|
1
|
package org.bgbm.biovel.drf.occurrences;
|
|
2
|
|
|
3
|
import java.net.URI;
|
|
4
|
import java.text.ParseException;
|
|
5
|
import java.text.SimpleDateFormat;
|
|
6
|
import java.util.ArrayList;
|
|
7
|
import java.util.Date;
|
|
8
|
import java.util.HashMap;
|
|
9
|
import java.util.Iterator;
|
|
10
|
import java.util.List;
|
|
11
|
import java.util.Map;
|
|
12
|
|
|
13
|
import org.apache.http.HttpHost;
|
|
14
|
import org.bgbm.biovel.drf.checklist.DRFChecklistException;
|
|
15
|
import org.bgbm.biovel.drf.client.ServiceProviderInfo;
|
|
16
|
import org.bgbm.biovel.drf.query.RestClient;
|
|
17
|
import org.bgbm.biovel.drf.utils.CSVUtils;
|
|
18
|
import org.bgbm.biovel.drf.utils.JSONUtils;
|
|
19
|
import org.json.simple.JSONArray;
|
|
20
|
import org.json.simple.JSONObject;
|
|
21
|
|
|
22
|
public class GBIFOccurrencesClient extends BaseOccurrencesClient<RestClient> {
|
|
23
|
|
|
24
|
|
|
25
|
/**
|
|
26
|
*
|
|
27
|
*/
|
|
28
|
private static final HttpHost HTTP_HOST = new HttpHost("api.gbif.org",80);
|
|
29
|
public static final String ID = "gbif";
|
|
30
|
public static final String LABEL = "GBIF Occurrence Bank";
|
|
31
|
public static final String URL = "http://uat.gbif.org/developer/species";
|
|
32
|
public static final String DATA_AGR_URL = "http://data.gbif.org/tutorial/datauseagreement";
|
|
33
|
// in v0.9 the max limit is 300
|
|
34
|
private static final String MAX_PAGING_LIMIT = "300";
|
|
35
|
private static final String VERSION = "v0.9";
|
|
36
|
private static final ServiceProviderInfo CINFO = new ServiceProviderInfo(ID,LABEL,ServiceProviderInfo.DEFAULT_SEARCH_MODE,URL,DATA_AGR_URL, VERSION);
|
|
37
|
|
|
38
|
private final Map<String, JSONObject> datasetCacheMap = new HashMap<String, JSONObject>();
|
|
39
|
private final Map<String, JSONObject> orgCacheMap = new HashMap<String, JSONObject>();
|
|
40
|
public final static List<String> nameidList = new ArrayList<String>();
|
|
41
|
|
|
42
|
|
|
43
|
/**
|
|
44
|
* {@inheritDoc}
|
|
45
|
*/
|
|
46
|
@Override
|
|
47
|
public void initQueryClient() {
|
|
48
|
queryClient = new RestClient(HTTP_HOST);
|
|
49
|
|
|
50
|
}
|
|
51
|
|
|
52
|
@Override
|
|
53
|
public int getMaxPageSize() {
|
|
54
|
// TODO Auto-generated method stub
|
|
55
|
return 0;
|
|
56
|
}
|
|
57
|
|
|
58
|
@Override
|
|
59
|
public ServiceProviderInfo buildServiceProviderInfo() {
|
|
60
|
ServiceProviderInfo ocbankInfo = CINFO;
|
|
61
|
return ocbankInfo;
|
|
62
|
}
|
|
63
|
|
|
64
|
@Override
|
|
65
|
public String getOccurrences(String nameid) throws DRFChecklistException {
|
|
66
|
|
|
67
|
URI namesUri = queryClient.buildUriFromQueryString(nameid, "/" + CINFO.getVersion() + "/species/match", "name", null);
|
|
68
|
String nameResponse = queryClient.processRESTService(namesUri);
|
|
69
|
JSONObject nameJsonResponse = JSONUtils.parseJsonToObject(nameResponse);
|
|
70
|
StringBuilder occurrences = new StringBuilder();
|
|
71
|
if(nameJsonResponse.get("usageKey") != null) {
|
|
72
|
String usageKey = String.valueOf(nameJsonResponse.get("usageKey"));
|
|
73
|
|
|
74
|
if(!nameidList.contains(usageKey)) {
|
|
75
|
nameidList.add(usageKey);
|
|
76
|
//http://api.gbif.org/v0.9/occurrence/search?offset=100&limit=100&taxonKey=2818622
|
|
77
|
Map<String, String> paramMap = new HashMap<String, String>();
|
|
78
|
paramMap.put("limit", MAX_PAGING_LIMIT);
|
|
79
|
boolean endOfRecords = false;
|
|
80
|
int offset = 0;
|
|
81
|
|
|
82
|
|
|
83
|
int count = 0;
|
|
84
|
do {
|
|
85
|
paramMap.put("offset", Integer.toString(offset));
|
|
86
|
URI occUri = queryClient.buildUriFromQueryString(usageKey, "/" + CINFO.getVersion() + "/occurrence/search", "taxonKey", paramMap);
|
|
87
|
|
|
88
|
String occResponse = queryClient.processRESTService(occUri);
|
|
89
|
|
|
90
|
JSONObject jsonOccResponse = JSONUtils.parseJsonToObject(occResponse);
|
|
91
|
JSONArray results = (JSONArray) jsonOccResponse.get("results");
|
|
92
|
System.out.println("actual results size : " + results.size());
|
|
93
|
if(results != null) {
|
|
94
|
Iterator<JSONObject> resIterator = results.iterator();
|
|
95
|
|
|
96
|
while (resIterator.hasNext()) {
|
|
97
|
JSONObject jsonOccurence = resIterator.next();
|
|
98
|
occurrences.append(",");
|
|
99
|
|
|
100
|
if(jsonOccurence.get("genus") != null) {
|
|
101
|
occurrences.append(CSVUtils.wrapWhenComma((String) jsonOccurence.get("genus")));
|
|
102
|
}
|
|
103
|
occurrences.append(",");
|
|
104
|
|
|
105
|
|
|
106
|
occurrences.append(",");
|
|
107
|
|
|
108
|
|
|
109
|
occurrences.append(",");
|
|
110
|
|
|
111
|
|
|
112
|
occurrences.append(",");
|
|
113
|
|
|
114
|
if(jsonOccurence.get("scientificName") != null) {
|
|
115
|
occurrences.append(CSVUtils.wrapWhenComma((String) jsonOccurence.get("scientificName")));
|
|
116
|
}
|
|
117
|
occurrences.append(",");
|
|
118
|
|
|
119
|
occurrences.append(",");
|
|
120
|
|
|
121
|
if(jsonOccurence.get("scientificName") != null) {
|
|
122
|
occurrences.append(CSVUtils.wrapWhenComma((String) jsonOccurence.get("scientificName")));
|
|
123
|
}
|
|
124
|
occurrences.append(",");
|
|
125
|
|
|
126
|
if(jsonOccurence.get("key") != null) {
|
|
127
|
occurrences.append(CSVUtils.wrapWhenComma(String.valueOf(jsonOccurence.get("key"))));
|
|
128
|
}
|
|
129
|
occurrences.append(",");
|
|
130
|
|
|
131
|
if(jsonOccurence.get("decimalLatitude") != null) {
|
|
132
|
occurrences.append(CSVUtils.wrapWhenComma(String.valueOf(jsonOccurence.get("decimalLatitude"))));
|
|
133
|
}
|
|
134
|
occurrences.append(",");
|
|
135
|
|
|
136
|
if(jsonOccurence.get("decimalLongitude") != null) {
|
|
137
|
occurrences.append(CSVUtils.wrapWhenComma(String.valueOf(jsonOccurence.get("decimalLongitude"))));
|
|
138
|
}
|
|
139
|
occurrences.append(",");
|
|
140
|
|
|
141
|
String formattedDate = "";
|
|
142
|
SimpleDateFormat gbifFormatter = new SimpleDateFormat("yyyy-MM-dd'T'HH:mm:ss.SSSZ");
|
|
143
|
SimpleDateFormat biovelFormatter = new SimpleDateFormat("yyyy-MM-dd");
|
|
144
|
|
|
145
|
if(jsonOccurence.get("occurrenceDate") != null) {
|
|
146
|
String strDate = (String) jsonOccurence.get("occurrenceDate");
|
|
147
|
try {
|
|
148
|
Date date = gbifFormatter.parse(strDate);
|
|
149
|
formattedDate = biovelFormatter.format(date);
|
|
150
|
} catch (ParseException e) {
|
|
151
|
formattedDate = "";
|
|
152
|
}
|
|
153
|
} else if((jsonOccurence.get("year") != null) && (jsonOccurence.get("month") != null) && (jsonOccurence.get("day") != null)) {
|
|
154
|
String year = String.valueOf(jsonOccurence.get("year"));
|
|
155
|
String month = String.valueOf(jsonOccurence.get("month"));
|
|
156
|
String day = String.valueOf(jsonOccurence.get("day"));
|
|
157
|
formattedDate = year + "-" + month + "-" + day;
|
|
158
|
System.out.println("date : " + formattedDate);
|
|
159
|
}
|
|
160
|
occurrences.append(CSVUtils.wrapWhenComma(formattedDate));
|
|
161
|
occurrences.append(",");
|
|
162
|
|
|
163
|
|
|
164
|
//if(jsonOccurence.get("occurrenceDate") != null) {
|
|
165
|
occurrences.append(CSVUtils.wrapWhenComma(formattedDate));
|
|
166
|
//}
|
|
167
|
occurrences.append(",");
|
|
168
|
|
|
169
|
|
|
170
|
if(jsonOccurence.get("coordinateAccurracyInMeters") != null) {
|
|
171
|
occurrences.append(CSVUtils.wrapWhenComma(Integer.toString((Integer) jsonOccurence.get("coordinateAccurracyInMeters"))));
|
|
172
|
}
|
|
173
|
occurrences.append(",");
|
|
174
|
|
|
175
|
if(jsonOccurence.get("country") != null) {
|
|
176
|
occurrences.append(CSVUtils.wrapWhenComma((String) jsonOccurence.get("country")));
|
|
177
|
}
|
|
178
|
occurrences.append(",");
|
|
179
|
|
|
180
|
if(jsonOccurence.get("collectorName") != null) {
|
|
181
|
occurrences.append(CSVUtils.wrapWhenComma((String) jsonOccurence.get("collectorName")));
|
|
182
|
}
|
|
183
|
occurrences.append(",");
|
|
184
|
|
|
185
|
occurrences.append(",");
|
|
186
|
|
|
187
|
if(jsonOccurence.get("locality") != null) {
|
|
188
|
occurrences.append(CSVUtils.wrapWhenComma((String) jsonOccurence.get("locality")));
|
|
189
|
}
|
|
190
|
occurrences.append(",");
|
|
191
|
|
|
192
|
if(jsonOccurence.get("depth") != null) {
|
|
193
|
occurrences.append(CSVUtils.wrapWhenComma(String.valueOf(jsonOccurence.get("depth"))));
|
|
194
|
}
|
|
195
|
occurrences.append(",");
|
|
196
|
|
|
197
|
if(jsonOccurence.get("altitude") != null) {
|
|
198
|
occurrences.append(CSVUtils.wrapWhenComma(String.valueOf(jsonOccurence.get("altitude"))));
|
|
199
|
}
|
|
200
|
occurrences.append(",");
|
|
201
|
|
|
202
|
if(jsonOccurence.get("depth") != null) {
|
|
203
|
String depth = CSVUtils.wrapWhenComma(String.valueOf(jsonOccurence.get("depth")));
|
|
204
|
occurrences.append(depth);
|
|
205
|
System.out.println("depth : " + depth);
|
|
206
|
}
|
|
207
|
occurrences.append(",");
|
|
208
|
|
|
209
|
if(jsonOccurence.get("altitude") != null) {
|
|
210
|
occurrences.append(CSVUtils.wrapWhenComma(String.valueOf(jsonOccurence.get("altitude"))));
|
|
211
|
}
|
|
212
|
occurrences.append(",");
|
|
213
|
|
|
214
|
occurrences.append(",");
|
|
215
|
|
|
216
|
JSONObject datasetJsonResponse = null;
|
|
217
|
if(jsonOccurence.get("datasetKey") != null) {
|
|
218
|
String datasetKey = (String) jsonOccurence.get("datasetKey");
|
|
219
|
datasetJsonResponse = datasetCacheMap.get(datasetKey);
|
|
220
|
if(datasetJsonResponse == null) {
|
|
221
|
URI datasetUri = queryClient.buildUriFromQueryString("/" + CINFO.getVersion() + "/dataset/" + datasetKey, null);
|
|
222
|
String datasetResponse = queryClient.processRESTService(datasetUri);
|
|
223
|
datasetJsonResponse = JSONUtils.parseJsonToObject(datasetResponse);
|
|
224
|
datasetCacheMap.put(datasetKey, datasetJsonResponse);
|
|
225
|
}
|
|
226
|
}
|
|
227
|
|
|
228
|
JSONObject orgJsonResponse = null;
|
|
229
|
if(datasetJsonResponse != null && datasetJsonResponse.get("owningOrganizationKey") != null) {
|
|
230
|
String owningOrganizationKey = (String) datasetJsonResponse.get("owningOrganizationKey");
|
|
231
|
orgJsonResponse = orgCacheMap.get(owningOrganizationKey);
|
|
232
|
if(orgJsonResponse == null) {
|
|
233
|
URI orgUri = queryClient.buildUriFromQueryString("/" + CINFO.getVersion() + "/organization/" + owningOrganizationKey, null);
|
|
234
|
String orgResponse = queryClient.processRESTService(orgUri);
|
|
235
|
orgJsonResponse = JSONUtils.parseJsonToObject(orgResponse);
|
|
236
|
orgCacheMap.put(owningOrganizationKey, orgJsonResponse);
|
|
237
|
}
|
|
238
|
if(orgJsonResponse != null && orgJsonResponse.get("title") != null) {
|
|
239
|
occurrences.append(CSVUtils.wrapWhenComma((String) orgJsonResponse.get("title")));
|
|
240
|
}
|
|
241
|
|
|
242
|
}
|
|
243
|
occurrences.append(",");
|
|
244
|
|
|
245
|
if(datasetJsonResponse != null && datasetJsonResponse.get("title") != null) {
|
|
246
|
occurrences.append(CSVUtils.wrapWhenComma((String) datasetJsonResponse.get("title")));
|
|
247
|
}
|
|
248
|
occurrences.append(",");
|
|
249
|
|
|
250
|
if(datasetJsonResponse != null && datasetJsonResponse.get("rights") != null) {
|
|
251
|
occurrences.append(CSVUtils.wrapWhenComma(((String) datasetJsonResponse.get("rights")).replaceAll("\r\n|\r|\n", " ")));
|
|
252
|
//System.out.println("rights : " + CSVUtils.wrapWhenComma(((String) datasetJsonResponse.get("rights")).replaceAll("\r\n|\r|\n", " ")));
|
|
253
|
}
|
|
254
|
occurrences.append(",");
|
|
255
|
|
|
256
|
if(datasetJsonResponse != null && datasetJsonResponse.get("citation") != null) {
|
|
257
|
JSONObject citationJson = (JSONObject) datasetJsonResponse.get("citation");
|
|
258
|
if(citationJson.get("text") != null) {
|
|
259
|
occurrences.append(CSVUtils.wrapWhenComma(((String) citationJson.get("text")).replaceAll("\r\n|\r|\n", " ")));
|
|
260
|
|
|
261
|
}
|
|
262
|
}
|
|
263
|
occurrences.append(System.getProperty("line.separator"));
|
|
264
|
count++;
|
|
265
|
|
|
266
|
}
|
|
267
|
}
|
|
268
|
endOfRecords = (Boolean) jsonOccResponse.get("endOfRecords");
|
|
269
|
System.out.println("usageKey : " + usageKey + ", count : " + String.valueOf(jsonOccResponse.get("count")) + ", offset : " + offset + ", + occ count : " + count);
|
|
270
|
offset = offset + Integer.parseInt(MAX_PAGING_LIMIT);
|
|
271
|
} while(!endOfRecords);
|
|
272
|
System.out.println("occ count : " + count);
|
|
273
|
}
|
|
274
|
}
|
|
275
|
return occurrences.toString();
|
|
276
|
}
|
|
277
|
|
|
278
|
}
|