fix: Handle empty lines in patch parser (#6208)

Co-authored-by: openhands <[email protected]>

openhands/resolver/patching/patch.py

@@ -24,7 +24,7 @@
 unified_header_old_line = re.compile(r'^--- ' + file_timestamp_str + '$')
 unified_header_new_line = re.compile(r'^\+\+\+ ' + file_timestamp_str + '$')
 unified_hunk_start = re.compile(r'^@@ -(\d+),?(\d*) \+(\d+),?(\d*) @@(.*)$')
-unified_change = re.compile('^([-+ ])(.*)$')
+unified_change = re.compile('^([-+ ])(.*)$', re.MULTILINE)
 
 context_header_old_line = re.compile(r'^\*\*\* ' + file_timestamp_str + '$')
 context_header_new_line = re.compile('^--- ' + file_timestamp_str + '$')
@@ -606,38 +606,39 @@ def parse_unified_diff(text):
             h = unified_hunk_start.match(hunk[0])
             del hunk[0]
             if h:
-                old = int(h.group(1))
-                if len(h.group(2)) > 0:
-                    old_len = int(h.group(2))
-                else:
-                    old_len = 0
+                # The hunk header @@ -1,6 +1,6 @@ means:
+                # - Start at line 1 in the old file and show 6 lines
+                # - Start at line 1 in the new file and show 6 lines
+                old = int(h.group(1))  # Starting line in old file
+                old_len = int(h.group(2)) if len(h.group(2)) > 0 else 1  # Number of lines in old file
 
-                new = int(h.group(3))
-                if len(h.group(4)) > 0:
-                    new_len = int(h.group(4))
-                else:
-                    new_len = 0
+                new = int(h.group(3))  # Starting line in new file
+                new_len = int(h.group(4)) if len(h.group(4)) > 0 else 1  # Number of lines in new file
 
                 h = None
                 break
 
+        # Process each line in the hunk
         for n in hunk:
-            c = unified_change.match(n)
-            if c:
-                kind = c.group(1)
-                line = c.group(2)
-
-                if kind == '-' and (r != old_len or r == 0):
-                    changes.append(Change(old + r, None, line, hunk_n))
-                    r += 1
-                elif kind == '+' and (i != new_len or i == 0):
-                    changes.append(Change(None, new + i, line, hunk_n))
-                    i += 1
-                elif kind == ' ':
-                    if r != old_len and i != new_len:
-                        changes.append(Change(old + r, new + i, line, hunk_n))
-                    r += 1
-                    i += 1
+            # Each line in a unified diff starts with a space (context), + (addition), or - (deletion)
+            # The first character is the kind, the rest is the line content
+            kind = n[0] if len(n) > 0 else ' '  # Empty lines in the hunk are treated as context lines
+            line = n[1:] if len(n) > 1 else ''
+
+            # Process the line based on its kind
+            if kind == '-' and (r != old_len or r == 0):
+                # Line was removed from the old file
+                changes.append(Change(old + r, None, line, hunk_n))
+                r += 1
+            elif kind == '+' and (i != new_len or i == 0):
+                # Line was added in the new file
+                changes.append(Change(None, new + i, line, hunk_n))
+                i += 1
+            elif kind == ' ':
+                # Context line - exists in both old and new file
+                changes.append(Change(old + r, new + i, line, hunk_n))
+                r += 1
+                i += 1
 
     if len(changes) > 0:
         return changes

tests/unit/resolver/test_patch_apply.py

@@ -0,0 +1,47 @@
+import pytest
+from openhands.resolver.patching.apply import apply_diff
+from openhands.resolver.patching.exceptions import HunkApplyException
+from openhands.resolver.patching.patch import parse_diff, diffobj
+
+
+def test_patch_apply_with_empty_lines():
+    # The original file has no indentation and uses \n line endings
+    original_content = "# PR Viewer\n\nThis React application allows you to view open pull requests from GitHub repositories in a GitHub organization. By default, it uses the All-Hands-AI organization.\n\n## Setup"
+
+    # The patch has spaces at the start of each line and uses \n line endings
+    patch = """diff --git a/README.md b/README.md
+index b760a53..5071727 100644
+--- a/README.md
++++ b/README.md
+@@ -1,3 +1,3 @@
+ # PR Viewer
+
+-This React application allows you to view open pull requests from GitHub repositories in a GitHub organization. By default, it uses the All-Hands-AI organization.
++This React application was created by Graham Neubig and OpenHands. It allows you to view open pull requests from GitHub repositories in a GitHub organization. By default, it uses the All-Hands-AI organization."""
+
+    print("Original content lines:")
+    for i, line in enumerate(original_content.splitlines(), 1):
+        print(f"{i}: {repr(line)}")
+
+    print("\nPatch lines:")
+    for i, line in enumerate(patch.splitlines(), 1):
+        print(f"{i}: {repr(line)}")
+
+    changes = parse_diff(patch)
+    print("\nParsed changes:")
+    for change in changes:
+        print(f"Change(old={change.old}, new={change.new}, line={repr(change.line)}, hunk={change.hunk})")
+    diff = diffobj(header=None, changes=changes, text=patch)
+
+    # Apply the patch
+    result = apply_diff(diff, original_content)
+
+    # The patch should be applied successfully
+    expected_result = [
+        "# PR Viewer",
+        "",
+        "This React application was created by Graham Neubig and OpenHands. It allows you to view open pull requests from GitHub repositories in a GitHub organization. By default, it uses the All-Hands-AI organization.",
+        "",
+        "## Setup"
+    ]
+    assert result == expected_result