feat(search): add line numbers to semantic search results (#2085)

Author: tusharmath

crates/forge_app/src/snapshots/forge_app__operation__tests__sem_search_with_results.snap

@@ -11,25 +11,25 @@ expression: to_value(actual)
 <file_chunk
   file_path="src/retry.rs"
   lines="10-19"
-><![CDATA[fn retry_with_backoff(max_attempts: u32) {
-    let mut delay = 100;
-    for attempt in 0..max_attempts {
-        if try_operation().is_ok() {
-            return;
-        }
-        thread::sleep(Duration::from_millis(delay));
-        delay *= 2;
-    }
-}]]>
+><![CDATA[10:fn retry_with_backoff(max_attempts: u32) {
+11:    let mut delay = 100;
+12:    for attempt in 0..max_attempts {
+13:        if try_operation().is_ok() {
+14:            return;
+15:        }
+16:        thread::sleep(Duration::from_millis(delay));
+17:        delay *= 2;
+18:    }
+19:}]]>
 </file_chunk>
 <file_chunk
   file_path="src/http/client.rs"
   lines="45-50"
-><![CDATA[async fn request_with_retry(&self, url: &str) -> Result<Response> {
-    const MAX_RETRIES: usize = 3;
-    let mut backoff = ExponentialBackoff::default();
-    // Implementation...
-}]]>
+><![CDATA[45:async fn request_with_retry(&self, url: &str) -> Result<Response> {
+46:    const MAX_RETRIES: usize = 3;
+47:    let mut backoff = ExponentialBackoff::default();
+48:    // Implementation...
+49:}]]>
 </file_chunk>
 </query_result>
 </sem_search_results>

crates/forge_app/src/snapshots/forge_app__operation__tests__sem_search_with_usecase.snap

@@ -11,9 +11,9 @@ expression: to_value(actual)
 <file_chunk
   file_path="src/auth.rs"
   lines="10-12"
-><![CDATA[fn authenticate_user(token: &str) -> Result<User> {
-    verify_jwt(token)
-}]]>
+><![CDATA[10:fn authenticate_user(token: &str) -> Result<User> {
+11:    verify_jwt(token)
+12:}]]>
 </file_chunk>
 </query_result>
 </sem_search_results>

crates/forge_domain/src/line_numbers.rs

@@ -10,10 +10,21 @@ pub trait LineNumbers {
 
 impl<T: AsRef<str>> LineNumbers for T {
     fn numbered_from(&self, start: usize) -> String {
-        self.as_ref()
-            .lines()
+        let text = self.as_ref();
+        let lines: Vec<&str> = text.lines().collect();
+
+        if lines.is_empty() {
+            return String::new();
+        }
+
+        // Calculate the width needed for the largest line number
+        let max_line_number = start + lines.len() - 1;
+        let width = max_line_number.to_string().len();
+
+        lines
+            .into_iter()
             .enumerate()
-            .map(|(i, line)| format!("{}:{}", start + i, line))
+            .map(|(i, line)| format!("{:>width$}:{}", start + i, line, width = width))
             .collect::<Vec<_>>()
             .join("\n")
     }
@@ -57,4 +68,51 @@ mod tests {
         let expected = "1:line1\n2:\n3:line3";
         assert_eq!(text.numbered(), expected);
     }
+
+    #[test]
+    fn test_numbered_right_aligned_single_digit() {
+        let text = "line1\nline2\nline3";
+        let expected = "1:line1\n2:line2\n3:line3";
+        assert_eq!(text.numbered(), expected);
+    }
+
+    #[test]
+    fn test_numbered_right_aligned_two_digits() {
+        let text = "a\nb\nc\nd\ne\nf\ng\nh\ni\nj\nk";
+        let expected = " 1:a\n 2:b\n 3:c\n 4:d\n 5:e\n 6:f\n 7:g\n 8:h\n 9:i\n10:j\n11:k";
+        assert_eq!(text.numbered(), expected);
+    }
+
+    #[test]
+    fn test_numbered_right_aligned_three_digits() {
+        let mut lines = Vec::new();
+        for i in 1..=100 {
+            lines.push(format!("line{}", i));
+        }
+        let text = lines.join("\n");
+        let actual = text.numbered();
+
+        // Check first line has 2 leading spaces (001 -> "  1")
+        assert!(actual.starts_with("  1:line1"));
+        // Check line 10 has 1 leading space (010 -> " 10")
+        assert!(actual.contains("\n 10:line10\n"));
+        // Check line 100 has no leading spaces (100 -> "100")
+        assert!(actual.contains("\n100:line100"));
+    }
+
+    #[test]
+    fn test_numbered_from_right_aligned() {
+        let text = "alpha\nbeta\ngamma\ndelta";
+        // Starting from 98, so max is 101 (3 digits)
+        let expected = " 98:alpha\n 99:beta\n100:gamma\n101:delta";
+        assert_eq!(text.numbered_from(98), expected);
+    }
+
+    #[test]
+    fn test_numbered_from_crosses_digit_boundary() {
+        let text = "line8\nline9\nline10\nline11";
+        // Starting from 8, max is 11 (2 digits)
+        let expected = " 8:line8\n 9:line9\n10:line10\n11:line11";
+        assert_eq!(text.numbered_from(8), expected);
+    }
 }

crates/forge_domain/src/node.rs

@@ -3,7 +3,7 @@ use derive_setters::Setters;
 use serde::{Deserialize, Serialize};
 use uuid::Uuid;
 
-use crate::WorkspaceId;
+use crate::{LineNumbers, WorkspaceId};
 
 /// Progress events emitted during codebase indexing
 #[derive(Debug, Clone, PartialEq)]
@@ -497,14 +497,18 @@ impl NodeData {
 
         match self {
             Self::FileChunk { file_path, content, start_line, end_line } => {
+                let numbered_content = content.numbered_from(*start_line as usize);
                 Element::new("file_chunk")
                     .attr("file_path", file_path)
                     .attr("lines", format!("{}-{}", start_line, end_line))
-                    .cdata(content)
+                    .cdata(numbered_content)
+            }
+            Self::File { file_path, content, .. } => {
+                let numbered_content = content.numbered();
+                Element::new("file")
+                    .attr("file_path", file_path)
+                    .cdata(numbered_content)
             }
-            Self::File { file_path, content, .. } => Element::new("file")
-                .attr("file_path", file_path)
-                .cdata(content),
             Self::FileRef { file_path, .. } => {
                 Element::new("file_ref").attr("file_path", file_path)
             }

crates/forge_domain/src/snapshots/forge_domain__node__tests__codebase_query_result_with_results.snap

@@ -10,6 +10,6 @@ expression: actual
 <file_chunk
   file_path="src/auth.rs"
   lines="10-15"
-><![CDATA[fn authenticate() {}]]>
+><![CDATA[10:fn authenticate() {}]]>
 </file_chunk>
 </query_result>