `good-MITM`.
 
 - [低端影视](https://ddrk.me/)

diff --git a/assets/rules/demo.yaml b/assets/rules/demo.yaml
@@ -0,0 +1,23 @@
+- name: "奈菲影视去广告"
+  filter:
+    url-regex: '(nfmovies)(?!.*?(\.css|\.js|\.jpeg|\.png|\.gif)).*'
+  action:
+    modify:
+      body:
+        origin: '<head>'
+        new: '<link rel="stylesheet" href="https://limbopro.com/CSS/nfmovies.css" type="text/css"><script type="text/javascript"  src="//limbopro.com/Adguard/nfmovies.js"></script></head>'
+
+- name: "低端影视去广告"
+  filter:
+    url-regex: '^https:?/\/ddrk.me.*'
+  action:
+    modify:
+      body:
+        origin: '<head>'
+        new: '<head><link rel="stylesheet" href="https://limbopro.com/CSS/ddrk.css" type="text/css"><script type="text/javascript" src="//limbopro.com/Adguard/ddrk.js"></script>'
+
+- name: "跳转博客"
+  filter:
+    domain: 'none.lgf.im'
+  action:
+    redirect: "https://lgf.im/"
diff --git a/src/action/mod.rs b/src/action/mod.rs
@@ -3,19 +3,17 @@ pub use modify::*;
 
 use hudsucker::hyper::{header, header::HeaderValue, Body, Request, Response, StatusCode};
 use hudsucker::RequestOrResponse;
+use serde::{Deserialize, Serialize};
 
-#[derive(Debug, Clone)]
+#[derive(Debug, Clone, Deserialize, Serialize)]
+#[serde(rename_all = "kebab-case")]
 pub enum Action {
     Reject,
     Redirect(String),
     Modify(Modify),
 }
 
 impl Action {
-    pub fn new() -> Self {
-        Self::Redirect("https://lgf.im/".into())
-    }
-
     pub async fn do_req(&self, req: Request<Body>) -> RequestOrResponse {
         match self {
             Action::Reject => {

diff --git a/src/action/modify.rs b/src/action/modify.rs
@@ -1,35 +1,25 @@
 use hudsucker::hyper::{body::*, header, Body, Request, Response, StatusCode};
+use serde::{Deserialize, Serialize};
 
-#[derive(Debug, Clone)]
+#[derive(Debug, Clone, Deserialize, Serialize)]
+#[serde(rename_all = "kebab-case")]
 pub struct BodyModify {
     pub origin: String,
     pub new: String,
 }
 
-impl BodyModify {
-    pub fn new(origin: &str, new: &str) -> Self {
-        Self {
-            origin: String::from(origin),
-            new: String::from(new),
-        }
-    }
-}
-
-#[derive(Debug, Clone)]
+#[derive(Debug, Clone, Deserialize, Serialize)]
+#[serde(rename_all = "kebab-case")]
 pub enum Modify {
     Header,
     Cookie,
     Body(BodyModify),
 }
 
 impl Modify {
-    pub fn new_modify_body(origin: &str, new: &str) -> Self {
-        Self::Body(BodyModify::new(origin, new))
-    }
-
     pub async fn modify_req(&self, req: Request<Body>) -> Request<Body> {
         match self {
-            Modify::Body(bm) => req,
+            Modify::Body(_bm) => req,
             _ => req,
         }
     }

diff --git a/src/main.rs b/src/main.rs
@@ -6,6 +6,7 @@ mod filter;
 mod handler;
 mod rule;
 
+use clap::{App, Arg};
 use hudsucker::{rustls::internal::pemfile, *};
 use log::*;
 use std::net::SocketAddr;
@@ -17,9 +18,9 @@ async fn shutdown_signal() {
 }
 
 #[tokio::main]
-async fn main() {
+async fn run() {
     env_logger::init();
-    rule::add_rule_examples();
+    // rule::add_rule_examples_internal();
 
     let mut private_key_bytes: &[u8] = include_bytes!("../assets/ca/private.key");
     let mut ca_cert_bytes: &[u8] = include_bytes!("../assets/ca/cert.crt");
@@ -47,3 +48,28 @@ async fn main() {
         error!("{}", e);
     }
 }
+
+fn main() {
+    let matches = App::new("Good Man in the Middle")
+        .author("zu1k <i@lgf.im>")
+        .about("Use MITM technology to provide features like rewrite, redirect.")
+        .arg(
+            Arg::with_name("rule")
+                .short("f")
+                .long("rule")
+                .help("rule file")
+                .long_help("load rules from file")
+                .takes_value(true)
+                .required(true),
+        )
+        .get_matches();
+
+    let rule_file = matches
+        .value_of("rule")
+        .expect("rule file path should not be none");
+    if let Err(err) = rule::add_rule_file(rule_file) {
+        error!("parse rule file failed, err: {}", err);
+        std::process::exit(3);
+    }
+    run()
+}
diff --git a/src/rule/file.rs b/src/rule/file.rs
@@ -0,0 +1,66 @@
+use crate::action;
+use serde::{Deserialize, Serialize};
+use std::error::Error;
+use std::path::Path;
+use std::{fs::File, io::BufReader};
+
+#[derive(Debug, Clone, Deserialize, Serialize)]
+pub struct Rule {
+    pub name: String,
+    pub filter: Filter,
+    pub action: action::Action,
+}
+
+#[derive(Debug, Clone, Deserialize, Serialize)]
+#[serde(rename_all = "kebab-case")]
+pub enum Filter {
+    Domain(String),
+    DomainKeyword(String),
+    DomainPrefix(String),
+    DomainSuffix(String),
+    UrlRegex(String),
+}
+
+pub fn read_rules_from_file<P: AsRef<Path>>(path: P) -> Result<Vec<Rule>, Box<dyn Error>> {
+    let file = File::open(path)?;
+    let reader = BufReader::new(file);
+    let rules = serde_yaml::from_reader(reader)?;
+    Ok(rules)
+}
+
+#[test]
+fn test_gen_yaml() {
+    let rules = vec![
+        Rule {
+            name: "demo rule 1".into(),
+            filter: Filter::Domain("domain.com".into()),
+            action: action::Action::Modify(action::Modify::new_modify_body(
+                "origin body",
+                "new body",
+            )),
+        },
+        Rule {
+            name: "demo rule 2".into(),
+            filter: Filter::UrlRegex(r".*\.domain.com".into()),
+            action: action::Action::Modify(action::Modify::new_modify_body(
+                "origin body",
+                "new body",
+            )),
+        },
+    ];
+
+    let s = serde_yaml::to_string(&rules).expect("serde yaml failed!");
+    println!("{}", s);
+
+    let f = include_str!("../../assets/rules/demo.yaml");
+    let rules: Vec<Rule> = serde_yaml::from_str(f).expect("parse failed!");
+    println!("{:#?}", rules);
+    if let Filter::UrlRegex(r) = rules[0].clone().filter {
+        println!("{}", r)
+    }
+
+    // match read_rules_from_file("assets/rules/demo.yaml") {
+    //     Ok(rules) => println!("{:#?}", rules),
+    //     Err(err) => panic!("{}", err),
+    // }
+}