fix(scrape): preserve ordered list markers

Author: OnlyYu1996

src/cortex-cli/src/agent_cmd/tests.rs

@@ -3,10 +3,9 @@
 #[cfg(test)]
 mod tests {
     use crate::agent_cmd::cli::{CopyArgs, ExportArgs};
-    use crate::agent_cmd::loader::{
-        load_builtin_agents, parse_frontmatter, read_file_with_encoding,
-    };
+    use crate::agent_cmd::loader::{load_builtin_agents, parse_frontmatter};
     use crate::agent_cmd::types::AgentMode;
+    use crate::utils::file::read_file_with_encoding;
 
     #[test]
     fn test_read_file_with_utf8() {

src/cortex-cli/src/scrape_cmd/html.rs

@@ -311,7 +311,8 @@ fn process_node_to_markdown(
                         }
                         "li" => {
                             let indent = "  ".repeat(list_depth.saturating_sub(1));
-                            output.push_str(&format!("\n{indent}- "));
+                            let marker = list_item_marker(element_ref);
+                            output.push_str(&format!("\n{indent}{marker}"));
                             process_node_to_markdown(
                                 element_ref,
                                 output,
@@ -448,6 +449,28 @@ fn process_node_to_markdown(
     }
 }
 
+fn list_item_marker(element_ref: scraper::ElementRef) -> String {
+    let Some(parent) = element_ref.parent().and_then(scraper::ElementRef::wrap) else {
+        return "- ".to_string();
+    };
+
+    if parent.value().name() != "ol" {
+        return "- ".to_string();
+    }
+
+    let start = parent
+        .attr("start")
+        .and_then(|value| value.parse::<usize>().ok())
+        .unwrap_or(1);
+    let previous_items = element_ref
+        .prev_siblings()
+        .filter_map(scraper::ElementRef::wrap)
+        .filter(|sibling| sibling.value().name() == "li")
+        .count();
+
+    format!("{}. ", start + previous_items)
+}
+
 /// Convert HTML to plain text.
 pub fn html_to_text(html: &str) -> String {
     let cleaned = remove_unwanted_elements(html);

src/cortex-cli/src/scrape_cmd/tests.rs

@@ -41,6 +41,28 @@ mod tests {
         assert!(!md_no_images.contains("!["));
     }
 
+    #[test]
+    fn test_html_to_markdown_ordered_lists() {
+        let html = r#"
+            <ol>
+                <li>Install Rust</li>
+                <li>Run cortex</li>
+                <li>Check output<ul><li>Keep nested unordered item</li></ul></li>
+            </ol>
+            <ol start="4">
+                <li>Continue numbering</li>
+            </ol>
+        "#;
+        let md = html_to_markdown(html, false, false);
+
+        assert!(md.contains("1. Install Rust"), "got: {md}");
+        assert!(md.contains("2. Run cortex"), "got: {md}");
+        assert!(md.contains("3. Check output"), "got: {md}");
+        assert!(md.contains("  - Keep nested unordered item"), "got: {md}");
+        assert!(md.contains("4. Continue numbering"), "got: {md}");
+        assert!(!md.contains("- Install Rust"), "got: {md}");
+    }
+
     #[test]
     fn test_html_to_text() {
         let html = "<h1>Title</h1><p>Hello <strong>world</strong>!</p>";