Accept # in tag URIs

src/scanner.c

@@ -2622,7 +2622,7 @@ yaml_parser_scan_tag_uri(yaml_parser_t *parser, int uri_char, int directive,
      * The set of characters that may appear in URI is as follows:
      *
      *      '0'-'9', 'A'-'Z', 'a'-'z', '_', '-', ';', '/', '?', ':', '@', '&',
-     *      '=', '+', '$', '.', '!', '~', '*', '\'', '(', ')', '%'.
+     *      '=', '+', '$', '.', '!', '~', '*', '\'', '(', ')', '%', '#'.
      *
      * If we are inside a verbatim tag <...> (parameter uri_char is true)
      * then also the following flow indicators are allowed:
@@ -2638,6 +2638,7 @@ yaml_parser_scan_tag_uri(yaml_parser_t *parser, int uri_char, int directive,
             || CHECK(parser->buffer, '!') || CHECK(parser->buffer, '~')
             || CHECK(parser->buffer, '*') || CHECK(parser->buffer, '\'')
             || CHECK(parser->buffer, '(') || CHECK(parser->buffer, ')')
+            || CHECK(parser->buffer, '#')
             || (uri_char && (
                 CHECK(parser->buffer, ',')
                 || CHECK(parser->buffer, '[') || CHECK(parser->buffer, ']')

tests/CMakeLists.txt

@@ -18,6 +18,7 @@ foreach(name IN ITEMS
   run-scanner
   test-nesting
   test-reader
+  test-tag-uri
   test-version
   )
   add_yaml_executable(${name})
@@ -26,4 +27,5 @@ endforeach()
 add_test(NAME version COMMAND test-version)
 add_test(NAME reader COMMAND test-reader)
 add_test(NAME nesting COMMAND test-nesting)
+add_test(NAME tag-uri COMMAND test-tag-uri)
 

tests/Makefile.am

@@ -1,8 +1,8 @@
 AM_CPPFLAGS = -I$(top_srcdir)/include -Wall
 #AM_CFLAGS = -Wno-pointer-sign
 LDADD = $(top_builddir)/src/libyaml.la
-TESTS = test-version test-reader test-nesting
-check_PROGRAMS = test-version test-reader test-nesting
+TESTS = test-version test-reader test-nesting test-tag-uri
+check_PROGRAMS = test-version test-reader test-nesting test-tag-uri
 noinst_PROGRAMS = run-scanner run-parser run-loader run-emitter run-dumper	\
 				  example-reformatter example-reformatter-alt	\
 				  example-deconstructor example-deconstructor-alt \

tests/test-tag-uri.c

@@ -0,0 +1,74 @@
+#include <yaml.h>
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+#ifdef NDEBUG
+#undef NDEBUG
+#endif
+#include <assert.h>
+
+static const struct {
+    const char *title;
+    const char *input;
+    const char *tag;
+} cases[] = {
+    {"fragment in tag directive prefix",
+        "%TAG ! http://www.w3.org/2001/XMLSchema#\n---\n!foo bar\n",
+        "http://www.w3.org/2001/XMLSchema#foo"},
+    {"fragment in verbatim tag",
+        "!<http://example.com/#foo> bar\n",
+        "http://example.com/#foo"},
+    {"fragment in local tag",
+        "!foo#fragment bar\n",
+        "!foo#fragment"},
+    {"fragment in named handle suffix",
+        "%TAG !e! http://example.com/\n---\n!e!foo#fragment bar\n",
+        "http://example.com/foo#fragment"},
+    {"escaped fragment in verbatim tag",
+        "!<http://example.com/%23foo> bar\n",
+        "http://example.com/#foo"},
+    {"comments after directive and scalar",
+        "%TAG !e! http://example.com/ # comment\n"
+        "---\n!e!foo bar # comment\n",
+        "http://example.com/foo"},
+    {"comment after local tag",
+        "!foo # comment\nbar\n",
+        "!foo"}
+};
+
+int
+main(void)
+{
+    size_t i;
+    int failures = 0;
+
+    for (i = 0; i < sizeof(cases)/sizeof(cases[0]); i++) {
+        yaml_parser_t parser;
+        yaml_document_t document;
+        yaml_node_t *root;
+
+        assert(yaml_parser_initialize(&parser));
+        yaml_parser_set_input_string(&parser,
+                (const unsigned char *)cases[i].input, strlen(cases[i].input));
+
+        if (!yaml_parser_load(&parser, &document)) {
+            fprintf(stderr, "%s: %s\n", cases[i].title,
+                    parser.problem ? parser.problem : "no error detail");
+            failures++;
+        }
+        else {
+            root = yaml_document_get_root_node(&document);
+            assert(root && root->type == YAML_SCALAR_NODE);
+            assert(strcmp((const char *)root->tag, cases[i].tag) == 0);
+            assert(root->data.scalar.length == 3);
+            assert(memcmp(root->data.scalar.value, "bar", 3) == 0);
+            yaml_document_delete(&document);
+        }
+
+        yaml_parser_delete(&parser);
+    }
+
+    return failures ? EXIT_FAILURE : EXIT_SUCCESS;
+}