From e34e2ef75f0976e8d09e26fcd89e4b0d4e5f65f7 Mon Sep 17 00:00:00 2001
From: Tom Alexander <tom@fizz.buzz>
Date: Sat, 22 Feb 2025 15:44:45 -0500
Subject: [PATCH] Implement the highlighted src intermediate format.

I am going to remove the enum because I realized plain src blocks can just be highlighted src blocks with only RawText entries.
---
 src/intermediate/src_block.rs | 61 +++++++++++++++++++++++++++--------
 1 file changed, 48 insertions(+), 13 deletions(-)

diff --git a/src/intermediate/src_block.rs b/src/intermediate/src_block.rs
index 5566cbf..ca88164 100644
--- a/src/intermediate/src_block.rs
+++ b/src/intermediate/src_block.rs
@@ -24,11 +24,23 @@ pub(crate) struct PlainSrcBlock {
 
 #[derive(Debug, Clone)]
 pub(crate) struct HighlightedSrcBlock {
-    pub(crate) lines: Vec<String>,
+    pub(crate) lines: Vec<HighlightedSrcLine>,
     pub(crate) language: Option<String>,
     pub(crate) post_blank: organic::types::PostBlank,
 }
 
+#[derive(Debug, Clone)]
+pub(crate) struct HighlightedSrcLine {
+    pub(crate) children: Vec<HighlightedSrcSegment>,
+}
+
+#[derive(Debug, Clone)]
+pub(crate) enum HighlightedSrcSegment {
+    RawText(String),
+    HighlightStart { name: String },
+    HighlightEnd,
+}
+
 intermediate!(
     ISrcBlock,
     &'orig organic::types::SrcBlock<'parse>,
@@ -97,6 +109,14 @@ intermediate!(
     }
 );
 
+impl HighlightedSrcLine {
+    pub(crate) fn new() -> HighlightedSrcLine {
+        HighlightedSrcLine {
+            children: Vec::new(),
+        }
+    }
+}
+
 fn ascii_whitespace_value(c: char) -> usize {
     match c {
         ' ' => 1,
@@ -107,7 +127,7 @@ fn ascii_whitespace_value(c: char) -> usize {
     }
 }
 
-fn highlight_nix<L>(lines: &[L]) -> Result<Vec<String>, CustomError>
+fn highlight_nix<L>(lines: &[L]) -> Result<Vec<HighlightedSrcLine>, CustomError>
 where
     L: Borrow<str>,
 {
@@ -126,28 +146,43 @@ where
         .highlight(&config, combined_text.as_bytes(), None, |_| None)
         .unwrap();
 
-    let mut highlighted_text = Vec::new();
+    let mut highlighted_text: Vec<HighlightedSrcLine> = Vec::with_capacity(lines.len());
+    let mut current_line = HighlightedSrcLine::new();
     for event in highlights {
         match event.unwrap() {
             HighlightEvent::Source { start, end } => {
-                highlighted_text.push(Cow::Borrowed(&combined_text[start..end]));
+                let mut span = &combined_text[start..end];
+                while let Some(line_break_index) = span.find('\n') {
+                    let first_line = &span[..(line_break_index + 1)];
+                    current_line
+                        .children
+                        .push(HighlightedSrcSegment::RawText(first_line.to_owned()));
+                    highlighted_text.push(current_line);
+                    current_line = HighlightedSrcLine::new();
+                    span = &span[(line_break_index + 1)..];
+                }
+                if !span.is_empty() {
+                    current_line
+                        .children
+                        .push(HighlightedSrcSegment::RawText(span.to_owned()));
+                }
             }
             HighlightEvent::HighlightStart(s) => {
-                let class_name = format!("srchl_{}", highlight_names[s.0]);
-                highlighted_text.push(Cow::Owned(format!(r#"<span class="{}">"#, class_name)));
+                current_line
+                    .children
+                    .push(HighlightedSrcSegment::HighlightStart {
+                        name: highlight_names[s.0].to_owned(),
+                    });
             }
             HighlightEvent::HighlightEnd => {
-                highlighted_text.push(Cow::Borrowed(r#"</span>"#));
+                current_line
+                    .children
+                    .push(HighlightedSrcSegment::HighlightEnd);
             }
         }
     }
 
-    let highlighted_text = highlighted_text.join("");
-    let lines = highlighted_text
-        .split_inclusive('\n')
-        .map(str::to_owned)
-        .collect();
-    Ok(lines)
+    Ok(highlighted_text)
 }
 
 // use tree_sitter::Parser;