fix(web): escape september and kineto proxy information

1 files changed, 35 insertions, 23 deletions

diff --git a/src/modules/web.rs b/src/modules/web.rs
index 2a0bce6..b0d6b40 100644
--- a/src/modules/web.rs
+++ b/src/modules/web.rs
@@ -82,13 +82,18 @@ To visit the web version of this exact page, <https://fuwn.me/web>, you would vi
         };
         let parser = dom.parser();
         let mut nodes = dom.nodes().iter().peekable();
+        let mut in_proxy_information = false;
 
         while let Some(element) = nodes.next() {
           let mut text = String::new();
 
           let () = element.as_tag().map_or((), |tag| {
             match tag.name().as_utf8_str().to_string().as_str() {
-              "p" => {
+              "p" if !in_proxy_information => {
+                if tag.inner_html(parser).contains("<img") {
+                  return;
+                }
+
                 contents.push(format!(
                   "{}\n\n",
                   tag
@@ -99,7 +104,7 @@ To visit the web version of this exact page, <https://fuwn.me/web>, you would vi
                     .unwrap_or(&"A parse error occurred in this location.")
                 ));
               }
-              "a" => {
+              "a" if !in_proxy_information => {
                 contents.pop();
                 contents.push(format!(
                   "=> {} {}\n{}",
@@ -121,52 +126,59 @@ To visit the web version of this exact page, <https://fuwn.me/web>, you would vi
                   })
                 ));
               }
-              "h1" => {
+              "h1" if !in_proxy_information => {
                 contents.push(format!("# {}\n\n", tag.inner_text(parser)));
               }
-              "h2" => {
+              "h2" if !in_proxy_information => {
                 contents.push(format!("## {}\n\n", tag.inner_text(parser)));
               }
-              "h3" => {
+              "h3" if !in_proxy_information => {
                 contents.push(format!("### {}\n\n", tag.inner_text(parser)));
               }
-              "pre" => {
+              "pre" if !in_proxy_information => {
                 contents.push(format!("```\n{}```\n", tag.inner_text(parser)));
               }
-              "blockquote" => {
+              "blockquote" if !in_proxy_information => {
                 contents.push(format!("> {}\n\n", tag.inner_text(parser)));
               }
-              "li" => {
+              "li" if !in_proxy_information => {
                 contents.push(format!("* {}\n", tag.inner_text(parser)));
               }
               "html" | "head" | "script" | "link" | "title" | "body" | "ul"
-              | "style" => {}
+              | "style"
+                if !in_proxy_information => {}
               _ => {
                 text = tag.inner_text(parser).to_string();
 
-                if text.contains("Proxy information")
-                  || text.contains("Proxied content from")
+                if (text.contains("Proxy Information")
+                  || text.contains("Proxied content from"))
+                  && tag.inner_html(parser).contains("<summary>")
                 {
+                  in_proxy_information = true;
+
                   return;
                 }
 
-                println!(
-                  "{}: {}",
-                  tag.name().as_utf8_str(),
-                  tag.inner_text(parser)
-                );
+                if (text.contains("This content has been proxied by September")
+                  || text.contains(
+                    "Be advised that no attempt was made to verify the remote \
+                     SSL certificate.",
+                  ))
+                  && in_proxy_information
+                {
+                  in_proxy_information = false;
+
+                  return;
+                }
+
+                if in_proxy_information {
+                  return;
+                }
 
                 contents.push(format!("{}\n\n", tag.inner_text(parser)));
               }
             }
           });
-
-          // Covers September and Kineto
-          if text == "Proxy information"
-            || text.contains("Proxied content from")
-          {
-            break;
-          }
         }
 
         contents.join("")