From 3d3ecb52b2dbcf74c22ad1aba0253f8beb3c76fe Mon Sep 17 00:00:00 2001
From: Evan Lloyd New-Schmidt <evan@new-schmidt.com>
Date: Tue, 15 Aug 2023 16:19:16 -0400
Subject: [PATCH] Minify whitespace between elements

Signed-off-by: Evan Lloyd New-Schmidt <evan@new-schmidt.com>
---
 src/html.rs | 25 +++++++++++++++++++++++++
 1 file changed, 25 insertions(+)

diff --git a/src/html.rs b/src/html.rs
index 8d75cc6..fa85ce5 100644
--- a/src/html.rs
+++ b/src/html.rs
@@ -134,6 +134,8 @@ pub fn simplify_html(document: &mut Html, lang: &str) {
     remove_attrs(document);
 
     final_expansions(document);
+
+    remove_toplevel_whitespace(document);
 }
 
 fn remove_ids(document: &mut Html, ids: impl IntoIterator<Item = NodeId>) {
@@ -158,6 +160,29 @@ fn remove_non_element_nodes(document: &mut Html) {
     remove_ids(document, to_remove.drain(..));
 }
 
+fn remove_toplevel_whitespace(document: &mut Html) {
+    let mut to_remove = Vec::new();
+
+    let parent = document.tree.root();
+
+    for el in parent.children() {
+        let Some(t) = el.value().as_text() else {
+            continue;
+        };
+
+        if t.chars().all(char::is_whitespace) {
+            to_remove.push(el.id());
+        }
+    }
+
+    trace!(
+        "Removing {} whitespace text nodes children from {:?}",
+        to_remove.len(),
+        parent.value(),
+    );
+    remove_ids(document, to_remove.drain(..));
+}
+
 fn remove_empty_sections(document: &mut Html) {
     let mut to_remove = Vec::new();
     for el in document.select(&HEADERS) {