Handle empty commit headers

In some cases, especially for some merges, a commit can contain zero files.  This can throw off the header parser by interpreting the following headers as the body text of the first commit.  Down-stream, this causes the GitLeaks tool to identify the wrong commit ID with the leaked data.

This patch currently fails the empty-commit check.  A minor issue with the expected data eludes me at the moment.

gitdiff/parser_test.go

@@ -630,7 +630,49 @@ Date:   Tue Apr 2 22:55:40 2019 -0700
 					TextFragments: textFragments,
 				},
 			},
-			Preamble: binaryPreamble,
+			Preamble: textPreamble,
+		},
+		"emptyCommit": {
+			InputFile: "testdata/empty-commit.patch",
+			Output: []*File{
+				{
+					PatchHeader: &PatchHeader{
+						SHA: "48f13f61afb200ad8386573632cba3abd1703af2",
+						Author: &PatchIdentity{
+							Name:  "Morton Haypenny",
+							Email: "mhaypenny@example.com",
+						},
+						AuthorDate: asTime("2017-11-03T12:29:49Z"),
+						Title:      "A simple change",
+						Body:       "The change is simple.",
+					},
+					OldName:       "dir/file1.txt",
+					NewName:       "dir/file1.txt",
+					OldMode:       os.FileMode(0100644),
+					OldOIDPrefix:  "422cce7",
+					NewOIDPrefix:  "24b39ed",
+					TextFragments: textFragments,
+				},
+				{
+					PatchHeader: &PatchHeader{
+						SHA: "183d9cdbecda47e6e95acf9e4e23fa3a71ba99ad",
+						Author: &PatchIdentity{
+							Name:  "Regina Smithee",
+							Email: "rsmithee@example.com",
+						},
+						AuthorDate: asTime("2017-09-14T19:46:12Z"),
+						Title:      "Simple change",
+						Body:       "Simple change body.",
+					},
+					OldName:       "d1/file2.txt",
+					NewName:       "d1/file2.txt",
+					OldMode:       os.FileMode(0100755),
+					OldOIDPrefix:  "c6513ff",
+					NewOIDPrefix:  "bf4c6cc",
+					TextFragments: textFragments,
+				},
+			},
+			Preamble: textPreamble,
 		},
 	}
 

gitdiff/patch_header.go

@@ -294,10 +294,14 @@ func parseHeaderPretty(prettyLine string, r io.Reader) (*PatchHeader, error) {
 
 	if title != "" {
 		// Don't check for an appendix
-		body, _ := scanMessageBody(s, indent, false)
+		body, _, remainder := scanMessageBody(s, indent, false)
 		if s.Err() != nil {
 			return nil, s.Err()
 		}
+		if remainder != "" {
+			// There was another header immediately after this one.
+			return ParsePatchHeader(remainder)
+		}
 		h.Body = body
 	}
 
@@ -326,22 +330,34 @@ func scanMessageTitle(s *bufio.Scanner) (title string, indent string) {
 	return b.String(), indent
 }
 
-func scanMessageBody(s *bufio.Scanner, indent string, separateAppendix bool) (string, string) {
+func scanMessageBody(s *bufio.Scanner, indent string, separateAppendix bool) (string, string, string) {
 	// Body and appendix
 	var body, appendix strings.Builder
 	c := &body
 	var empty int
 	for i := 0; s.Scan(); i++ {
-		line := s.Text()
+		baseLine := s.Text()
 
-		line = strings.TrimRightFunc(line, unicode.IsSpace)
+		line := strings.TrimRightFunc(baseLine, unicode.IsSpace)
 		line = strings.TrimPrefix(line, indent)
 
 		if line == "" {
 			empty++
 			continue
 		}
 
+		if baseLine == line && indent != "" {
+			// The line does not start with the indent.
+			var remainder strings.Builder
+			remainder.WriteString(baseLine)
+			remainder.WriteByte('\n')
+			for ; s.Scan(); i++ {
+				remainder.WriteString(s.Text())
+				remainder.WriteByte('\n')
+			}
+			return body.String(), appendix.String(), remainder.String()
+		}
+
 		// If requested, parse out "appendix" information (often added
 		// by `git format-patch` and removed by `git am`).
 		if separateAppendix && c == &body && line == "---" {
@@ -359,7 +375,7 @@ func scanMessageBody(s *bufio.Scanner, indent string, separateAppendix bool) (st
 
 		c.WriteString(line)
 	}
-	return body.String(), appendix.String()
+	return body.String(), appendix.String(), ""
 }
 
 func parseHeaderMail(mailLine string, r io.Reader) (*PatchHeader, error) {
@@ -402,7 +418,7 @@ func parseHeaderMail(mailLine string, r io.Reader) (*PatchHeader, error) {
 	h.SubjectPrefix, h.Title = parseSubject(subject)
 
 	s := bufio.NewScanner(msg.Body)
-	h.Body, h.BodyAppendix = scanMessageBody(s, "", true)
+	h.Body, h.BodyAppendix, _ = scanMessageBody(s, "", true)
 	if s.Err() != nil {
 		return nil, s.Err()
 	}

gitdiff/patch_header_test.go

@@ -441,6 +441,39 @@ Author: Morton Haypenny <mhaypenny@example.com>
 				Title:  expectedTitle,
 			},
 		},
+		"emptyCommit": {
+			Input: `commit 989f970ba7e43cd0eac6fcf71acfd9c92effc047
+Author: Morton Haypenny <mhaypenny@example.com>
+Date:   Sat Apr 12 05:20:49 2020 -0700
+
+    An empty commit
+
+    With a body.
+
+
+commit 61f5cd90bed4d204ee3feb3aa41ee91d4734855b
+Author: Morton Haypenny <mhaypenny@example.com>
+Date:   Sat Apr 11 15:21:23 2020 -0700
+
+    A sample commit to test header parsing
+
+
+    The medium format shows the body, which
+    may wrap on to multiple lines.
+
+
+    Another body line.
+
+
+`,
+			Header: PatchHeader{
+				SHA:        expectedSHA,
+				Author:     expectedIdentity,
+				AuthorDate: expectedDate,
+				Title:      expectedTitle,
+				Body:       expectedBody,
+			},
+		},
 	}
 
 	for name, test := range tests {

gitdiff/testdata/empty-commit.patch

@@ -0,0 +1,83 @@
+
+commit 48f13f61afb200ad8386573632cba3abd1703af2
+Author: Morton Haypenny <mhaypenny@example.com>
+Date:   Thu Nov 3 12:29:49 2017 +0000
+
+    A simple change
+
+    The change is simple.
+
+diff --git a/dir/file1.txt b/dir/file1.txt
+index 422cce7..24b39ed 100644
+--- a/dir/file1.txt
++++ b/dir/file1.txt
+@@ -3,6 +3,8 @@ fragment 1
+ context line
+-old line 1
+-old line 2
+ context line
++new line 1
++new line 2
++new line 3
+ context line
+-old line 3
++new line 4
++new line 5
+@@ -31,2 +33,2 @@ fragment 2
+ context line
+-old line 4
++new line 6
+
+commit dbbc5f1e926391d737e397ec60cc1ff94787e105
+Author: Regina Smithee <rsmithee@example.com>
+Date:   Thu Sep 29 12:58:35 2017 +0000
+
+    Merged commit 3
+
+    Merged body 3.
+
+
+commit f90c2e23af6618388cb4082f39849476e39105af
+Merge: 989f970 e84c7ab
+Author: Regina Smithee <rsmithee@example.com>
+Date:   Thu Sep 15 16:37:30 2017 +0000
+
+    Merged commit 2
+
+
+commit 989f970ba7e43cd0eac6fcf71acfd9c92effc047
+Merge: 183d9cd e84c7ab
+Author: Regina Smithee <rsmithee@example.com>
+Date:   Thu Sep 15 14:41:57 2017 +0000
+
+    Merged commit 1
+
+
+commit 183d9cdbecda47e6e95acf9e4e23fa3a71ba99ad
+Author: Regina Smithee <rsmithee@example.com>
+Date:   Wed Sep 14 19:46:12 2017 +0000
+
+    Simple change
+
+    Simple change body.
+
+diff --git a/d1/file2.txt b/d1/file2.txt
+index c6513ff..bf4c6cc 100755
+--- a/d1/file2.txt
++++ b/d1/file2.txt
+@@ -3,6 +3,8 @@ fragment 1
+ context line
+-old line 1
+-old line 2
+ context line
++new line 1
++new line 2
++new line 3
+ context line
+-old line 3
++new line 4
++new line 5
+@@ -31,2 +33,2 @@ fragment 2
+ context line
+-old line 4
++new line 6