[9070] | 1 |
|
---|
[9406] | 2 | #include <stdio.h>
|
---|
[9070] | 3 | #include <string.h>
|
---|
| 4 | #include <expat.h>
|
---|
| 5 |
|
---|
| 6 | #if defined(GSDL_USE_OBJECTSPACE)
|
---|
| 7 | # include <ospace\std\iostream>
|
---|
| 8 | #elif defined(GSDL_USE_IOS_H)
|
---|
| 9 | # include <iostream.h>
|
---|
| 10 | #else
|
---|
| 11 | # include <iostream>
|
---|
| 12 | using namespace std;
|
---|
| 13 | #endif
|
---|
| 14 |
|
---|
| 15 | #include "text_t.h"
|
---|
| 16 | #include "queryinfo.h"
|
---|
[12262] | 17 | #include "gsdlunicode.h"
|
---|
[9070] | 18 |
|
---|
| 19 | struct queryresultpack {
|
---|
| 20 | queryresultsclass* queryresult_ptr;
|
---|
| 21 | int match_count;
|
---|
| 22 | };
|
---|
| 23 |
|
---|
| 24 | char* get_attribute(const char** attr, char* att_name)
|
---|
| 25 | {
|
---|
| 26 | char* att_val = NULL;
|
---|
| 27 |
|
---|
| 28 | for (int i = 0; attr[i]; i += 2) {
|
---|
| 29 | if (strcmp(attr[i],att_name)==0) {
|
---|
| 30 | att_val = strdup(attr[i+1]);
|
---|
| 31 | }
|
---|
| 32 | }
|
---|
| 33 |
|
---|
| 34 | return att_val;
|
---|
| 35 | }
|
---|
| 36 |
|
---|
| 37 | static void XMLCALL
|
---|
| 38 | startElement(void *userData, const char *name, const char **attributes)
|
---|
| 39 | {
|
---|
| 40 | queryresultpack* qrpack_ptr = (queryresultpack*)userData;
|
---|
| 41 | queryresultsclass* queryresult_ptr = qrpack_ptr->queryresult_ptr;
|
---|
| 42 |
|
---|
| 43 | text_t element_name = (char*)name;
|
---|
| 44 |
|
---|
[13000] | 45 | if (element_name == "ResultSet") {
|
---|
| 46 | char* cached_attribute_str = get_attribute(attributes, "cached");
|
---|
| 47 | if (cached_attribute_str != NULL) {
|
---|
| 48 | cerr << "Cached: " << cached_attribute_str << endl;
|
---|
| 49 | }
|
---|
| 50 | }
|
---|
| 51 |
|
---|
[12420] | 52 | if (element_name == "Error") {
|
---|
| 53 | char* error_type_str = get_attribute(attributes, "type");
|
---|
[12421] | 54 | queryresult_ptr->error_message = error_type_str;
|
---|
[12420] | 55 | if ((text_t) error_type_str == "PARSE_EXCEPTION") {
|
---|
| 56 | queryresult_ptr->syntax_error = true;
|
---|
| 57 | }
|
---|
| 58 | }
|
---|
| 59 |
|
---|
[12262] | 60 | if (element_name == "Term") {
|
---|
| 61 | char* term_value_str = get_attribute(attributes, "value");
|
---|
| 62 | char* term_frequency_str = get_attribute(attributes, "freq");
|
---|
| 63 |
|
---|
| 64 | termfreqclass termfreqobj;
|
---|
| 65 | termfreqobj.termstr = to_uni(term_value_str);
|
---|
[16311] | 66 |
|
---|
[12262] | 67 | termfreqobj.termfreq = atoi(term_frequency_str);
|
---|
| 68 | queryresult_ptr->orgterms.push_back(termfreqobj);
|
---|
| 69 | }
|
---|
| 70 |
|
---|
[12380] | 71 | if (element_name == "StopWord") {
|
---|
| 72 | char* stop_word_value_str = get_attribute(attributes, "value");
|
---|
| 73 | queryresult_ptr->stopwords.insert((text_t) stop_word_value_str);
|
---|
| 74 | }
|
---|
| 75 |
|
---|
[9070] | 76 | if (element_name=="MatchingDocsInfo") {
|
---|
| 77 | char* num_match_docs_str = get_attribute(attributes,"num");
|
---|
| 78 | if (num_match_docs_str != NULL) {
|
---|
| 79 | int num_match_docs = atoi(num_match_docs_str);
|
---|
| 80 |
|
---|
| 81 | queryresult_ptr->docs_matched = num_match_docs;
|
---|
| 82 | free(num_match_docs_str);
|
---|
| 83 | }
|
---|
| 84 | else {
|
---|
[12262] | 85 | queryresult_ptr->docs_matched = 0;
|
---|
[9070] | 86 | }
|
---|
| 87 | }
|
---|
| 88 |
|
---|
| 89 |
|
---|
| 90 | if (element_name=="Match") {
|
---|
| 91 |
|
---|
| 92 | char* id = get_attribute(attributes,"id");
|
---|
[16947] | 93 | if (id != NULL)
|
---|
| 94 | {
|
---|
[9070] | 95 | docresultclass doc;
|
---|
| 96 | doc.clear();
|
---|
[16947] | 97 | doc.docid = id;
|
---|
[9070] | 98 | doc.docweight = qrpack_ptr->match_count;
|
---|
[16915] | 99 |
|
---|
| 100 | char* termfreq = get_attribute(attributes, "termfreq");
|
---|
| 101 | if (termfreq != NULL)
|
---|
| 102 | {
|
---|
| 103 | doc.num_query_terms_matched = atoi(termfreq);
|
---|
| 104 | }
|
---|
| 105 |
|
---|
[16947] | 106 | queryresult_ptr->docs.docset[doc.docid] = doc;
|
---|
| 107 | queryresult_ptr->docs.docorder.push_back(doc.docid);
|
---|
[9620] | 108 | ++qrpack_ptr->match_count;
|
---|
[9070] | 109 |
|
---|
| 110 | free(id);
|
---|
| 111 | }
|
---|
| 112 |
|
---|
| 113 | }
|
---|
| 114 | }
|
---|
| 115 |
|
---|
| 116 | static void XMLCALL
|
---|
| 117 | endElement(void *userData, const char *name)
|
---|
| 118 | {
|
---|
| 119 | // no need to do anything
|
---|
| 120 | }
|
---|
| 121 |
|
---|
| 122 |
|
---|
| 123 |
|
---|
| 124 |
|
---|
| 125 | int expat_resultset(text_t& xml_text, queryresultsclass& queryresult)
|
---|
| 126 | {
|
---|
| 127 | queryresult.clear();
|
---|
| 128 | queryresultpack qrpack = { &queryresult, 0 };
|
---|
| 129 |
|
---|
| 130 | char * xml_text_cstr = xml_text.getcstr();
|
---|
| 131 |
|
---|
| 132 | XML_Parser parser = XML_ParserCreate(NULL);
|
---|
| 133 |
|
---|
| 134 | XML_SetUserData(parser, &qrpack);
|
---|
| 135 | XML_SetElementHandler(parser, startElement, endElement);
|
---|
| 136 |
|
---|
| 137 | int return_status = 0;
|
---|
| 138 | const int parse_status
|
---|
| 139 | = XML_Parse(parser, xml_text_cstr, strlen(xml_text_cstr), XML_TRUE);
|
---|
| 140 |
|
---|
| 141 | if (parse_status == XML_STATUS_ERROR) {
|
---|
| 142 | fprintf(stderr,
|
---|
| 143 | "%s at line %d\n",
|
---|
| 144 | XML_ErrorString(XML_GetErrorCode(parser)),
|
---|
| 145 | XML_GetCurrentLineNumber(parser));
|
---|
| 146 | return_status = 1;
|
---|
| 147 | }
|
---|
| 148 |
|
---|
| 149 | XML_ParserFree(parser);
|
---|
[20733] | 150 | delete []xml_text_cstr;
|
---|
[9070] | 151 |
|
---|
| 152 | return return_status;
|
---|
| 153 | }
|
---|