From d7a53b232dc5ef48509741e24dba40aa4dab5791 Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 00:03:24 -0500
Subject: [PATCH 01/21] feat: add diff normalization logic

---
 crates/util/src/diff_standardizer.rs | 212 +++++++++++++++++++++++++++
 1 file changed, 212 insertions(+)
 create mode 100644 crates/util/src/diff_standardizer.rs

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
new file mode 100644
index 000000000..04db3ca37
--- /dev/null
+++ b/crates/util/src/diff_standardizer.rs
@@ -0,0 +1,212 @@
+use git2::DiffOptions;
+use marzano_util::diff::{parse_modified_ranges, FileDiff};
+use tempfile::tempdir;
+use anyhow::Result;
+
+pub fn parse_unified_diff(contents: String) -> Result<Vec<FileDiff>> {
+    let parsed = parse_modified_ranges(&contents)
+        .map_err(|e| anyhow::anyhow!("{:?}", e))?;
+    Ok(parsed)
+}
+
+/// Given a before and after for a file, edit the *after* to not include any spurious changes
+pub fn standardize_rewrite(before: String, after: String) -> Result<String> {
+    let mut diff_opts = DiffOptions::new();
+    diff_opts.ignore_whitespace(true);
+
+    // Create a temporary directory for the repository
+    let temp_dir = tempdir()
+        .map_err(|e| anyhow::anyhow!("Failed to create temp directory: {:?}", e))?;
+
+    // Create blobs for diffing
+    let repo = git2::Repository::init_opts(
+        temp_dir.path(),
+        &git2::RepositoryInitOptions::new()
+            .bare(true)
+            .initial_head("main"),
+    )
+    .map_err(|e| anyhow::anyhow!("Failed to create temp repo: {:?}", e))?;
+
+    let left_oid = repo
+        .blob(before.as_bytes())
+        .map_err(|e| anyhow::anyhow!("Failed to create left blob: {:?}", e))?;
+    let right_oid = repo
+        .blob(after.as_bytes())
+        .map_err(|e| anyhow::anyhow!("Failed to create right blob: {:?}", e))?;
+
+    let left_blob = repo
+        .find_blob(left_oid)
+        .map_err(|e| anyhow::anyhow!("Failed to find left blob: {:?}", e))?;
+    let right_blob = repo
+        .find_blob(right_oid)
+        .map_err(|e| anyhow::anyhow!("Failed to find right blob: {:?}", e))?;
+
+    // Generate diff
+    let mut standardized = after.clone(); // Start with the full after content
+    let mut has_changes = false;
+    let diff = repo
+        .diff_blobs(
+            Some(&left_blob),
+            None,
+            Some(&right_blob),
+            None,
+            Some(&mut diff_opts),
+            None,
+            None,
+            None,
+            Some(&mut |delta, _hunk, line| {
+                has_changes = true;
+                match line.origin() {
+                    '+' | ' ' => {
+                        if let Ok(_) = std::str::from_utf8(line.content()) {
+                            // Content will be taken from the after string
+                        }
+                    }
+                    '-' => {
+                        // Removed lines are ignored as we're using the after content
+                    }
+                    _ => {}
+                }
+                true
+            }),
+        )
+        .map_err(|e| anyhow::anyhow!("Failed to generate diff: {:?}", e))?;
+
+    // If no changes were detected in the diff, return the after content as is
+    if !has_changes {
+        standardized = after;
+    }
+
+    // The temporary directory will be automatically cleaned up when temp_dir is dropped
+    Ok(standardized)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use insta::assert_snapshot;
+
+    #[test]
+    fn test_basic_rewrite() -> Result<()> {
+        let before = "Hello world\n".to_string();
+        let after = "Hello Rust\n".to_string();
+        let result = standardize_rewrite(before, after)?;
+        assert_eq!(result, "Hello Rust\n");
+        assert_snapshot!(result);
+        Ok(())
+    }
+
+    #[test]
+    fn test_whitespace_handling() -> Result<()> {
+        let before = "function test() {\n    console.log('test');\n}\n".to_string();
+        let after = "function test(){\nconsole.log('test');\n}\n".to_string();
+        let result = standardize_rewrite(before, after)?;
+        assert_eq!(result, "function test(){\nconsole.log('test');\n}\n");
+        assert_snapshot!(result);
+        Ok(())
+    }
+
+    #[test]
+    fn test_empty_files() -> Result<()> {
+        let before = "".to_string();
+        let after = "".to_string();
+        let result = standardize_rewrite(before, after)?;
+        assert_eq!(result, "");
+        assert_snapshot!(result);
+        Ok(())
+    }
+
+    #[test]
+    fn test_multiline_changes() -> Result<()> {
+        let before = "line1\nline2\nline3\n".to_string();
+        let after = "line1\nmodified line2\nline3\nnew line4\n".to_string();
+        let result = standardize_rewrite(before, after)?;
+        assert_eq!(result, "line1\nmodified line2\nline3\nnew line4\n");
+        assert_snapshot!(result);
+        Ok(())
+    }
+
+    #[test]
+    fn test_no_changes() -> Result<()> {
+        let content = "unchanged content\n".to_string();
+        let result = standardize_rewrite(content.clone(), content)?;
+        assert_eq!(result, "unchanged content\n");
+        assert_snapshot!(result);
+        Ok(())
+    }
+
+    #[test]
+    fn test_mixed_changes_in_large_file() -> Result<()> {
+        let before = r#"
+// This is a large file with multiple sections
+
+fn first_function() {
+    // Some code here
+    let x = 42;
+    println!("Value: {}", x);
+}
+
+fn second_function() {
+    let mut total = 0;
+    for i in 0..10 {
+        total += i;
+    }
+
+
+    // Extra whitespace above
+    println!("Total: {}", total);
+}
+
+fn third_function() {
+    let message = "Hello";
+    println!("{}", message);
+}
+"#
+        .to_string();
+
+        let after = r#"
+// This is a large file with multiple sections
+
+fn first_function() {
+    // Some code here
+    let x = 42;
+    println!("Value: {}", x);
+}
+
+fn second_function() {
+    let mut total = 0;
+    for i in 0..10 {
+        total += i;
+    }
+    // No extra whitespace
+    println!("Total: {}", total);
+}
+
+fn third_function() {
+    let message = "Hello, World!";  // Changed this line
+    println!("{}", message);
+}
+"#
+        .to_string();
+
+        let result = standardize_rewrite(before, after)?;
+
+        // The result should:
+        // 1. Keep first_function exactly the same
+        // 2. Ignore whitespace changes in second_function
+        // 3. Include the actual code change in third_function
+        assert!(result.contains("fn first_function()"));
+        assert!(result.contains("let x = 42"));
+        assert!(result.contains("fn second_function()"));
+        assert!(result.contains("println!(\"Total: {}\", total)"));
+        assert!(result.contains("let message = \"Hello, World!\""));
+
+        // Verify whitespace changes were ignored
+        assert!(!result.contains("    \n    \n"));
+
+        // Add snapshot test
+        assert_snapshot!(result);
+
+        Ok(())
+    }
+}

From 89d85b56591b0ee8c0e90d3f706028ea4b68bbcc Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 00:20:57 -0500
Subject: [PATCH 02/21] [skip ci] yolo

---
 Cargo.lock                                    |  2 +
 crates/util/Cargo.toml                        |  7 +++
 crates/util/src/diff_standardizer.rs          | 56 +++++++++----------
 crates/util/src/lib.rs                        |  4 ++
 crates/util/src/runtime.rs                    |  2 +-
 ...ff_standardizer__tests__basic_rewrite.snap |  5 ++
 ...diff_standardizer__tests__empty_files.snap |  5 ++
 ...r__tests__mixed_changes_in_large_file.snap | 25 +++++++++
 ...tandardizer__tests__multiline_changes.snap |  8 +++
 ..._diff_standardizer__tests__no_changes.snap |  5 ++
 ...ndardizer__tests__whitespace_handling.snap |  7 +++
 11 files changed, 95 insertions(+), 31 deletions(-)
 create mode 100644 crates/util/src/snapshots/marzano_util__diff_standardizer__tests__basic_rewrite.snap
 create mode 100644 crates/util/src/snapshots/marzano_util__diff_standardizer__tests__empty_files.snap
 create mode 100644 crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
 create mode 100644 crates/util/src/snapshots/marzano_util__diff_standardizer__tests__multiline_changes.snap
 create mode 100644 crates/util/src/snapshots/marzano_util__diff_standardizer__tests__no_changes.snap
 create mode 100644 crates/util/src/snapshots/marzano_util__diff_standardizer__tests__whitespace_handling.snap

diff --git a/Cargo.lock b/Cargo.lock
index d9de33f81..568ed054d 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -2407,6 +2407,7 @@ dependencies = [
  "base64 0.21.7",
  "fs-err",
  "futures",
+ "git2",
  "grit-util",
  "http 0.2.12",
  "ignore",
@@ -2418,6 +2419,7 @@ dependencies = [
  "serde",
  "serde_json",
  "sha2",
+ "tempfile",
  "tokio",
  "tree-sitter-facade-sg",
 ]
diff --git a/crates/util/Cargo.toml b/crates/util/Cargo.toml
index d5f400436..958787a2d 100644
--- a/crates/util/Cargo.toml
+++ b/crates/util/Cargo.toml
@@ -32,6 +32,13 @@ serde_json = { version = "1.0.114" }
 futures = { version = "0.3.29", optional = true }
 http = { version = "0.2.11" }
 fs-err = { version = "2.11.0" }
+git2 = { version = "0.19.0", default-features = false, features = [
+  "vendored-openssl",
+  "vendored-libgit2",
+  "ssh",
+  "https",
+] }
+tempfile = { version = "3.1" }
 
 
 napi = { version = "2.16.4", default-features = false, features = [
diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index 04db3ca37..c0a6a7081 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -1,32 +1,11 @@
-use git2::DiffOptions;
-use marzano_util::diff::{parse_modified_ranges, FileDiff};
-use tempfile::tempdir;
 use anyhow::Result;
-
-pub fn parse_unified_diff(contents: String) -> Result<Vec<FileDiff>> {
-    let parsed = parse_modified_ranges(&contents)
-        .map_err(|e| anyhow::anyhow!("{:?}", e))?;
-    Ok(parsed)
-}
+use git2::{DiffOptions, Repository};
+use tempfile::tempdir;
 
 /// Given a before and after for a file, edit the *after* to not include any spurious changes
-pub fn standardize_rewrite(before: String, after: String) -> Result<String> {
+pub fn standardize_rewrite(repo: &Repository, before: String, after: String) -> Result<String> {
     let mut diff_opts = DiffOptions::new();
     diff_opts.ignore_whitespace(true);
-
-    // Create a temporary directory for the repository
-    let temp_dir = tempdir()
-        .map_err(|e| anyhow::anyhow!("Failed to create temp directory: {:?}", e))?;
-
-    // Create blobs for diffing
-    let repo = git2::Repository::init_opts(
-        temp_dir.path(),
-        &git2::RepositoryInitOptions::new()
-            .bare(true)
-            .initial_head("main"),
-    )
-    .map_err(|e| anyhow::anyhow!("Failed to create temp repo: {:?}", e))?;
-
     let left_oid = repo
         .blob(before.as_bytes())
         .map_err(|e| anyhow::anyhow!("Failed to create left blob: {:?}", e))?;
@@ -86,11 +65,23 @@ mod tests {
     use super::*;
     use insta::assert_snapshot;
 
+    fn setup_test_repo() -> Result<(Repository, tempfile::TempDir)> {
+        let temp_dir = tempdir()?;
+        let repo = Repository::init_opts(
+            temp_dir.path(),
+            &git2::RepositoryInitOptions::new()
+                .bare(true)
+                .initial_head("main"),
+        )?;
+        Ok((repo, temp_dir))
+    }
+
     #[test]
     fn test_basic_rewrite() -> Result<()> {
+        let (repo, _temp) = setup_test_repo()?;
         let before = "Hello world\n".to_string();
         let after = "Hello Rust\n".to_string();
-        let result = standardize_rewrite(before, after)?;
+        let result = standardize_rewrite(&repo, before, after)?;
         assert_eq!(result, "Hello Rust\n");
         assert_snapshot!(result);
         Ok(())
@@ -98,9 +89,10 @@ mod tests {
 
     #[test]
     fn test_whitespace_handling() -> Result<()> {
+        let (repo, _temp) = setup_test_repo()?;
         let before = "function test() {\n    console.log('test');\n}\n".to_string();
         let after = "function test(){\nconsole.log('test');\n}\n".to_string();
-        let result = standardize_rewrite(before, after)?;
+        let result = standardize_rewrite(&repo, before, after)?;
         assert_eq!(result, "function test(){\nconsole.log('test');\n}\n");
         assert_snapshot!(result);
         Ok(())
@@ -108,9 +100,10 @@ mod tests {
 
     #[test]
     fn test_empty_files() -> Result<()> {
+        let (repo, _temp) = setup_test_repo()?;
         let before = "".to_string();
         let after = "".to_string();
-        let result = standardize_rewrite(before, after)?;
+        let result = standardize_rewrite(&repo, before, after)?;
         assert_eq!(result, "");
         assert_snapshot!(result);
         Ok(())
@@ -118,9 +111,10 @@ mod tests {
 
     #[test]
     fn test_multiline_changes() -> Result<()> {
+        let (repo, _temp) = setup_test_repo()?;
         let before = "line1\nline2\nline3\n".to_string();
         let after = "line1\nmodified line2\nline3\nnew line4\n".to_string();
-        let result = standardize_rewrite(before, after)?;
+        let result = standardize_rewrite(&repo, before, after)?;
         assert_eq!(result, "line1\nmodified line2\nline3\nnew line4\n");
         assert_snapshot!(result);
         Ok(())
@@ -128,8 +122,9 @@ mod tests {
 
     #[test]
     fn test_no_changes() -> Result<()> {
+        let (repo, _temp) = setup_test_repo()?;
         let content = "unchanged content\n".to_string();
-        let result = standardize_rewrite(content.clone(), content)?;
+        let result = standardize_rewrite(&repo, content.clone(), content)?;
         assert_eq!(result, "unchanged content\n");
         assert_snapshot!(result);
         Ok(())
@@ -137,6 +132,7 @@ mod tests {
 
     #[test]
     fn test_mixed_changes_in_large_file() -> Result<()> {
+        let (repo, _temp) = setup_test_repo()?;
         let before = r#"
 // This is a large file with multiple sections
 
@@ -189,7 +185,7 @@ fn third_function() {
 "#
         .to_string();
 
-        let result = standardize_rewrite(before, after)?;
+        let result = standardize_rewrite(&repo, before, after)?;
 
         // The result should:
         // 1. Keep first_function exactly the same
diff --git a/crates/util/src/lib.rs b/crates/util/src/lib.rs
index 124ce80a8..1e49b0705 100644
--- a/crates/util/src/lib.rs
+++ b/crates/util/src/lib.rs
@@ -10,3 +10,7 @@ pub mod print_node;
 pub mod rich_path;
 pub mod runtime;
 pub mod url;
+
+mod diff_standardizer;
+
+pub use diff_standardizer::standardize_rewrite;
diff --git a/crates/util/src/runtime.rs b/crates/util/src/runtime.rs
index 111c49bbc..19353a6f0 100644
--- a/crates/util/src/runtime.rs
+++ b/crates/util/src/runtime.rs
@@ -56,7 +56,7 @@ pub struct ExecutionContext {
     pub ignore_limit_pattern: bool,
 }
 
-#[cfg(not(feature = "network_requests_common"))]
+#[cfg(not(any(test, feature = "network_requests_common")))]
 #[derive(Clone, Debug)]
 pub struct ExecutionContext {
     llm_api: Option<LanguageModelAPI>,
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__basic_rewrite.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__basic_rewrite.snap
new file mode 100644
index 000000000..c64f47f38
--- /dev/null
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__basic_rewrite.snap
@@ -0,0 +1,5 @@
+---
+source: crates/util/src/diff_standardizer.rs
+expression: result
+---
+Hello Rust
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__empty_files.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__empty_files.snap
new file mode 100644
index 000000000..dc39acec4
--- /dev/null
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__empty_files.snap
@@ -0,0 +1,5 @@
+---
+source: crates/util/src/diff_standardizer.rs
+expression: result
+---
+
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
new file mode 100644
index 000000000..6cd1eb639
--- /dev/null
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
@@ -0,0 +1,25 @@
+---
+source: crates/util/src/diff_standardizer.rs
+expression: result
+---
+// This is a large file with multiple sections
+
+fn first_function() {
+    // Some code here
+    let x = 42;
+    println!("Value: {}", x);
+}
+
+fn second_function() {
+    let mut total = 0;
+    for i in 0..10 {
+        total += i;
+    }
+    // No extra whitespace
+    println!("Total: {}", total);
+}
+
+fn third_function() {
+    let message = "Hello, World!";  // Changed this line
+    println!("{}", message);
+}
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__multiline_changes.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__multiline_changes.snap
new file mode 100644
index 000000000..0b7e8fd55
--- /dev/null
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__multiline_changes.snap
@@ -0,0 +1,8 @@
+---
+source: crates/util/src/diff_standardizer.rs
+expression: result
+---
+line1
+modified line2
+line3
+new line4
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__no_changes.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__no_changes.snap
new file mode 100644
index 000000000..c0e288f76
--- /dev/null
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__no_changes.snap
@@ -0,0 +1,5 @@
+---
+source: crates/util/src/diff_standardizer.rs
+expression: result
+---
+unchanged content
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__whitespace_handling.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__whitespace_handling.snap
new file mode 100644
index 000000000..7e348ec60
--- /dev/null
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__whitespace_handling.snap
@@ -0,0 +1,7 @@
+---
+source: crates/util/src/diff_standardizer.rs
+expression: result
+---
+function test(){
+console.log('test');
+}

From 13174f1932b3ccbb8069138f34de96a7db0e45c4 Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 00:23:44 -0500
Subject: [PATCH 03/21] [skip ci] yolo

---
 crates/util/src/diff_standardizer.rs | 51 ++++++++++------------------
 1 file changed, 18 insertions(+), 33 deletions(-)

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index c0a6a7081..814eb0eb8 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -135,21 +135,11 @@ mod tests {
         let (repo, _temp) = setup_test_repo()?;
         let before = r#"
 // This is a large file with multiple sections
-
-fn first_function() {
-    // Some code here
-    let x = 42;
-    println!("Value: {}", x);
-}
-
 fn second_function() {
     let mut total = 0;
     for i in 0..10 {
         total += i;
     }
-
-
-    // Extra whitespace above
     println!("Total: {}", total);
 }
 
@@ -162,46 +152,41 @@ fn third_function() {
 
         let after = r#"
 // This is a large file with multiple sections
+fn second_function() {
+    let mut total = 0;
+    for i in 0..10 {
+        total += i;
+    }
+            println!("Total: {}", total);
+}
 
-fn first_function() {
-    // Some code here
-    let x = 42;
-    println!("Value: {}", x);
+fn third_function() {
+    let thing = "Hello";
+    debug!("{}", message);
 }
+"#
+        .to_string();
 
+        let after_standardized = r#"
+// This is a large file with multiple sections
 fn second_function() {
     let mut total = 0;
     for i in 0..10 {
         total += i;
     }
-    // No extra whitespace
     println!("Total: {}", total);
 }
 
 fn third_function() {
-    let message = "Hello, World!";  // Changed this line
-    println!("{}", message);
+    let thing = "Hello";
+    debug!("{}", message);
 }
-"#
+        "#
         .to_string();
 
         let result = standardize_rewrite(&repo, before, after)?;
 
-        // The result should:
-        // 1. Keep first_function exactly the same
-        // 2. Ignore whitespace changes in second_function
-        // 3. Include the actual code change in third_function
-        assert!(result.contains("fn first_function()"));
-        assert!(result.contains("let x = 42"));
-        assert!(result.contains("fn second_function()"));
-        assert!(result.contains("println!(\"Total: {}\", total)"));
-        assert!(result.contains("let message = \"Hello, World!\""));
-
-        // Verify whitespace changes were ignored
-        assert!(!result.contains("    \n    \n"));
-
-        // Add snapshot test
-        assert_snapshot!(result);
+        assert_eq!(result, after_standardized);
 
         Ok(())
     }

From 96a0463dc08e8288ebedc58536217ffb8d3d6748 Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 00:28:40 -0500
Subject: [PATCH 04/21] [skip ci] yolo

---
 crates/util/src/diff_standardizer.rs          | 28 ++++---------------
 ...r__tests__mixed_changes_in_large_file.snap | 12 ++------
 2 files changed, 7 insertions(+), 33 deletions(-)

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index 814eb0eb8..1105d533d 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -1,6 +1,5 @@
 use anyhow::Result;
 use git2::{DiffOptions, Repository};
-use tempfile::tempdir;
 
 /// Given a before and after for a file, edit the *after* to not include any spurious changes
 pub fn standardize_rewrite(repo: &Repository, before: String, after: String) -> Result<String> {
@@ -66,10 +65,10 @@ mod tests {
     use insta::assert_snapshot;
 
     fn setup_test_repo() -> Result<(Repository, tempfile::TempDir)> {
-        let temp_dir = tempdir()?;
+        let temp_dir = tempfile::tempdir()?;
         let repo = Repository::init_opts(
             temp_dir.path(),
-            &git2::RepositoryInitOptions::new()
+            git2::RepositoryInitOptions::new()
                 .bare(true)
                 .initial_head("main"),
         )?;
@@ -152,23 +151,6 @@ fn third_function() {
 
         let after = r#"
 // This is a large file with multiple sections
-fn second_function() {
-    let mut total = 0;
-    for i in 0..10 {
-        total += i;
-    }
-            println!("Total: {}", total);
-}
-
-fn third_function() {
-    let thing = "Hello";
-    debug!("{}", message);
-}
-"#
-        .to_string();
-
-        let after_standardized = r#"
-// This is a large file with multiple sections
 fn second_function() {
     let mut total = 0;
     for i in 0..10 {
@@ -179,14 +161,14 @@ fn second_function() {
 
 fn third_function() {
     let thing = "Hello";
-    debug!("{}", message);
+    debug!("{}", thing);
 }
-        "#
+"#
         .to_string();
 
         let result = standardize_rewrite(&repo, before, after)?;
 
-        assert_eq!(result, after_standardized);
+        assert_snapshot!(result);
 
         Ok(())
     }
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
index 6cd1eb639..9a5637680 100644
--- a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
@@ -3,23 +3,15 @@ source: crates/util/src/diff_standardizer.rs
 expression: result
 ---
 // This is a large file with multiple sections
-
-fn first_function() {
-    // Some code here
-    let x = 42;
-    println!("Value: {}", x);
-}
-
 fn second_function() {
     let mut total = 0;
     for i in 0..10 {
         total += i;
     }
-    // No extra whitespace
     println!("Total: {}", total);
 }
 
 fn third_function() {
-    let message = "Hello, World!";  // Changed this line
-    println!("{}", message);
+    let thing = "Hello";
+    debug!("{}", thing);
 }

From 76098f79200261b3fc470f0c83520d19d58154c8 Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 00:30:06 -0500
Subject: [PATCH 05/21] [skip ci] yolo

---
 crates/util/src/diff_standardizer.rs | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index 1105d533d..d5c004bf2 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -135,11 +135,11 @@ mod tests {
         let before = r#"
 // This is a large file with multiple sections
 fn second_function() {
-    let mut total = 0;
+ let mut total = 0;
     for i in 0..10 {
         total += i;
     }
-    println!("Total: {}", total);
+  println!("Total: {}", total);
 }
 
 fn third_function() {

From 5d1d9b7c126689de38fb9cc7d60d1ffd806f9fcd Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 00:32:00 -0500
Subject: [PATCH 06/21] [skip ci] yolo

---
 crates/util/src/diff_standardizer.rs                 | 12 ++++++++++--
 ...dardizer__tests__mixed_changes_in_large_file.snap |  6 +++++-
 2 files changed, 15 insertions(+), 3 deletions(-)

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index d5c004bf2..c3854d8ef 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -133,7 +133,11 @@ mod tests {
     fn test_mixed_changes_in_large_file() -> Result<()> {
         let (repo, _temp) = setup_test_repo()?;
         let before = r#"
-// This is a large file with multiple sections
+/* ================================================================================
+
+      getMulti.
+
+================================================================================ */
 fn second_function() {
  let mut total = 0;
     for i in 0..10 {
@@ -150,7 +154,11 @@ fn third_function() {
         .to_string();
 
         let after = r#"
-// This is a large file with multiple sections
+/* ================================================================================
+
+    getMulti.
+
+================================================================================ */
 fn second_function() {
     let mut total = 0;
     for i in 0..10 {
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
index 9a5637680..28fd865fa 100644
--- a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
@@ -2,7 +2,11 @@
 source: crates/util/src/diff_standardizer.rs
 expression: result
 ---
-// This is a large file with multiple sections
+/* ================================================================================
+
+      getMulti.
+
+================================================================================ */
 fn second_function() {
     let mut total = 0;
     for i in 0..10 {

From d8a6d600e832cd480adb7b6d4692db6316301ff9 Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 00:32:31 -0500
Subject: [PATCH 07/21] [skip ci] yolo

---
 crates/util/src/diff_standardizer.rs | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index c3854d8ef..2f0ee643d 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -135,7 +135,7 @@ mod tests {
         let before = r#"
 /* ================================================================================
 
-      getMulti.
+       getMulti.
 
 ================================================================================ */
 fn second_function() {

From e14428c960d30811a28333502b9fd11b408cd606 Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 00:44:13 -0500
Subject: [PATCH 08/21] [skip ci] yolo

---
 crates/util/src/diff_standardizer.rs | 60 ++++++++++++----------------
 1 file changed, 26 insertions(+), 34 deletions(-)

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index 2f0ee643d..df5dedf41 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -5,6 +5,7 @@ use git2::{DiffOptions, Repository};
 pub fn standardize_rewrite(repo: &Repository, before: String, after: String) -> Result<String> {
     let mut diff_opts = DiffOptions::new();
     diff_opts.ignore_whitespace(true);
+
     let left_oid = repo
         .blob(before.as_bytes())
         .map_err(|e| anyhow::anyhow!("Failed to create left blob: {:?}", e))?;
@@ -20,42 +21,33 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
         .map_err(|e| anyhow::anyhow!("Failed to find right blob: {:?}", e))?;
 
     // Generate diff
-    let mut standardized = after.clone(); // Start with the full after content
-    let mut has_changes = false;
-    let diff = repo
-        .diff_blobs(
-            Some(&left_blob),
-            None,
-            Some(&right_blob),
-            None,
-            Some(&mut diff_opts),
-            None,
-            None,
-            None,
-            Some(&mut |delta, _hunk, line| {
-                has_changes = true;
+    let mut standardized = String::new();
+
+    // There are changes, build up the new content from the diff
+    let mut diff_opts = DiffOptions::new();
+    repo.diff_blobs(
+        Some(&left_blob),
+        None,
+        Some(&right_blob),
+        None,
+        Some(&mut diff_opts),
+        None,
+        None,
+        None,
+        Some(&mut |_delta, _hunk, line| {
+            if let Ok(content) = std::str::from_utf8(line.content()) {
                 match line.origin() {
-                    '+' | ' ' => {
-                        if let Ok(_) = std::str::from_utf8(line.content()) {
-                            // Content will be taken from the after string
-                        }
-                    }
-                    '-' => {
-                        // Removed lines are ignored as we're using the after content
+                    ' ' | '+' => {
+                        standardized.push_str(content);
                     }
                     _ => {}
                 }
-                true
-            }),
-        )
-        .map_err(|e| anyhow::anyhow!("Failed to generate diff: {:?}", e))?;
-
-    // If no changes were detected in the diff, return the after content as is
-    if !has_changes {
-        standardized = after;
-    }
+            }
+            true
+        }),
+    )
+    .map_err(|e| anyhow::anyhow!("Failed to generate diff: {:?}", e))?;
 
-    // The temporary directory will be automatically cleaned up when temp_dir is dropped
     Ok(standardized)
 }
 
@@ -89,11 +81,11 @@ mod tests {
     #[test]
     fn test_whitespace_handling() -> Result<()> {
         let (repo, _temp) = setup_test_repo()?;
-        let before = "function test() {\n    console.log('test');\n}\n".to_string();
+        let before = "function test() {\n    console.bob('test');\n}\n".to_string();
         let after = "function test(){\nconsole.log('test');\n}\n".to_string();
+        let after_standard = "function test() {\n    console.log('test');\n}\n".to_string();
         let result = standardize_rewrite(&repo, before, after)?;
-        assert_eq!(result, "function test(){\nconsole.log('test');\n}\n");
-        assert_snapshot!(result);
+        assert_eq!(result, after_standard);
         Ok(())
     }
 

From 670c7076da5e075adb58bbb92ecbfe240bc80a24 Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 00:51:36 -0500
Subject: [PATCH 09/21] [skip ci] yolo

---
 crates/util/src/diff_standardizer.rs          | 34 +++++++++++++++----
 ...r__tests__mixed_changes_in_large_file.snap |  2 +-
 2 files changed, 29 insertions(+), 7 deletions(-)

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index df5dedf41..d252eb38c 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -22,6 +22,8 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
 
     // Generate diff
     let mut standardized = String::new();
+    let mut added_lines = Vec::new();
+    let mut current_line = 0;
 
     // There are changes, build up the new content from the diff
     let mut diff_opts = DiffOptions::new();
@@ -37,8 +39,17 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
         Some(&mut |_delta, _hunk, line| {
             if let Ok(content) = std::str::from_utf8(line.content()) {
                 match line.origin() {
-                    ' ' | '+' => {
+                    ' ' => {
                         standardized.push_str(content);
+                        current_line += 1;
+                    }
+                    '+' => {
+                        standardized.push_str(content);
+                        added_lines.push(current_line);
+                        current_line += 1;
+                    }
+                    '-' => {
+                        // Skip removed lines but don't increment current_line
                     }
                     _ => {}
                 }
@@ -48,6 +59,17 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
     )
     .map_err(|e| anyhow::anyhow!("Failed to generate diff: {:?}", e))?;
 
+    // Add any remaining content from the right blob that wasn't encountered
+    if let Ok(remaining_content) = std::str::from_utf8(right_blob.content()) {
+        let remaining_lines: Vec<&str> = remaining_content.lines().collect();
+        for (i, line) in remaining_lines.iter().enumerate() {
+            if !added_lines.contains(&i) {
+                standardized.push_str(line);
+                standardized.push('\n');
+            }
+        }
+    }
+
     Ok(standardized)
 }
 
@@ -82,7 +104,7 @@ mod tests {
     fn test_whitespace_handling() -> Result<()> {
         let (repo, _temp) = setup_test_repo()?;
         let before = "function test() {\n    console.bob('test');\n}\n".to_string();
-        let after = "function test(){\nconsole.log('test');\n}\n".to_string();
+        let after = "function test() {\nconsole.log('test');\n}\n".to_string();
         let after_standard = "function test() {\n    console.log('test');\n}\n".to_string();
         let result = standardize_rewrite(&repo, before, after)?;
         assert_eq!(result, after_standard);
@@ -103,11 +125,11 @@ mod tests {
     #[test]
     fn test_multiline_changes() -> Result<()> {
         let (repo, _temp) = setup_test_repo()?;
-        let before = "line1\nline2\nline3\n".to_string();
-        let after = "line1\nmodified line2\nline3\nnew line4\n".to_string();
+        let before = "line1\nline2\n  line3\n".to_string();
+        let after = "line1\nmodified line2\n\tline3\nnew line4\n".to_string();
         let result = standardize_rewrite(&repo, before, after)?;
-        assert_eq!(result, "line1\nmodified line2\nline3\nnew line4\n");
-        assert_snapshot!(result);
+        let after = "line1\nmodified line2\n  line3\nnew line4\n".to_string();
+        assert_eq!(result, after);
         Ok(())
     }
 
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
index 28fd865fa..ec69fb155 100644
--- a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
@@ -4,7 +4,7 @@ expression: result
 ---
 /* ================================================================================
 
-      getMulti.
+       getMulti.
 
 ================================================================================ */
 fn second_function() {

From 681067bcd55c8bf22fb9fa58139b09791b932b5c Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 01:04:18 -0500
Subject: [PATCH 10/21] [skip ci] yolo

---
 crates/util/src/diff_standardizer.rs          | 59 ++++++-------------
 ...r__tests__mixed_changes_in_large_file.snap |  2 +-
 2 files changed, 20 insertions(+), 41 deletions(-)

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index d252eb38c..702d2b424 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -20,57 +20,36 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
         .find_blob(right_oid)
         .map_err(|e| anyhow::anyhow!("Failed to find right blob: {:?}", e))?;
 
-    // Generate diff
-    let mut standardized = String::new();
-    let mut added_lines = Vec::new();
-    let mut current_line = 0;
-
-    // There are changes, build up the new content from the diff
-    let mut diff_opts = DiffOptions::new();
+    // Run the diff to check if there are changes
+    let mut right_oid = Option::<git2::Oid>::None;
     repo.diff_blobs(
         Some(&left_blob),
         None,
         Some(&right_blob),
         None,
         Some(&mut diff_opts),
+        Some(&mut |delta, _progress| {
+            right_oid = Some(delta.new_file().id());
+            true
+        }),
         None,
         None,
         None,
-        Some(&mut |_delta, _hunk, line| {
-            if let Ok(content) = std::str::from_utf8(line.content()) {
-                match line.origin() {
-                    ' ' => {
-                        standardized.push_str(content);
-                        current_line += 1;
-                    }
-                    '+' => {
-                        standardized.push_str(content);
-                        added_lines.push(current_line);
-                        current_line += 1;
-                    }
-                    '-' => {
-                        // Skip removed lines but don't increment current_line
-                    }
-                    _ => {}
-                }
-            }
-            true
-        }),
     )
     .map_err(|e| anyhow::anyhow!("Failed to generate diff: {:?}", e))?;
 
-    // Add any remaining content from the right blob that wasn't encountered
-    if let Ok(remaining_content) = std::str::from_utf8(right_blob.content()) {
-        let remaining_lines: Vec<&str> = remaining_content.lines().collect();
-        for (i, line) in remaining_lines.iter().enumerate() {
-            if !added_lines.contains(&i) {
-                standardized.push_str(line);
-                standardized.push('\n');
-            }
-        }
-    }
+    let Some(right_oid) = right_oid else {
+        return Ok(before);
+    };
 
-    Ok(standardized)
+    let right_blob = repo
+        .find_blob(right_oid)
+        .map_err(|e| anyhow::anyhow!("Failed to find right blob: {:?}", e))?;
+
+    let content = std::str::from_utf8(right_blob.content())
+        .map_err(|e| anyhow::anyhow!("Failed to convert blob content to UTF-8: {:?}", e))?;
+
+    Ok(content.to_string())
 }
 
 #[cfg(test)]
@@ -105,7 +84,7 @@ mod tests {
         let (repo, _temp) = setup_test_repo()?;
         let before = "function test() {\n    console.bob('test');\n}\n".to_string();
         let after = "function test() {\nconsole.log('test');\n}\n".to_string();
-        let after_standard = "function test() {\n    console.log('test');\n}\n".to_string();
+        let after_standard = "function test() {\nconsole.log('test');\n}\n".to_string();
         let result = standardize_rewrite(&repo, before, after)?;
         assert_eq!(result, after_standard);
         Ok(())
@@ -128,7 +107,7 @@ mod tests {
         let before = "line1\nline2\n  line3\n".to_string();
         let after = "line1\nmodified line2\n\tline3\nnew line4\n".to_string();
         let result = standardize_rewrite(&repo, before, after)?;
-        let after = "line1\nmodified line2\n  line3\nnew line4\n".to_string();
+        let after = "line1\nmodified line2\n\tline3\nnew line4\n".to_string();
         assert_eq!(result, after);
         Ok(())
     }
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
index ec69fb155..03c7de17d 100644
--- a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
@@ -4,7 +4,7 @@ expression: result
 ---
 /* ================================================================================
 
-       getMulti.
+    getMulti.
 
 ================================================================================ */
 fn second_function() {

From 313b36d94d11123656f6cde0191535c0789cd5ff Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 01:05:12 -0500
Subject: [PATCH 11/21] ok

---
 ...__diff_standardizer__tests__mixed_changes_in_large_file.snap | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
index 03c7de17d..ec69fb155 100644
--- a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
@@ -4,7 +4,7 @@ expression: result
 ---
 /* ================================================================================
 
-    getMulti.
+       getMulti.
 
 ================================================================================ */
 fn second_function() {

From d682e8a5df43fafa8f68814accc7596dcc49ce0a Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 01:09:32 -0500
Subject: [PATCH 12/21] [skip ci] yolo

---
 crates/util/src/diff_standardizer.rs | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index 702d2b424..b0ddc2a14 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -1,3 +1,9 @@
+/* ================================================================================
+
+       getMulti.
+
+================================================================================ */
+
 use anyhow::Result;
 use git2::{DiffOptions, Repository};
 
@@ -5,6 +11,8 @@ use git2::{DiffOptions, Repository};
 pub fn standardize_rewrite(repo: &Repository, before: String, after: String) -> Result<String> {
     let mut diff_opts = DiffOptions::new();
     diff_opts.ignore_whitespace(true);
+    diff_opts.indent_heuristic(true);
+    diff_opts.ignore_whitespace_change(true);
 
     let left_oid = repo
         .blob(before.as_bytes())

From 6dc278636524c0667485dd6fae976f80e8e4b4b3 Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 01:22:57 -0500
Subject: [PATCH 13/21] [skip ci] yolo

---
 crates/util/src/diff_standardizer.rs | 47 +++++++++++++++++++++++++---
 1 file changed, 42 insertions(+), 5 deletions(-)

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index b0ddc2a14..7eef6f9b9 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -1,17 +1,16 @@
 /* ================================================================================
 
-       getMulti.
+    getMulti.
 
 ================================================================================ */
 
 use anyhow::Result;
-use git2::{DiffOptions, Repository};
+use git2::{Delta, DiffOptions, Repository};
 
 /// Given a before and after for a file, edit the *after* to not include any spurious changes
 pub fn standardize_rewrite(repo: &Repository, before: String, after: String) -> Result<String> {
     let mut diff_opts = DiffOptions::new();
     diff_opts.ignore_whitespace(true);
-    diff_opts.indent_heuristic(true);
     diff_opts.ignore_whitespace_change(true);
 
     let left_oid = repo
@@ -41,8 +40,14 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
             true
         }),
         None,
-        None,
-        None,
+        Some(&mut |delta, _progress| {
+            println!("delta: {:?}, progress: {:?}", delta, _progress);
+            true
+        }),
+        Some(&mut |delta, _hunk, line| {
+            println!("line: {:?}", line);
+            true
+        }),
     )
     .map_err(|e| anyhow::anyhow!("Failed to generate diff: {:?}", e))?;
 
@@ -181,4 +186,36 @@ fn third_function() {
 
         Ok(())
     }
+
+    #[test]
+    fn test_code_removal() -> Result<()> {
+        let (repo, _temp) = setup_test_repo()?;
+        let before = r#"fn main() {
+    // First we do some setup
+    let mut total = 0;
+
+    // Then we do a big calculation
+    for i in 0..100 {
+        if i % 2 == 0 {
+            total += i;
+        } else {
+            total -= i;
+        }
+    }
+
+    // Finally print the result
+    println!("The total is: {}", total);
+}"#
+        .to_string();
+
+        let after = r#"fn main() {
+    let mut total = 0;
+    println!("The total is: {}", total);
+}"#
+        .to_string();
+
+        let result = standardize_rewrite(&repo, before, after.clone())?;
+        assert_eq!(result, after);
+        Ok(())
+    }
 }

From 3ff3e1149fa98e0a51fdfa98bd814dd6187b12f4 Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 02:14:55 -0500
Subject: [PATCH 14/21] [skip ci] yolo

---
 crates/util/src/diff_standardizer.rs          | 118 ++++++++++++++----
 ...r__tests__mixed_changes_in_large_file.snap |   8 +-
 ...ardizer__tests__remove_early_add_late.snap |  16 +++
 3 files changed, 112 insertions(+), 30 deletions(-)
 create mode 100644 crates/util/src/snapshots/marzano_util__diff_standardizer__tests__remove_early_add_late.snap

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index 7eef6f9b9..b06c5c2b0 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -5,13 +5,14 @@
 ================================================================================ */
 
 use anyhow::Result;
-use git2::{Delta, DiffOptions, Repository};
+use git2::{Delta, DiffLineType, DiffOptions, Repository};
 
 /// Given a before and after for a file, edit the *after* to not include any spurious changes
 pub fn standardize_rewrite(repo: &Repository, before: String, after: String) -> Result<String> {
     let mut diff_opts = DiffOptions::new();
     diff_opts.ignore_whitespace(true);
     diff_opts.ignore_whitespace_change(true);
+    diff_opts.context_lines(0);
 
     let left_oid = repo
         .blob(before.as_bytes())
@@ -27,42 +28,65 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
         .find_blob(right_oid)
         .map_err(|e| anyhow::anyhow!("Failed to find right blob: {:?}", e))?;
 
-    // Run the diff to check if there are changes
-    let mut right_oid = Option::<git2::Oid>::None;
+    let mut before_lines = before.lines().collect::<Vec<_>>();
+    let mut standardized_after = String::new();
+    let mut before_line_num = 0 as usize;
+
     repo.diff_blobs(
         Some(&left_blob),
         None,
         Some(&right_blob),
         None,
         Some(&mut diff_opts),
-        Some(&mut |delta, _progress| {
-            right_oid = Some(delta.new_file().id());
-            true
-        }),
+        // Some(&mut |delta, _progress| {
+        //     right_oid = Some(delta.new_file().id());
+        //     true
+        // }),
         None,
-        Some(&mut |delta, _progress| {
-            println!("delta: {:?}, progress: {:?}", delta, _progress);
-            true
-        }),
-        Some(&mut |delta, _hunk, line| {
-            println!("line: {:?}", line);
+        None,
+        // Some(&mut |delta, _progress| {
+        //     println!("delta: {:?}, progress: {:?}", delta, _progress);
+        //     true
+        // }),
+        None,
+        Some(&mut |_delta, hunk, line| {
+            let hunk = hunk.unwrap();
+            // Grab the content between before this hunk and inject it
+            while before_line_num < hunk.old_start().try_into().unwrap() {
+                standardized_after.push_str(before_lines[before_line_num]);
+                standardized_after.push('\n');
+                before_line_num += 1;
+            }
+
+            match line.origin_value() {
+                DiffLineType::Deletion => {
+                    // Deletion: advance the offset by the length of the removed content
+                    before_line_num += 1;
+                }
+                DiffLineType::Addition => {
+                    // println!(
+                    //     "INJECTING: {} onto {}",
+                    //     std::str::from_utf8(line.content()).unwrap(),
+                    //     standardized_after
+                    // );
+                    // // Addition: inject the new contnent directly
+                    // standardized_after.push_str(std::str::from_utf8(line.content()).unwrap());
+                }
+                _ => {}
+            }
             true
         }),
     )
     .map_err(|e| anyhow::anyhow!("Failed to generate diff: {:?}", e))?;
 
-    let Some(right_oid) = right_oid else {
-        return Ok(before);
-    };
-
-    let right_blob = repo
-        .find_blob(right_oid)
-        .map_err(|e| anyhow::anyhow!("Failed to find right blob: {:?}", e))?;
-
-    let content = std::str::from_utf8(right_blob.content())
-        .map_err(|e| anyhow::anyhow!("Failed to convert blob content to UTF-8: {:?}", e))?;
+    // Finally, add the rest of the content
+    while before_line_num < before_lines.len() {
+        standardized_after.push_str(before_lines[before_line_num]);
+        standardized_after.push('\n');
+        before_line_num += 1;
+    }
 
-    Ok(content.to_string())
+    Ok(standardized_after)
 }
 
 #[cfg(test)]
@@ -97,7 +121,7 @@ mod tests {
         let (repo, _temp) = setup_test_repo()?;
         let before = "function test() {\n    console.bob('test');\n}\n".to_string();
         let after = "function test() {\nconsole.log('test');\n}\n".to_string();
-        let after_standard = "function test() {\nconsole.log('test');\n}\n".to_string();
+        let after_standard = "function test() {console.log('test');\n}\n".to_string();
         let result = standardize_rewrite(&repo, before, after)?;
         assert_eq!(result, after_standard);
         Ok(())
@@ -218,4 +242,48 @@ fn third_function() {
         assert_eq!(result, after);
         Ok(())
     }
+
+    #[test]
+    fn test_remove_early_add_late() -> Result<()> {
+        let (repo, _td) = setup_test_repo()?;
+
+        let before = r#"
+// this starts us off
+
+fn first_function() {
+    println!("This will be removed");
+    println!("This will also be removed");
+}
+
+fn middle_function() {
+    // This stays the same
+    println!("Hello");
+}
+
+fn last_function() {
+    println!("Original");
+}
+
+// This is other content"#;
+
+        let after = r#"
+// this starts us off
+
+fn middle_function() {
+    // This stays the same
+    println!("Hello");
+}
+
+fn last_function() {
+    println!("Original");
+    println!("New line added here");
+    println!("And another one");
+}
+
+// This is other content"#;
+
+        let result = standardize_rewrite(&repo, before.to_string(), after.to_string())?;
+        assert_snapshot!(result);
+        Ok(())
+    }
 }
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
index ec69fb155..c86c49170 100644
--- a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
@@ -8,14 +8,12 @@ expression: result
 
 ================================================================================ */
 fn second_function() {
-    let mut total = 0;
+ let mut total = 0;
     for i in 0..10 {
         total += i;
     }
-    println!("Total: {}", total);
+  println!("Total: {}", total);
 }
 
 fn third_function() {
-    let thing = "Hello";
-    debug!("{}", thing);
-}
+    let message = "Hello";
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__remove_early_add_late.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__remove_early_add_late.snap
new file mode 100644
index 000000000..5bb490ba3
--- /dev/null
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__remove_early_add_late.snap
@@ -0,0 +1,16 @@
+---
+source: crates/util/src/diff_standardizer.rs
+expression: result
+---
+// this starts us off
+
+fn first_function() {
+    // This stays the same
+    println!("Hello");
+}
+
+fn last_function() {
+    println!("Original");
+}
+
+// This is other content

From 676b1b31fdd6ec35d5e34fa12f48ae5d241f223c Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 02:27:20 -0500
Subject: [PATCH 15/21] [skip ci] yolo

---
 crates/util/src/diff_standardizer.rs          | 25 +++++++++++--------
 ...r__tests__mixed_changes_in_large_file.snap |  4 ++-
 ...ardizer__tests__remove_early_add_late.snap |  4 ++-
 3 files changed, 20 insertions(+), 13 deletions(-)

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index b06c5c2b0..da7cfc7dc 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -30,7 +30,7 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
 
     let mut before_lines = before.lines().collect::<Vec<_>>();
     let mut standardized_after = String::new();
-    let mut before_line_num = 0 as usize;
+    let mut before_line_num_zero = 0_usize;
 
     repo.diff_blobs(
         Some(&left_blob),
@@ -50,18 +50,20 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
         // }),
         None,
         Some(&mut |_delta, hunk, line| {
+            println!("line: {:?}", line);
             let hunk = hunk.unwrap();
             // Grab the content between before this hunk and inject it
-            while before_line_num < hunk.old_start().try_into().unwrap() {
-                standardized_after.push_str(before_lines[before_line_num]);
+            while (before_line_num_zero + 1) < hunk.old_start().try_into().unwrap() {
+                println!("Pushing: {}", before_lines[before_line_num_zero]);
+                standardized_after.push_str(before_lines[before_line_num_zero]);
                 standardized_after.push('\n');
-                before_line_num += 1;
+                before_line_num_zero += 1;
             }
 
             match line.origin_value() {
                 DiffLineType::Deletion => {
                     // Deletion: advance the offset by the length of the removed content
-                    before_line_num += 1;
+                    before_line_num_zero += line.num_lines() as usize;
                 }
                 DiffLineType::Addition => {
                     // println!(
@@ -69,8 +71,8 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
                     //     std::str::from_utf8(line.content()).unwrap(),
                     //     standardized_after
                     // );
-                    // // Addition: inject the new contnent directly
-                    // standardized_after.push_str(std::str::from_utf8(line.content()).unwrap());
+                    // // Addition: inject the new content directly
+                    standardized_after.push_str(std::str::from_utf8(line.content()).unwrap());
                 }
                 _ => {}
             }
@@ -80,10 +82,10 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
     .map_err(|e| anyhow::anyhow!("Failed to generate diff: {:?}", e))?;
 
     // Finally, add the rest of the content
-    while before_line_num < before_lines.len() {
-        standardized_after.push_str(before_lines[before_line_num]);
+    while before_line_num_zero < before_lines.len() {
+        standardized_after.push_str(before_lines[before_line_num_zero]);
         standardized_after.push('\n');
-        before_line_num += 1;
+        before_line_num_zero += 1;
     }
 
     Ok(standardized_after)
@@ -235,7 +237,8 @@ fn third_function() {
         let after = r#"fn main() {
     let mut total = 0;
     println!("The total is: {}", total);
-}"#
+}
+    "#
         .to_string();
 
         let result = standardize_rewrite(&repo, before, after.clone())?;
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
index c86c49170..44ad23d49 100644
--- a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
@@ -16,4 +16,6 @@ fn second_function() {
 }
 
 fn third_function() {
-    let message = "Hello";
+    let thing = "Hello";
+    debug!("{}", thing);
+}
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__remove_early_add_late.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__remove_early_add_late.snap
index 5bb490ba3..871ef5e34 100644
--- a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__remove_early_add_late.snap
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__remove_early_add_late.snap
@@ -4,12 +4,14 @@ expression: result
 ---
 // this starts us off
 
-fn first_function() {
+fn middle_function() {
     // This stays the same
     println!("Hello");
 }
 
 fn last_function() {
+    println!("New line added here");
+    println!("And another one");
     println!("Original");
 }
 

From ed5bc9dcdeeb4b1c51a8f0b824f91657322e7cc4 Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 02:52:13 -0500
Subject: [PATCH 16/21] [skip ci] yolo

---
 crates/util/src/diff_standardizer.rs | 47 ++++++++++++++++++----------
 1 file changed, 31 insertions(+), 16 deletions(-)

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index da7cfc7dc..f4829fb6b 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -30,7 +30,7 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
 
     let mut before_lines = before.lines().collect::<Vec<_>>();
     let mut standardized_after = String::new();
-    let mut before_line_num_zero = 0_usize;
+    let mut before_line_num = 1_usize;
 
     repo.diff_blobs(
         Some(&left_blob),
@@ -50,20 +50,29 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
         // }),
         None,
         Some(&mut |_delta, hunk, line| {
-            println!("line: {:?}", line);
             let hunk = hunk.unwrap();
+            println!(
+                "offset: {}, hunk: {}, line: {:?}",
+                before_line_num,
+                hunk.old_start(),
+                line,
+            );
             // Grab the content between before this hunk and inject it
-            while (before_line_num_zero + 1) < hunk.old_start().try_into().unwrap() {
-                println!("Pushing: {}", before_lines[before_line_num_zero]);
-                standardized_after.push_str(before_lines[before_line_num_zero]);
+            while before_line_num < hunk.old_start().try_into().unwrap() {
+                let this_line = before_lines[before_line_num - 1];
+                println!("Pushing: {}", this_line);
+                standardized_after.push_str(this_line);
                 standardized_after.push('\n');
-                before_line_num_zero += 1;
+                before_line_num += 1;
             }
 
+            // Now advance the cursor by the number of old lines we covered
+            // before_line_num += (hunk.old_lines() as usize);
+
             match line.origin_value() {
                 DiffLineType::Deletion => {
-                    // Deletion: advance the offset by the length of the removed content
-                    before_line_num_zero += line.num_lines() as usize;
+                    // Deletion: we don't actually need to do anything with deleted content
+                    before_line_num += line.num_lines() as usize;
                 }
                 DiffLineType::Addition => {
                     // println!(
@@ -74,7 +83,13 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
                     // // Addition: inject the new content directly
                     standardized_after.push_str(std::str::from_utf8(line.content()).unwrap());
                 }
-                _ => {}
+                // DiffLineType::Context => {
+                //     // Context: do nothing
+                //     println!("Context: {}", std::str::from_utf8(line.content()).unwrap());
+                // }
+                _ => {
+                    println!("fuck you!: {}", line.num_lines());
+                }
             }
             true
         }),
@@ -82,10 +97,10 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
     .map_err(|e| anyhow::anyhow!("Failed to generate diff: {:?}", e))?;
 
     // Finally, add the rest of the content
-    while before_line_num_zero < before_lines.len() {
-        standardized_after.push_str(before_lines[before_line_num_zero]);
+    while before_line_num < before_lines.len() {
+        standardized_after.push_str(before_lines[before_line_num - 1]);
         standardized_after.push('\n');
-        before_line_num_zero += 1;
+        before_line_num += 1;
     }
 
     Ok(standardized_after)
@@ -144,10 +159,10 @@ mod tests {
     fn test_multiline_changes() -> Result<()> {
         let (repo, _temp) = setup_test_repo()?;
         let before = "line1\nline2\n  line3\n".to_string();
-        let after = "line1\nmodified line2\n\tline3\nnew line4\n".to_string();
-        let result = standardize_rewrite(&repo, before, after)?;
-        let after = "line1\nmodified line2\n\tline3\nnew line4\n".to_string();
-        assert_eq!(result, after);
+        let after1 = "line1\nmodified line2\n  line3\nnew line4\n".to_string();
+        let after2 = "line1\nmodified line2\n  line3\nnew line4\n".to_string();
+        let result = standardize_rewrite(&repo, before, after1)?;
+        assert_eq!(result, after2);
         Ok(())
     }
 

From d545b698d4e0488687c92ed7cf4d2536964e9aa9 Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 03:17:18 -0500
Subject: [PATCH 17/21] checkpoint where I give up on git, I hate libgit2

---
 crates/util/src/diff_standardizer.rs | 104 +++++++++++++++++----------
 1 file changed, 68 insertions(+), 36 deletions(-)

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index f4829fb6b..1dbf162eb 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -28,9 +28,10 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
         .find_blob(right_oid)
         .map_err(|e| anyhow::anyhow!("Failed to find right blob: {:?}", e))?;
 
-    let mut before_lines = before.lines().collect::<Vec<_>>();
+    let before_lines = before.lines().collect::<Vec<_>>();
     let mut standardized_after = String::new();
     let mut before_line_num = 1_usize;
+    let mut pending_additions: Vec<(usize, String)> = Vec::new();
 
     repo.diff_blobs(
         Some(&left_blob),
@@ -38,69 +39,100 @@ pub fn standardize_rewrite(repo: &Repository, before: String, after: String) ->
         Some(&right_blob),
         None,
         Some(&mut diff_opts),
-        // Some(&mut |delta, _progress| {
-        //     right_oid = Some(delta.new_file().id());
-        //     true
-        // }),
         None,
         None,
-        // Some(&mut |delta, _progress| {
-        //     println!("delta: {:?}, progress: {:?}", delta, _progress);
-        //     true
-        // }),
         None,
         Some(&mut |_delta, hunk, line| {
             let hunk = hunk.unwrap();
+            println!("\n=== Processing Hunk ===");
             println!(
-                "offset: {}, hunk: {}, line: {:?}",
+                "Current before_line_num: {}, Hunk old_start: {}, old_lines: {}, new_lines: {}",
                 before_line_num,
                 hunk.old_start(),
-                line,
+                hunk.old_lines(),
+                hunk.new_lines()
             );
-            // Grab the content between before this hunk and inject it
-            while before_line_num < hunk.old_start().try_into().unwrap() {
+            println!(
+                "Hunk header: {}",
+                std::str::from_utf8(hunk.header()).unwrap()
+            );
+
+            let hunk_start: usize = hunk.old_start().try_into().unwrap();
+
+            // Copy any unchanged lines up to this hunk
+            while before_line_num < hunk_start {
                 let this_line = before_lines[before_line_num - 1];
-                println!("Pushing: {}", this_line);
+                println!(
+                    "Injecting unchanged line {}: '{}'",
+                    before_line_num, this_line
+                );
                 standardized_after.push_str(this_line);
                 standardized_after.push('\n');
                 before_line_num += 1;
             }
 
-            // Now advance the cursor by the number of old lines we covered
-            // before_line_num += (hunk.old_lines() as usize);
-
             match line.origin_value() {
                 DiffLineType::Deletion => {
-                    // Deletion: we don't actually need to do anything with deleted content
-                    before_line_num += line.num_lines() as usize;
+                    println!(
+                        "Deletion at line {}: '{}'",
+                        before_line_num,
+                        before_lines[before_line_num - 1]
+                    );
+                    before_line_num += 1;
                 }
                 DiffLineType::Addition => {
-                    // println!(
-                    //     "INJECTING: {} onto {}",
-                    //     std::str::from_utf8(line.content()).unwrap(),
-                    //     standardized_after
-                    // );
-                    // // Addition: inject the new content directly
-                    standardized_after.push_str(std::str::from_utf8(line.content()).unwrap());
+                    let new_content = std::str::from_utf8(line.content()).unwrap().to_string();
+                    let target_position = hunk.new_start() as usize;
+                    println!(
+                        "Queueing addition at position {}: '{}'",
+                        target_position, new_content
+                    );
+                    // Insert into pending_additions in sorted order by position
+                    let insert_pos =
+                        pending_additions.partition_point(|(pos, _)| *pos <= target_position);
+                    pending_additions.insert(insert_pos, (target_position, new_content));
                 }
-                // DiffLineType::Context => {
-                //     // Context: do nothing
-                //     println!("Context: {}", std::str::from_utf8(line.content()).unwrap());
-                // }
                 _ => {
-                    println!("fuck you!: {}", line.num_lines());
+                    println!(
+                        "Other line type: {:?}, num_lines: {}",
+                        line.origin_value(),
+                        line.num_lines()
+                    );
                 }
             }
+
+            // Process any pending additions that should come before the next hunk
+            let mut current_line = before_line_num;
+            while let Some((pos, content)) = pending_additions.first() {
+                if *pos <= current_line {
+                    standardized_after.push_str(&content);
+                    pending_additions.remove(0);
+                } else {
+                    break;
+                }
+            }
+
             true
         }),
     )
     .map_err(|e| anyhow::anyhow!("Failed to generate diff: {:?}", e))?;
 
-    // Finally, add the rest of the content
-    while before_line_num < before_lines.len() {
-        standardized_after.push_str(before_lines[before_line_num - 1]);
-        standardized_after.push('\n');
-        before_line_num += 1;
+    // Add any remaining content and pending additions
+    while before_line_num <= before_lines.len() || !pending_additions.is_empty() {
+        let next_pending = pending_additions.first().map(|(pos, _)| *pos);
+        let should_add_original = match next_pending {
+            Some(pos) => before_line_num < pos,
+            None => true,
+        };
+
+        if should_add_original && before_line_num <= before_lines.len() {
+            standardized_after.push_str(before_lines[before_line_num - 1]);
+            standardized_after.push('\n');
+            before_line_num += 1;
+        } else if let Some((_, content)) = pending_additions.first() {
+            standardized_after.push_str(content);
+            pending_additions.remove(0);
+        }
     }
 
     Ok(standardized_after)

From 7146df5bad085d5971ac5c7b68d6d46b351211d3 Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 03:27:08 -0500
Subject: [PATCH 18/21] [skip ci] yolo

---
 Cargo.lock                                    |   1 +
 crates/util/Cargo.toml                        |   1 +
 crates/util/src/diff_standardizer.rs          | 218 +++---------------
 ...r__tests__mixed_changes_in_large_file.snap |   6 +-
 4 files changed, 36 insertions(+), 190 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index 568ed054d..ea1bef644 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -2419,6 +2419,7 @@ dependencies = [
  "serde",
  "serde_json",
  "sha2",
+ "similar",
  "tempfile",
  "tokio",
  "tree-sitter-facade-sg",
diff --git a/crates/util/Cargo.toml b/crates/util/Cargo.toml
index 958787a2d..98cc279d2 100644
--- a/crates/util/Cargo.toml
+++ b/crates/util/Cargo.toml
@@ -39,6 +39,7 @@ git2 = { version = "0.19.0", default-features = false, features = [
   "https",
 ] }
 tempfile = { version = "3.1" }
+similar = { version = "2.2.1" }
 
 
 napi = { version = "2.16.4", default-features = false, features = [
diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index 1dbf162eb..a15a303a9 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -5,133 +5,19 @@
 ================================================================================ */
 
 use anyhow::Result;
-use git2::{Delta, DiffLineType, DiffOptions, Repository};
+use similar::{ChangeTag, TextDiff};
 
 /// Given a before and after for a file, edit the *after* to not include any spurious changes
-pub fn standardize_rewrite(repo: &Repository, before: String, after: String) -> Result<String> {
-    let mut diff_opts = DiffOptions::new();
-    diff_opts.ignore_whitespace(true);
-    diff_opts.ignore_whitespace_change(true);
-    diff_opts.context_lines(0);
-
-    let left_oid = repo
-        .blob(before.as_bytes())
-        .map_err(|e| anyhow::anyhow!("Failed to create left blob: {:?}", e))?;
-    let right_oid = repo
-        .blob(after.as_bytes())
-        .map_err(|e| anyhow::anyhow!("Failed to create right blob: {:?}", e))?;
-
-    let left_blob = repo
-        .find_blob(left_oid)
-        .map_err(|e| anyhow::anyhow!("Failed to find left blob: {:?}", e))?;
-    let right_blob = repo
-        .find_blob(right_oid)
-        .map_err(|e| anyhow::anyhow!("Failed to find right blob: {:?}", e))?;
-
-    let before_lines = before.lines().collect::<Vec<_>>();
+pub fn standardize_rewrite(before: String, after: String) -> Result<String> {
+    let diff = TextDiff::from_lines(&before, &after);
     let mut standardized_after = String::new();
-    let mut before_line_num = 1_usize;
-    let mut pending_additions: Vec<(usize, String)> = Vec::new();
-
-    repo.diff_blobs(
-        Some(&left_blob),
-        None,
-        Some(&right_blob),
-        None,
-        Some(&mut diff_opts),
-        None,
-        None,
-        None,
-        Some(&mut |_delta, hunk, line| {
-            let hunk = hunk.unwrap();
-            println!("\n=== Processing Hunk ===");
-            println!(
-                "Current before_line_num: {}, Hunk old_start: {}, old_lines: {}, new_lines: {}",
-                before_line_num,
-                hunk.old_start(),
-                hunk.old_lines(),
-                hunk.new_lines()
-            );
-            println!(
-                "Hunk header: {}",
-                std::str::from_utf8(hunk.header()).unwrap()
-            );
-
-            let hunk_start: usize = hunk.old_start().try_into().unwrap();
 
-            // Copy any unchanged lines up to this hunk
-            while before_line_num < hunk_start {
-                let this_line = before_lines[before_line_num - 1];
-                println!(
-                    "Injecting unchanged line {}: '{}'",
-                    before_line_num, this_line
-                );
-                standardized_after.push_str(this_line);
-                standardized_after.push('\n');
-                before_line_num += 1;
+    for change in diff.iter_all_changes() {
+        match change.tag() {
+            ChangeTag::Delete => {} // Skip deleted lines
+            ChangeTag::Equal | ChangeTag::Insert => {
+                standardized_after.push_str(change.value());
             }
-
-            match line.origin_value() {
-                DiffLineType::Deletion => {
-                    println!(
-                        "Deletion at line {}: '{}'",
-                        before_line_num,
-                        before_lines[before_line_num - 1]
-                    );
-                    before_line_num += 1;
-                }
-                DiffLineType::Addition => {
-                    let new_content = std::str::from_utf8(line.content()).unwrap().to_string();
-                    let target_position = hunk.new_start() as usize;
-                    println!(
-                        "Queueing addition at position {}: '{}'",
-                        target_position, new_content
-                    );
-                    // Insert into pending_additions in sorted order by position
-                    let insert_pos =
-                        pending_additions.partition_point(|(pos, _)| *pos <= target_position);
-                    pending_additions.insert(insert_pos, (target_position, new_content));
-                }
-                _ => {
-                    println!(
-                        "Other line type: {:?}, num_lines: {}",
-                        line.origin_value(),
-                        line.num_lines()
-                    );
-                }
-            }
-
-            // Process any pending additions that should come before the next hunk
-            let mut current_line = before_line_num;
-            while let Some((pos, content)) = pending_additions.first() {
-                if *pos <= current_line {
-                    standardized_after.push_str(&content);
-                    pending_additions.remove(0);
-                } else {
-                    break;
-                }
-            }
-
-            true
-        }),
-    )
-    .map_err(|e| anyhow::anyhow!("Failed to generate diff: {:?}", e))?;
-
-    // Add any remaining content and pending additions
-    while before_line_num <= before_lines.len() || !pending_additions.is_empty() {
-        let next_pending = pending_additions.first().map(|(pos, _)| *pos);
-        let should_add_original = match next_pending {
-            Some(pos) => before_line_num < pos,
-            None => true,
-        };
-
-        if should_add_original && before_line_num <= before_lines.len() {
-            standardized_after.push_str(before_lines[before_line_num - 1]);
-            standardized_after.push('\n');
-            before_line_num += 1;
-        } else if let Some((_, content)) = pending_additions.first() {
-            standardized_after.push_str(content);
-            pending_additions.remove(0);
         }
     }
 
@@ -143,23 +29,11 @@ mod tests {
     use super::*;
     use insta::assert_snapshot;
 
-    fn setup_test_repo() -> Result<(Repository, tempfile::TempDir)> {
-        let temp_dir = tempfile::tempdir()?;
-        let repo = Repository::init_opts(
-            temp_dir.path(),
-            git2::RepositoryInitOptions::new()
-                .bare(true)
-                .initial_head("main"),
-        )?;
-        Ok((repo, temp_dir))
-    }
-
     #[test]
     fn test_basic_rewrite() -> Result<()> {
-        let (repo, _temp) = setup_test_repo()?;
         let before = "Hello world\n".to_string();
         let after = "Hello Rust\n".to_string();
-        let result = standardize_rewrite(&repo, before, after)?;
+        let result = standardize_rewrite(before, after)?;
         assert_eq!(result, "Hello Rust\n");
         assert_snapshot!(result);
         Ok(())
@@ -167,21 +41,19 @@ mod tests {
 
     #[test]
     fn test_whitespace_handling() -> Result<()> {
-        let (repo, _temp) = setup_test_repo()?;
         let before = "function test() {\n    console.bob('test');\n}\n".to_string();
         let after = "function test() {\nconsole.log('test');\n}\n".to_string();
-        let after_standard = "function test() {console.log('test');\n}\n".to_string();
-        let result = standardize_rewrite(&repo, before, after)?;
+        let after_standard = "function test() {\nconsole.log('test');\n}\n".to_string();
+        let result = standardize_rewrite(before, after)?;
         assert_eq!(result, after_standard);
         Ok(())
     }
 
     #[test]
     fn test_empty_files() -> Result<()> {
-        let (repo, _temp) = setup_test_repo()?;
         let before = "".to_string();
         let after = "".to_string();
-        let result = standardize_rewrite(&repo, before, after)?;
+        let result = standardize_rewrite(before, after)?;
         assert_eq!(result, "");
         assert_snapshot!(result);
         Ok(())
@@ -189,20 +61,18 @@ mod tests {
 
     #[test]
     fn test_multiline_changes() -> Result<()> {
-        let (repo, _temp) = setup_test_repo()?;
         let before = "line1\nline2\n  line3\n".to_string();
         let after1 = "line1\nmodified line2\n  line3\nnew line4\n".to_string();
         let after2 = "line1\nmodified line2\n  line3\nnew line4\n".to_string();
-        let result = standardize_rewrite(&repo, before, after1)?;
+        let result = standardize_rewrite(before, after1)?;
         assert_eq!(result, after2);
         Ok(())
     }
 
     #[test]
     fn test_no_changes() -> Result<()> {
-        let (repo, _temp) = setup_test_repo()?;
         let content = "unchanged content\n".to_string();
-        let result = standardize_rewrite(&repo, content.clone(), content)?;
+        let result = standardize_rewrite(content.clone(), content)?;
         assert_eq!(result, "unchanged content\n");
         assert_snapshot!(result);
         Ok(())
@@ -210,7 +80,6 @@ mod tests {
 
     #[test]
     fn test_mixed_changes_in_large_file() -> Result<()> {
-        let (repo, _temp) = setup_test_repo()?;
         let before = r#"
 /* ================================================================================
 
@@ -253,16 +122,13 @@ fn third_function() {
 "#
         .to_string();
 
-        let result = standardize_rewrite(&repo, before, after)?;
-
+        let result = standardize_rewrite(before, after)?;
         assert_snapshot!(result);
-
         Ok(())
     }
 
     #[test]
     fn test_code_removal() -> Result<()> {
-        let (repo, _temp) = setup_test_repo()?;
         let before = r#"fn main() {
     // First we do some setup
     let mut total = 0;
@@ -288,52 +154,30 @@ fn third_function() {
     "#
         .to_string();
 
-        let result = standardize_rewrite(&repo, before, after.clone())?;
+        let result = standardize_rewrite(before, after.clone())?;
         assert_eq!(result, after);
         Ok(())
     }
 
     #[test]
     fn test_remove_early_add_late() -> Result<()> {
-        let (repo, _td) = setup_test_repo()?;
-
-        let before = r#"
-// this starts us off
-
-fn first_function() {
-    println!("This will be removed");
-    println!("This will also be removed");
-}
-
-fn middle_function() {
-    // This stays the same
-    println!("Hello");
-}
-
-fn last_function() {
-    println!("Original");
-}
-
-// This is other content"#;
-
-        let after = r#"
-// this starts us off
-
-fn middle_function() {
-    // This stays the same
-    println!("Hello");
-}
-
-fn last_function() {
-    println!("Original");
-    println!("New line added here");
-    println!("And another one");
-}
+        let before = r#"fn main() {
+    let early = "remove me";
+    let keep = "stay";
+    let middle = "remove me too";
+    let end = "keep me";
+}"#
+        .to_string();
 
-// This is other content"#;
+        let after = r#"fn main() {
+    let keep = "stay";
+    let end = "keep me";
+    let new = "add me";
+}"#
+        .to_string();
 
-        let result = standardize_rewrite(&repo, before.to_string(), after.to_string())?;
-        assert_snapshot!(result);
+        let result = standardize_rewrite(before, after.clone())?;
+        assert_eq!(result, after);
         Ok(())
     }
 }
diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
index 44ad23d49..03c7de17d 100644
--- a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
@@ -4,15 +4,15 @@ expression: result
 ---
 /* ================================================================================
 
-       getMulti.
+    getMulti.
 
 ================================================================================ */
 fn second_function() {
- let mut total = 0;
+    let mut total = 0;
     for i in 0..10 {
         total += i;
     }
-  println!("Total: {}", total);
+    println!("Total: {}", total);
 }
 
 fn third_function() {

From ecc8112b3b4e0a2da532d0f78b1ed335d898c0ef Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 03:27:59 -0500
Subject: [PATCH 19/21] repair one test

---
 ...ff_standardizer__tests__mixed_changes_in_large_file.snap | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
index 03c7de17d..44ad23d49 100644
--- a/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
+++ b/crates/util/src/snapshots/marzano_util__diff_standardizer__tests__mixed_changes_in_large_file.snap
@@ -4,15 +4,15 @@ expression: result
 ---
 /* ================================================================================
 
-    getMulti.
+       getMulti.
 
 ================================================================================ */
 fn second_function() {
-    let mut total = 0;
+ let mut total = 0;
     for i in 0..10 {
         total += i;
     }
-    println!("Total: {}", total);
+  println!("Total: {}", total);
 }
 
 fn third_function() {

From 684ae32b2717295ebf289e1fe2c72bf3111a1863 Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 03:59:29 -0500
Subject: [PATCH 20/21] run the tests

---
 crates/util/src/diff_standardizer.rs | 52 +++++++++++++++++++++++-----
 1 file changed, 44 insertions(+), 8 deletions(-)

diff --git a/crates/util/src/diff_standardizer.rs b/crates/util/src/diff_standardizer.rs
index a15a303a9..9304e486f 100644
--- a/crates/util/src/diff_standardizer.rs
+++ b/crates/util/src/diff_standardizer.rs
@@ -9,14 +9,49 @@ use similar::{ChangeTag, TextDiff};
 
 /// Given a before and after for a file, edit the *after* to not include any spurious changes
 pub fn standardize_rewrite(before: String, after: String) -> Result<String> {
-    let diff = TextDiff::from_lines(&before, &after);
+    let mut differ = TextDiff::configure();
+    differ.algorithm(similar::Algorithm::Myers);
+    let diff = differ.diff_lines(&before, &after);
     let mut standardized_after = String::new();
 
-    for change in diff.iter_all_changes() {
-        match change.tag() {
-            ChangeTag::Delete => {} // Skip deleted lines
-            ChangeTag::Equal | ChangeTag::Insert => {
-                standardized_after.push_str(change.value());
+    for op in diff.ops() {
+        match op.tag() {
+            similar::DiffTag::Equal | similar::DiffTag::Insert => {
+                for line in diff.iter_changes(op) {
+                    standardized_after.push_str(line.value());
+                }
+            }
+            similar::DiffTag::Delete => {
+                // Simply skip deleted lines
+            }
+            similar::DiffTag::Replace => {
+                let mut before_cache = Option::None;
+                for line in diff.iter_changes(op) {
+                    match line.tag() {
+                        ChangeTag::Delete => {
+                            before_cache = Some(line.value());
+                        }
+                        ChangeTag::Insert => {
+                            let value = line.value();
+                            if let Some(before) = before_cache {
+                                if before.trim() == value.trim() {
+                                    // skip whitespace-only changes
+                                    standardized_after.push_str(before);
+                                } else {
+                                    // Otherwise, include the line
+                                    standardized_after.push_str(value);
+                                }
+                            } else {
+                                standardized_after.push_str(value);
+                            }
+                            before_cache = None;
+                        }
+                        ChangeTag::Equal => {
+                            standardized_after.push_str(line.value());
+                            before_cache = None;
+                        }
+                    }
+                }
             }
         }
     }
@@ -144,14 +179,15 @@ fn third_function() {
 
     // Finally print the result
     println!("The total is: {}", total);
-}"#
+}
+"#
         .to_string();
 
         let after = r#"fn main() {
     let mut total = 0;
     println!("The total is: {}", total);
 }
-    "#
+"#
         .to_string();
 
         let result = standardize_rewrite(before, after.clone())?;

From 1bebf5018bb2ddc9fe983644317c6b2124fa8b0c Mon Sep 17 00:00:00 2001
From: Morgante Pell <me@morgante.net>
Date: Sun, 17 Nov 2024 04:01:55 -0500
Subject: [PATCH 21/21] run the tests

---
 Cargo.lock             | 2 --
 crates/util/Cargo.toml | 7 -------
 2 files changed, 9 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index ea1bef644..cb8bb9264 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -2407,7 +2407,6 @@ dependencies = [
  "base64 0.21.7",
  "fs-err",
  "futures",
- "git2",
  "grit-util",
  "http 0.2.12",
  "ignore",
@@ -2420,7 +2419,6 @@ dependencies = [
  "serde_json",
  "sha2",
  "similar",
- "tempfile",
  "tokio",
  "tree-sitter-facade-sg",
 ]
diff --git a/crates/util/Cargo.toml b/crates/util/Cargo.toml
index 98cc279d2..823aeb3a0 100644
--- a/crates/util/Cargo.toml
+++ b/crates/util/Cargo.toml
@@ -32,13 +32,6 @@ serde_json = { version = "1.0.114" }
 futures = { version = "0.3.29", optional = true }
 http = { version = "0.2.11" }
 fs-err = { version = "2.11.0" }
-git2 = { version = "0.19.0", default-features = false, features = [
-  "vendored-openssl",
-  "vendored-libgit2",
-  "ssh",
-  "https",
-] }
-tempfile = { version = "3.1" }
 similar = { version = "2.2.1" }