Add LYD_PARSE_ANYDATA_STRICT option

src/parser_data.h

@@ -195,6 +195,10 @@ struct ly_in;
                                                        format according to RFC 7951 based on their type. Using this
                                                        option the validation can be softened to accept boolean and
                                                        number type values enclosed in quotes. */
+#define LYD_PARSE_ANYDATA_STRICT 0x10000000 /**< Apply strict parsing (::LYD_PARSE_STRICT) also to anydata
+                                                content. By default, unknown elements in anydata are parsed
+                                                as opaque nodes. With this flag, an error is raised for any unknown
+                                                elements within anydata/anyxml subtrees. */
 #define LYD_PARSE_OPTS_MASK 0xFFFF0000      /**< Mask for all the LYD_PARSE_ options. */
 
 /** @} dataparseroptions */

src/parser_json.c

@@ -1310,7 +1310,11 @@ lydjson_parse_any(struct lyd_json_ctx *lydctx, const struct lysc_node *snode, st
 
         /* parse any data tree with correct options, first backup the current options and then make the parser
          * process data as opaq nodes */
-        lydctx->parse_opts &= ~LYD_PARSE_STRICT;
+        if (lydctx->parse_opts & LYD_PARSE_ANYDATA_STRICT) {
+            lydctx->parse_opts |= LYD_PARSE_STRICT;
+        } else {
+            lydctx->parse_opts &= ~LYD_PARSE_STRICT;
+        }
         lydctx->parse_opts |= LYD_PARSE_OPAQ | (ext ? LYD_PARSE_ONLY : 0);
         lydctx->int_opts |= LYD_INTOPT_ANY | LYD_INTOPT_WITH_SIBLINGS;
         lydctx->any_schema = snode;

src/parser_xml.c

@@ -954,8 +954,12 @@ lydxml_subtree_any(struct lyd_xml_ctx *lydctx, const struct lysc_node *snode, co
         r = lyxml_ctx_next(xmlctx);
         LY_CHECK_ERR_GOTO(r, rc = r, cleanup);
 
-        /* update options so that generic data can be parsed */
-        lydctx->parse_opts &= ~LYD_PARSE_STRICT;
+        if (lydctx->parse_opts & LYD_PARSE_ANYDATA_STRICT) {
+            lydctx->parse_opts |= LYD_PARSE_STRICT;
+        } else {
+            /* update options so that generic data can be parsed */
+            lydctx->parse_opts &= ~LYD_PARSE_STRICT;
+        }
         lydctx->parse_opts |= LYD_PARSE_OPAQ | (ext ? LYD_PARSE_ONLY : 0);
         lydctx->int_opts |= LYD_INTOPT_ANY | LYD_INTOPT_WITH_SIBLINGS;
 

tests/utests/data/test_parser_json.c

@@ -376,6 +376,35 @@ test_anydata(void **state)
     assert_null(tree);
 }
 
+static void
+test_anydata_strict_validation(void **state)
+{
+    const char *data_without_schema;
+    const char *data_invalid;
+    const char *data_valid;
+    struct lyd_node *tree;
+
+    // no shcema defiend for "x" in the parsing context
+    data_without_schema = "{\"a:any\":{\"x:element1\":{\"element2\":\"/a:some/a:path\",\"list\":[{},{\"key\":\"a\"}]}}}";
+
+    PARSER_CHECK_ERROR(data_without_schema, LYD_PARSE_ANYDATA_STRICT, LYD_VALIDATE_PRESENT, tree, LY_EVALID,
+            "No module named \"x\" in the context.", "/a:any", 1);
+
+    data_invalid = "{\"a:any\":{\"a:fooA\":{\"element2\":\"/a:some/a:path\",\"list\":[{},{\"key\":\"a\"}]}}}";
+
+    PARSER_CHECK_ERROR(data_invalid, LYD_PARSE_ANYDATA_STRICT, LYD_VALIDATE_PRESENT, tree, LY_EVALID,
+            "Node \"fooA\" not found in the \"a\" module.", "/a:any", 1);
+
+    data_valid = "{\"a:any\":{\"foo\":\"default-val\"}}";
+    CHECK_PARSE_LYD(data_valid, 0, LYD_VALIDATE_PRESENT, tree);
+    assert_non_null(tree);
+    tree = tree->next;
+    CHECK_LYSC_NODE(tree->schema, NULL, 0, LYS_STATUS_CURR | LYS_CONFIG_R | LYS_SET_CONFIG, 1, "any",
+            1, LYS_ANYDATA, 0, 0, NULL, 0);
+    CHECK_LYD_STRING(tree, LYD_PRINT_SHRINK | LYD_PRINT_SIBLINGS, data_valid);
+    lyd_free_all(tree);
+}
+
 static void
 test_anyxml(void **state)
 {
@@ -1031,6 +1060,7 @@ main(void)
         UTEST(test_leaf, setup),
         UTEST(test_leaflist, setup),
         UTEST(test_anydata, setup),
+        UTEST(test_anydata_strict_validation, setup),
         UTEST(test_anyxml, setup),
         UTEST(test_list, setup),
         UTEST(test_container, setup),

tests/utests/data/test_parser_xml.c

@@ -159,6 +159,61 @@ test_anydata(void **state)
     lyd_free_all(tree);
 }
 
+static void
+test_anydata_strict_validation(void **state)
+{
+    const char *data_without_schema;
+    const char *data_invalid;
+    const char *data_valid;
+    char *str;
+    struct lyd_node *tree;
+
+    // no shcema defiend for "urn:tests:no:schema" in the parsing context
+    data_without_schema = "<any xmlns=\"urn:tests:a\">\n"
+            "  <x:element1 xmlns:x=\"urn:tests:no:schema\">\n"
+            "    <x:element2>default-val</x:element2>\n"
+            "  </x:element1>\n"
+            "</any>\n";
+
+    PARSER_CHECK_ERROR(data_without_schema, LYD_PARSE_ANYDATA_STRICT, LYD_VALIDATE_PRESENT, tree, LY_EVALID,
+            "No module with namespace \"urn:tests:no:schema\" in the context.", "/a:any", 3);
+
+    // anydata value are based on "module a" defined in the setup function and loaded into the parsing context.
+    // However, the value passed in the anydata subtree is not defined in "module a".
+    data_invalid = "<any xmlns=\"urn:tests:a\">\n"
+            "  <element1>default-val</element1>\n"
+            "</any>\n";
+
+    PARSER_CHECK_ERROR(data_invalid, LYD_PARSE_ANYDATA_STRICT, LYD_VALIDATE_PRESENT, tree, LY_EVALID,
+            "Node \"element1\" not found in the \"a\" module.", "/a:any", 2);
+
+    // anydata value are based on "module a" defined in the setup function and loaded into the parsing context
+    data_valid = "<any xmlns=\"urn:tests:a\">\n"
+            "  <foo>default-val</foo>\n"
+            "</any>\n";
+
+    CHECK_PARSE_LYD(data_valid, LYD_PARSE_ANYDATA_STRICT, LYD_VALIDATE_PRESENT, tree);
+    assert_non_null(tree);
+    tree = tree->next;
+    CHECK_LYSC_NODE(tree->schema, NULL, 0, LYS_CONFIG_R | LYS_STATUS_CURR | LYS_SET_CONFIG, 1, "any",
+            1, LYS_ANYDATA, 0, 0, NULL, 0);
+
+    const char *data_expected =
+            "<any xmlns=\"urn:tests:a\">\n"
+            "  <foo>default-val</foo>\n"
+            "</any>\n";
+
+    CHECK_LYD_STRING(tree, LYD_PRINT_SIBLINGS, data_expected);
+
+    assert_int_equal(LY_SUCCESS, lyd_any_value_str(tree, &str));
+    lyd_free_all(tree);
+
+    assert_int_equal(LY_SUCCESS, lyd_new_path2(NULL, UTEST_LYCTX, "/a:any", str, strlen(str), LYD_ANYDATA_XML, 0, &tree, NULL));
+    free(str);
+    CHECK_LYD_STRING(tree, LYD_PRINT_SIBLINGS, data_expected);
+    lyd_free_all(tree);
+}
+
 static void
 test_anyxml(void **state)
 {
@@ -1042,6 +1097,7 @@ main(void)
     const struct CMUnitTest tests[] = {
         UTEST(test_leaf, setup),
         UTEST(test_anydata, setup),
+        UTEST(test_anydata_strict_validation, setup),
         UTEST(test_anyxml, setup),
         UTEST(test_list, setup),
         UTEST(test_container, setup),

tools/lint/cmd_data.c

@@ -146,7 +146,9 @@ cmd_data_help(void)
             "                is supposed to contain the source 'nc-rpc' operation of the reply.\n"
             "  -k, --ext-inst <name>\n"
             "                Name of extension instance in format:\n"
-            "                <module-name>:<extension-name>:<argument>\n");
+            "                <module-name>:<extension-name>:<argument>\n"
+            "  -A, --anydata-strict\n"
+            "                Enable strict parsing of anydata content\n");
     cmd_data_help_format();
     cmd_data_help_in_format();
     printf("  -o OUTFILE, --output=OUTFILE\n"
@@ -161,19 +163,20 @@ cmd_data_opt(struct yl_opt *yo, const char *cmdline, char ***posv, int *posc)
     int rc = 0, argc = 0;
     int opt, opt_index;
     struct option options[] = {
-        {"defaults",    required_argument, NULL, 'd'},
-        {"present",     no_argument,       NULL, 'e'},
-        {"format",      required_argument, NULL, 'f'},
-        {"in-format",   required_argument, NULL, 'F'},
-        {"help",        no_argument,       NULL, 'h'},
-        {"merge",       no_argument,       NULL, 'm'},
-        {"output",      required_argument, NULL, 'o'},
-        {"operational", required_argument, NULL, 'O'},
-        {"reply-rpc",   required_argument, NULL, 'R'},
-        {"not-strict",  no_argument,       NULL, 'n'},
-        {"type",        required_argument, NULL, 't'},
-        {"xpath",       required_argument, NULL, 'x'},
-        {"ext-inst",    required_argument, NULL, 'k'},
+        {"defaults",        required_argument, NULL, 'd'},
+        {"present",         no_argument,       NULL, 'e'},
+        {"format",          required_argument, NULL, 'f'},
+        {"in-format",       required_argument, NULL, 'F'},
+        {"help",            no_argument,       NULL, 'h'},
+        {"merge",           no_argument,       NULL, 'm'},
+        {"output",          required_argument, NULL, 'o'},
+        {"operational",     required_argument, NULL, 'O'},
+        {"reply-rpc",       required_argument, NULL, 'R'},
+        {"not-strict",      no_argument,       NULL, 'n'},
+        {"anydata-strict",  no_argument      , NULL, 'A'},
+        {"type",            required_argument, NULL, 't'},
+        {"xpath",           required_argument, NULL, 'x'},
+        {"ext-inst",        required_argument, NULL, 'k'},
         {NULL, 0, NULL, 0}
     };
 
@@ -242,6 +245,10 @@ cmd_data_opt(struct yl_opt *yo, const char *cmdline, char ***posv, int *posc)
         case 'n': /* --not-strict */
             yo->data_parse_options &= ~LYD_PARSE_STRICT;
             break;
+        case 'A':   /* --anydata-strict */
+            yo->data_parse_options |= LYD_PARSE_ANYDATA_STRICT;
+            break;
+
         case 't': /* --type */
             if (data_type_set) {
                 YLMSG_E("The data type (-t) cannot be set multiple times.");

tools/lint/main_ni.c

@@ -128,6 +128,9 @@ help(int shortout)
             "                Do not require strict data parsing (silently skip unknown data),\n"
             "                has no effect for schemas.\n\n");
 
+    printf("  -A, --anydata-strict\n"
+            "                Enable strict parsing of anydata content.\n\n");
+
     printf("  -e, --present Validate only with the schema modules whose data actually\n"
             "                exist in the provided input data files. Takes effect only\n"
             "                with the 'data' or 'config' TYPEs. Used to avoid requiring\n"
@@ -486,6 +489,7 @@ process_args(int argc, char *argv[], struct yl_opt *yo, struct ly_ctx **ctx)
         {"submodule",         required_argument, NULL, 's'},
         {"ext-data",          required_argument, NULL, 'x'},
         {"not-strict",        no_argument,       NULL, 'n'},
+        {"anydata-strict",    no_argument,       NULL, 'A'},
         {"present",           no_argument,       NULL, 'e'},
         {"type",              required_argument, NULL, 't'},
         {"default",           required_argument, NULL, 'd'},
@@ -511,7 +515,7 @@ process_args(int argc, char *argv[], struct yl_opt *yo, struct ly_ctx **ctx)
     yo->line_length = 0;
 
     opterr = 0;
-    while ((opt = getopt_long(argc, argv, "hvVQf:I:p:DF:iP:qs:neE:t:d:lL:o:O:R:myY:XJx:G:k:", options, &opt_index)) != -1) {
+    while ((opt = getopt_long(argc, argv, "hvVQf:I:p:DF:iP:qs:neEA:t:d:lL:o:O:R:myY:XJx:G:k:", options, &opt_index)) != -1) {
         switch (opt) {
         case 'h': /* --help */
             help(0);
@@ -604,6 +608,10 @@ process_args(int argc, char *argv[], struct yl_opt *yo, struct ly_ctx **ctx)
             yo->data_parse_options &= ~LYD_PARSE_STRICT;
             break;
 
+        case 'A':   /* --anydata-strict */
+            yo->data_parse_options |= LYD_PARSE_ANYDATA_STRICT;
+            break;
+
         case 'e': /* --present */
             yo->data_validate_options |= LYD_VALIDATE_PRESENT;
             break;