tests: internal: utils: adjusts expected utf8 encoded bytes

Signed-off-by: Eduardo Silva <[email protected]>
fluent · Dec 4, 2024 · d06b3fd · d06b3fd
1 parent ca70432
commit d06b3fd
Showing 1 changed file with 38 additions and 15 deletions.
diff --git a/tests/internal/utils.c b/tests/internal/utils.c
@@ -127,7 +127,8 @@ static void write_str_test_cases(struct write_str_case *cases) {
 }
 
 /* test case loop for flb_utils_write_str */
-static void write_str_test_cases_w_buf_size(struct write_str_case *cases, int buf_size) {
+static void write_str_test_cases_w_buf_size(struct write_str_case *cases, int buf_size)
+{
     char *buf = flb_calloc(buf_size + 1, sizeof(char));
     int size = buf_size + 1;
     int off;
@@ -170,11 +171,13 @@ static void write_str_test_cases_w_buf_size(struct write_str_case *cases, int bu
 
 void test_write_str()
 {
-    char buf[10];
-    char japanese_a[4] = {0xe3, 0x81, 0x82};
-    int size = sizeof(buf);
     int off;
     int ret;
+    char buf[10] = {0};
+    int size = sizeof(buf);
+
+    /* escaped Unicode representation of あ */
+    char jp_expected_output[] = "\\u3042";
 
     off = 0;
     ret = flb_utils_write_str(buf, &off, size, "a", 1);
@@ -189,15 +192,16 @@ void test_write_str()
     off = 0;
     ret = flb_utils_write_str(buf, &off, size, "\xe3\x81\x82", 3);
     TEST_CHECK(ret == FLB_TRUE);
-    TEST_CHECK(memcmp(buf, japanese_a, off) == 0);
+    TEST_CHECK(memcmp(buf, jp_expected_output, off) == 0);
 
-    // Truncated bytes
+    /* Truncated bytes: 'buf' should not be touched and off == 0 */
     off = 0;
     ret = flb_utils_write_str(buf, &off, size, "\xe3\x81\x82\xe3", 1);
     TEST_CHECK(ret == FLB_TRUE);
-    TEST_CHECK(memcmp(buf, japanese_a, off) == 0);
+    TEST_CHECK(off == 0);
+    TEST_CHECK(memcmp(buf, jp_expected_output, off) == 0);
 
-    // Error: buffer too small
+    /* Error: buffer too small */
     off = 0;
     ret = flb_utils_write_str(buf, &off, size, "aaaaaaaaaaa", 11);
     TEST_CHECK(ret == FLB_FALSE);
@@ -238,7 +242,7 @@ void test_write_str_invalid_leading_byte()
          */
         {
             "\x00\x01\xe3\x81\x82""abc", 8,  /* note that 0x01 is an invalid byte */
-            "\\u0000\\u0001""\xe3\x81\x82""abc",  /* escape hex */
+            "\\u0000\\u0001\\u3042""abc",  /* escape hex */
             FLB_TRUE
         },
         /*
@@ -252,7 +256,7 @@ void test_write_str_invalid_leading_byte()
             "\xee\x83\xb3" /* f3 fragment */ /* replace invalid unicode */
             "\xee\x82\x81" /* 81 fragment */
             "\xee\x82\x81" /* 81 fragment */
-            "\xe3\x81\x82""abc", /* valid unicode */
+            "\\u3042""abc", /* valid unicode */
             FLB_TRUE
         },
         /*
@@ -263,7 +267,25 @@ void test_write_str_invalid_leading_byte()
             "\xf3\x81\x01\xe3\x81\x82""abc", 9,  /* note that 0x01 is an invalid byte */
             "\xee\x83\xb3" /* f3 fragment */ /* replace invalid unicode */
             "\xee\x82\x81" /* 81 fragment */
-            "\\u0001""\xe3\x81\x82""abc",
+            "\\u0001""\\u3042""abc",
+            FLB_TRUE
+        },
+        { 0 }
+    };
+
+    write_str_test_cases(cases);
+}
+
+void test_write_str_special_bytes()
+{
+
+    struct write_str_case cases[] = {
+        /*
+         * Escaped leading hex (two hex, one valid unicode)
+         */
+        {
+            "你好世界", 12,
+            "\\u4f60\\u597d\\u4e16\\u754c",
             FLB_TRUE
         },
         { 0 }
@@ -342,10 +364,10 @@ void test_write_str_buffer_overrun()
             FLB_FALSE
         },
         {
-            "\""
+            "a"
             "\xe3\x81\x82", 4, /* valid unicode */
-            "\\\"""\xe3\x81\x82", /* just enough space for valid unicode */
-            FLB_TRUE
+            "a", /* just enough space for valid ascii */
+            FLB_FALSE /* no space for \u3042 */
         },
         {
             "\x81"
@@ -662,7 +684,7 @@ struct size_to_bytes_check size_to_bytes_checks[] = {
     {"9223372036.78G", -1},
 };
 
-void test_size_to_bytes() 
+void test_size_to_bytes()
 {
     int i;
     int size;
@@ -682,6 +704,7 @@ TEST_LIST = {
     /* JSON maps iteration */
     { "url_split", test_url_split },
     { "write_str", test_write_str },
+    { "write_str_special_bytes", test_write_str_special_bytes },
     { "test_write_str_invalid_trailing_bytes", test_write_str_invalid_trailing_bytes },
     { "test_write_str_invalid_leading_byte", test_write_str_invalid_leading_byte },
     { "test_write_str_edge_cases", test_write_str_edge_cases },