1 |
50066
|
jochen.sch
|
/**
|
2 |
|
|
*
|
3 |
52983
|
andreas.cz
|
*
|
4 |
|
|
* log.debug(...) equal to log.trace(...) in the application-logs
|
5 |
52997
|
andreas.cz
|
*
|
6 |
|
|
* known bug: at resumptionType 'discover' if the (resultTotal % resultSizeValue) == 0 the collecting fails -> change the resultSizeValue
|
7 |
50066
|
jochen.sch
|
*/
|
8 |
|
|
package eu.dnetlib.data.collector.plugins.rest;
|
9 |
|
|
|
10 |
|
|
import java.io.InputStream;
|
11 |
50582
|
jochen.sch
|
import java.io.StringWriter;
|
12 |
50066
|
jochen.sch
|
import java.net.URL;
|
13 |
|
|
import java.util.Iterator;
|
14 |
|
|
import java.util.Queue;
|
15 |
50582
|
jochen.sch
|
import java.util.concurrent.PriorityBlockingQueue;
|
16 |
50066
|
jochen.sch
|
|
17 |
|
|
import javax.xml.transform.OutputKeys;
|
18 |
|
|
import javax.xml.transform.Transformer;
|
19 |
|
|
import javax.xml.transform.TransformerConfigurationException;
|
20 |
|
|
import javax.xml.transform.TransformerFactory;
|
21 |
50582
|
jochen.sch
|
import javax.xml.transform.dom.DOMSource;
|
22 |
|
|
import javax.xml.transform.stream.StreamResult;
|
23 |
50066
|
jochen.sch
|
import javax.xml.xpath.XPath;
|
24 |
|
|
import javax.xml.xpath.XPathConstants;
|
25 |
|
|
import javax.xml.xpath.XPathExpression;
|
26 |
|
|
import javax.xml.xpath.XPathExpressionException;
|
27 |
|
|
import javax.xml.xpath.XPathFactory;
|
28 |
|
|
|
29 |
|
|
import org.apache.commons.io.IOUtils;
|
30 |
50662
|
alessia.ba
|
import org.apache.commons.lang3.StringUtils;
|
31 |
50582
|
jochen.sch
|
import org.apache.commons.logging.Log;
|
32 |
|
|
import org.apache.commons.logging.LogFactory;
|
33 |
50066
|
jochen.sch
|
import org.w3c.dom.Node;
|
34 |
50582
|
jochen.sch
|
import org.w3c.dom.NodeList;
|
35 |
50066
|
jochen.sch
|
import org.xml.sax.InputSource;
|
36 |
|
|
|
37 |
50582
|
jochen.sch
|
import eu.dnetlib.data.collector.plugins.oai.OaiIterator;
|
38 |
53116
|
andreas.cz
|
import eu.dnetlib.data.collector.plugins.oai.engine.XmlCleaner;
|
39 |
50582
|
jochen.sch
|
import eu.dnetlib.data.collector.rmi.CollectorServiceException;
|
40 |
|
|
|
41 |
50066
|
jochen.sch
|
/**
|
42 |
52970
|
andreas.cz
|
* @author Jochen Schirrwagen, Aenne Loehden, Andreas Czerniak
|
43 |
52997
|
andreas.cz
|
* @date 2018-09-03
|
44 |
50066
|
jochen.sch
|
*
|
45 |
|
|
*/
|
46 |
|
|
public class RestIterator implements Iterator<String> {
|
47 |
|
|
|
48 |
52971
|
andreas.cz
|
// TODO: clean up the comments of replaced source code
|
49 |
50582
|
jochen.sch
|
private static final Log log = LogFactory.getLog(OaiIterator.class); // NOPMD by marko on 11/24/08 5:02 PM
|
50 |
|
|
|
51 |
50066
|
jochen.sch
|
private static final String wrapName = "recordWrap";
|
52 |
|
|
private String baseUrl;
|
53 |
|
|
private String resumptionType;
|
54 |
|
|
private String resumptionParam;
|
55 |
|
|
private String resultFormatValue;
|
56 |
|
|
private String queryParams;
|
57 |
52970
|
andreas.cz
|
private int resultSizeValue;
|
58 |
50066
|
jochen.sch
|
private int resumptionInt = 0; // integer resumption token (first record to harvest)
|
59 |
|
|
private int resultTotal = -1;
|
60 |
|
|
private String resumptionStr = Integer.toString(resumptionInt); // string resumption token (first record to harvest or token scanned from results)
|
61 |
|
|
private InputStream resultStream;
|
62 |
|
|
private Transformer transformer;
|
63 |
|
|
private XPath xpath;
|
64 |
50582
|
jochen.sch
|
private String query;
|
65 |
50066
|
jochen.sch
|
private XPathExpression xprResultTotalPath;
|
66 |
|
|
private XPathExpression xprResumptionPath;
|
67 |
50582
|
jochen.sch
|
private XPathExpression xprEntity;
|
68 |
50066
|
jochen.sch
|
private String queryFormat;
|
69 |
|
|
private String querySize;
|
70 |
50582
|
jochen.sch
|
private Queue<String> recordQueue = new PriorityBlockingQueue<String>();
|
71 |
52983
|
andreas.cz
|
private int discoverResultSize = 0;
|
72 |
53183
|
andreas.cz
|
private int pagination = 1;
|
73 |
52983
|
andreas.cz
|
|
74 |
50066
|
jochen.sch
|
public RestIterator(
|
75 |
|
|
final String baseUrl,
|
76 |
|
|
final String resumptionType,
|
77 |
|
|
final String resumptionParam,
|
78 |
|
|
final String resumptionXpath,
|
79 |
|
|
final String resultTotalXpath,
|
80 |
|
|
final String resultFormatParam,
|
81 |
|
|
final String resultFormatValue,
|
82 |
|
|
final String resultSizeParam,
|
83 |
52971
|
andreas.cz
|
final String resultSizeValueStr,
|
84 |
50582
|
jochen.sch
|
final String queryParams,
|
85 |
|
|
final String entityXpath
|
86 |
50066
|
jochen.sch
|
) {
|
87 |
|
|
this.baseUrl = baseUrl;
|
88 |
|
|
this.resumptionType = resumptionType;
|
89 |
|
|
this.resumptionParam = resumptionParam;
|
90 |
|
|
this.resultFormatValue = resultFormatValue;
|
91 |
|
|
this.queryParams = queryParams;
|
92 |
52971
|
andreas.cz
|
this.resultSizeValue = Integer.valueOf(resultSizeValueStr);
|
93 |
|
|
|
94 |
50584
|
claudio.at
|
queryFormat = StringUtils.isNotBlank(resultFormatParam) ? "&" + resultFormatParam + "=" + resultFormatValue : "";
|
95 |
52971
|
andreas.cz
|
querySize = StringUtils.isNotBlank(resultSizeParam) ? "&" + resultSizeParam + "=" + resultSizeValueStr : "";
|
96 |
50066
|
jochen.sch
|
|
97 |
|
|
try {
|
98 |
50582
|
jochen.sch
|
initXmlTransformation(resultTotalXpath, resumptionXpath, entityXpath);
|
99 |
50584
|
claudio.at
|
} catch(Exception e) {
|
100 |
|
|
throw new IllegalStateException("xml transformation init failed: " + e.getMessage());
|
101 |
50066
|
jochen.sch
|
}
|
102 |
|
|
initQueue();
|
103 |
|
|
}
|
104 |
|
|
|
105 |
50582
|
jochen.sch
|
private void initXmlTransformation(String resultTotalXpath, String resumptionXpath, String entityXpath) throws TransformerConfigurationException, XPathExpressionException{
|
106 |
50066
|
jochen.sch
|
transformer = TransformerFactory.newInstance().newTransformer();
|
107 |
52970
|
andreas.cz
|
transformer.setOutputProperty(OutputKeys.INDENT,"yes");
|
108 |
|
|
transformer.setOutputProperty("{http://xml.apache.org/xslt}indent-amount","3");
|
109 |
|
|
xpath = XPathFactory.newInstance().newXPath();
|
110 |
50066
|
jochen.sch
|
xprResultTotalPath = xpath.compile(resultTotalXpath);
|
111 |
52970
|
andreas.cz
|
xprResumptionPath = xpath.compile(StringUtils.isBlank(resumptionXpath) ? "/" : resumptionXpath);
|
112 |
|
|
xprEntity = xpath.compile(entityXpath);
|
113 |
50066
|
jochen.sch
|
}
|
114 |
|
|
|
115 |
|
|
private void initQueue() {
|
116 |
52997
|
andreas.cz
|
// query = baseUrl + "?" + queryParams + querySize + "&" + resumptionParam + "=" + resumptionStr + queryFormat;
|
117 |
|
|
// change at 2018-08-30 and remove resumptionParam&-Type from "query" URL because first request starts mostly at the beginning.
|
118 |
|
|
query = baseUrl + "?" + queryParams + querySize + queryFormat;
|
119 |
|
|
|
120 |
50066
|
jochen.sch
|
}
|
121 |
|
|
|
122 |
|
|
private void disconnect() {
|
123 |
|
|
// TODO close inputstream
|
124 |
|
|
}
|
125 |
|
|
|
126 |
|
|
/* (non-Javadoc)
|
127 |
|
|
* @see java.util.Iterator#hasNext()
|
128 |
|
|
*/
|
129 |
|
|
@Override
|
130 |
|
|
public boolean hasNext() {
|
131 |
50582
|
jochen.sch
|
if (recordQueue.isEmpty() && query.isEmpty()) {
|
132 |
50066
|
jochen.sch
|
disconnect();
|
133 |
|
|
return false;
|
134 |
|
|
} else {
|
135 |
|
|
return true;
|
136 |
|
|
}
|
137 |
|
|
}
|
138 |
|
|
|
139 |
|
|
/* (non-Javadoc)
|
140 |
|
|
* @see java.util.Iterator#next()
|
141 |
|
|
*/
|
142 |
|
|
@Override
|
143 |
|
|
public String next() {
|
144 |
50582
|
jochen.sch
|
synchronized (recordQueue) {
|
145 |
|
|
while (recordQueue.isEmpty() && !query.isEmpty() ) {
|
146 |
|
|
try {
|
147 |
52983
|
andreas.cz
|
log.info("get Query: " + query);
|
148 |
50582
|
jochen.sch
|
query = downloadPage(query);
|
149 |
53116
|
andreas.cz
|
log.debug("next queryURL from downloadPage(): " + query);
|
150 |
50584
|
claudio.at
|
} catch(CollectorServiceException e) {
|
151 |
52983
|
andreas.cz
|
log.debug("CollectorPlugin.next()-Exception: " + e);
|
152 |
50582
|
jochen.sch
|
throw new RuntimeException(e);
|
153 |
|
|
}
|
154 |
|
|
}
|
155 |
|
|
return recordQueue.poll();
|
156 |
|
|
}
|
157 |
|
|
}
|
158 |
|
|
|
159 |
|
|
|
160 |
|
|
/*
|
161 |
|
|
* download page and return nextQuery
|
162 |
|
|
*/
|
163 |
|
|
private String downloadPage(String query) throws CollectorServiceException{
|
164 |
50066
|
jochen.sch
|
String resultJson;
|
165 |
53116
|
andreas.cz
|
String resultXml = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>";
|
166 |
50582
|
jochen.sch
|
String nextQuery = "";
|
167 |
53116
|
andreas.cz
|
String emptyXml = resultXml + "<"+wrapName+"></"+wrapName+">";
|
168 |
52979
|
andreas.cz
|
Node resultNode = null;
|
169 |
52982
|
andreas.cz
|
NodeList nodeList = null;
|
170 |
52983
|
andreas.cz
|
String qUrlArgument = "";
|
171 |
|
|
int urlOldResumptionSize = 0;
|
172 |
52979
|
andreas.cz
|
|
173 |
50066
|
jochen.sch
|
try {
|
174 |
52983
|
andreas.cz
|
URL qUrl = new URL(query);
|
175 |
|
|
|
176 |
|
|
resultStream = qUrl.openStream();
|
177 |
53116
|
andreas.cz
|
if("json".equals(resultFormatValue.toLowerCase())){
|
178 |
|
|
|
179 |
50066
|
jochen.sch
|
resultJson = IOUtils.toString(resultStream,"UTF-8");
|
180 |
52970
|
andreas.cz
|
resultJson = syntaxConvertJsonKeyNamens(resultJson);
|
181 |
50066
|
jochen.sch
|
org.json.JSONObject jsonObject = new org.json.JSONObject(resultJson);
|
182 |
53116
|
andreas.cz
|
resultXml += org.json.XML.toString(jsonObject,wrapName); // wrap xml in single root element
|
183 |
|
|
log.trace("before inputStream: " + resultXml);
|
184 |
53123
|
andreas.cz
|
resultXml = XmlCleaner.cleanAllEntities(resultXml);
|
185 |
|
|
log.trace("after cleaning: " + resultXml);
|
186 |
50066
|
jochen.sch
|
resultStream = IOUtils.toInputStream(resultXml,"UTF-8");
|
187 |
|
|
}
|
188 |
|
|
|
189 |
52979
|
andreas.cz
|
if (!(emptyXml.toLowerCase()).equals(resultXml.toLowerCase())) {
|
190 |
|
|
resultNode = (Node) xpath.evaluate("/", new InputSource(resultStream), XPathConstants.NODE);
|
191 |
|
|
nodeList = (NodeList) xprEntity.evaluate(resultNode, XPathConstants.NODESET);
|
192 |
52982
|
andreas.cz
|
log.debug("nodeList.length: " + nodeList.getLength());
|
193 |
52979
|
andreas.cz
|
for (int i = 0; i < nodeList.getLength(); i++) {
|
194 |
|
|
StringWriter sw = new StringWriter();
|
195 |
|
|
transformer.transform(new DOMSource(nodeList.item(i)), new StreamResult(sw));
|
196 |
|
|
recordQueue.add(sw.toString());
|
197 |
|
|
}
|
198 |
52982
|
andreas.cz
|
} else { log.info("resultXml is equal with emptyXml"); }
|
199 |
|
|
|
200 |
50066
|
jochen.sch
|
resumptionInt += resultSizeValue;
|
201 |
52970
|
andreas.cz
|
|
202 |
52971
|
andreas.cz
|
/* replaced by switch statement as follow
|
203 |
|
|
if("scan".equals(resumptionType.toLowerCase())) { resumptionStr = xprResumptionPath.evaluate(resultNode);}
|
204 |
|
|
if("count".equals(resumptionType.toLowerCase())){ resumptionStr = Integer.toString(resumptionInt); }
|
205 |
|
|
*/
|
206 |
52970
|
andreas.cz
|
switch(resumptionType.toLowerCase()) {
|
207 |
52997
|
andreas.cz
|
case "scan": // read of resumptionToken , evaluate next results, e.g. OAI, iterate over items
|
208 |
52970
|
andreas.cz
|
resumptionStr = xprResumptionPath.evaluate(resultNode);
|
209 |
|
|
break;
|
210 |
52979
|
andreas.cz
|
|
211 |
52997
|
andreas.cz
|
case "count": // begin at one step for all records, iterate over items
|
212 |
52970
|
andreas.cz
|
resumptionStr = Integer.toString(resumptionInt);
|
213 |
|
|
break;
|
214 |
52979
|
andreas.cz
|
|
215 |
53116
|
andreas.cz
|
case "discover": // size of result items unknown, iterate over items (for openDOAR - 201808)
|
216 |
52997
|
andreas.cz
|
if (resultSizeValue < 2 ) {throw new CollectorServiceException("Mode: discover, Param 'resultSizeValue' is less than 2");}
|
217 |
52983
|
andreas.cz
|
qUrlArgument = qUrl.getQuery();
|
218 |
|
|
String[] arrayQUrlArgument = qUrlArgument.split("&");
|
219 |
|
|
for(String arrayUrlArgStr : arrayQUrlArgument ) {
|
220 |
|
|
if(arrayUrlArgStr.startsWith(resumptionParam)) {
|
221 |
|
|
String[] resumptionKeyValue = arrayUrlArgStr.split("=");
|
222 |
|
|
urlOldResumptionSize = Integer.parseInt(resumptionKeyValue[1]);
|
223 |
|
|
log.debug("discover OldResumptionSize from Url: " + urlOldResumptionSize);
|
224 |
|
|
}
|
225 |
|
|
}
|
226 |
|
|
|
227 |
|
|
//
|
228 |
|
|
if( ( (emptyXml.toLowerCase()).equals(resultXml.toLowerCase()) )
|
229 |
|
|
|| ( (nodeList != null) && (nodeList.getLength() < resultSizeValue) )
|
230 |
|
|
) {
|
231 |
52982
|
andreas.cz
|
// resumptionStr = "";
|
232 |
52983
|
andreas.cz
|
if(nodeList != null) { discoverResultSize += nodeList.getLength(); }
|
233 |
52982
|
andreas.cz
|
resultTotal = discoverResultSize;
|
234 |
52970
|
andreas.cz
|
} else {
|
235 |
|
|
resumptionStr = Integer.toString(resumptionInt);
|
236 |
|
|
resultTotal = resumptionInt+1;
|
237 |
52982
|
andreas.cz
|
if(nodeList != null) { discoverResultSize += nodeList.getLength(); }
|
238 |
52970
|
andreas.cz
|
}
|
239 |
52982
|
andreas.cz
|
log.info("discoverResultSize: " + discoverResultSize);
|
240 |
52970
|
andreas.cz
|
break;
|
241 |
52979
|
andreas.cz
|
|
242 |
53183
|
andreas.cz
|
case "pagination": // pagination, iterate over pages
|
243 |
|
|
case "page":
|
244 |
|
|
pagination += 1;
|
245 |
|
|
if (nodeList != null) {
|
246 |
|
|
discoverResultSize += nodeList.getLength();
|
247 |
|
|
} else {
|
248 |
|
|
resultTotal = discoverResultSize;
|
249 |
|
|
pagination = discoverResultSize;
|
250 |
|
|
}
|
251 |
|
|
resumptionInt = pagination;
|
252 |
|
|
resumptionStr = Integer.toString(resumptionInt);
|
253 |
52997
|
andreas.cz
|
break;
|
254 |
|
|
|
255 |
52979
|
andreas.cz
|
default: // otherwise: abort
|
256 |
52982
|
andreas.cz
|
// resultTotal = resumptionInt;
|
257 |
52971
|
andreas.cz
|
break;
|
258 |
52970
|
andreas.cz
|
}
|
259 |
52971
|
andreas.cz
|
|
260 |
50066
|
jochen.sch
|
if (resultTotal == -1) {
|
261 |
|
|
resultTotal = Integer.parseInt(xprResultTotalPath.evaluate(resultNode));
|
262 |
52979
|
andreas.cz
|
log.info("resultTotal was -1 is now: " + resultTotal);
|
263 |
50066
|
jochen.sch
|
}
|
264 |
50584
|
claudio.at
|
log.info("resultTotal: " + resultTotal);
|
265 |
|
|
log.info("resInt: " + resumptionInt);
|
266 |
50066
|
jochen.sch
|
if (resumptionInt < resultTotal) {
|
267 |
50582
|
jochen.sch
|
nextQuery = baseUrl + "?" + queryParams + querySize + "&" + resumptionParam + "=" + resumptionStr + queryFormat;
|
268 |
50584
|
claudio.at
|
} else
|
269 |
50582
|
jochen.sch
|
nextQuery = "";
|
270 |
52971
|
andreas.cz
|
|
271 |
52979
|
andreas.cz
|
log.debug("nextQueryUrl: " + nextQuery);
|
272 |
50582
|
jochen.sch
|
return nextQuery;
|
273 |
50066
|
jochen.sch
|
|
274 |
50584
|
claudio.at
|
} catch(Exception e) {
|
275 |
|
|
log.error(e);
|
276 |
|
|
throw new IllegalStateException("collection failed: " + e.getMessage());
|
277 |
50066
|
jochen.sch
|
}
|
278 |
|
|
}
|
279 |
52970
|
andreas.cz
|
|
280 |
|
|
/**
|
281 |
52997
|
andreas.cz
|
* convert in JSON-KeyName 'whitespace(s)' to '_' and '/' to '_', '(' and ')' to ''
|
282 |
|
|
* check W3C XML syntax: https://www.w3.org/TR/2006/REC-xml11-20060816/#sec-starttags for valid tag names
|
283 |
|
|
* and work-around for the JSON to XML converting of org.json.XML-package.
|
284 |
52970
|
andreas.cz
|
*
|
285 |
53163
|
andreas.cz
|
* known bugs: doesn't prevent "key name":" ["sexy name",": penari","erotic dance"],
|
286 |
|
|
*
|
287 |
52970
|
andreas.cz
|
* @param jsonInput
|
288 |
52997
|
andreas.cz
|
* @return convertedJsonKeynameOutput
|
289 |
52970
|
andreas.cz
|
*/
|
290 |
|
|
private String syntaxConvertJsonKeyNamens(String jsonInput) {
|
291 |
50066
|
jochen.sch
|
|
292 |
52979
|
andreas.cz
|
log.trace("before convertJsonKeyNames: " + jsonInput);
|
293 |
53116
|
andreas.cz
|
// pre-clean json - rid spaces of element names (misinterpreted as elements with attributes in xml)
|
294 |
52970
|
andreas.cz
|
// replace ' 's in JSON Namens with '_'
|
295 |
|
|
while (jsonInput.matches(".*\"([^\"]*)\\s+([^\"]*)\":.*")) {
|
296 |
|
|
jsonInput = jsonInput.replaceAll("\"([^\"]*)\\s+([^\"]*)\":", "\"$1_$2\":");
|
297 |
|
|
}
|
298 |
|
|
|
299 |
|
|
// replace forward-slash (sign '/' ) in JSON Names with '_'
|
300 |
|
|
while (jsonInput.matches(".*\"([^\"]*)/([^\"]*)\":.*")) {
|
301 |
|
|
jsonInput = jsonInput.replaceAll("\"([^\"]*)/([^\"]*)\":", "\"$1_$2\":");
|
302 |
|
|
}
|
303 |
|
|
|
304 |
|
|
// replace '(' in JSON Names with ''
|
305 |
|
|
while (jsonInput.matches(".*\"([^\"]*)[(]([^\"]*)\":.*")) {
|
306 |
|
|
jsonInput = jsonInput.replaceAll("\"([^\"]*)[(]([^\"]*)\":", "\"$1$2\":");
|
307 |
|
|
}
|
308 |
52997
|
andreas.cz
|
|
309 |
52970
|
andreas.cz
|
// replace ')' in JSON Names with ''
|
310 |
|
|
while (jsonInput.matches(".*\"([^\"]*)[)]([^\"]*)\":.*")) {
|
311 |
|
|
jsonInput = jsonInput.replaceAll("\"([^\"]*)[)]([^\"]*)\":", "\"$1$2\":");
|
312 |
|
|
}
|
313 |
|
|
|
314 |
52997
|
andreas.cz
|
// replace startNumbers in JSON Keynames with 'n_'
|
315 |
|
|
while (jsonInput.matches(".*\"([^\"][0-9])([^\"]*)\":.*")) {
|
316 |
|
|
jsonInput = jsonInput.replaceAll("\"([^\"][0-9])([^\"]*)\":", "\"n_$1$2\":");
|
317 |
|
|
}
|
318 |
|
|
|
319 |
|
|
// replace ':' between number like '2018-08-28T11:05:00Z' in JSON keynames with ''
|
320 |
|
|
while (jsonInput.matches(".*\"([^\"]*[0-9]):([0-9][^\"]*)\":.*")) {
|
321 |
|
|
jsonInput = jsonInput.replaceAll("\"([^\"]*[0-9]):([0-9][^\"]*)\":", "\"$1$2\":");
|
322 |
|
|
}
|
323 |
|
|
|
324 |
53163
|
andreas.cz
|
// replace ',' in JSON Keynames with '.' to prevent , in xml tagnames.
|
325 |
|
|
// while (jsonInput.matches(".*\"([^\"]*),([^\"]*)\":.*")) {
|
326 |
|
|
// jsonInput = jsonInput.replaceAll("\"([^\"]*),([^\"]*)\":", "\"$1.$2\":");
|
327 |
|
|
// }
|
328 |
52997
|
andreas.cz
|
|
329 |
|
|
// replace '=' in JSON Keynames with '-'
|
330 |
|
|
while (jsonInput.matches(".*\"([^\"]*)=([^\"]*)\":.*")) {
|
331 |
|
|
jsonInput = jsonInput.replaceAll("\"([^\"]*)=([^\"]*)\":", "\"$1-$2\":");
|
332 |
|
|
}
|
333 |
|
|
|
334 |
52979
|
andreas.cz
|
log.trace("after syntaxConvertJsonKeyNames: " + jsonInput);
|
335 |
52970
|
andreas.cz
|
return jsonInput;
|
336 |
|
|
}
|
337 |
53123
|
andreas.cz
|
|
338 |
|
|
/**
|
339 |
|
|
*
|
340 |
53163
|
andreas.cz
|
* https://www.w3.org/TR/REC-xml/#charencoding shows character enoding in entities
|
341 |
53123
|
andreas.cz
|
* *
|
342 |
53163
|
andreas.cz
|
* @param bufferStr - XML string
|
343 |
53123
|
andreas.cz
|
* @return
|
344 |
|
|
*/
|
345 |
53163
|
andreas.cz
|
private static String cleanUnwantedJsonCharsInXmlTagnames( String bufferStr ) {
|
346 |
|
|
|
347 |
|
|
while ( bufferStr.matches(".*<([^<>].*),(.)>.*") ) {
|
348 |
|
|
bufferStr = bufferStr.replaceAll("<([^<>.*),(.*)>", "<$1$2>");
|
349 |
|
|
}
|
350 |
52970
|
andreas.cz
|
|
351 |
53123
|
andreas.cz
|
// replace [#x10-#x1f] with ''
|
352 |
53163
|
andreas.cz
|
// while (bufferStr.matches(".*[0-9a-f].*")) {
|
353 |
|
|
// bufferStr = bufferStr.replaceAll("([0-9a-fA-F])", "");
|
354 |
|
|
// }
|
355 |
53123
|
andreas.cz
|
|
356 |
|
|
return bufferStr;
|
357 |
|
|
}
|
358 |
|
|
|
359 |
50066
|
jochen.sch
|
}
|