1 | #include "oaiconfig.h"
|
---|
2 |
|
---|
3 | #include <algorithm>
|
---|
4 | #include <iostream>
|
---|
5 |
|
---|
6 | #include "fileutil.h"
|
---|
7 |
|
---|
8 | /**
|
---|
9 | * The mapping works as follows in the collect.cfg file.
|
---|
10 | *
|
---|
11 | * A line is in the format oaimapping <collection field> <oai field>
|
---|
12 | *
|
---|
13 | * The map here is used to look up the "Greenstone" name which is mapped from
|
---|
14 | * a given OAI field name, the reverse direction to that given in the
|
---|
15 | * Greenstone collect.cfg file. The oairecordaction class instance which
|
---|
16 | * produces output for an OAI record information request thus uses the map
|
---|
17 | * to work from the field in the collection it has on hand which OAI
|
---|
18 | * record name it should use instead.
|
---|
19 | *
|
---|
20 | * An extension is to be used for this in which the OAI field name in the
|
---|
21 | * collect.cfg file can be made specific for a particular record format.
|
---|
22 | * This is done using the OAI field name in the format of
|
---|
23 | * <OAI format>:<OAI field name>
|
---|
24 | * Thus, an rfc1807 Title field would be referred to as rfc1807:Title
|
---|
25 | *
|
---|
26 | * A collection-level mapping is not needed - to configure the behaviour
|
---|
27 | * of a collection, ensure that you place the appropriate configuration in
|
---|
28 | * its etc/collect.cfg file - the oaimapping stuff IS NOT picked up from
|
---|
29 | * the central main.cfg file at the moment.
|
---|
30 | *
|
---|
31 | * In the absence of a particular format name, the mapping is taken to be
|
---|
32 | * universal.
|
---|
33 | */
|
---|
34 |
|
---|
35 | oaiconfig::oaiconfig() : configurable () {
|
---|
36 | this->resumptionSize = -1; // Default = do not use resumption tokens
|
---|
37 | }
|
---|
38 |
|
---|
39 | oaiconfig::oaiconfig(text_t &gsdlhome, text_t &gsdlcollect)
|
---|
40 | {
|
---|
41 | // read main configuration file to get oai collections
|
---|
42 | text_t mainconfig = filename_cat(gsdlhome, "etc", "oai.cfg");
|
---|
43 | this->collection = "";
|
---|
44 | this->resumptionSize = -1;
|
---|
45 | this->read_configfile(mainconfig);
|
---|
46 |
|
---|
47 | // then if we've not got a specified collection in the gsdlcollect
|
---|
48 | // parameter, read in all the collection's individual configurations
|
---|
49 | if (gsdlcollect == "") {
|
---|
50 | text_tarray::iterator here = this->collectList.begin();
|
---|
51 | text_tarray::iterator end = this->collectList.end();
|
---|
52 | while (here != end) {
|
---|
53 | this->configureCollection(gsdlhome, *here);
|
---|
54 | ++here;
|
---|
55 | }
|
---|
56 | }
|
---|
57 | else {
|
---|
58 | this->configureCollection(gsdlhome, gsdlcollect);
|
---|
59 | }
|
---|
60 | }
|
---|
61 |
|
---|
62 | oaiconfig::~oaiconfig()
|
---|
63 | {
|
---|
64 | oaicollectmap::iterator here = this->collectMap.begin();
|
---|
65 | oaicollectmap::iterator end = this->collectMap.end();
|
---|
66 | while (here != end) {
|
---|
67 | delete here->second;
|
---|
68 | ++here;
|
---|
69 | }
|
---|
70 | }
|
---|
71 |
|
---|
72 | int oaiconfig::resumeAfter()
|
---|
73 | { return this->resumptionSize;
|
---|
74 | }
|
---|
75 |
|
---|
76 | int oaiconfig::getOAIVersion()
|
---|
77 | {
|
---|
78 | if (this->oaiVersion == "1.1") {
|
---|
79 | return 110;
|
---|
80 | }
|
---|
81 | return 200;
|
---|
82 | }
|
---|
83 |
|
---|
84 | void oaiconfig::configureCollection(const text_t &gsdlhome, const text_t &gsdlcollect)
|
---|
85 | {
|
---|
86 | text_t cnfgfile = filename_cat(gsdlhome, "collect", gsdlcollect, "etc", "collect.cfg");
|
---|
87 | this->collection = gsdlcollect;
|
---|
88 | this->read_configfile(cnfgfile);
|
---|
89 | }
|
---|
90 |
|
---|
91 | void oaiconfig::configure (const text_t &key, const text_tarray &cfgline)
|
---|
92 | {
|
---|
93 | // we've got an oai mapping item, and at least two fields
|
---|
94 | if (key == "oaimapping" && cfgline.size() > 1) {
|
---|
95 | text_t::const_iterator colonAt;
|
---|
96 | text_t index, name, configCollection;
|
---|
97 |
|
---|
98 | // Take a default collection as being whatever the collection being configured is...
|
---|
99 | configCollection = this->collection;
|
---|
100 |
|
---|
101 | // get the name of the (collection) field to map; this may actually
|
---|
102 | // be in a colon separated format of the type
|
---|
103 | // <collection name>:<field name>
|
---|
104 | index = cfgline[0];
|
---|
105 | if ((colonAt = find(index.begin(), index.end(), ':')) != index.end()) {
|
---|
106 | configCollection = substr(index.begin(), colonAt);
|
---|
107 |
|
---|
108 | if (this->collection != "" && configCollection != this->collection) {
|
---|
109 | cerr << "Attempt to configure OAI mappings for " << configCollection << " in " << this->collection << endl;
|
---|
110 | }
|
---|
111 |
|
---|
112 | colonAt += 1;
|
---|
113 | index = substr(colonAt, index.end());
|
---|
114 | }
|
---|
115 |
|
---|
116 | // the second parameter is the metadata field to map the collection
|
---|
117 | // field onto. It may be provided with a metadata protocol (which
|
---|
118 | // will be given first and separated by a period or full stop). In
|
---|
119 | // the case of format.field name, the splitting is done here.
|
---|
120 | if ((colonAt = find(cfgline[1].begin(), cfgline[1].end(), '.')) != cfgline[1].end()) {
|
---|
121 | text_t stub = substr(cfgline[1].begin(), colonAt);
|
---|
122 | colonAt += 1;
|
---|
123 | name = substr(colonAt, cfgline[1].end());
|
---|
124 | index.append(":");
|
---|
125 | index.append(stub);
|
---|
126 | }
|
---|
127 | else {
|
---|
128 | name = cfgline[1];
|
---|
129 | }
|
---|
130 |
|
---|
131 | // now 'index' is in the form <collectionfield>:(formatname)
|
---|
132 | // 'name' is simply the fieldname within the format
|
---|
133 | // 'configCollection' is the collection to be configured
|
---|
134 |
|
---|
135 | // now simply map the field name (index) onto the collection name (name)
|
---|
136 | if (this->collectMap[configCollection] == NULL) {
|
---|
137 | this->collectMap[configCollection] = new oaicollectconfig(configCollection);
|
---|
138 | }
|
---|
139 | this->collectMap[configCollection]->fieldMap[index] = name;
|
---|
140 |
|
---|
141 | // cerr << "Mapping " << index << " to " << name << " in " << configCollection << endl;
|
---|
142 |
|
---|
143 | // TODO: check that the mapped field is actually in use
|
---|
144 | }
|
---|
145 | else if (key == "oaicollection" && cfgline.size() >= 1) {
|
---|
146 | // Configure a collection to be used as part of the OAI archive.
|
---|
147 | // This line should read:
|
---|
148 | //
|
---|
149 | // oaicollection <collectionname>
|
---|
150 | //
|
---|
151 | // Where <collectionname> is the name of the directory inside the
|
---|
152 | // gsdl/collect folder which contains the collection.
|
---|
153 | //
|
---|
154 | // To configure several collections, merely repeat this line,
|
---|
155 | // or alternatively use additional collection names after the
|
---|
156 | // first one.
|
---|
157 | //
|
---|
158 | // This configuration should only appear in main.cfg
|
---|
159 | //
|
---|
160 | if (this->collection != "") {
|
---|
161 | cerr << "Attempt to configure an oai collection outside of main.cfg" << endl;
|
---|
162 | cerr << "Configuration attempted in " << this->collection << " collection." << endl;
|
---|
163 | exit(1);
|
---|
164 | }
|
---|
165 | for (int c = 0; c < cfgline.size(); ++c) {
|
---|
166 | this->collectList.push_back(cfgline[c]);
|
---|
167 | }
|
---|
168 | }
|
---|
169 | else if (key == "oaiinfo" && cfgline.size() >= 1) {
|
---|
170 | // Get a piece of information for the oai repository information
|
---|
171 | // request. The line should read:
|
---|
172 | //
|
---|
173 | // oaiinfo <information field name> <value>
|
---|
174 | //
|
---|
175 | // This configuration should only be attempted in main.cfg
|
---|
176 | //
|
---|
177 | if (this->collection != "") {
|
---|
178 | cerr << "Attempt to set oai information outside of main.cfg" << endl;
|
---|
179 | cerr << "Configuration attempted in " << this->collection << " collection." << endl;
|
---|
180 | exit(1);
|
---|
181 | }
|
---|
182 |
|
---|
183 | // if no second parameter is given, then the first parameter
|
---|
184 | if (cfgline.size() == 1) {
|
---|
185 | this->infoMap[cfgline[0]] = cfgline[0];
|
---|
186 | }
|
---|
187 | else {
|
---|
188 | this->infoMap[cfgline[0]] = cfgline[1];
|
---|
189 | }
|
---|
190 | }
|
---|
191 | else if (key == "oaiversion" && cfgline.size() >= 1) {
|
---|
192 | this->oaiVersion = cfgline[0];
|
---|
193 | }
|
---|
194 | else if (key == "resumeafter" && cfgline.size() >= 1) {
|
---|
195 | this->resumptionSize = cfgline[0].getint();
|
---|
196 | }
|
---|
197 | // get and note a maintainer item to support the Identify Verb of OAI
|
---|
198 | else if (key == "maintainer" && cfgline.size() >= 1) {
|
---|
199 | int line = 0;
|
---|
200 |
|
---|
201 | // TODO: exhaustive checks for empty or default values of maintainer
|
---|
202 | while (line < cfgline.size()) {
|
---|
203 | if (cfgline[line] != "NULL" &&
|
---|
204 | cfgline[line] != "") {
|
---|
205 | // do something
|
---|
206 | break;
|
---|
207 | }
|
---|
208 | else {
|
---|
209 | ++line;
|
---|
210 | }
|
---|
211 | }
|
---|
212 |
|
---|
213 | // Only try to set the configuration if we have a legitimate value ...
|
---|
214 | if (line < cfgline.size()) {
|
---|
215 | // ensure we have a map to write to
|
---|
216 | if (this->collectMap[this->collection] == NULL) {
|
---|
217 | this->collectMap[this->collection] = new oaicollectconfig(this->collection);
|
---|
218 | }
|
---|
219 | this->collectMap[this->collection]->maintainer = cfgline[line];
|
---|
220 | }
|
---|
221 | }
|
---|
222 | else if (key == "repositoryName" && cfgline.size() >= 1) {
|
---|
223 | int line = 0;
|
---|
224 |
|
---|
225 | // TODO: exhaustive checks for empty or default values of repositoryName
|
---|
226 | while (line < cfgline.size()) {
|
---|
227 | if (cfgline[line] != "NULL" &&
|
---|
228 | cfgline[line] != "") {
|
---|
229 | // do something
|
---|
230 | break;
|
---|
231 | }
|
---|
232 | else {
|
---|
233 | ++line;
|
---|
234 | }
|
---|
235 | }
|
---|
236 |
|
---|
237 | // Only try to set the configuration if we have a legitimate value ...
|
---|
238 | if (line < cfgline.size()) {
|
---|
239 | // ensure we have a map to write to
|
---|
240 | if (this->collectMap[this->collection] == NULL) {
|
---|
241 | this->collectMap[this->collection] = new oaicollectconfig(this->collection);
|
---|
242 | }
|
---|
243 | this->collectMap[this->collection]->repositoryName = cfgline[line];
|
---|
244 | }
|
---|
245 | }
|
---|
246 | else if (key == "baseURL" && cfgline.size() >= 1) {
|
---|
247 | int line = 0;
|
---|
248 |
|
---|
249 | while (line < cfgline.size()) {
|
---|
250 | if (cfgline[line] != "NULL" &&
|
---|
251 | cfgline[line] != "") {
|
---|
252 | // do something
|
---|
253 | break;
|
---|
254 | }
|
---|
255 | else {
|
---|
256 | ++line;
|
---|
257 | }
|
---|
258 | }
|
---|
259 |
|
---|
260 | // Only try to set the configuration if we have a legitimate value ...
|
---|
261 | if (line < cfgline.size()) {
|
---|
262 | // ensure we have a map to write to
|
---|
263 | if (this->collectMap[this->collection] == NULL) {
|
---|
264 | this->collectMap[this->collection] = new oaicollectconfig(this->collection);
|
---|
265 | }
|
---|
266 | this->collectMap[this->collection]->baseURL = cfgline[line];
|
---|
267 | }
|
---|
268 | }
|
---|
269 | }
|
---|
270 |
|
---|
271 | /**
|
---|
272 | * TODO: store all field values in a map per collection
|
---|
273 | */
|
---|
274 | text_t oaiconfig::getCollectionConfig(const text_t &collection, const text_t &field)
|
---|
275 | {
|
---|
276 | if (this->collectMap[collection] == NULL) {
|
---|
277 | return "";
|
---|
278 | }
|
---|
279 | if (field == "maintainer") {
|
---|
280 | return this->collectMap[collection]->maintainer;
|
---|
281 | }
|
---|
282 |
|
---|
283 | if (field == "repositoryName") {
|
---|
284 | return this->collectMap[collection]->repositoryName;
|
---|
285 | }
|
---|
286 |
|
---|
287 | if (field == "baseURL") {
|
---|
288 | return this->collectMap[collection]->baseURL;
|
---|
289 | }
|
---|
290 |
|
---|
291 | return "";
|
---|
292 | }
|
---|
293 |
|
---|
294 | text_t oaiconfig::getMapping(const text_t &collection, const text_t &collectfield)
|
---|
295 | {
|
---|
296 | if (this->collectMap[collection] == NULL) {
|
---|
297 | return "";
|
---|
298 | }
|
---|
299 | return this->collectMap[collection]->fieldMap[collectfield];
|
---|
300 | }
|
---|
301 |
|
---|
302 | /**
|
---|
303 | * Get the mapping for a field in a given collection; if no mapping
|
---|
304 | * exists, the result will be a blank string.
|
---|
305 | */
|
---|
306 | text_t oaiconfig::getMapping(const text_t &collection, const text_t &collectfield, const text_t &formatname)
|
---|
307 | {
|
---|
308 | text_t fullName = collectfield;
|
---|
309 | fullName.append(":");
|
---|
310 | fullName.append(formatname);
|
---|
311 |
|
---|
312 | // try the collection-specific options first
|
---|
313 | if (this->collectMap[collection] != NULL) {
|
---|
314 | // first try the most specific item - this collection, and given that protocol
|
---|
315 | if (this->collectMap[collection]->fieldMap.count(fullName) >= 1) {
|
---|
316 | return this->collectMap[collection]->fieldMap[fullName];
|
---|
317 | }
|
---|
318 | // otherwise, fall back to this collection, and all protocols
|
---|
319 | else if (this->collectMap[collection]->fieldMap.count(collectfield) >= 1) {
|
---|
320 | return this->collectMap[collection]->fieldMap[collectfield];
|
---|
321 | }
|
---|
322 | }
|
---|
323 |
|
---|
324 | // if no mappings exist, return an empty item
|
---|
325 | if (this->collectMap[""] == NULL) {
|
---|
326 | return "";
|
---|
327 | }
|
---|
328 |
|
---|
329 | // then try generic rules
|
---|
330 | if (this->collectMap[""]->fieldMap.count(fullName) >= 1) {
|
---|
331 | return this->collectMap[""]->fieldMap[fullName];
|
---|
332 | }
|
---|
333 | else {
|
---|
334 | return this->collectMap[""]->fieldMap[collectfield];
|
---|
335 | }
|
---|
336 | }
|
---|