|
1
|
/**
|
|
2
|
* Copyright (C) 2009 EDIT
|
|
3
|
* European Distributed Institute of Taxonomy
|
|
4
|
* http://www.e-taxonomy.eu
|
|
5
|
*
|
|
6
|
* The contents of this file are subject to the Mozilla Public License Version 1.1
|
|
7
|
* See LICENSE.TXT at the top of this package for the full license terms.
|
|
8
|
*/
|
|
9
|
package eu.etaxonomy.cdm.io.excel.stream;
|
|
10
|
|
|
11
|
import java.util.HashMap;
|
|
12
|
import java.util.Map;
|
|
13
|
import java.util.Set;
|
|
14
|
|
|
15
|
import org.apache.commons.lang.StringUtils;
|
|
16
|
import org.apache.log4j.Logger;
|
|
17
|
import org.apache.poi.ss.usermodel.Cell;
|
|
18
|
import org.apache.poi.ss.usermodel.Row;
|
|
19
|
import org.apache.poi.ss.usermodel.Sheet;
|
|
20
|
|
|
21
|
import eu.etaxonomy.cdm.common.ExcelUtils;
|
|
22
|
import eu.etaxonomy.cdm.io.common.events.IIoObserver;
|
|
23
|
import eu.etaxonomy.cdm.io.dwca.TermUri;
|
|
24
|
import eu.etaxonomy.cdm.io.stream.IItemStream;
|
|
25
|
import eu.etaxonomy.cdm.io.stream.StreamItem;
|
|
26
|
|
|
27
|
/**
|
|
28
|
* @author a.oppermann
|
|
29
|
* @date 16.05.2013
|
|
30
|
*
|
|
31
|
*/
|
|
32
|
public class ExcelRecordStream implements IItemStream{
|
|
33
|
private static Logger logger = Logger.getLogger(ExcelRecordStream.class);
|
|
34
|
|
|
35
|
private ExcelStreamImportState state;
|
|
36
|
|
|
37
|
private Sheet sheet;
|
|
38
|
// private ArchiveEntryBase archiveEntry;
|
|
39
|
private TermUri term;
|
|
40
|
private int line = 0;
|
|
41
|
|
|
42
|
private Map<Integer,String> mapping;
|
|
43
|
|
|
44
|
private StreamItem next;
|
|
45
|
|
|
46
|
|
|
47
|
|
|
48
|
|
|
49
|
/**
|
|
50
|
* Constructor.
|
|
51
|
* @param term
|
|
52
|
*/
|
|
53
|
public ExcelRecordStream(ExcelStreamImportState state, Sheet sheet, TermUri term) {
|
|
54
|
this.state = state;
|
|
55
|
this.sheet = sheet;
|
|
56
|
this.term = term;
|
|
57
|
}
|
|
58
|
|
|
59
|
/* (non-Javadoc)
|
|
60
|
* @see eu.etaxonomy.cdm.io.dwca.in.IReader#read()
|
|
61
|
*/
|
|
62
|
@Override
|
|
63
|
public StreamItem read() {
|
|
64
|
//FIXME:
|
|
65
|
return readMe();
|
|
66
|
}
|
|
67
|
|
|
68
|
/* (non-Javadoc)
|
|
69
|
* @see eu.etaxonomy.cdm.io.dwca.in.IReader#hasNext()
|
|
70
|
*/
|
|
71
|
@Override
|
|
72
|
public boolean hasNext() {
|
|
73
|
if (next != null){
|
|
74
|
return true;
|
|
75
|
}else{
|
|
76
|
next = readMe();
|
|
77
|
return (next != null);
|
|
78
|
}
|
|
79
|
}
|
|
80
|
|
|
81
|
|
|
82
|
private StreamItem readMe() {
|
|
83
|
StreamItem resultItem;
|
|
84
|
if (next != null){
|
|
85
|
resultItem = next;
|
|
86
|
next = null;
|
|
87
|
return resultItem;
|
|
88
|
}else{
|
|
89
|
|
|
90
|
if (mapping == null){
|
|
91
|
mapping = getHeaderMapping(sheet.getRow(line++));
|
|
92
|
}
|
|
93
|
// int rows = sheet.getPhysicalNumberOfRows();
|
|
94
|
Row row = sheet.getRow(line++);
|
|
95
|
int i = 0;
|
|
96
|
while (row == null && i++ < 10){
|
|
97
|
row = sheet.getRow(line++);
|
|
98
|
continue;
|
|
99
|
}
|
|
100
|
if (row == null){
|
|
101
|
return null;
|
|
102
|
}else{
|
|
103
|
int cells = row.getPhysicalNumberOfCells();
|
|
104
|
logger.info("\nROW " + row.getRowNum() + " has " + cells + " cell(s).");
|
|
105
|
Map<String, String> map = new HashMap<String, String>();
|
|
106
|
|
|
107
|
for (int c :mapping.keySet()){
|
|
108
|
Cell cell = row.getCell(c);
|
|
109
|
String value = ExcelUtils.getCellValue(cell);
|
|
110
|
map.put(mapping.get(c), value);
|
|
111
|
logger.info("CELL col=" + cell.getColumnIndex() + " VALUE=" + value);
|
|
112
|
}
|
|
113
|
|
|
114
|
resultItem = new StreamItem(term, map, String.valueOf(line));
|
|
115
|
|
|
116
|
return resultItem;
|
|
117
|
}
|
|
118
|
}
|
|
119
|
}
|
|
120
|
|
|
121
|
/**
|
|
122
|
* @param row
|
|
123
|
* @return
|
|
124
|
*/
|
|
125
|
private Map<Integer,String> getHeaderMapping(Row row) {
|
|
126
|
Map<Integer,String> result = new HashMap<Integer, String>();
|
|
127
|
|
|
128
|
int cells = row.getPhysicalNumberOfCells();
|
|
129
|
logger.info("\nROW " + row.getRowNum() + " has " + cells + " cell(s).");
|
|
130
|
for (int c = 0; c < cells; c++) {
|
|
131
|
Cell cell = row.getCell(c);
|
|
132
|
String value = ExcelUtils.getCellValue(cell);
|
|
133
|
String termUri = convert2semanticKey(value);
|
|
134
|
if (termUri != null){
|
|
135
|
result.put(c, termUri);
|
|
136
|
}else{
|
|
137
|
//TODO fire Event
|
|
138
|
String message = "No mapping defined for column %d '%s'";
|
|
139
|
logger.warn(String.format(message, c + 1, value));
|
|
140
|
}
|
|
141
|
}
|
|
142
|
|
|
143
|
return result;
|
|
144
|
}
|
|
145
|
|
|
146
|
/**
|
|
147
|
* @param value
|
|
148
|
* @return
|
|
149
|
*/
|
|
150
|
private String convert2semanticKey(String key) {
|
|
151
|
if (key.equalsIgnoreCase("id")){
|
|
152
|
return "id";
|
|
153
|
}else if (key.equalsIgnoreCase("ParentId")){
|
|
154
|
return TermUri.DWC_PARENT_NAME_USAGE_ID.getUriString();
|
|
155
|
}else if (key.equalsIgnoreCase("NameStatus")){
|
|
156
|
return TermUri.DWC_TAXONOMIC_STATUS.getUriString();
|
|
157
|
}else if (key.equalsIgnoreCase("Rank")){
|
|
158
|
return TermUri.DWC_TAXON_RANK.getUriString();
|
|
159
|
}else if (key.equalsIgnoreCase("ScientificName")){
|
|
160
|
return TermUri.DWC_SCIENTIFIC_NAME.getUriString();
|
|
161
|
}else if (key.equalsIgnoreCase("Author")){
|
|
162
|
return TermUri.DWC_SCIENTIFIC_NAME_AUTHORS.getUriString();
|
|
163
|
//TODO Taxon Remarks
|
|
164
|
}else if (key.equalsIgnoreCase("Comments")){
|
|
165
|
return TermUri.DWC_TAXON_REMARKS.getUriString();
|
|
166
|
}else if (key.equalsIgnoreCase("Language")){
|
|
167
|
return TermUri.DC_LANGUAGE.getUriString();
|
|
168
|
}else if (key.equalsIgnoreCase("TDWG_1")){
|
|
169
|
return TermUri.DWC_COUNTRY_CODE.getUriString();
|
|
170
|
}else if (key.equalsIgnoreCase("VernacularName")){
|
|
171
|
return TermUri.DWC_VERNACULAR_NAME.getUriString();
|
|
172
|
}else if (key.equalsIgnoreCase("ExternalId_sysCode")){
|
|
173
|
return TermUri.CDM_SOURCE_IDINSOURCE.getUriString();
|
|
174
|
}else if (key.equalsIgnoreCase("External_source")){
|
|
175
|
return TermUri.CDM_SOURCE_REFERENCE.getUriString();
|
|
176
|
}else if (key.equalsIgnoreCase("IdNamespace")){
|
|
177
|
return TermUri.CDM_SOURCE_IDNAMESPACE.getUriString();
|
|
178
|
}else{
|
|
179
|
//TODO fire Event
|
|
180
|
String message = "Key '%s' does not (yet) exist for import";
|
|
181
|
logger.warn(String.format(message, key));
|
|
182
|
return null;
|
|
183
|
}
|
|
184
|
|
|
185
|
//Language TDWG_1 TDWG_2 VernacularName SysCode
|
|
186
|
|
|
187
|
}
|
|
188
|
|
|
189
|
/* (non-Javadoc)
|
|
190
|
* @see eu.etaxonomy.cdm.io.stream.IItemStream#getTerm()
|
|
191
|
*/
|
|
192
|
@Override
|
|
193
|
public TermUri getTerm() {
|
|
194
|
return this.term;
|
|
195
|
}
|
|
196
|
|
|
197
|
/* (non-Javadoc)
|
|
198
|
* @see eu.etaxonomy.cdm.io.stream.IItemStream#getItemLocation()
|
|
199
|
*/
|
|
200
|
@Override
|
|
201
|
public String getItemLocation() {
|
|
202
|
// TODO Auto-generated method stub
|
|
203
|
return null;
|
|
204
|
}
|
|
205
|
|
|
206
|
/* (non-Javadoc)
|
|
207
|
* @see eu.etaxonomy.cdm.io.stream.IItemStream#getStreamLocation()
|
|
208
|
*/
|
|
209
|
@Override
|
|
210
|
public String getStreamLocation() {
|
|
211
|
// TODO Auto-generated method stub
|
|
212
|
return null;
|
|
213
|
}
|
|
214
|
|
|
215
|
/* (non-Javadoc)
|
|
216
|
* @see eu.etaxonomy.cdm.io.stream.IItemStream#addObservers(java.util.Set)
|
|
217
|
*/
|
|
218
|
@Override
|
|
219
|
public void addObservers(Set<IIoObserver> observers) {
|
|
220
|
// TODO Auto-generated method stub
|
|
221
|
logger.warn("addObservers Not yet implemented");
|
|
222
|
}
|
|
223
|
|
|
224
|
//******************** toString *******************************************
|
|
225
|
|
|
226
|
/* (non-Javadoc)
|
|
227
|
* @see java.lang.Object#toString()
|
|
228
|
*/
|
|
229
|
@Override
|
|
230
|
public String toString() {
|
|
231
|
if (sheet != null && StringUtils.isNotBlank(sheet.getSheetName())){
|
|
232
|
return sheet.getSheetName();
|
|
233
|
}else{
|
|
234
|
return super.toString();
|
|
235
|
}
|
|
236
|
}
|
|
237
|
|
|
238
|
|
|
239
|
|
|
240
|
|
|
241
|
}
|