| 1 | /*
|
|---|
| 2 | * testparser.c: Additional parser tests
|
|---|
| 3 | *
|
|---|
| 4 | * See Copyright for the status of this software.
|
|---|
| 5 | */
|
|---|
| 6 |
|
|---|
| 7 | #include <libxml/parser.h>
|
|---|
| 8 | #include <libxml/HTMLparser.h>
|
|---|
| 9 |
|
|---|
| 10 | static int
|
|---|
| 11 | testStandaloneWithEncoding(void) {
|
|---|
| 12 | xmlDocPtr doc;
|
|---|
| 13 | const char *str =
|
|---|
| 14 | "<?xml version=\"1.0\" standalone=\"yes\"?>\n"
|
|---|
| 15 | "<doc></doc>\n";
|
|---|
| 16 | int err = 0;
|
|---|
| 17 |
|
|---|
| 18 | xmlResetLastError();
|
|---|
| 19 |
|
|---|
| 20 | doc = xmlReadDoc(BAD_CAST str, NULL, "UTF-8", 0);
|
|---|
| 21 | if (doc == NULL) {
|
|---|
| 22 | fprintf(stderr, "xmlReadDoc failed\n");
|
|---|
| 23 | err = 1;
|
|---|
| 24 | }
|
|---|
| 25 | xmlFreeDoc(doc);
|
|---|
| 26 |
|
|---|
| 27 | return err;
|
|---|
| 28 | }
|
|---|
| 29 |
|
|---|
| 30 | #ifdef LIBXML_PUSH_ENABLED
|
|---|
| 31 | static int
|
|---|
| 32 | testHugePush(void) {
|
|---|
| 33 | xmlParserCtxtPtr ctxt;
|
|---|
| 34 | int err, i;
|
|---|
| 35 |
|
|---|
| 36 | ctxt = xmlCreatePushParserCtxt(NULL, NULL, NULL, 0, NULL);
|
|---|
| 37 |
|
|---|
| 38 | /*
|
|---|
| 39 | * Push parse a document larger than XML_MAX_LOOKUP_LIMIT
|
|---|
| 40 | * (10,000,000 bytes). This mainly tests whether shrinking the
|
|---|
| 41 | * buffer works when push parsing.
|
|---|
| 42 | */
|
|---|
| 43 | xmlParseChunk(ctxt, "<doc>", 5, 0);
|
|---|
| 44 | for (i = 0; i < 1000000; i++)
|
|---|
| 45 | xmlParseChunk(ctxt, "<elem>text</elem>", 17, 0);
|
|---|
| 46 | xmlParseChunk(ctxt, "</doc>", 6, 1);
|
|---|
| 47 |
|
|---|
| 48 | err = ctxt->wellFormed ? 0 : 1;
|
|---|
| 49 | xmlFreeDoc(ctxt->myDoc);
|
|---|
| 50 | xmlFreeParserCtxt(ctxt);
|
|---|
| 51 |
|
|---|
| 52 | return err;
|
|---|
| 53 | }
|
|---|
| 54 |
|
|---|
| 55 | static int
|
|---|
| 56 | testHugeEncodedChunk(void) {
|
|---|
| 57 | xmlBufferPtr buf;
|
|---|
| 58 | xmlChar *chunk;
|
|---|
| 59 | xmlParserCtxtPtr ctxt;
|
|---|
| 60 | int err, i;
|
|---|
| 61 |
|
|---|
| 62 | /*
|
|---|
| 63 | * Test the push parser with a built-in encoding handler like ISO-8859-1
|
|---|
| 64 | * and a chunk larger than the initial decoded buffer (currently 4 KB).
|
|---|
| 65 | */
|
|---|
| 66 | buf = xmlBufferCreate();
|
|---|
| 67 | xmlBufferCat(buf,
|
|---|
| 68 | BAD_CAST "<?xml version='1.0' encoding='ISO-8859-1'?>\n");
|
|---|
| 69 | xmlBufferCat(buf, BAD_CAST "<doc><!-- ");
|
|---|
| 70 | for (i = 0; i < 2000; i++)
|
|---|
| 71 | xmlBufferCat(buf, BAD_CAST "0123456789");
|
|---|
| 72 | xmlBufferCat(buf, BAD_CAST " --></doc>");
|
|---|
| 73 | chunk = xmlBufferDetach(buf);
|
|---|
| 74 | xmlBufferFree(buf);
|
|---|
| 75 |
|
|---|
| 76 | ctxt = xmlCreatePushParserCtxt(NULL, NULL, NULL, 0, NULL);
|
|---|
| 77 |
|
|---|
| 78 | xmlParseChunk(ctxt, (char *) chunk, xmlStrlen(chunk), 0);
|
|---|
| 79 | xmlParseChunk(ctxt, NULL, 0, 1);
|
|---|
| 80 |
|
|---|
| 81 | err = ctxt->wellFormed ? 0 : 1;
|
|---|
| 82 | xmlFreeDoc(ctxt->myDoc);
|
|---|
| 83 | xmlFreeParserCtxt(ctxt);
|
|---|
| 84 | xmlFree(chunk);
|
|---|
| 85 |
|
|---|
| 86 | return err;
|
|---|
| 87 | }
|
|---|
| 88 | #endif
|
|---|
| 89 |
|
|---|
| 90 | #if defined(LIBXML_HTML_ENABLED) && defined(LIBXML_PUSH_ENABLED)
|
|---|
| 91 | static int
|
|---|
| 92 | testHtmlPushWithEncoding(void) {
|
|---|
| 93 | htmlParserCtxtPtr ctxt;
|
|---|
| 94 | htmlDocPtr doc;
|
|---|
| 95 | htmlNodePtr node;
|
|---|
| 96 | int err = 0;
|
|---|
| 97 |
|
|---|
| 98 | ctxt = htmlCreatePushParserCtxt(NULL, NULL, NULL, 0, NULL,
|
|---|
| 99 | XML_CHAR_ENCODING_UTF8);
|
|---|
| 100 | htmlParseChunk(ctxt, "-\xC3\xA4-", 4, 1);
|
|---|
| 101 |
|
|---|
| 102 | doc = ctxt->myDoc;
|
|---|
| 103 | if (!xmlStrEqual(doc->encoding, BAD_CAST "UTF-8")) {
|
|---|
| 104 | fprintf(stderr, "testHtmlPushWithEncoding failed\n");
|
|---|
| 105 | err = 1;
|
|---|
| 106 | }
|
|---|
| 107 |
|
|---|
| 108 | node = xmlDocGetRootElement(doc)->children->children->children;
|
|---|
| 109 | if (!xmlStrEqual(node->content, BAD_CAST "-\xC3\xA4-")) {
|
|---|
| 110 | fprintf(stderr, "testHtmlPushWithEncoding failed\n");
|
|---|
| 111 | err = 1;
|
|---|
| 112 | }
|
|---|
| 113 |
|
|---|
| 114 | xmlFreeDoc(doc);
|
|---|
| 115 | htmlFreeParserCtxt(ctxt);
|
|---|
| 116 | return err;
|
|---|
| 117 | }
|
|---|
| 118 | #endif
|
|---|
| 119 |
|
|---|
| 120 | int
|
|---|
| 121 | main(void) {
|
|---|
| 122 | int err = 0;
|
|---|
| 123 |
|
|---|
| 124 | err |= testStandaloneWithEncoding();
|
|---|
| 125 | #ifdef LIBXML_PUSH_ENABLED
|
|---|
| 126 | err |= testHugePush();
|
|---|
| 127 | err |= testHugeEncodedChunk();
|
|---|
| 128 | #endif
|
|---|
| 129 | #if defined(LIBXML_HTML_ENABLED) && defined(LIBXML_PUSH_ENABLED)
|
|---|
| 130 | err |= testHtmlPushWithEncoding();
|
|---|
| 131 | #endif
|
|---|
| 132 |
|
|---|
| 133 | return err;
|
|---|
| 134 | }
|
|---|
| 135 |
|
|---|
