From 1d6392cb288765d17c39f75d49586e18b0fa3ece Mon Sep 17 00:00:00 2001
From: Sunshine <sunshine@uberspace.net>
Date: Sat, 1 Aug 2020 01:44:09 -0400
Subject: [PATCH] implement support for BASE tag

---
 src/html.rs | 92 ++++++++++++++++++++++++++++++++++++++++++++++++++++-
 src/main.rs | 16 +++++++---
 2 files changed, 102 insertions(+), 6 deletions(-)

diff --git a/src/html.rs b/src/html.rs
index 3739d3d..5a1ac4c 100644
--- a/src/html.rs
+++ b/src/html.rs
@@ -29,6 +29,31 @@ struct SrcSetItem<'a> {
 
 const ICON_VALUES: &[&str] = &["icon", "shortcut icon"];
 
+pub fn add_base_tag(document: &Handle, url: String) -> RcDom {
+    let mut buf: Vec<u8> = Vec::new();
+    serialize(&mut buf, document, SerializeOpts::default())
+        .expect("unable to serialize DOM into buffer");
+    let result = String::from_utf8(buf).unwrap();
+
+    let mut dom = html_to_dom(&result);
+    let doc = dom.get_document();
+    let html = get_child_node_by_name(&doc, "html");
+    let head = get_child_node_by_name(&html, "head");
+    let favicon_node = dom.create_element(
+        QualName::new(None, ns!(), local_name!("base")),
+        vec![Attribute {
+            name: QualName::new(None, ns!(), local_name!("href")),
+            value: format_tendril!("{}", url),
+        }],
+        Default::default(),
+    );
+
+    // Insert BASE tag into HEAD
+    head.children.borrow_mut().push(favicon_node.clone());
+
+    dom
+}
+
 pub fn add_favicon(document: &Handle, favicon_data_url: String) -> RcDom {
     let mut buf: Vec<u8> = Vec::new();
     serialize(&mut buf, document, SerializeOpts::default())
@@ -54,7 +79,7 @@ pub fn add_favicon(document: &Handle, favicon_data_url: String) -> RcDom {
         Default::default(),
     );
 
-    // Append favicon node to HEAD
+    // Insert favicon LINK tag into HEAD
     head.children.borrow_mut().push(favicon_node.clone());
 
     dom
@@ -205,6 +230,56 @@ pub fn has_proper_integrity(data: &[u8], integrity: &str) -> bool {
     }
 }
 
+pub fn has_base_tag(handle: &Handle) -> bool {
+    let mut found_base_tag: bool = false;
+
+    match handle.data {
+        NodeData::Document => {
+            // Dig deeper
+            for child in handle.children.borrow().iter() {
+                if has_base_tag(child) {
+                    found_base_tag = true;
+                    break;
+                }
+            }
+        }
+        NodeData::Element {
+            ref name,
+            ref attrs,
+            ..
+        } => {
+            match name.local.as_ref() {
+                "base" => {
+                    let attrs_mut = &mut attrs.borrow_mut();
+
+                    for attr in attrs_mut.iter_mut() {
+                        if &attr.name.local == "href" {
+                            if !attr.value.trim().is_empty() {
+                                found_base_tag = true;
+                                break;
+                            }
+                        }
+                    }
+                }
+                _ => {}
+            }
+
+            if !found_base_tag {
+                // Dig deeper
+                for child in handle.children.borrow().iter() {
+                    if has_base_tag(child) {
+                        found_base_tag = true;
+                        break;
+                    }
+                }
+            }
+        }
+        _ => {}
+    }
+
+    found_base_tag
+}
+
 pub fn has_favicon(handle: &Handle) -> bool {
     let mut found_favicon: bool = false;
 
@@ -600,6 +675,7 @@ pub fn walk_and_embed_assets(
                             }
                         }
                         LinkType::Unknown => {
+                            // Make sure that all other LINKs' href attributes are full URLs
                             for attr in attrs_mut.iter_mut() {
                                 let attr_name: &str = &attr.name.local;
                                 if attr_name.eq_ignore_ascii_case("href") {
@@ -612,6 +688,20 @@ pub fn walk_and_embed_assets(
                         }
                     }
                 }
+                "base" => {
+                    if is_http_url(url) {
+                        // Ensure BASE href is a full URL, not a relative one
+                        for attr in attrs_mut.iter_mut() {
+                            let attr_name: &str = &attr.name.local;
+                            if attr_name.eq_ignore_ascii_case("href") {
+                                let href_full_url =
+                                    resolve_url(&url, attr.value.trim()).unwrap_or_default();
+                                attr.value.clear();
+                                attr.value.push_slice(&href_full_url.as_str());
+                            }
+                        }
+                    }
+                }
                 "body" => {
                     // Find and remove background attribute(s), keep value of the last found one
                     let mut background: String = str!();
diff --git a/src/main.rs b/src/main.rs
index aa54333..e8bba9e 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -9,7 +9,8 @@ use std::process;
 use std::time::Duration;
 
 use monolith::html::{
-    add_favicon, has_favicon, html_to_dom, metadata_tag, stringify_document, walk_and_embed_assets,
+    add_base_tag, add_favicon, has_base_tag, has_favicon, html_to_dom, metadata_tag,
+    stringify_document, walk_and_embed_assets,
 };
 use monolith::opts::Options;
 use monolith::url::{
@@ -141,6 +142,14 @@ fn main() {
         process::exit(1);
     }
 
+    // Embed remote assets
+    walk_and_embed_assets(&mut cache, &client, &base_url, &dom.document, &options, 0);
+
+    // Take care of BASE tag
+    if is_http_url(base_url.clone()) && !has_base_tag(&dom.document) {
+        dom = add_base_tag(&dom.document, base_url.clone());
+    }
+
     // Request and embed /favicon.ico (unless it's already linked in the document)
     if !options.no_images && is_http_url(target_url) && !has_favicon(&dom.document) {
         let favicon_ico_url: String = resolve_url(&base_url, "/favicon.ico").unwrap();
@@ -163,15 +172,12 @@ fn main() {
         }
     }
 
-    // Embed remote assets
-    walk_and_embed_assets(&mut cache, &client, &base_url, &dom.document, &options, 0);
-
     // Serialize DOM tree
     let mut result: String = stringify_document(&dom.document, &options);
 
     // Add metadata tag
     if !options.no_metadata {
-        let metadata_comment = metadata_tag(&base_url);
+        let metadata_comment: String = metadata_tag(&base_url);
         result.insert_str(0, &metadata_comment);
         if metadata_comment.len() > 0 {
             result.insert_str(metadata_comment.len(), "\n");