fix: use null-delimited git diff --raw output to avoid Windows escaping issues (#189)

* logging for windows

* use git diff --raw -z

* strip debug code

* update tests with null delimiters

Author: danvk

testdata/unified/manyfiles.txt

@@ -148,7 +148,7 @@ def test_add_replaces():
 
 
 def test_parse_raw_diff_many():
-    # git diff --no-index --raw testdata/manyfiles/{left,right}
+    # git diff --no-index --raw -z testdata/manyfiles/{left,right}
     diff = open('testdata/unified/manyfiles.txt').read()
     mod644 = ['100644', '100644', '0000000', '0000000']
     assert parse_raw_diff(diff) == [
@@ -174,7 +174,7 @@ def test_parse_raw_diff_many():
 
 
 def test_parse_raw_diff_rename():
-    # git diff --no-index --raw testdata/rename+change/{left,right}
+    # git diff --no-index --raw -z testdata/rename+change/{left,right}
     diff = open('testdata/unified/rename+change.txt').read()
     assert parse_raw_diff(diff) == [
         RawDiffLine(

testdata/unified/rename+change.txt

@@ -44,7 +44,7 @@ def make_resolved_dir(dir: str) -> str:
 
 def gitdiff(a_dir, b_dir, webdiff_config):
     extra_args = webdiff_config['extraDirDiffArgs']
-    cmd = 'git diff --raw --no-index'
+    cmd = 'git diff --raw -z --no-index'
     if extra_args:
         cmd += ' ' + extra_args
     a_dir_nosym = a_dir

tests/unified_diff_test.py

@@ -162,8 +162,7 @@ class RawDiffLine:
     """Only set when status=C or R."""
 
 
-def parse_raw_diff_line(line: str) -> RawDiffLine:
-    parts = line.split('\t')
+def parse_raw_diff_line(parts: list[str]) -> RawDiffLine:
     meta = parts[0]
     path = parts[1]
     dst_path = parts[2] if len(parts) > 2 else None
@@ -188,4 +187,13 @@ def parse_raw_diff_line(line: str) -> RawDiffLine:
 
 
 def parse_raw_diff(diff: str) -> List[RawDiffLine]:
-    return [parse_raw_diff_line(line) for line in diff.split('\n') if line]
+    # each diff line can be two or three parts. The parts and lines are both null-delimited.
+    # The "lines" start with ":".
+    parts = diff.strip('\0').split('\0')
+    lines = []
+    for part in parts:
+        if part.startswith(':'):
+            lines.append([part])
+        else:
+            lines[-1].append(part)
+    return [parse_raw_diff_line(line) for line in lines]