1 | #include "oaiconfig.h"
|
---|
2 | #include <iostream>
|
---|
3 | #include "fileutil.h"
|
---|
4 |
|
---|
5 |
|
---|
6 | /**
|
---|
7 | * The mapping works as follows in the oai.cfg or collect.cfg file.
|
---|
8 | *
|
---|
9 | * A line is in the format oaimapping <collection field> <oai field>
|
---|
10 | *
|
---|
11 | * The map here is used to look up the "Greenstone" name which is mapped from
|
---|
12 | * a given OAI field name, the reverse direction to that given in the
|
---|
13 | * Greenstone collect.cfg file. The oairecordaction class instance which
|
---|
14 | * produces output for an OAI record information request thus uses the map
|
---|
15 | * to work from the field in the collection it has on hand which OAI
|
---|
16 | * record name it should use instead.
|
---|
17 | *
|
---|
18 | * An extension is to be used for this in which the OAI field name in the
|
---|
19 | * collect.cfg file can be made specific for a particular record format.
|
---|
20 | * This is done using the OAI field name in the format of
|
---|
21 | * <OAI format>.<OAI field name>
|
---|
22 | * Thus, an rfc1807 Title field would be referred to as rfc1807.Title
|
---|
23 | *
|
---|
24 | * A collection-level mapping can be provided in oai.cfg by prepending
|
---|
25 | * collname to collection field:
|
---|
26 | * <collname:field> <oai field>
|
---|
27 |
|
---|
28 | * In the absence of a particular format name, the mapping is taken to be
|
---|
29 | * universal.
|
---|
30 | */
|
---|
31 |
|
---|
32 | oaiconfig::oaiconfig() : configurable () {
|
---|
33 | this->resumptionSize = -1; // Default = do not use resumption tokens
|
---|
34 | }
|
---|
35 |
|
---|
36 | oaiconfig::oaiconfig(text_t &gsdlhome, text_t &gsdlcollect)
|
---|
37 | {
|
---|
38 | // read main configuration file (oai.cfg) to get oai collections
|
---|
39 | text_t mainconfig = filename_cat(gsdlhome, "etc", "oai.cfg");
|
---|
40 | this->collection = "";
|
---|
41 | this->resumptionSize = -1;
|
---|
42 | this->read_configfile(mainconfig);
|
---|
43 |
|
---|
44 | // then if we've not got a specified collection in the gsdlcollect
|
---|
45 | // parameter, read in all the collection's individual configurations
|
---|
46 | if (gsdlcollect == "") {
|
---|
47 | text_tarray::iterator start = this->collectList.begin();
|
---|
48 | text_tarray::iterator here = this->collectList.end()-1;
|
---|
49 | while (here != start) {
|
---|
50 | if (!this->configureCollection(gsdlhome, *here)) {
|
---|
51 | this->collectList.erase(here);
|
---|
52 | }
|
---|
53 | --here;
|
---|
54 | }
|
---|
55 | // and do the first one
|
---|
56 | if (!this->configureCollection(gsdlhome, *here)) {
|
---|
57 | this->collectList.erase(here);
|
---|
58 | }
|
---|
59 |
|
---|
60 | }
|
---|
61 | else {
|
---|
62 | // what do we do if this fails?
|
---|
63 | this->configureCollection(gsdlhome, gsdlcollect);
|
---|
64 | }
|
---|
65 | }
|
---|
66 |
|
---|
67 | oaiconfig::~oaiconfig()
|
---|
68 | {
|
---|
69 | oaicollectmap::iterator here = this->collectMap.begin();
|
---|
70 | oaicollectmap::iterator end = this->collectMap.end();
|
---|
71 | while (here != end) {
|
---|
72 | delete here->second;
|
---|
73 | ++here;
|
---|
74 | }
|
---|
75 | }
|
---|
76 |
|
---|
77 | int oaiconfig::resumeAfter()
|
---|
78 | { return this->resumptionSize;
|
---|
79 | }
|
---|
80 |
|
---|
81 | int oaiconfig::getOAIVersion()
|
---|
82 | {
|
---|
83 | if (this->oaiVersion == "1.1") {
|
---|
84 | return 110;
|
---|
85 | }
|
---|
86 | return 200;
|
---|
87 | }
|
---|
88 |
|
---|
89 | bool oaiconfig::configureCollection(const text_t &gsdlhome, const text_t &gsdlcollect)
|
---|
90 | {
|
---|
91 | text_t cnfgfile = filename_cat(gsdlhome, "collect", gsdlcollect, "etc", "collect.cfg");
|
---|
92 | if (!file_exists(cnfgfile)) {
|
---|
93 | return false;
|
---|
94 | }
|
---|
95 | this->collection = gsdlcollect;
|
---|
96 | this->read_configfile(cnfgfile);
|
---|
97 |
|
---|
98 | return true;
|
---|
99 | }
|
---|
100 | void oaiconfig::configure (const text_t &key, const text_tarray &cfgline)
|
---|
101 | {
|
---|
102 | // we've got an oai mapping item, and at least two fields
|
---|
103 | if (key == "oaimapping" && cfgline.size() > 1) {
|
---|
104 | text_t::const_iterator colonAt;
|
---|
105 | text_t index, name, configCollection;
|
---|
106 |
|
---|
107 | // Take a default collection as being whatever the collection being configured is...
|
---|
108 | configCollection = this->collection;
|
---|
109 |
|
---|
110 | // get the name of the (collection) field to map; this may actually
|
---|
111 | // be in a colon separated format of the type
|
---|
112 | // <collection name>:<field name>
|
---|
113 | index = cfgline[0];
|
---|
114 | if ((colonAt = findchar(index.begin(), index.end(), ':')) != index.end()) {
|
---|
115 | configCollection = substr(index.begin(), colonAt);
|
---|
116 |
|
---|
117 | if (this->collection != "" && configCollection != this->collection) {
|
---|
118 | cerr << "Attempt to configure OAI mappings for " << configCollection << " in " << this->collection << endl;
|
---|
119 | }
|
---|
120 |
|
---|
121 | colonAt += 1;
|
---|
122 | index = substr(colonAt, index.end());
|
---|
123 | }
|
---|
124 |
|
---|
125 | // the second parameter is the metadata field to map the collection
|
---|
126 | // field onto. It may be provided with a metadata protocol (which
|
---|
127 | // will be given first and separated by a period or full stop). In
|
---|
128 | // the case of format.field name, the splitting is done here.
|
---|
129 | if ((colonAt = findchar(cfgline[1].begin(), cfgline[1].end(), '.')) != cfgline[1].end()) {
|
---|
130 | text_t stub = substr(cfgline[1].begin(), colonAt);
|
---|
131 | colonAt += 1;
|
---|
132 | name = substr(colonAt, cfgline[1].end());
|
---|
133 | index.append(":");
|
---|
134 | index.append(stub);
|
---|
135 | }
|
---|
136 | else {
|
---|
137 | name = cfgline[1];
|
---|
138 | }
|
---|
139 |
|
---|
140 | // now 'index' is in the form <collectionfield>:(formatname)
|
---|
141 | // 'name' is simply the fieldname within the format
|
---|
142 | // 'configCollection' is the collection to be configured
|
---|
143 |
|
---|
144 | // now simply map the field name (index) onto the collection name (name)
|
---|
145 | if (this->collectMap[configCollection] == NULL) {
|
---|
146 | this->collectMap[configCollection] = new oaicollectconfig(configCollection);
|
---|
147 | }
|
---|
148 | this->collectMap[configCollection]->fieldMap[index] = name;
|
---|
149 |
|
---|
150 | // cerr << "Mapping " << index << " to " << name << " in " << configCollection << endl;
|
---|
151 |
|
---|
152 | // TODO: check that the mapped field is actually in use
|
---|
153 | }
|
---|
154 | else if (key == "oaicollection" && cfgline.size() >= 1) {
|
---|
155 | // Configure a collection to be used as part of the OAI archive.
|
---|
156 | // This line should read:
|
---|
157 | //
|
---|
158 | // oaicollection <collectionname>
|
---|
159 | //
|
---|
160 | // Where <collectionname> is the name of the directory inside the
|
---|
161 | // gsdl/collect folder which contains the collection.
|
---|
162 | //
|
---|
163 | // To configure several collections, merely repeat this line,
|
---|
164 | // or alternatively use additional collection names after the
|
---|
165 | // first one.
|
---|
166 | //
|
---|
167 | // This configuration should only appear in oai.cfg
|
---|
168 | //
|
---|
169 | if (this->collection != "") {
|
---|
170 | cerr << "Attempt to configure an oai collection outside of oai.cfg" << endl;
|
---|
171 | cerr << "Configuration attempted in " << this->collection << " collection." << endl;
|
---|
172 | exit(1);
|
---|
173 | }
|
---|
174 | for (int c = 0; c < cfgline.size(); ++c) {
|
---|
175 | this->collectList.push_back(cfgline[c]);
|
---|
176 | }
|
---|
177 | }
|
---|
178 | else if (key == "oaimetadata" && cfgline.size() >= 1) {
|
---|
179 | // List of metadata prefixes to suuport
|
---|
180 | // This line should read:
|
---|
181 | //
|
---|
182 | // oaicollection <metadataname> <metadataname>...
|
---|
183 | //
|
---|
184 | //
|
---|
185 | // This configuration should only appear in oai.cfg
|
---|
186 | //
|
---|
187 | if (this->collection != "") {
|
---|
188 | cerr << "Attempt to configure oai metadata outside of oai.cfg" << endl;
|
---|
189 | cerr << "Configuration attempted in " << this->collection << " collection." << endl;
|
---|
190 | exit(1);
|
---|
191 | }
|
---|
192 | for (int c = 0; c < cfgline.size(); ++c) {
|
---|
193 | // todo: check that the set name is valid
|
---|
194 | this->metadataSet.insert(cfgline[c]);
|
---|
195 | }
|
---|
196 | }
|
---|
197 | else if (key == "oaiinfo" && cfgline.size() >= 1) {
|
---|
198 | // Get a piece of information for the oai repository information
|
---|
199 | // request. The line should read:
|
---|
200 | //
|
---|
201 | // oaiinfo <information field name> <value>
|
---|
202 | //
|
---|
203 | // This configuration should only be attempted in oai.cfg
|
---|
204 | //
|
---|
205 | if (this->collection != "") {
|
---|
206 | cerr << "Attempt to set oai information outside of oai.cfg" << endl;
|
---|
207 | cerr << "Configuration attempted in " << this->collection << " collection." << endl;
|
---|
208 | exit(1);
|
---|
209 | }
|
---|
210 |
|
---|
211 | // if no second parameter is given, then the first parameter
|
---|
212 | if (cfgline.size() == 1) {
|
---|
213 | this->infoMap[cfgline[0]] = cfgline[0];
|
---|
214 | }
|
---|
215 | else {
|
---|
216 | this->infoMap[cfgline[0]] = cfgline[1];
|
---|
217 | }
|
---|
218 | }
|
---|
219 | else if (key == "oaisetname" && cfgline.size() == 2) {
|
---|
220 | this->setNameMap[cfgline[0]] = cfgline[1];
|
---|
221 | }
|
---|
222 | else if (key == "oaisetdescription" && cfgline.size() == 2) {
|
---|
223 | this->setDescriptionMap[cfgline[0]] = cfgline[1];
|
---|
224 | }
|
---|
225 | else if (key == "oaiversion" && cfgline.size() >= 1) {
|
---|
226 | this->oaiVersion = cfgline[0];
|
---|
227 | }
|
---|
228 | else if (key == "resumeafter" && cfgline.size() >= 1) {
|
---|
229 | this->resumptionSize = cfgline[0].getint();
|
---|
230 | }
|
---|
231 | // get and note a maintainer item to support the Identify Verb of OAI
|
---|
232 | else if (key == "maintainer" && cfgline.size() >= 1) {
|
---|
233 | int line = 0;
|
---|
234 |
|
---|
235 | // TODO: exhaustive checks for empty or default values of maintainer
|
---|
236 | while (line < cfgline.size()) {
|
---|
237 | if (cfgline[line] != "NULL" &&
|
---|
238 | cfgline[line] != "") {
|
---|
239 | // do something
|
---|
240 | break;
|
---|
241 | }
|
---|
242 | else {
|
---|
243 | ++line;
|
---|
244 | }
|
---|
245 | }
|
---|
246 |
|
---|
247 | // Only try to set the configuration if we have a legitimate value ...
|
---|
248 | if (line < cfgline.size()) {
|
---|
249 | // ensure we have a map to write to
|
---|
250 | if (this->collectMap[this->collection] == NULL) {
|
---|
251 | this->collectMap[this->collection] = new oaicollectconfig(this->collection);
|
---|
252 | }
|
---|
253 | this->collectMap[this->collection]->maintainer = cfgline[line];
|
---|
254 | }
|
---|
255 | }
|
---|
256 | else if (key == "repositoryName" && cfgline.size() >= 1) {
|
---|
257 | int line = 0;
|
---|
258 |
|
---|
259 | // TODO: exhaustive checks for empty or default values of repositoryName
|
---|
260 | while (line < cfgline.size()) {
|
---|
261 | if (cfgline[line] != "NULL" &&
|
---|
262 | cfgline[line] != "") {
|
---|
263 | // do something
|
---|
264 | break;
|
---|
265 | }
|
---|
266 | else {
|
---|
267 | ++line;
|
---|
268 | }
|
---|
269 | }
|
---|
270 |
|
---|
271 | // Only try to set the configuration if we have a legitimate value ...
|
---|
272 | if (line < cfgline.size()) {
|
---|
273 | // ensure we have a map to write to
|
---|
274 | if (this->collectMap[this->collection] == NULL) {
|
---|
275 | this->collectMap[this->collection] = new oaicollectconfig(this->collection);
|
---|
276 | }
|
---|
277 | this->collectMap[this->collection]->repositoryName = cfgline[line];
|
---|
278 | }
|
---|
279 | }
|
---|
280 | else if (key == "baseURL" && cfgline.size() >= 1) {
|
---|
281 | int line = 0;
|
---|
282 |
|
---|
283 | while (line < cfgline.size()) {
|
---|
284 | if (cfgline[line] != "NULL" &&
|
---|
285 | cfgline[line] != "") {
|
---|
286 | // do something
|
---|
287 | break;
|
---|
288 | }
|
---|
289 | else {
|
---|
290 | ++line;
|
---|
291 | }
|
---|
292 | }
|
---|
293 |
|
---|
294 | // Only try to set the configuration if we have a legitimate value ...
|
---|
295 | if (line < cfgline.size()) {
|
---|
296 | // ensure we have a map to write to
|
---|
297 | if (this->collectMap[this->collection] == NULL) {
|
---|
298 | this->collectMap[this->collection] = new oaicollectconfig(this->collection);
|
---|
299 | }
|
---|
300 | this->collectMap[this->collection]->baseURL = cfgline[line];
|
---|
301 | }
|
---|
302 | }
|
---|
303 | else if (key == "baseDocRoot" && cfgline.size() >= 1) {
|
---|
304 | int line = 0;
|
---|
305 |
|
---|
306 | while (line < cfgline.size()) {
|
---|
307 | if (cfgline[line] != "NULL" &&
|
---|
308 | cfgline[line] != "") {
|
---|
309 | // do something
|
---|
310 | break;
|
---|
311 | }
|
---|
312 | else {
|
---|
313 | ++line;
|
---|
314 | }
|
---|
315 | }
|
---|
316 |
|
---|
317 | // Only try to set the configuration if we have a legitimate value ...
|
---|
318 | if (line < cfgline.size()) {
|
---|
319 | // ensure we have a map to write to
|
---|
320 | if (this->collectMap[this->collection] == NULL) {
|
---|
321 | this->collectMap[this->collection] = new oaicollectconfig(this->collection);
|
---|
322 | }
|
---|
323 | this->collectMap[this->collection]->baseDocRoot = cfgline[line];
|
---|
324 | }
|
---|
325 | }
|
---|
326 | }
|
---|
327 |
|
---|
328 | /**
|
---|
329 | * TODO: store all field values in a map per collection
|
---|
330 | */
|
---|
331 | text_t oaiconfig::getCollectionConfig(const text_t &collection, const text_t &field)
|
---|
332 | {
|
---|
333 | if (this->collectMap[collection] == NULL) {
|
---|
334 | return "";
|
---|
335 | }
|
---|
336 | if (field == "maintainer") {
|
---|
337 | return this->collectMap[collection]->maintainer;
|
---|
338 | }
|
---|
339 |
|
---|
340 | if (field == "repositoryName") {
|
---|
341 | return this->collectMap[collection]->repositoryName;
|
---|
342 | }
|
---|
343 |
|
---|
344 | if (field == "baseURL") {
|
---|
345 | return this->collectMap[collection]->baseURL;
|
---|
346 | }
|
---|
347 |
|
---|
348 | if (field == "baseDocRoot") {
|
---|
349 | return this->collectMap[collection]->baseDocRoot;
|
---|
350 | }
|
---|
351 |
|
---|
352 | return "";
|
---|
353 | }
|
---|
354 |
|
---|
355 | text_t oaiconfig::getMapping(const text_t &collection, const text_t &collectfield)
|
---|
356 | {
|
---|
357 | if (this->collectMap[collection] == NULL) {
|
---|
358 | return "";
|
---|
359 | }
|
---|
360 | return this->collectMap[collection]->fieldMap[collectfield];
|
---|
361 | }
|
---|
362 |
|
---|
363 | /**
|
---|
364 | * Get the mapping for a field in a given collection; if no mapping
|
---|
365 | * exists, the result will be a blank string.
|
---|
366 | */
|
---|
367 | text_t oaiconfig::getMapping(const text_t &collection, const text_t &collectfield, const text_t &formatname)
|
---|
368 | {
|
---|
369 | text_t fullName = collectfield;
|
---|
370 | fullName.append(":");
|
---|
371 | fullName.append(formatname);
|
---|
372 |
|
---|
373 | // try the collection-specific options first
|
---|
374 | if (this->collectMap[collection] != NULL) {
|
---|
375 | // first try the most specific item - this collection, and given that protocol
|
---|
376 | if (this->collectMap[collection]->fieldMap.count(fullName) >= 1) {
|
---|
377 | return this->collectMap[collection]->fieldMap[fullName];
|
---|
378 | }
|
---|
379 | // otherwise, fall back to this collection, and all protocols
|
---|
380 | else if (this->collectMap[collection]->fieldMap.count(collectfield) >= 1) {
|
---|
381 | return this->collectMap[collection]->fieldMap[collectfield];
|
---|
382 | }
|
---|
383 | }
|
---|
384 |
|
---|
385 | // if no mappings exist, return an empty item
|
---|
386 | if (this->collectMap[""] == NULL) {
|
---|
387 | return "";
|
---|
388 | }
|
---|
389 |
|
---|
390 | // then try generic rules
|
---|
391 | if (this->collectMap[""]->fieldMap.count(fullName) >= 1) {
|
---|
392 | return this->collectMap[""]->fieldMap[fullName];
|
---|
393 | }
|
---|
394 | else {
|
---|
395 | return this->collectMap[""]->fieldMap[collectfield];
|
---|
396 | }
|
---|
397 | }
|
---|
398 |
|
---|
399 |
|
---|
400 | text_t oaiconfig::getSetName(const text_t &setSpec)
|
---|
401 | {
|
---|
402 | if (setNameMap.find(setSpec) != setNameMap.end())
|
---|
403 | {
|
---|
404 | return setNameMap[setSpec];
|
---|
405 | }
|
---|
406 |
|
---|
407 | return "";
|
---|
408 | }
|
---|
409 |
|
---|
410 |
|
---|
411 | text_t oaiconfig::getSetDescription(const text_t &setSpec)
|
---|
412 | {
|
---|
413 | if (setDescriptionMap.find(setSpec) != setDescriptionMap.end())
|
---|
414 | {
|
---|
415 | return setDescriptionMap[setSpec];
|
---|
416 | }
|
---|
417 |
|
---|
418 | return "";
|
---|
419 | }
|
---|