From 766498ebfb382b0edd87030639c10b23da44775d Mon Sep 17 00:00:00 2001
From: Wilson Lin <code@wilsonl.in>
Date: Fri, 6 Aug 2021 23:07:55 +1000
Subject: [PATCH] Fix next sibling calc

---
 src/ast/mod.rs           |  1 +
 src/lib.rs               |  9 +++++++--
 src/minify/content.rs    | 18 ++++++++++++++++--
 src/minify/tests/attr.rs |  2 +-
 src/parse/content.rs     | 25 +------------------------
 5 files changed, 26 insertions(+), 29 deletions(-)

diff --git a/src/ast/mod.rs b/src/ast/mod.rs
index 6272065..9cebffc 100644
--- a/src/ast/mod.rs
+++ b/src/ast/mod.rs
@@ -39,6 +39,7 @@ pub enum NodeData {
         closing_tag: ElementClosingTag,
         name: Vec<u8>,
         namespace: Namespace,
+        // WARNING: This should only be set during minification, as minification can alter tree (e.g. remove text nodes completely).
         // If the next text or element sibling is an element, this will be set to its tag name.
         // Otherwise, this will be empty. It should be empty on creation.
         next_sibling_element_name: Vec<u8>,
diff --git a/src/lib.rs b/src/lib.rs
index 8835a0d..46ac75a 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -1,4 +1,4 @@
-use crate::cfg::Cfg;
+pub use crate::cfg::Cfg;
 use crate::minify::content::minify_content;
 use crate::parse::content::parse_content;
 use crate::parse::Code;
@@ -31,8 +31,13 @@ mod whitespace;
 ///
 /// let mut code: &[u8] = b"<p>  Hello, world!  </p>";
 /// let cfg = &Cfg {
-///     minify_js: false,
 ///     minify_css: false,
+///     minify_js: false,
+///     omit_closing_tags: true,
+///     remove_bangs: false,
+///     remove_comments: true,
+///     remove_processing_instructions: false,
+///     remove_spaces_between_attributes: true,
 /// };
 /// let minified = minify(&code, cfg);
 /// assert_eq!(minified, b"<p>Hello, world!".to_vec());
diff --git a/src/minify/content.rs b/src/minify/content.rs
index e7fd407..80afde4 100644
--- a/src/minify/content.rs
+++ b/src/minify/content.rs
@@ -59,9 +59,23 @@ pub fn minify_content(
     let mut found_first_text_or_elem = false;
     let mut index_of_last_nonempty_text_or_elem: isize = -1;
     let mut index_of_last_text_or_elem: isize = -1;
-    for (i, n) in nodes.iter_mut().enumerate() {
+    for i in 0..nodes.len() {
+        let (previous_nodes, next_nodes) = nodes.split_at_mut(i);
+        let n = &mut next_nodes[0];
         match n {
-            NodeData::Element { .. } => {
+            NodeData::Element { name, .. } => {
+                if index_of_last_nonempty_text_or_elem > -1 {
+                    match &mut previous_nodes[index_of_last_nonempty_text_or_elem as usize] {
+                        NodeData::Element {
+                            next_sibling_element_name,
+                            ..
+                        } => {
+                            debug_assert!(next_sibling_element_name.is_empty());
+                            next_sibling_element_name.extend_from_slice(name);
+                        }
+                        _ => {}
+                    };
+                };
                 found_first_text_or_elem = true;
                 index_of_last_nonempty_text_or_elem = i as isize;
                 index_of_last_text_or_elem = i as isize;
diff --git a/src/minify/tests/attr.rs b/src/minify/tests/attr.rs
index 2d41ce6..4dfe178 100644
--- a/src/minify/tests/attr.rs
+++ b/src/minify/tests/attr.rs
@@ -13,7 +13,7 @@ fn test_encode_using_double_quotes() {
 
 #[test]
 fn test_encode_using_single_quotes() {
-    let min = encode_using_single_quotes(br#"'abr'aca'dab   &amp&amp;'10';'8'$4 a'"#);
+    let min = encode_using_single_quotes(br#"'abr'aca'dab   &amp&amp;''10';''8'$4 a'"#);
     assert_eq!(
         min.str(),
         r#"'&#39abr&#39aca&#39dab   &amp&amp;&#39&#39;10&#39;;&#39&#39;8&#39$4 a&#39'"#,
diff --git a/src/parse/content.rs b/src/parse/content.rs
index 002fb54..a6b211f 100644
--- a/src/parse/content.rs
+++ b/src/parse/content.rs
@@ -82,9 +82,6 @@ pub fn parse_content(
     // We assume the closing tag has been omitted until we see one explicitly before EOF (or it has been omitted as per the spec).
     let mut closing_tag_omitted = true;
     let mut nodes = Vec::<NodeData>::new();
-    // This is set to the index of the last text or element node that is an element node.
-    // If it's not an element node, this is set to -1.
-    let mut last_elem_node_pos: isize = -1;
     loop {
         let (text_len, mut typ) = match CONTENT_TYPE_MATCHER.0.find(&code.str()) {
             Some(m) => (m.start(), CONTENT_TYPE_MATCHER.1[m.pattern()]),
@@ -98,7 +95,6 @@ pub fn parse_content(
                 Some(NodeData::Text { value }) => value.extend_from_slice(&text),
                 _ => nodes.push(NodeData::Text { value: text }),
             };
-            last_elem_node_pos = -1;
         };
         // Check using Parsing.md tag rules.
         if typ == OpeningTag || typ == ClosingTag {
@@ -130,26 +126,7 @@ pub fn parse_content(
         };
         match typ {
             Text => break,
-            OpeningTag => {
-                let node = parse_element(code, ns, parent);
-                if last_elem_node_pos > -1 {
-                    match (&mut nodes[last_elem_node_pos as usize], &node) {
-                        (
-                            NodeData::Element {
-                                next_sibling_element_name,
-                                ..
-                            },
-                            NodeData::Element { name, .. },
-                        ) => {
-                            debug_assert!(next_sibling_element_name.is_empty());
-                            next_sibling_element_name.extend_from_slice(name);
-                        }
-                        _ => unreachable!(),
-                    }
-                }
-                last_elem_node_pos = nodes.len() as isize;
-                nodes.push(node);
-            }
+            OpeningTag => nodes.push(parse_element(code, ns, parent)),
             ClosingTag => {
                 closing_tag_omitted = false;
                 break;