Trim highlights

For example, if you add a number of spaces followed by a comment at the end of a line, the spaces will not be highlighted. I like this better.
walles · Oct 23, 2024 · 983a5ce · 983a5ce
1 parent dd62ad7
commit 983a5ce
Show file tree

Hide file tree

Showing 13 changed files with 93 additions and 30 deletions.
diff --git a/src/hunk_highlighter.rs b/src/hunk_highlighter.rs
@@ -316,7 +316,7 @@ mod tests {
             result.highlighted[0].get(),
             concat!(
                 "\u{1b}[31m-\u{1b}[33mHello, my name is Johan\u{1b}[0m\n",
-                "\u{1b}[32m+\u{1b}[33mHello, my \u{1b}[7m\u{1b}[32mfirst \u{1b}[27m\u{1b}[33mname is Johan\u{1b}[0m\n"
+                "\u{1b}[32m+\u{1b}[33mHello, my \u{1b}[7m\u{1b}[32mfirst\u{1b}[27m \u{1b}[33mname is Johan\u{1b}[0m\n"
             )
         );
         assert_eq!(result.highlighted[1].get(), " I like pie.\n");

diff --git a/src/refiner.rs b/src/refiner.rs
@@ -217,9 +217,34 @@ fn is_whitepace_replacement(old_run: &[&str], new_run: &[&str]) -> bool {
     return old_whitespace_only && new_whitespace_only;
 }
 
-/// Returns two vectors for old and new sections. The first bool is true if
-/// there were any highlights found in the old text. The second bool is true if
-/// any highlights were removed for readability in the new text.
+fn push_styled_tokens(destination: &mut Vec<StyledToken>, run: Vec<&str>, style: Style) {
+    // Except for just pushing the tokens, any leading or trailing
+    // whitespace-only tokens in the run should always be midlighted.
+
+    let first_non_leading_whitespace_index = run
+        .iter()
+        .position(|token| !token.chars().all(|c| c.is_whitespace()));
+
+    let last_non_trailing_whitespace_index = run
+        .iter()
+        .rposition(|token| !token.chars().all(|c| c.is_whitespace()));
+
+    for (index, token) in run.iter().enumerate() {
+        let in_leading_whitespace = first_non_leading_whitespace_index.is_some()
+            && index < first_non_leading_whitespace_index.unwrap();
+        let in_trailing_whitespace = last_non_trailing_whitespace_index.is_some()
+            && index > last_non_trailing_whitespace_index.unwrap();
+        let style = if in_leading_whitespace || in_trailing_whitespace {
+            Style::DiffPartMidlighted
+        } else {
+            style
+        };
+
+        destination.push(StyledToken::new(token.to_string(), style));
+    }
+}
+
+/// Returns two vectors for old and new sections.
 ///
 /// `old_text` and `new_text` are multi line strings. Having or not having
 /// trailing newlines will affect tokenization. The lines are not expected to
@@ -280,9 +305,7 @@ pub fn to_highlighted_tokens(
                 } else {
                     Style::DiffPartMidlighted
                 };
-                for token in run.iter() {
-                    new_tokens.push(StyledToken::new(token.to_string(), style));
-                }
+                push_styled_tokens(&mut new_tokens, run, style);
             }
 
             similar::DiffOp::Delete {
@@ -296,9 +319,7 @@ pub fn to_highlighted_tokens(
                 } else {
                     Style::DiffPartMidlighted
                 };
-                for token in run.iter() {
-                    old_tokens.push(StyledToken::new(token.to_string(), style));
-                }
+                push_styled_tokens(&mut old_tokens, run, style);
             }
 
             similar::DiffOp::Replace {
@@ -319,13 +340,8 @@ pub fn to_highlighted_tokens(
                     Style::DiffPartMidlighted
                 };
 
-                for token in old_run.iter() {
-                    old_tokens.push(StyledToken::new(token.to_string(), style));
-                }
-
-                for token in new_run.iter() {
-                    new_tokens.push(StyledToken::new(token.to_string(), style));
-                }
+                push_styled_tokens(&mut old_tokens, old_run, style);
+                push_styled_tokens(&mut new_tokens, new_run, style);
             }
         }
 
@@ -761,4 +777,51 @@ pub(crate) mod tests {
             [StyledToken::new("\t".to_string(), Style::DiffPartUnchanged),]
         );
     }
+
+    #[test]
+    fn test_push_styled_tokens() {
+        let mut tokens = Vec::new();
+        push_styled_tokens(&mut tokens, vec!["a", "b", "c"], Style::DiffPartHighlighted);
+        assert_eq!(
+            tokens,
+            vec![
+                StyledToken::new("a".to_string(), Style::DiffPartHighlighted),
+                StyledToken::new("b".to_string(), Style::DiffPartHighlighted),
+                StyledToken::new("c".to_string(), Style::DiffPartHighlighted),
+            ]
+        );
+
+        let mut tokens = Vec::new();
+        push_styled_tokens(&mut tokens, vec![" ", "b", "c"], Style::DiffPartHighlighted);
+        assert_eq!(
+            tokens,
+            vec![
+                StyledToken::new(" ".to_string(), Style::DiffPartMidlighted),
+                StyledToken::new("b".to_string(), Style::DiffPartHighlighted),
+                StyledToken::new("c".to_string(), Style::DiffPartHighlighted),
+            ]
+        );
+
+        let mut tokens = Vec::new();
+        push_styled_tokens(&mut tokens, vec!["a", "b", " "], Style::DiffPartHighlighted);
+        assert_eq!(
+            tokens,
+            vec![
+                StyledToken::new("a".to_string(), Style::DiffPartHighlighted),
+                StyledToken::new("b".to_string(), Style::DiffPartHighlighted),
+                StyledToken::new(" ".to_string(), Style::DiffPartMidlighted),
+            ]
+        );
+
+        let mut tokens = Vec::new();
+        push_styled_tokens(&mut tokens, vec![" ", "b", " "], Style::DiffPartHighlighted);
+        assert_eq!(
+            tokens,
+            vec![
+                StyledToken::new(" ".to_string(), Style::DiffPartMidlighted),
+                StyledToken::new("b".to_string(), Style::DiffPartHighlighted),
+                StyledToken::new(" ".to_string(), Style::DiffPartMidlighted),
+            ]
+        );
+    }
 }
diff --git a/testdata/adds-only.riff-output b/testdata/adds-only.riff-output
@@ -2,5 +2,5 @@
 [1m+++ [0m/tmp/[7m[32mb[27m[1m[39m.txt[22m[2m  2023-08-06 16:07:39.000000000 +0200[0m
 [36m@@ -1,2 +1,2 @@[0m
 [31m-Hello, my name is Johan[0m
-[32m+Hello, my [7mfirst [27mname is Johan[0m
+[32m+Hello, my [7mfirst[27m name is Johan[0m
  I like pie.
diff --git a/testdata/conflict-markers-diff3.txt.riff-output b/testdata/conflict-markers-diff3.txt.riff-output
@@ -3,7 +3,7 @@ This is an example of git conflict markers.
 [7m<<<<<<< HEAD
 [7m[0mThis line is [7m[32mchanged[0m on the main branch.
 [7m||||||| 07ffb9b
-[7m[0mThis line is [7m[31mfrom the initial commit on the main [0mbranch.
+[7m[0mThis line is [7m[31mfrom the initial commit on the main[27m [0mbranch.
 [7m=======
 [7m[0mThis line is from the [7m[32mbranch named "[0mbranch[7m[32m"[0m.
 [7m>>>>>>> branch
diff --git a/testdata/conflict-with-context.riff-output b/testdata/conflict-with-context.riff-output
@@ -20,7 +20,7 @@
 [7m +[27m[32m	}[0m
 [7m +[0m
 [7m +[27m[32m	firstHitPosition := p.findFirstHit(*lineNumber[0m, nil, false)
-[7m +[0m	if firstHitPosition == nil [7m[32m&& (*lineNumber != linenumbers.LineNumber{}) [0m{
+[7m +[0m	if firstHitPosition == nil [7m[32m&& (*lineNumber != linenumbers.LineNumber{})[27m [0m{
 [7m +[0m		// Try again from the top
 [7m +[0m		firstHitPosition = p.findFirstHit(linenumbers.LineNumber{}, lineNumber, false)
 [7m +[0m	}
@@ -37,7 +37,7 @@
 [7m +[0m	p.scrollPosition = *firstHitPosition
 [7m +[0m}
 [7m +[0m
-[7m +[0m// NOTE: When we search, we do that by looping over the *input lines*, not[7m[32m the[0m
+[7m +[0m// NOTE: When we search, we do that by looping over the *input lines*, not[32m [7mthe[0m
 [7m +[0m// screen lines. That's why [7m[32mstartPosition is[0m a [7m[32mLineNumber[0m rather than a
 [7m +[0m// scrollPosition[32m.[0m
 [7m +[27m[32m//[0m
@@ -72,7 +72,7 @@
 [7m++[27m[31m}[0m
 [7m++[0m
 [7m+ [0m// NOTE: When we search, we do that by looping over the *input lines*, not
-[7m+ [0m//[7m[31m the[0m screen lines. That's why [7m[31mwe're using[0m a[7m[31m line number[0m rather than a
+[7m+ [0m//[31m [7mthe[0m screen lines. That's why [7m[31mwe're using[0m a[31m [7mline number[0m rather than a
 [7m+ [0m// scrollPosition [31mfor[0m searching.
 [7m++=======[0m
 [7m+ [0m// NOTE: When we search, we do that by looping over the *input lines*, not

diff --git a/testdata/copies-and-renames.riff-output b/testdata/copies-and-renames.riff-output
@@ -16,7 +16,7 @@
  copies of the Software, and to permit persons to whom the Software is
  furnished to do so, subject to the following conditions:
 
-[31m-The above [7mcopyright [27mnotice and this permission notice shall be included in all[0m
+[31m-The above [7mcopyright[27m notice and this permission notice shall be included in all[0m
 [32m+The above notice and this permission notice shall be included in all[0m
  copies or substantial portions of the Software.
 

diff --git a/testdata/dont-highlight-complete-lines.riff-output b/testdata/dont-highlight-complete-lines.riff-output
@@ -2,7 +2,7 @@
 [1m+++ [7m[22m[32mafter[27m[1m[39m.txt[22m[2m   2023-09-24 16:22:42[0m
 [36m@@ -1,1 +1,6 @@[0m
 [31m-        let tokens = std::mem::take(&mut self.tokens);[0m
-[32m+        let [7mmut [27mtokens = std::mem::take(&mut self.tokens);[0m
+[32m+        let [7mmut[27m tokens = std::mem::take(&mut self.tokens);[0m
 [32m+[0m
 [32m+        // FIXME: Maybe do highlight_space_between_words() before this one? And[0m
 [32m+        // not do that for each line?[0m

diff --git a/testdata/git-diff-conflict-diff3.riff-output b/testdata/git-diff-conflict-diff3.riff-output
@@ -8,7 +8,7 @@
 [7m++<<<<<<< HEAD[0m
 [7m +[0mThis line is [7m[32mchanged[0m on the main branch.
 [7m++||||||| c26bbf8[0m
-[7m++[0mThis line is [7m[31mfrom the initial commit on the main [0mbranch.
+[7m++[0mThis line is [7m[31mfrom the initial commit on the main[27m [0mbranch.
 [7m++=======[0m
 [7m+ [0mThis line is from the [7m[32mbranch named "[0mbranch[7m[32m"[0m.
 [7m++>>>>>>> branch[0m
diff --git a/testdata/issue-with-incomplete-output.riff-output b/testdata/issue-with-incomplete-output.riff-output
@@ -7,7 +7,7 @@
  use crate::string_future::StringFuture;
  use crate::token_collector::{
 [31m-    lowlight_timestamp, render, unhighlight_git_prefix, LINE_STYLE_NEW_FILENAME,[0m
-[32m+    [7malign_tabs, [27mlowlight_timestamp, render, unhighlight_git_prefix, LINE_STYLE_NEW_FILENAME,[0m
+[32m+    [7malign_tabs,[27m lowlight_timestamp, render, unhighlight_git_prefix, LINE_STYLE_NEW_FILENAME,[0m
      LINE_STYLE_OLD_FILENAME,
  };
 

diff --git a/testdata/merge-two-way.riff-output b/testdata/merge-two-way.riff-output
@@ -29,5 +29,5 @@ Date:   Thu Jan 25 08:57:37 2024 +0100
   This is placeholder line six.
 [31m- -> A prefix will be added to this line on main, a suffix on branch and the middle [7mwill be[27m changed during conflict resolution.[0m
 [31m -A prefix will be added to this line on main, a suffix on branch and the middle [7mwill be[27m changed during conflict resolution. <-[0m
-[32m++[7m-> [27mA prefix will be added to this line on main, a suffix on branch and the middle [7mwas[27m changed during conflict resolution.[7m <-[0m
+[32m++[7m->[27m A prefix will be added to this line on main, a suffix on branch and the middle [7mwas[27m changed during conflict resolution. [7m<-[0m
   This is placeholder line seven.
diff --git a/testdata/no-newline-in-context.riff-output b/testdata/no-newline-in-context.riff-output
@@ -2,6 +2,6 @@
 [1m+++ [0m/tmp/dom/[7m[32mbepa[27m[1m[39m.txt[22m[2m  2024-02-04 08:37:13[0m
 [36m@@ -1,2 +1,2 @@[0m
 [31m-[7mChange this[27m line[0m
-[32m+[7mThis[27m line[7m was changed[0m
+[32m+[7mThis[27m line [7mwas changed[0m
  This line has no EOL
 [2m\ No newline at end of file[0m
diff --git a/testdata/partial-refine.riff-output b/testdata/partial-refine.riff-output
@@ -7,7 +7,7 @@
 
  fn print(stream: &mut BufWriter<&mut dyn Write>, text: &str) {
 [31m-    stream.write_all(text.as_bytes()).[7munwrap[27m();[0m
-[32m+    [7mif let Err(error) = [27mstream.write_all(text.as_bytes()) {[0m
+[32m+    [7mif let Err(error) =[27m stream.write_all(text.as_bytes()) {[0m
 [32m+        if error.[7mkind[27m() == ErrorKind::BrokenPipe {[0m
 [32m+            // This is fine, somebody probably just quit their pager before it[0m
 [32m+            // was done reading our output.[0m

diff --git a/testdata/screenshot.riff-output b/testdata/screenshot.riff-output
@@ -13,7 +13,7 @@ Date:   Tue Nov 17 08:44:29 2020 +0100
    search if your search string is a valid regexp
  * Supports displaying ANSI color coded texts (like the output from
 [31m-  `git diff` for example)[0m
-[32m+  `git diff` [7m[| `riff`](https://github.com/walles/riff) [27mfor example)[0m
+[32m+  `git diff` [7m[| `riff`](https://github.com/walles/riff)[27m for example)[0m
  * Supports UTF-8 input and output
  * The position in the file is always shown