remove /jump.php track links

Cargo.lock

@@ -1075,6 +1075,7 @@ dependencies = [
  "reqwest",
  "spin_on",
  "tokio",
+ "urlparse",
  "utf16string",
  "winapi",
 ]
@@ -1702,6 +1703,12 @@ dependencies = [
  "percent-encoding",
 ]
 
+[[package]]
+name = "urlparse"
+version = "0.7.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "110352d4e9076c67839003c7788d8604e24dcded13e0b375af3efaa8cf468517"
+
 [[package]]
 name = "utf16string"
 version = "0.2.0"

Cargo.toml

@@ -16,12 +16,13 @@ html_parser = "0.6.2"
 int-enum = { version = "0.4", optional = true }
 lazy_static = "1.4"
 json = "0.12"
-utf16string = { version= "0.2", optional = true }
 regex = "1"
 reqwest = { version = "0.11", features = ["brotli", "deflate", "gzip", "rustls-tls", "socks", "stream"] }
 RustyXML = "0.3"
 spin_on = "0.1.1"
 tokio = { version = "1.17", features = ["rt", "macros", "rt-multi-thread", "time"] }
+urlparse = "0.7"
+utf16string = { version= "0.2", optional = true }
 
 [build-dependencies]
 cmake = "0.1"

src/main.rs

@@ -11,6 +11,7 @@ extern crate lazy_static;
 extern crate tokio;
 extern crate regex;
 extern crate reqwest;
+extern crate urlparse;
 #[cfg(feature = "utf16string")]
 extern crate utf16string;
 extern crate xml;

src/parser/description.rs

@@ -1,4 +1,5 @@
 use crate::gettext;
+use crate::pixiv_link::remove_track;
 use html_parser::Dom;
 use html_parser::Node;
 use std::collections::HashMap;
@@ -98,7 +99,8 @@ impl DescriptionParser {
                     if href.is_some() {
                         let href = href.unwrap();
                         if href.is_some() {
-                            node.add_attr("href", href.as_ref().unwrap());
+                            let link = remove_track(href.as_ref().unwrap());
+                            node.add_attr("href", link.as_str());
                         }
                     }
                 }
@@ -171,4 +173,8 @@ fn test_parse_description() {
         Some(String::from("a [a\n[bc](a.com)d](b.com)\ndata")),
         parse_description("a <a href=\"b.com\">a<br/><a href=\"a.com\">bc</a>d</a><br>data")
     );
+    assert_eq!(
+        Some(String::from("https://a.com")),
+        parse_description("<a href=\"/jump.php?https%3A%2F%2Fa.com\">https://a.com</a>")
+    )
 }

src/pixiv_link.rs

@@ -1,6 +1,7 @@
 use crate::data::json::ToJson;
 use json::JsonValue;
 use regex::Regex;
+use reqwest::IntoUrl;
 use std::convert::TryInto;
 
 lazy_static! {
@@ -98,3 +99,19 @@ impl TryInto<u64> for &PixivID {
         }
     }
 }
+
+pub fn remove_track<U: IntoUrl>(url: U) -> String {
+    let s = String::from(url.as_str());
+    let u = urlparse::urlparse(s.as_str());
+    let path = u.path.as_str();
+    if path.ends_with("/jump.php") {
+        if u.query.is_some() {
+            let q = u.query.as_ref().unwrap();
+            let re = urlparse::unquote(q);
+            if re.is_ok() {
+                return re.unwrap();
+            }
+        }
+    }
+    s
+}