wip: pr file diffing

src/util/file.rs

@@ -0,0 +1,48 @@
+use memchr::memchr;
+
+pub(crate) enum ContentType {
+    Text,
+    Binary,
+}
+
+pub(crate) struct ContentDiff {
+    old_content: bytes::Bytes,
+    new_content: bytes::Bytes,
+}
+
+pub(crate) struct LineDiff {
+    old_line: Option<usize>,
+    old_content_range: std::ops::Range<usize>,
+    new_line: Option<usize>,
+    new_content_range: std::ops::Range<usize>,
+}
+
+pub(crate) fn classify_content(content: &[u8]) -> ContentType {
+    if content.is_empty() {
+        ContentType::Text
+    } else if content.starts_with(&[0xEF, 0xBB, 0xBF])        // UTF-8
+             || content.starts_with(&[0x00, 0x00, 0xFE, 0xFF]) // UTF-32 BE
+             || content.starts_with(&[0xFF, 0xFE, 0x00, 0x00]) // UTF-32 LE
+             || content.starts_with(&[0xFE, 0xFF])       // UTF-16 BE
+             || content.starts_with(&[0xFF, 0xFE])
+    {
+        ContentType::Text
+    } else {
+        match memchr(0, &content[0..8192]) {
+        | None => ContentType::Text,
+        | Some(_) => ContentType::Binary,
+        }
+    }
+}
+
+pub(crate) fn diff_content(old: &[u8], new: &[u8]) -> ContentDiff {
+    similar::TextDiff::from_lines::<[u8]>(old, new)
+        .iter_all_changes()
+        .map(|change| LineDiff {
+            old_line: change.old_index(),
+            old_content_range: change.old_range,
+            new_line: change.new_index(),
+            new_content_range: change.new_range,
+        })
+        .collect()
+}

src/util/mod.rs

@@ -1 +1,2 @@
+pub(crate) mod file;
 pub(crate) mod timeout;