1
|
<%@page
|
2
|
import="java.util.Iterator"%><%@page
|
3
|
import="java.util.ArrayList"%><%@page
|
4
|
import="edu.ucsb.nceas.utilities.OrderedMap"%><%@page
|
5
|
import="java.util.List"%><%@page
|
6
|
import="java.util.Enumeration"%><%@page
|
7
|
import="java.sql.SQLException"%><%@page
|
8
|
import="org.ecoinformatics.datamanager.transpose.DataTranspose"%><%@page
|
9
|
import="au.com.bytecode.opencsv.CSVWriter"%><%@page
|
10
|
import="java.io.OutputStreamWriter"%><%@page
|
11
|
import="java.io.Writer"%><%@page
|
12
|
import="java.sql.ResultSet"%><%@page
|
13
|
import="edu.ucsb.nceas.metacat.dataquery.DataQuery"%><%@page
|
14
|
import="java.io.IOException"%><%@page
|
15
|
import="edu.ucsb.nceas.utilities.PropertyNotFoundException"%><%@page
|
16
|
import="java.util.Hashtable"%><%@ page
|
17
|
language="java" %><%
|
18
|
/**
|
19
|
*
|
20
|
* '$RCSfile$'
|
21
|
* Copyright: 2008 Regents of the University of California and the
|
22
|
* National Center for Ecological Analysis and Synthesis
|
23
|
* '$Author: leinfelder $'
|
24
|
* '$Date: 2008-08-22 16:48:56 -0700 (Fri, 22 Aug 2008) $'
|
25
|
* '$Revision: 4305 $'
|
26
|
*
|
27
|
* This program is free software; you can redistribute it and/or modify
|
28
|
* it under the terms of the GNU General Public License as published by
|
29
|
* the Free Software Foundation; either version 2 of the License, or
|
30
|
* (at your option) any later version.
|
31
|
*
|
32
|
* This program is distributed in the hope that it will be useful,
|
33
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
34
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
35
|
* GNU General Public License for more details.
|
36
|
|
37
|
* You should have received a copy of the GNU General Public License
|
38
|
* along with this program; if not, write to the Free Software
|
39
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
40
|
*/
|
41
|
%><%
|
42
|
Hashtable params = getParams(request);
|
43
|
handleDataquery(params,response,request.getSession().getId());
|
44
|
%><%!
|
45
|
private Hashtable getParams(HttpServletRequest request) {
|
46
|
Hashtable params = new Hashtable();
|
47
|
Enumeration<String> paramlist = (Enumeration<String>)request.getParameterNames();
|
48
|
while (paramlist.hasMoreElements()) {
|
49
|
|
50
|
String name = paramlist.nextElement();
|
51
|
String[] value = request.getParameterValues(name);
|
52
|
params.put(name, value);
|
53
|
}
|
54
|
|
55
|
return params;
|
56
|
}
|
57
|
%><%!
|
58
|
private List transpose(ResultSet rs, int idCol, int pivotCol, List pivotAttributes, boolean omitIdValues) throws SQLException {
|
59
|
//map keyed by id column - data
|
60
|
OrderedMap table = new OrderedMap();
|
61
|
//track all the data columns
|
62
|
OrderedMap widestRow = new OrderedMap();
|
63
|
//map keyed by the pivot column - metadata
|
64
|
OrderedMap headerRows = new OrderedMap();
|
65
|
|
66
|
//vocab columns
|
67
|
String vocabNameCol = "qmetadatavocabulary";
|
68
|
String vocabValueCol = "qmetadatavalue";
|
69
|
//maps for the vocab lists
|
70
|
OrderedMap vocabNames = new OrderedMap();
|
71
|
OrderedMap vocabValues = new OrderedMap();
|
72
|
//all vocab names/values
|
73
|
List allVocabNames = new ArrayList();
|
74
|
List allVocabValues = new ArrayList();
|
75
|
boolean twoColumnMetadata = true;
|
76
|
|
77
|
int colCount = rs.getMetaData().getColumnCount();
|
78
|
String idColName = rs.getMetaData().getColumnName(idCol);
|
79
|
|
80
|
while (rs.next()) {
|
81
|
String id = rs.getString(idCol);
|
82
|
String pivotValue = rs.getString(pivotCol);
|
83
|
|
84
|
//look up the data row we are working on
|
85
|
OrderedMap row = (OrderedMap) table.get(id);
|
86
|
if (row == null) {
|
87
|
row = new OrderedMap();
|
88
|
}
|
89
|
//look up the metadata row we are working on
|
90
|
OrderedMap metadataRow = (OrderedMap) table.get(pivotValue);
|
91
|
if (metadataRow == null) {
|
92
|
metadataRow = new OrderedMap();
|
93
|
}
|
94
|
|
95
|
//get the values for this pivot
|
96
|
for (int i = 1; i <= colCount; i++) {
|
97
|
if (i != pivotCol) {
|
98
|
String colName = rs.getMetaData().getColumnName(i);
|
99
|
String value = rs.getString(i);
|
100
|
//clean up the value
|
101
|
if (value != null) {
|
102
|
value = value.replaceAll("\n", " ");
|
103
|
value = value.replaceAll("\\s+", " ");
|
104
|
value = value.replaceAll("<html>", " ");
|
105
|
value = value.replaceAll("</html>", " ");
|
106
|
value = value.replaceAll("<head>", " ");
|
107
|
value = value.replaceAll("</head>", " ");
|
108
|
value = value.replaceAll("<body>", " ");
|
109
|
value = value.replaceAll("</body>", " ");
|
110
|
}
|
111
|
//do we include this column in the pivot?
|
112
|
if (pivotAttributes.contains(colName)) {
|
113
|
//annotate the column name with the pivot column value if not the id column
|
114
|
if (i != idCol) {
|
115
|
colName = pivotValue + "_" + colName;
|
116
|
}
|
117
|
row.put(colName, value);
|
118
|
}
|
119
|
else {
|
120
|
if (colName.startsWith(vocabNameCol) || colName.startsWith(vocabValueCol)) {
|
121
|
//don't add it to the normal metadata
|
122
|
}
|
123
|
else {
|
124
|
metadataRow.put(colName, value);
|
125
|
}
|
126
|
}
|
127
|
//names
|
128
|
if (colName.startsWith(vocabNameCol)) {
|
129
|
List list = (List) vocabNames.get(pivotValue);
|
130
|
if (list == null) {
|
131
|
list = new ArrayList();
|
132
|
}
|
133
|
list.add(value);
|
134
|
vocabNames.put(pivotValue, list);
|
135
|
allVocabNames.add(value);
|
136
|
}
|
137
|
//values
|
138
|
if (colName.startsWith(vocabValueCol)) {
|
139
|
List list = (List) vocabValues.get(pivotValue);
|
140
|
if (list == null) {
|
141
|
list = new ArrayList();
|
142
|
}
|
143
|
list.add(value);
|
144
|
vocabValues.put(pivotValue, list);
|
145
|
allVocabValues.add(value);
|
146
|
}
|
147
|
}
|
148
|
}
|
149
|
//track the data columns - the values are junk
|
150
|
widestRow.putAll(row);
|
151
|
|
152
|
//put the row back (or maybe it's the first time)
|
153
|
table.put(id, row);
|
154
|
|
155
|
//put the metadata header back
|
156
|
headerRows.put(pivotValue, metadataRow);
|
157
|
|
158
|
}
|
159
|
|
160
|
/** Construct the table structure for returning **/
|
161
|
|
162
|
//now make it into a list
|
163
|
List retTable = new ArrayList();
|
164
|
|
165
|
//map keyed by metadata labels
|
166
|
OrderedMap metadataHeaders = new OrderedMap();
|
167
|
|
168
|
//do the data header - drives the other columns - based on widest entry
|
169
|
List header = new ArrayList(widestRow.keySet());
|
170
|
|
171
|
//do the metadata header rows (basically rotate them around)
|
172
|
Iterator headerIter = header.iterator();
|
173
|
String lastValue = "";
|
174
|
while (headerIter.hasNext()) {
|
175
|
String column = (String) headerIter.next();
|
176
|
//get the pivotValue part of column name
|
177
|
String pivotValue = column;
|
178
|
try {
|
179
|
pivotValue = column.substring(0, column.indexOf("_"));
|
180
|
}
|
181
|
catch (Exception e) {}
|
182
|
//look up the row from the metadata - keyed by pivot value
|
183
|
OrderedMap metadataRow = (OrderedMap) headerRows.get(pivotValue);
|
184
|
if (metadataRow != null) {
|
185
|
//go through the values
|
186
|
Iterator metadataIter = metadataRow.keySet().iterator();
|
187
|
while (metadataIter.hasNext()) {
|
188
|
String key = (String) metadataIter.next();
|
189
|
String value = (String) metadataRow.get(key);
|
190
|
OrderedMap newMetadataRow = (OrderedMap) metadataHeaders.get(key);
|
191
|
if (newMetadataRow == null) {
|
192
|
newMetadataRow = new OrderedMap();
|
193
|
}
|
194
|
//if it's the same as the last one, just use null value
|
195
|
if (lastValue.equals(pivotValue)) {
|
196
|
value = null;
|
197
|
}
|
198
|
newMetadataRow.put(column, value);
|
199
|
metadataHeaders.put(key, newMetadataRow);
|
200
|
}
|
201
|
}
|
202
|
|
203
|
lastValue = pivotValue;
|
204
|
|
205
|
}
|
206
|
|
207
|
//make metadata rows as list/arrays on the reteurn table
|
208
|
Iterator metadataLabelIter = metadataHeaders.keySet().iterator();
|
209
|
while (metadataLabelIter.hasNext()) {
|
210
|
String label = (String) metadataLabelIter.next();
|
211
|
OrderedMap row = (OrderedMap) metadataHeaders.get(label);
|
212
|
List rowValues = new ArrayList(row.values());
|
213
|
rowValues.add(0, label);
|
214
|
if (twoColumnMetadata) {
|
215
|
//add extra column
|
216
|
rowValues.add(1, null);
|
217
|
}
|
218
|
retTable.add(rowValues.toArray(new String[0]));
|
219
|
}
|
220
|
|
221
|
//create the special vocab matrix rows
|
222
|
List vocabTable = new ArrayList();
|
223
|
List uniqueVocabs = new ArrayList();
|
224
|
for (int i = 0; i < allVocabNames.size(); i++) {
|
225
|
List vocabRow = new ArrayList();
|
226
|
String vocabName = (String) allVocabNames.get(i);
|
227
|
String vocabValue = (String) allVocabValues.get(i);
|
228
|
String key = vocabName + "/" + vocabValue;
|
229
|
//check if we've processed this already, skip if so
|
230
|
if (uniqueVocabs.contains(key)) {
|
231
|
continue;
|
232
|
}
|
233
|
uniqueVocabs.add(key);
|
234
|
if (twoColumnMetadata) {
|
235
|
vocabRow.add(vocabName);
|
236
|
vocabRow.add(vocabValue);
|
237
|
}
|
238
|
else {
|
239
|
vocabRow.add(key);
|
240
|
}
|
241
|
//go through the questions now, again
|
242
|
headerIter = header.iterator();
|
243
|
while (headerIter.hasNext()) {
|
244
|
String column = (String) headerIter.next();
|
245
|
//get the pivotValue part of column name if it exists
|
246
|
String pivotValue = null;
|
247
|
try {
|
248
|
pivotValue = column.substring(0, column.indexOf("_"));
|
249
|
}
|
250
|
catch (Exception e) {}
|
251
|
if (pivotValue == null) {
|
252
|
continue;
|
253
|
}
|
254
|
//check to see if this question has that keyword
|
255
|
List names = (List) vocabNames.get(pivotValue);
|
256
|
List values = (List) vocabValues.get(pivotValue);
|
257
|
if (names != null && names.indexOf(vocabName) > -1 && names.indexOf(vocabName) == values.indexOf(vocabValue) ) {
|
258
|
vocabRow.add("true");
|
259
|
}
|
260
|
else {
|
261
|
vocabRow.add("false");
|
262
|
}
|
263
|
}
|
264
|
//put the row on
|
265
|
vocabTable.add(vocabRow.toArray(new String[0]));
|
266
|
}
|
267
|
|
268
|
//put the vocab matrix on the table
|
269
|
retTable.addAll(vocabTable);
|
270
|
|
271
|
if (twoColumnMetadata) {
|
272
|
//add column to data header row
|
273
|
header.add(1, null);
|
274
|
}
|
275
|
//put the data header row on the table
|
276
|
retTable.add(header.toArray(new String[0]));
|
277
|
|
278
|
//now the value rows in the table
|
279
|
Iterator rowIter = table.values().iterator();
|
280
|
int rowCount = 1;
|
281
|
while (rowIter.hasNext()) {
|
282
|
OrderedMap rowMap = (OrderedMap) rowIter.next();
|
283
|
List row = new ArrayList();
|
284
|
//iterate over the widest row's columns
|
285
|
Iterator columnIter = widestRow.keySet().iterator();
|
286
|
while (columnIter.hasNext()) {
|
287
|
Object key = columnIter.next();
|
288
|
Object value = rowMap.get(key);
|
289
|
//hide the value used for Ids - just increment row
|
290
|
if (key.equals(idColName) && omitIdValues) {
|
291
|
value = String.valueOf(rowCount);
|
292
|
}
|
293
|
row.add(value);
|
294
|
}
|
295
|
rowCount++;
|
296
|
if (twoColumnMetadata) {
|
297
|
//add extra column
|
298
|
row.add(1, null);
|
299
|
}
|
300
|
retTable.add(row.toArray(new String[0]));
|
301
|
}
|
302
|
|
303
|
return retTable;
|
304
|
}
|
305
|
%><%!
|
306
|
private void handleDataquery(
|
307
|
Hashtable<String, String[]> params,
|
308
|
HttpServletResponse response,
|
309
|
String sessionId) throws PropertyNotFoundException, IOException {
|
310
|
|
311
|
DataQuery dq = null;
|
312
|
if (sessionId != null) {
|
313
|
dq = new DataQuery(sessionId);
|
314
|
}
|
315
|
else {
|
316
|
dq = new DataQuery();
|
317
|
}
|
318
|
|
319
|
String dataqueryXML = (params.get("dataquery"))[0];
|
320
|
|
321
|
ResultSet rs = null;
|
322
|
try {
|
323
|
rs = dq.executeQuery(dataqueryXML);
|
324
|
} catch (Exception e) {
|
325
|
//probably need to do something here
|
326
|
e.printStackTrace();
|
327
|
return;
|
328
|
}
|
329
|
|
330
|
//process the result set
|
331
|
String qformat = "csv";
|
332
|
String[] temp = params.get("qformat");
|
333
|
if (temp != null && temp.length > 0) {
|
334
|
qformat = temp[0];
|
335
|
}
|
336
|
String fileName = "query-results." + qformat;
|
337
|
|
338
|
boolean transpose = false;
|
339
|
temp = params.get("transpose");
|
340
|
if (temp != null && temp.length > 0) {
|
341
|
transpose = Boolean.parseBoolean(temp[0]);
|
342
|
}
|
343
|
int observation = 0;
|
344
|
temp = params.get("observation");
|
345
|
if (temp != null && temp.length > 0) {
|
346
|
observation = Integer.parseInt(temp[0]);
|
347
|
}
|
348
|
int pivot = 0;
|
349
|
temp = params.get("pivot");
|
350
|
if (temp != null && temp.length > 0) {
|
351
|
pivot = Integer.parseInt(temp[0]);
|
352
|
}
|
353
|
|
354
|
//get the results as csv file
|
355
|
if (qformat != null && qformat.equalsIgnoreCase("csv")) {
|
356
|
response.setContentType("text/csv");
|
357
|
//response.setContentType("application/csv");
|
358
|
response.setHeader("Content-Disposition", "attachment; filename=" + fileName);
|
359
|
|
360
|
Writer writer = new OutputStreamWriter(response.getOutputStream());
|
361
|
//CSVWriter csv = new CSVWriter(writer, CSVWriter.DEFAULT_SEPARATOR, CSVWriter.NO_QUOTE_CHARACTER);
|
362
|
CSVWriter csv = new CSVWriter(writer, CSVWriter.DEFAULT_SEPARATOR, CSVWriter.DEFAULT_QUOTE_CHARACTER, CSVWriter.DEFAULT_ESCAPE_CHARACTER);
|
363
|
try {
|
364
|
if (transpose) {
|
365
|
List pivotCols = new ArrayList();
|
366
|
pivotCols.add("studentid");
|
367
|
pivotCols.add("score");
|
368
|
pivotCols.add("response");
|
369
|
List transposedTable = transpose(rs, observation, pivot, pivotCols, true);
|
370
|
csv.writeAll(transposedTable);
|
371
|
} else {
|
372
|
csv.writeAll(rs, true);
|
373
|
}
|
374
|
|
375
|
csv.flush();
|
376
|
response.flushBuffer();
|
377
|
|
378
|
rs.close();
|
379
|
|
380
|
} catch (SQLException e) {
|
381
|
e.printStackTrace();
|
382
|
}
|
383
|
|
384
|
return;
|
385
|
}
|
386
|
|
387
|
}
|
388
|
%>
|