[PATCH v6 3/5] fast-export: avoid stripping encoding header if we cannot reencode

[Elijah Newren <newren@gmail.com>]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=UTF-8, Size: 3048 bytes --]

When fast-export encounters a commit with an 'encoding' header, it tries
to reencode in UTF-8 and then drops the encoding header.  However, if it
fails to reencode in UTF-8 because e.g. one of the characters in the
commit message was invalid in the old encoding, then we need to retain
the original encoding or otherwise we lose information needed to
understand all the other (valid) characters in the original commit
message.

Signed-off-by: Elijah Newren <newren@gmail.com>
---
 builtin/fast-export.c                        |  7 +++++--
 t/t9350-fast-export.sh                       | 21 ++++++++++++++++++++
 t/t9350/broken-iso-8859-7-commit-message.txt |  1 +
 3 files changed, 27 insertions(+), 2 deletions(-)
 create mode 100644 t/t9350/broken-iso-8859-7-commit-message.txt

diff --git a/builtin/fast-export.c b/builtin/fast-export.c
index 9e283482ef..7734a9f5a5 100644
--- a/builtin/fast-export.c
+++ b/builtin/fast-export.c
@@ -642,9 +642,12 @@ static void handle_commit(struct commit *commit, struct rev_info *rev,
 	printf("commit %s\nmark :%"PRIu32"\n", refname, last_idnum);
 	if (show_original_ids)
 		printf("original-oid %s\n", oid_to_hex(&commit->object.oid));
-	printf("%.*s\n%.*s\ndata %u\n%s",
+	printf("%.*s\n%.*s\n",
 	       (int)(author_end - author), author,
-	       (int)(committer_end - committer), committer,
+	       (int)(committer_end - committer), committer);
+	if (!reencoded && encoding)
+		printf("encoding %s\n", encoding);
+	printf("data %u\n%s",
 	       (unsigned)(reencoded
 			  ? strlen(reencoded) : message
 			  ? strlen(message) : 0),
diff --git a/t/t9350-fast-export.sh b/t/t9350-fast-export.sh
index 756d6a9905..e2ab8eddc0 100755
--- a/t/t9350-fast-export.sh
+++ b/t/t9350-fast-export.sh
@@ -118,6 +118,27 @@ test_expect_success 'iso-8859-7' '
 		 ! grep ^encoding actual)
 '
 
+test_expect_success 'encoding preserved if reencoding fails' '
+
+	test_when_finished "git reset --hard HEAD~1" &&
+	test_config i18n.commitencoding iso-8859-7 &&
+	echo rosten >file &&
+	git commit -s -F "$TEST_DIRECTORY/t9350/broken-iso-8859-7-commit-message.txt" file &&
+	git fast-export wer^..wer >iso-8859-7.fi &&
+	sed "s/wer/i18n-invalid/" iso-8859-7.fi |
+		(cd new &&
+		 git fast-import &&
+		 git cat-file commit i18n-invalid >actual &&
+		 # Make sure the commit still has the encoding header
+		 grep ^encoding actual &&
+		 # Verify that the commit has the expected size; i.e.
+		 # that no bytes were re-encoded to a different encoding.
+		 test 252 -eq "$(git cat-file -s i18n-invalid)" &&
+		 # ...and check for the original special bytes
+		 grep $(printf "\360") actual &&
+		 grep $(printf "\377") actual)
+'
+
 test_expect_success 'import/export-marks' '
 
 	git checkout -b marks master &&
diff --git a/t/t9350/broken-iso-8859-7-commit-message.txt b/t/t9350/broken-iso-8859-7-commit-message.txt
new file mode 100644
index 0000000000..d06ad75b44
--- /dev/null
+++ b/t/t9350/broken-iso-8859-7-commit-message.txt
@@ -0,0 +1 @@
+Pi: ð; Invalid: ÿ
\ No newline at end of file
-- 
2.21.0.782.gd18f03d1bf