From 8cdb59fa5edd896cd639592a9ee94673992bcb21 Mon Sep 17 00:00:00 2001
From: Wilfred Hughes <me@wilfred.me.uk>
Date: Mon, 21 Oct 2024 08:34:39 -0700
Subject: [PATCH] Normalise newlines before diffing

This produces good, consistent results on different trailing newlines
before, consistent with the behaviour before 7edd2a82cd (see #755).
---
 sample_files/compare.expected | 10 +++++-----
 src/display/side_by_side.rs   | 11 +++++++++--
 src/main.rs                   | 23 ++++++++++++++---------
 3 files changed, 28 insertions(+), 16 deletions(-)

diff --git a/sample_files/compare.expected b/sample_files/compare.expected
index 3f1a3b27ee..77b7ce1f05 100644
--- a/sample_files/compare.expected
+++ b/sample_files/compare.expected
@@ -5,7 +5,7 @@ sample_files/ada_1.adb sample_files/ada_2.adb
 eb3b0e12e239ae33789136380e81406b  -
 
 sample_files/added_line_1.txt sample_files/added_line_2.txt
-f935214c79833318d39364145d36d8e2  -
+8a1587e6b5fc53f2ec2ac665a5d00372  -
 
 sample_files/align_footer_1.txt sample_files/align_footer_2.txt
 d640bd2de31e56a39f0efb92aff0f379  -
@@ -23,7 +23,7 @@ sample_files/bad_combine_1.rs sample_files/bad_combine_2.rs
 f5051bf7d2b8afa3a677388cbd458891  -
 
 sample_files/big_text_hunk_1.txt sample_files/big_text_hunk_2.txt
-dcc22684c2a200afdd583d621b47dfa3  -
+fc26d41a5ff771670e04033b177973d2  -
 
 sample_files/change_outer_1.el sample_files/change_outer_2.el
 2b9334a4cc72da63bba28eff958f0038  -
@@ -74,7 +74,7 @@ sample_files/erlang_1.erl sample_files/erlang_2.erl
 dccdb8f65d2f099ab1a8cb66011376a2  -
 
 sample_files/f_sharp_1.fs sample_files/f_sharp_2.fs
-1a9173c15a42c1ebb9522109df172faf  -
+0a6651925acadf366f213d15968ae353  -
 
 sample_files/hack_1.php sample_files/hack_2.php
 c2bb0aa7d7b07d6ced79f6a5363e878b  -
@@ -215,7 +215,7 @@ sample_files/racket_1.rkt sample_files/racket_2.rkt
 605aec93fa7b89d08e5d8ed56ad3c1be  -
 
 sample_files/repeated_line_no_eol_1.txt sample_files/repeated_line_no_eol_2.txt
-ac714893a2d28dc0204855d308972ccd  -
+3786f55d2c9b1897e866b4602e50408d  -
 
 sample_files/ruby_1.rb sample_files/ruby_2.rb
 d4d591902030355656f5c18c78f965a6  -
@@ -302,5 +302,5 @@ sample_files/yaml_1.yaml sample_files/yaml_2.yaml
 f068239fc7bade0e6de96d81136c1ac5  -
 
 sample_files/zig_1.zig sample_files/zig_2.zig
-e36d1ea126b8b68e3344434bb63f205e  -
+4516796003b81f35bfa57d84bb7c0cbe  -
 
diff --git a/src/display/side_by_side.rs b/src/display/side_by_side.rs
index 055fa655f4..c3407bd300 100644
--- a/src/display/side_by_side.rs
+++ b/src/display/side_by_side.rs
@@ -407,8 +407,15 @@ pub(crate) fn print(
     let mut prev_lhs_line_num = None;
     let mut prev_rhs_line_num = None;
 
-    let lhs_lines = split_on_newlines(lhs_src).collect::<Vec<_>>();
-    let rhs_lines = split_on_newlines(rhs_src).collect::<Vec<_>>();
+    let mut lhs_lines = split_on_newlines(lhs_src).collect::<Vec<_>>();
+    let mut rhs_lines = split_on_newlines(rhs_src).collect::<Vec<_>>();
+
+    if lhs_lines.last() == Some(&"") && lhs_lines.len() > 1 {
+        lhs_lines.pop();
+    }
+    if rhs_lines.last() == Some(&"") && rhs_lines.len() > 1 {
+        rhs_lines.pop();
+    }
 
     let matched_lines = all_matched_lines_filled(lhs_mps, rhs_mps, &lhs_lines, &rhs_lines);
     let mut matched_lines_to_print = &matched_lines[..];
diff --git a/src/main.rs b/src/main.rs
index 4aad367612..6fd6b3fe1c 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -373,16 +373,21 @@ fn diff_file(
         rhs_src.retain(|c| c != '\r');
     }
 
-    // If "foo" is one line, is "foo\n" two lines? Generally we want
-    // to care about newlines when deciding whether content differs.
+    // Ensure that lhs_src and rhs_src both have trailing
+    // newlines.
     //
-    // Ending a file with a trailing newline is extremely common
-    // though. If both files have a trailing newline, consider "foo\n"
-    // to be "foo" so we don't end up displaying a blank line on both
-    // sides.
-    if lhs_src.ends_with('\n') && rhs_src.ends_with('\n') {
-        lhs_src.pop();
-        rhs_src.pop();
+    // This is important when textually diffing files that don't have
+    // a trailing newline, e.g. "foo\n\bar\n" versus "foo". We want to
+    // consider `foo` to be unchanged in this case.
+    //
+    // Theoretically a tree-sitter parser coud change its AST due to
+    // the additional trailing newline, but it seems vanishingly
+    // unlikely.
+    if !lhs_src.is_empty() && !lhs_src.ends_with('\n') {
+        lhs_src.push('\n');
+    }
+    if !rhs_src.is_empty() && !rhs_src.ends_with('\n') {
+        rhs_src.push('\n');
     }
 
     let mut extra_info = renamed;