Add support for rewriting hunks if only specific lines match

bin/mechanicaldiff.py

@@ -4,9 +4,9 @@ import re
 import sys
 
 
-def should_include_change(
+def filter_change_block(
     change_lines: list[str], search: str, replace: str
-) -> bool:
+) -> tuple[list[str] | None, bool]:
     removed_lines = []
     added_lines = []
     phase = "minus"
@@ -23,33 +23,66 @@ def should_include_change(
             added_lines.append(line[1:])
             continue
         raise ValueError("Unexpected non-change line in change block.")
+
     transformed_removed = [
         re.sub(search, replace, line) for line in removed_lines
     ]
-    return transformed_removed == added_lines
+    if transformed_removed == added_lines:
+        return change_lines, False
+
+    if len(removed_lines) != len(added_lines):
+        return None, True
+
+    kept_lines = []
+    for removed, added in zip(removed_lines, added_lines):
+        if re.sub(search, replace, removed) == added:
+            kept_lines.append(f"-{removed}")
+            kept_lines.append(f"+{added}")
+        else:
+            kept_lines.append(f" {removed}")
+
+    if not kept_lines:
+        return None, True
+
+    return kept_lines, True
 
 
-def should_include_hunk(hunk_text: str, search: str, replace: str) -> bool:
-    lines = hunk_text.splitlines()
-    if not lines:
-        return True
+def filter_hunk(
+    hunk_lines: list[str], search: str, replace: str
+) -> tuple[list[str] | None, bool]:
+    if not hunk_lines:
+        return None, False
 
-    change = []
-    for line in lines[1:]:
+    header = hunk_lines[0]
+    body = hunk_lines[1:]
+    output_body: list[str] = []
+    left_out = False
+
+    i = 0
+    while i < len(body):
+        line = body[i]
         if line.startswith(("+", "-")):
-            change.append(line)
+            block = []
+            while i < len(body) and body[i].startswith(("+", "-")):
+                block.append(body[i])
+                i += 1
+            kept_block, block_left_out = filter_change_block(
+                block, search, replace
+            )
+            if kept_block:
+                output_body.extend(kept_block)
+            else:
+                left_out = True
+            if block_left_out:
+                left_out = True
             continue
+        output_body.append(line)
+        i += 1
 
-        if change:
-            if not should_include_change(change, search, replace):
-                return False
-            change = []
+    if not any(line.startswith(("+", "-")) for line in output_body):
+        return None, True
 
-    if change:
-        if not should_include_change(change, search, replace):
-            return False
-
-    return True
+    return [header] + output_body, left_out
 
 
 def main() -> None:
@@ -97,11 +130,15 @@ def main() -> None:
 
         kept_hunks = []
         for hunk_lines in section["hunks"]:
-            hunk_text = "".join(hunk_lines)
-            if should_include_hunk(hunk_text, search, replace):
-                kept_hunks.append(hunk_lines)
+            filtered, hunk_left_out = filter_hunk(
+                hunk_lines, search, replace
+            )
+            if filtered:
+                kept_hunks.append(filtered)
             else:
                 left_out = True
+            if hunk_left_out:
+                left_out = True
 
         if not kept_hunks:
             continue

test/mechanicaldiff.bats

@@ -128,3 +128,31 @@ setup() {
   assert_success
   assert_output "$(cat "$BATS_TEST_TMPDIR/diff")"
 }
+
+@test "keeps only matching line in a consecutive change block" {
+  cat >"$BATS_TEST_TMPDIR/old" <<-'EOF'
+	foo
+	bar
+	baz
+	EOF
+  sed \
+    -e 's/foo/foo_changed/' \
+    -e 's/bar/bar_changed/' \
+    "$BATS_TEST_TMPDIR/old" >"$BATS_TEST_TMPDIR/new"
+  sed 's/foo/foo_changed/' "$BATS_TEST_TMPDIR/old" \
+    >"$BATS_TEST_TMPDIR/new_kept"
+
+  git diff --no-index "$BATS_TEST_TMPDIR/old" "$BATS_TEST_TMPDIR/new" \
+    >"$BATS_TEST_TMPDIR/diff_full" || true
+  git diff --no-index "$BATS_TEST_TMPDIR/old" "$BATS_TEST_TMPDIR/new_kept" \
+    >"$BATS_TEST_TMPDIR/diff_expected" || true
+
+  run mechanicaldiff.py foo foo_changed <"$BATS_TEST_TMPDIR/diff_full"
+  assert_failure
+  expected="$(sed \
+    -e 's/new_kept/new/g' \
+    -e '/^index /d' \
+    "$BATS_TEST_TMPDIR/diff_expected")"
+  output_normalized="$(printf '%s\n' "$output" | sed -e '/^index /d')"
+  assert_equal "$expected" "$output_normalized"
+}