commit 117d208261459f982b3e68c93140e161af1f0ca0
author Radek Krejci <rkrejci@cesnet.cz> Wed Sep 26 10:05:14 2018 +0200
committer Radek Krejci <rkrejci@cesnet.cz> Wed Sep 26 10:05:14 2018 +0200
tree 1a516496a3565fbb2899241b1d50438eb4368053
parent ed6c6add12ea14df5403ac386462e20df5e8d77b

tests CHANGE nehance xml tests - processing n-bytes UTF characters

src/xml.c

diff --git a/src/xml.c b/src/xml.c
index 78b259f..bd26f16 100644
--- a/src/xml.c
+++ b/src/xml.c
@@ -168,7 +168,7 @@
  * Includes checking for valid characters (following RFC 7950, sec 9.4)
  */
 static LY_ERR
-lyxml_pututf8(char *dst, int32_t value, size_t *bytes_written)
+lyxml_pututf8(char *dst, uint32_t value, size_t *bytes_written)
 {
     if (value < 0x80) {
         /* one byte character */
@@ -283,8 +283,8 @@
     size_t len;     /* write offset in output buffer */
     size_t size;    /* size of the output buffer */
     void *p;
-    int32_t n;
-    size_t u;
+    uint32_t n;
+    size_t u, newlines;
     bool empty_content = false;
     LY_ERR rc;
 
@@ -302,8 +302,13 @@
     if (empty_content) {
         /* only when processing element's content - try to ignore whitespaces used to format XML data
          * before element's child or closing tag */
-        for (offset = 0; in[offset] && is_xmlws(in[offset]); ++offset);
+        for (offset = newlines = 0; in[offset] && is_xmlws(in[offset]); ++offset) {
+            if (in[offset] == '\n') {
+                ++newlines;
+            }
+        }
         LY_CHECK_ERR_RET(!in[offset], LOGVAL(ctx, LY_VLOG_LINE, &context->line, LY_VCODE_EOF), LY_EVALID);
+        context->line += newlines;
         if (in[offset] == '<') {
             (*input) = in + offset;
             return LY_EINVAL;
@@ -395,7 +400,7 @@
                 ++offset;
                 rc = lyxml_pututf8(&buf[len], n, &u);
                 LY_CHECK_ERR_GOTO(rc, LOGVAL(ctx, LY_VLOG_LINE, &context->line, LYVE_SYNTAX,
-                                             "Invalid character reference %.*s (0x%08x).", 12, p, n),
+                                             "Invalid character reference \"%.*s\" (0x%08x).", 12, p, n),
                                   error);
                 len += u;
                 in += offset;

tests/src/xml.c

diff --git a/tests/src/xml.c b/tests/src/xml.c
index d299116..4ebdbbd 100644
--- a/tests/src/xml.c
+++ b/tests/src/xml.c
@@ -298,14 +298,14 @@
     /* empty element content is invalid - missing content terminating character < */
     str = "";
     assert_int_equal(LY_EVALID, lyxml_get_string(&ctx, &str, &out, &out_len));
-    logbuf_assert("Unexpected end-of-file. Line number 1.");
-    str = p = "xxx";
+    logbuf_assert("Unexpected end-of-file. Line number 2.");
 
     free(out);
     out = NULL;
 
+    str = p = "xxx";
     assert_int_equal(LY_EVALID, lyxml_get_string(&ctx, &str, &out, &out_len));
-    logbuf_assert("Unexpected end-of-file. Line number 1.");
+    logbuf_assert("Unexpected end-of-file. Line number 2.");
     assert_ptr_equal(p, str); /* input data not eaten */
 
     free(out);
@@ -318,22 +318,40 @@
     assert_string_equal("€𠜎Øn \n<&\"\'> ROK", out);
     assert_string_equal("<", str);
 
+    /* test using n-bytes UTF8 hexadecimal code points */
+    str = "\'&#x0024;&#x00A2;&#x20ac;&#x10348;\'";
+    assert_int_equal(LY_SUCCESS, lyxml_get_string(&ctx, &str, &out, &out_len));
+    assert_string_equal("$¢€𐍈", out);
+
     /* invalid characters in string */
     str = p = "\'&#x52\'";
     assert_int_equal(LY_EVALID, lyxml_get_string(&ctx, &str, &out, &out_len));
-    logbuf_assert("Invalid character sequence \"'\", expected ;. Line number 2.");
+    logbuf_assert("Invalid character sequence \"'\", expected ;. Line number 3.");
     assert_ptr_equal(p, str); /* input data not eaten */
     str = p = "\"&#82\"";
     assert_int_equal(LY_EVALID, lyxml_get_string(&ctx, &str, &out, &out_len));
-    logbuf_assert("Invalid character sequence \"\"\", expected ;. Line number 2.");
+    logbuf_assert("Invalid character sequence \"\"\", expected ;. Line number 3.");
     assert_ptr_equal(p, str); /* input data not eaten */
     str = p = "\"&nonsence;\"";
     assert_int_equal(LY_EVALID, lyxml_get_string(&ctx, &str, &out, &out_len));
-    logbuf_assert("Entity reference \"&nonsence;\" not supported, only predefined references allowed. Line number 2.");
+    logbuf_assert("Entity reference \"&nonsence;\" not supported, only predefined references allowed. Line number 3.");
     assert_ptr_equal(p, str); /* input data not eaten */
     str = p = "&#o122;";
     assert_int_equal(LY_EVALID, lyxml_get_string(&ctx, &str, &out, &out_len));
-    logbuf_assert("Invalid character reference \"&#o122;\". Line number 2.");
+    logbuf_assert("Invalid character reference \"&#o122;\". Line number 3.");
+    assert_ptr_equal(p, str); /* input data not eaten */
+
+    str = p = "\'&#x06;\'";
+    assert_int_equal(LY_EVALID, lyxml_get_string(&ctx, &str, &out, &out_len));
+    logbuf_assert("Invalid character reference \"&#x06;\'\" (0x00000006). Line number 3.");
+    assert_ptr_equal(p, str); /* input data not eaten */
+    str = p = "\'&#xfdd0;\'";
+    assert_int_equal(LY_EVALID, lyxml_get_string(&ctx, &str, &out, &out_len));
+    logbuf_assert("Invalid character reference \"&#xfdd0;\'\" (0x0000fdd0). Line number 3.");
+    assert_ptr_equal(p, str); /* input data not eaten */
+    str = p = "\'&#xffff;\'";
+    assert_int_equal(LY_EVALID, lyxml_get_string(&ctx, &str, &out, &out_len));
+    logbuf_assert("Invalid character reference \"&#xffff;\'\" (0x0000ffff). Line number 3.");
     assert_ptr_equal(p, str); /* input data not eaten */
 
     free(out);