| 1 | #include <assert.h> |
| 2 | #include <check.h> |
| 3 | #include <stdlib.h> |
| 4 | #include <stdio.h> |
| 5 | #include <string.h> |
| 6 | |
| 7 | #include "expat.h" |
| 8 | #include "chardata.h" |
| 9 | |
| 10 | |
| 11 | static XML_Parser parser; |
| 12 | |
| 13 | |
| 14 | static void |
| 15 | basic_setup(void) |
| 16 | { |
| 17 | parser = XML_ParserCreate(NULL); |
| 18 | if (parser == NULL) |
| 19 | fail("Parser not created."); |
| 20 | } |
| 21 | |
| 22 | static void |
| 23 | basic_teardown(void) |
| 24 | { |
| 25 | if (parser != NULL) |
| 26 | XML_ParserFree(parser); |
| 27 | } |
| 28 | |
| 29 | /* Generate a failure using the parser state to create an error message; |
| 30 | this should be used when the parser reports an error we weren't |
| 31 | expecting. |
| 32 | */ |
| 33 | static void |
| 34 | _xml_failure(XML_Parser parser, const char *file, int line) |
| 35 | { |
| 36 | char buffer[1024]; |
| 37 | sprintf(buffer, |
| 38 | "\n %s (line %d, offset %d)\n reported from %s, line %d", |
| 39 | XML_ErrorString(XML_GetErrorCode(parser)), |
| 40 | XML_GetCurrentLineNumber(parser), |
| 41 | XML_GetCurrentColumnNumber(parser), |
| 42 | file, line); |
| 43 | _fail_unless(0, file, line, buffer); |
| 44 | } |
| 45 | |
| 46 | #define xml_failure(parser) _xml_failure((parser), __FILE__, __LINE__) |
| 47 | |
| 48 | static void |
| 49 | _expect_failure(char *text, enum XML_Error errorCode, char *errorMessage, |
| 50 | char *file, int lineno) |
| 51 | { |
| 52 | if (XML_Parse(parser, text, strlen(text), XML_TRUE) == XML_STATUS_OK) |
| 53 | /* Hackish use of _fail_unless() macro, but let's us report |
| 54 | the right filename and line number. */ |
| 55 | _fail_unless(0, file, lineno, errorMessage); |
| 56 | if (XML_GetErrorCode(parser) != errorCode) |
| 57 | _xml_failure(parser, file, lineno); |
| 58 | } |
| 59 | |
| 60 | #define expect_failure(text, errorCode, errorMessage) \ |
| 61 | _expect_failure((text), (errorCode), (errorMessage), \ |
| 62 | __FILE__, __LINE__) |
| 63 | |
| 64 | /* Dummy handlers for when we need to set a handler to tickle a bug, |
| 65 | but it doesn't need to do anything. |
| 66 | */ |
| 67 | |
| 68 | static void |
| 69 | dummy_start_doctype_handler(void *userData, |
| 70 | const XML_Char *doctypeName, |
| 71 | const XML_Char *sysid, |
| 72 | const XML_Char *pubid, |
| 73 | int has_internal_subset) |
| 74 | {} |
| 75 | |
| 76 | static void |
| 77 | dummy_end_doctype_handler(void *userData) |
| 78 | {} |
| 79 | |
| 80 | static void |
| 81 | dummy_entity_decl_handler(void *userData, |
| 82 | const XML_Char *entityName, |
| 83 | int is_parameter_entity, |
| 84 | const XML_Char *value, |
| 85 | int value_length, |
| 86 | const XML_Char *base, |
| 87 | const XML_Char *systemId, |
| 88 | const XML_Char *publicId, |
| 89 | const XML_Char *notationName) |
| 90 | {} |
| 91 | |
| 92 | static void |
| 93 | dummy_notation_decl_handler(void *userData, |
| 94 | const XML_Char *notationName, |
| 95 | const XML_Char *base, |
| 96 | const XML_Char *systemId, |
| 97 | const XML_Char *publicId) |
| 98 | {} |
| 99 | |
| 100 | static void |
| 101 | dummy_element_decl_handler(void *userData, |
| 102 | const XML_Char *name, |
| 103 | XML_Content *model) |
| 104 | {} |
| 105 | |
| 106 | static void |
| 107 | dummy_attlist_decl_handler(void *userData, |
| 108 | const XML_Char *elname, |
| 109 | const XML_Char *attname, |
| 110 | const XML_Char *att_type, |
| 111 | const XML_Char *dflt, |
| 112 | int isrequired) |
| 113 | {} |
| 114 | |
| 115 | static void |
| 116 | dummy_comment_handler(void *userData, const XML_Char *data) |
| 117 | {} |
| 118 | |
| 119 | static void |
| 120 | dummy_pi_handler(void *userData, const XML_Char *target, const XML_Char *data) |
| 121 | {} |
| 122 | |
| 123 | static void |
| 124 | dummy_start_element(void *userData, |
| 125 | const XML_Char *name, const XML_Char **atts) |
| 126 | {} |
| 127 | |
| 128 | |
| 129 | /* |
| 130 | * Character & encoding tests. |
| 131 | */ |
| 132 | |
| 133 | START_TEST(test_nul_byte) |
| 134 | { |
| 135 | char text[] = "<doc>\0</doc>"; |
| 136 | |
| 137 | /* test that a NUL byte (in US-ASCII data) is an error */ |
| 138 | if (XML_Parse(parser, text, sizeof(text) - 1, XML_TRUE) == XML_STATUS_OK) |
| 139 | fail("Parser did not report error on NUL-byte."); |
| 140 | if (XML_GetErrorCode(parser) != XML_ERROR_INVALID_TOKEN) |
| 141 | xml_failure(parser); |
| 142 | } |
| 143 | END_TEST |
| 144 | |
| 145 | |
| 146 | START_TEST(test_u0000_char) |
| 147 | { |
| 148 | /* test that a NUL byte (in US-ASCII data) is an error */ |
| 149 | expect_failure("<doc>�</doc>", |
| 150 | XML_ERROR_BAD_CHAR_REF, |
| 151 | "Parser did not report error on NUL-byte."); |
| 152 | } |
| 153 | END_TEST |
| 154 | |
| 155 | START_TEST(test_bom_utf8) |
| 156 | { |
| 157 | /* This test is really just making sure we don't core on a UTF-8 BOM. */ |
| 158 | char *text = "\357\273\277<e/>"; |
| 159 | |
| 160 | if (XML_Parse(parser, text, strlen(text), XML_TRUE) == XML_STATUS_ERROR) |
| 161 | xml_failure(parser); |
| 162 | } |
| 163 | END_TEST |
| 164 | |
| 165 | START_TEST(test_bom_utf16_be) |
| 166 | { |
| 167 | char text[] = "\376\377\0<\0e\0/\0>"; |
| 168 | |
| 169 | if (XML_Parse(parser, text, sizeof(text)-1, XML_TRUE) == XML_STATUS_ERROR) |
| 170 | xml_failure(parser); |
| 171 | } |
| 172 | END_TEST |
| 173 | |
| 174 | START_TEST(test_bom_utf16_le) |
| 175 | { |
| 176 | char text[] = "\377\376<\0e\0/\0>\0"; |
| 177 | |
| 178 | if (XML_Parse(parser, text, sizeof(text)-1, XML_TRUE) == XML_STATUS_ERROR) |
| 179 | xml_failure(parser); |
| 180 | } |
| 181 | END_TEST |
| 182 | |
| 183 | static void |
| 184 | accumulate_characters(void *userData, const XML_Char *s, int len) |
| 185 | { |
| 186 | CharData_AppendXMLChars((CharData *)userData, s, len); |
| 187 | } |
| 188 | |
| 189 | static void |
| 190 | accumulate_attribute(void *userData, const XML_Char *name, |
| 191 | const XML_Char **atts) |
| 192 | { |
| 193 | CharData *storage = (CharData *)userData; |
| 194 | if (storage->count < 0 && atts != NULL && atts[0] != NULL) { |
| 195 | /* "accumulate" the value of the first attribute we see */ |
| 196 | CharData_AppendXMLChars(storage, atts[1], -1); |
| 197 | } |
| 198 | } |
| 199 | |
| 200 | |
| 201 | static void |
| 202 | _run_character_check(XML_Char *text, XML_Char *expected, |
| 203 | const char *file, int line) |
| 204 | { |
| 205 | CharData storage; |
| 206 | |
| 207 | CharData_Init(&storage); |
| 208 | XML_SetUserData(parser, &storage); |
| 209 | XML_SetCharacterDataHandler(parser, accumulate_characters); |
| 210 | if (XML_Parse(parser, text, strlen(text), XML_TRUE) == XML_STATUS_ERROR) |
| 211 | _xml_failure(parser, file, line); |
| 212 | CharData_CheckXMLChars(&storage, expected); |
| 213 | } |
| 214 | |
| 215 | #define run_character_check(text, expected) \ |
| 216 | _run_character_check(text, expected, __FILE__, __LINE__) |
| 217 | |
| 218 | static void |
| 219 | _run_attribute_check(XML_Char *text, XML_Char *expected, |
| 220 | const char *file, int line) |
| 221 | { |
| 222 | CharData storage; |
| 223 | |
| 224 | CharData_Init(&storage); |
| 225 | XML_SetUserData(parser, &storage); |
| 226 | XML_SetStartElementHandler(parser, accumulate_attribute); |
| 227 | if (XML_Parse(parser, text, strlen(text), XML_TRUE) == XML_STATUS_ERROR) |
| 228 | _xml_failure(parser, file, line); |
| 229 | CharData_CheckXMLChars(&storage, expected); |
| 230 | } |
| 231 | |
| 232 | #define run_attribute_check(text, expected) \ |
| 233 | _run_attribute_check(text, expected, __FILE__, __LINE__) |
| 234 | |
| 235 | /* Regression test for SF bug #491986. */ |
| 236 | START_TEST(test_danish_latin1) |
| 237 | { |
| 238 | char *text = |
| 239 | "<?xml version='1.0' encoding='iso-8859-1'?>\n" |
| 240 |