From: Rhodri James <rhodri@kynesim.co.uk>
Date: Fri, 26 May 2017 11:13:56 +0000 (+0100)
Subject: Check latin1 char expanding to two bytes UTF-8 across a buffer boundary.
X-Git-Tag: R_2_2_3~22^2~105
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=fddefa068028b597ed04a3d67101f8ab8d67bb2c;p=libexpat

Check latin1 char expanding to two bytes UTF-8 across a buffer boundary.
---

diff --git a/expat/tests/runtests.c b/expat/tests/runtests.c
index 9ca0d767..e5b041b4 100644
--- a/expat/tests/runtests.c
+++ b/expat/tests/runtests.c
@@ -836,6 +836,59 @@ START_TEST(test_long_utf8_character)
 }
 END_TEST
 
+/* Test that a long latin-1 attribute (too long to convert in one go)
+ * is correctly converted
+ */
+START_TEST(test_long_latin1_attribute)
+{
+    const char *text =
+        "<?xml version='1.0' encoding='iso-8859-1'?>\n"
+        "<doc att='"
+        /* 64 characters per line */
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNO"
+        /* Last character splits across a buffer boundary */
+        "\xe4'>\n</doc>";
+    const char *expected =
+        /* 64 characters per line */
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOP"
+        "ABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNOPABCDEFGHIJKLMNO"
+        "\xc3\xa4";
+
+    run_attribute_check(text, expected);
+}
+END_TEST
+
+
+
 
 /* Regression test #1 for SF bug #653180. */
 START_TEST(test_line_number_after_parse)
@@ -10227,6 +10280,7 @@ make_suite(void)
     tcase_add_test(tc_basic, test_bad_encoding);
     tcase_add_test(tc_basic, test_latin1_umlauts);
     tcase_add_test(tc_basic, test_long_utf8_character);
+    tcase_add_test(tc_basic, test_long_latin1_attribute);
     /* Regression test for SF bug #491986. */
     tcase_add_test(tc_basic, test_danish_latin1);
     /* Regression test for SF bug #514281. */