fix: should trim bullet even when spaces in bullet chars

ee/tabby-webserver/src/service/page.rs

@@ -29,7 +29,7 @@ use tabby_schema::{
 use super::graphql_pagination_to_filter;
 use crate::service::utils::{
     convert_messages_to_chat_completion_request, convert_user_message_to_chat_completion_request,
-    prompt::{request_llm_stream, request_llm_with_message, transform_line_items},
+    prompt::{request_llm_stream, request_llm_with_message, transform_line_items, trim_title},
 };
 
 struct PageServiceImpl {
@@ -367,7 +367,7 @@ impl PageServiceImpl {
         .await?;
 
         let title = request_llm_with_message(self.chat.clone(), messages).await?;
-        let title = trim_title(title.as_ref());
+        let title = trim_title(title.as_str());
 
         self.db
             .update_page_title(page_id.as_rowid()?, title)
@@ -376,12 +376,6 @@ impl PageServiceImpl {
     }
 }
 
-fn trim_title(title: &str) -> &str {
-    // take first line.
-    let title = title.lines().next().unwrap_or(title);
-    title.trim_matches(&['"', '#', ' ', '-', '*'][..]).trim()
-}
-
 async fn build_chat_messages(
     context: Arc<dyn ContextService>,
     policy: &AccessPolicy,
@@ -447,7 +441,7 @@ pub async fn generate_page_sections(
     let titles = request_llm_with_message(chat.clone(), messages).await?;
     Ok(transform_line_items(&titles)
         .into_iter()
-        .map(|x| trim_title(&x).to_owned())
+        .map(|x| trim_title(x.as_str()).to_owned())
         .collect())
 }
 

ee/tabby-webserver/src/service/utils/prompt.rs

@@ -103,17 +103,32 @@ pub fn transform_line_items(content: &str) -> Vec<String> {
         .lines()
         .map(trim_bullet)
         .filter(|x| !x.is_empty())
+        .map(|x| x.to_owned())
         .collect()
 }
 
 /// Trims leading and trailing bullet-like characters or digits from the provided string and returns the trimmed result.
-pub fn trim_bullet(s: &str) -> String {
-    let is_bullet = |c: char| c == '-' || c == '*' || c == '.' || c.is_numeric();
-    s.trim()
-        .trim_start_matches(is_bullet)
-        .trim_end_matches(is_bullet)
-        .trim()
-        .to_owned()
+fn trim_bullet(s: &str) -> &str {
+    let s = s.trim();
+    if s.starts_with('-') || s.starts_with('+') || s.starts_with('*') {
+        return s.trim_matches(&['-', '+', '*', ' '][..]);
+    }
+
+    // Check for numbered list
+    if let Some(pos) = s.find(". ") {
+        let prefix = &s[..pos];
+        if prefix.chars().all(|c| c.is_numeric()) {
+            return trim_bullet(s[pos + 2..].trim_start());
+        }
+    }
+
+    s
+}
+
+pub fn trim_title(title: &str) -> &str {
+    // take first line.
+    let title = title.lines().next().unwrap_or(title).trim();
+    trim_bullet(title.trim_matches(&['"', '#', ' ', '-', '*'][..]))
 }
 
 /// Checks if the `check` string is contained within `content` in a case-insensitive manner.
@@ -130,10 +145,12 @@ mod tests {
         assert_eq!(trim_bullet("- Hello"), "Hello");
         assert_eq!(trim_bullet("* World"), "World");
         assert_eq!(trim_bullet("1. Test"), "Test");
-        assert_eq!(trim_bullet(".Dot"), "Dot");
 
         assert_eq!(trim_bullet("- Hello -"), "Hello");
-        assert_eq!(trim_bullet("1. Test 1"), "Test");
+        assert_eq!(trim_bullet("1. Test 1"), "Test 1");
+        assert_eq!(trim_bullet("12. Test 1"), "Test 1");
+        assert_eq!(trim_bullet("1. Test 1."), "Test 1.");
+        assert_eq!(trim_bullet("1 Test "), "1 Test");
 
         assert_eq!(trim_bullet("--** Mixed"), "Mixed");
 
@@ -145,6 +162,20 @@ mod tests {
 
         assert_eq!(trim_bullet("Hello World"), "Hello World");
 
-        assert_eq!(trim_bullet("1. *Bold* and -italic-"), "*Bold* and -italic");
+        assert_eq!(trim_bullet("1. *Bold* and -italic-"), "Bold* and -italic");
+        assert_eq!(trim_bullet("1. **Bold* and -italic"), "Bold* and -italic");
+    }
+
+    #[test]
+    fn test_trim_title() {
+        assert_eq!(trim_title("# - Hello"), "Hello");
+        assert_eq!(trim_title("# * World"), "World");
+        assert_eq!(trim_title("# 1. Test"), "Test");
+
+        assert_eq!(trim_title("## - Hello -"), "Hello");
+        assert_eq!(trim_title("  # 1. Test 1"), "Test 1");
+        assert_eq!(trim_title("12. Test 1"), "Test 1");
+        assert_eq!(trim_title(r#""1. Test 1.""#), "Test 1.");
+        assert_eq!(trim_title("## **Test "), "Test");
     }
 }