source: gsdl/trunk/runtime-src/src/oaiservr/metaformat.cpp@ 18896

Last change on this file since 18896 was 18896, checked in by kjdon, 13 years ago

convert CoverageSpatial to Spatial. Just deletes anything before the . This may not be enough in future if sets validly have

  • Property svn:keywords set to Author Date Id Revision
File size: 5.7 KB
Line 
1#include <fstream>
2#include "metaformat.h"
3#include "gsdltools.h"
4#include "gsdlunicode.h"
5#include "recptprototools.h"
6
7
8metaformat::metaformat()
9{
10}
11
12text_t metaformat::get_mapping(const text_t &collection, const text_t &collectionField)
13{
14 if (this->oaiConfigure == NULL) {
15 return "";
16 }
17
18 return this->oaiConfigure->getMapping(collection, collectionField, this->formatName());
19}
20
21void metaformat::output_item(ostream &output, outconvertclass &outconvert,
22 bool &headerDone, const text_t &label,
23 const text_tarray &values)
24{
25
26 if (!headerDone && (values.size() > 0)) {
27 this->output_metadata_header(output);
28 headerDone = true;
29 }
30
31 for (int item = 0; item < values.size(); ++item) {
32 if (this->oaiConfigure->getOAIVersion() >= 200) { // TODO: GRB: This code may need to be subclassed by dc for 200 and later...
33 output << outconvert << " <" << this->formatPrefix() << ":" << label << ">" << xml_safe(values[item]) << "</" << this->formatPrefix() << ":" << label << ">\n";
34 }
35 else {
36 output << outconvert << " <" << label << ">" << xml_safe(values[item]) << "</" << label << ">\n";
37 }
38 }
39}
40
41bool metaformat::scan_metadata(ostream &output, const text_t &collection, ResultDocInfo_t &docInfo,
42 bool doOutput)
43{
44 bool headerDone = false;
45 MetadataInfo_tmap::iterator here = docInfo.metadata.begin();
46 MetadataInfo_tmap::iterator end = docInfo.metadata.end();
47
48 utf8outconvertclass utf8convert; // we want to output metadata in utf8
49
50 // metaItem is used initially to identify the rfc1807 (etc) metadata items. It is
51 // then used to hold the name of the metadata item, such as "title" or "subject".
52 text_t metaItem;
53 text_t::const_iterator start, last; // Use two iterators to go through metaItem
54
55 while (here != end) {
56 start = last = here->first.begin();
57
58 if (here->first.size() < this->formatPrefix().size() ||
59 here->first[this->formatPrefix().size()] != '.') {
60 metaItem == "";
61 }
62 else {
63 last += this->formatPrefix().size(); // Move last so that it is at the
64 // '.'
65 metaItem = substr(start, last); // Gets the substring starting at start and going up to (but
66 // not including) last. This should be "dc" (for example)
67 }
68
69 if (metaItem == this->formatPrefix()) {
70 metaItem = substr(last+1, here->first.end()); // Get the rest of the metadata tag (it's name) but without the '.'
71 // remove xxx^ eg Coverage^Spatial becomes spatial
72 // this is for qualified dublin core. May affect other sets later if they
73 // validly have ^ in them.
74 text_t::iterator hat = findchar(metaItem.begin(), metaItem.end(), '^');
75 if (hat != metaItem.end()) {
76 metaItem = substr(hat+1, metaItem.end());
77 }
78 lc(metaItem.begin(),metaItem.begin()+1); // We want lowercase, but some of the fields in qualified dublin core have internal upper case, eg instructionalMethod. So we assume that lowercasing the first letter is enough
79 if (doOutput) {
80 if (this->is_valid_element(metaItem)) {
81
82 this->output_item(output, utf8convert, headerDone, metaItem, here->second.values);
83 }
84 }
85 else {
86 if (here->second.values.size() > 0) {
87 return true;
88 }
89 }
90 }
91 else {
92 text_t mapTo = this->get_mapping(collection, here->first);
93 if (mapTo != "") {
94 // Do we actually want to do anything here? Doesn't getting here imply that this
95 // particular metadata is stuff we don't want?
96 if (doOutput) {
97 if (this->is_valid_element(mapTo)) {
98 this->output_item(output, utf8convert, headerDone, mapTo, here->second.values);
99 }
100 }
101 else {
102 if (here->second.values.size() > 0) {
103 return true;
104 }
105 }
106 }
107 }
108
109 ++here;
110 }
111
112 if (!doOutput) {
113 return false;
114 }
115
116 if (headerDone) {
117 this->output_metadata_footer(output);
118 }
119
120 return headerDone;
121}
122
123
124bool metaformat::is_available(const text_t &collection, ResultDocInfo_t &docInfo)
125{
126 ofstream o("dummy", ios::out);
127 return this->scan_metadata(o, collection, docInfo, false);
128}
129
130bool metaformat::is_valid_element(text_t &meta_name)
131{
132 if (elementSet.count(meta_name)==1) return true;
133 return false;
134
135}
136
137bool metaformat::output_metadata(ostream &output, const text_t &collection, ResultDocInfo_t &docInfo)
138{
139 return this->scan_metadata(output, collection, docInfo, true);
140}
141
142bool metaformat::output_record(ostream &output, recptproto *protocol, const text_t &collection,
143 const text_t &OID)
144{
145 FilterResponse_t response;
146 text_tset metadata;
147 ofstream logout("oai.log", ios::app);
148
149 // get the document information
150 if (!get_info(OID, collection, "", metadata, false, protocol, response, logout)) {
151 // TODO: error, bad request
152 // cerr << "Bad identifier or protocol " << OID << endl;
153 return false;
154 }
155
156 // check to see if it's a classifier
157 text_t childHead;
158 // int oaiVersion = this->oaiConfigure->getOAIVersion();
159 text_t::const_iterator start = OID.begin();
160 text_t::const_iterator here = OID.begin();
161 here += 2;
162 childHead = substr(start, here);
163
164 // if it isn't a document, kill it now
165 if (childHead == "CL") {
166 // cerr << "Not a document" << endl;
167 return false;
168 }
169
170 // output record header
171 output << "<record>\n";
172
173 // output header part of oai response
174 output << "<header>" << endl;
175 output << " <identifier>" << OID << "</identifier>" << endl;
176 // TODO: add modified date
177
178 output << "</header>" << endl;
179
180 // output metadata part of oai response
181 this->output_metadata(output, collection, response.docInfo[0]);
182
183 // output the description of the document
184 // output << "<about>\n";
185 // output << "</about>\n";
186
187 // close record
188 output << "</record>\n";
189
190 return true;
191}
Note: See TracBrowser for help on using the repository browser.