Revision 1553
Added by Jing Tao over 21 years ago
| src/edu/ucsb/nceas/metacat/ContentTypeProvider.java | ||
|---|---|---|
| 34 | 34 |
import java.util.Hashtable; |
| 35 | 35 |
import java.util.Vector; |
| 36 | 36 |
|
| 37 |
import org.apache.xpath.objects.XObject; |
|
| 37 | 38 |
import org.apache.xpath.XPathAPI; |
| 38 | 39 |
import org.apache.xerces.parsers.DOMParser; |
| 40 |
import org.apache.xerces.dom.DocumentTypeImpl; |
|
| 39 | 41 |
import org.w3c.dom.Attr; |
| 40 | 42 |
import org.w3c.dom.NamedNodeMap; |
| 41 | 43 |
import org.w3c.dom.NodeList; |
| ... | ... | |
| 43 | 45 |
import org.w3c.dom.Node; |
| 44 | 46 |
import org.w3c.dom.NodeList; |
| 45 | 47 |
import org.w3c.dom.DocumentType; |
| 46 |
import org.apache.xerces.dom.DocumentTypeImpl; |
|
| 47 |
import org.apache.xpath.objects.XObject; |
|
| 48 |
import org.xml.sax.InputSource; |
|
| 48 | 49 |
|
| 50 |
import javax.xml.parsers.DocumentBuilder; |
|
| 51 |
import javax.xml.parsers.DocumentBuilderFactory; |
|
| 52 |
import javax.xml.parsers.ParserConfigurationException; |
|
| 53 |
import javax.xml.transform.*; |
|
| 54 |
import javax.xml.transform.stream.*; |
|
| 55 |
import javax.xml.transform.dom.*; |
|
| 56 |
|
|
| 49 | 57 |
import org.ecoinformatics.eml.EMLParser; |
| 50 | 58 |
/** |
| 51 | 59 |
* This class will figure out which content type it is for a given data file. |
| ... | ... | |
| 73 | 81 |
private String HTML = "HTML"; |
| 74 | 82 |
private String HTMLTYPE = "text/html"; |
| 75 | 83 |
private String GIF = "gif"; |
| 84 |
private String JPEG = "jpeg"; |
|
| 85 |
private String JPEGTYPE = "image/jpeg"; |
|
| 76 | 86 |
private String GIFTYPE = "image/gif"; |
| 77 | 87 |
private String BMP = "bmp"; |
| 78 | 88 |
private String BMPTYPE = "image/bmp"; |
| ... | ... | |
| 85 | 95 |
|
| 86 | 96 |
private String ENTITYDOCTYPE = "entitydoctype"; |
| 87 | 97 |
private String PHYSICALDOCTYPE = "physicaldoctype"; |
| 98 |
private String EML2DOCTYPE = "eml2namespace"; |
|
| 99 |
private String DATAFORMAT = "dataFormat"; |
|
| 100 |
private String TEXTFORMAT = "textFormat"; |
|
| 101 |
private String EXTENALFORMAT = "externallyDefinedFormat"; |
|
| 102 |
private String FORMATNAME = "formatName"; |
|
| 103 |
private String BINARYRASTERFORMAT = "binaryRasterFormat"; |
|
| 88 | 104 |
|
| 105 |
private String DATAFILEPATH ="//physical/distribution/online/url"; |
|
| 89 | 106 |
|
| 90 | 107 |
/** |
| 91 | 108 |
* Constructor of ContentTypeProvider |
| ... | ... | |
| 124 | 141 |
else if (packageType.equals(EML2)) |
| 125 | 142 |
{
|
| 126 | 143 |
// for eml2 package |
| 144 |
// get eml document for data file |
|
| 145 |
String eml2Docid = getTargetDocIdForBeta(docLists, EML2DOCTYPE); |
|
| 146 |
findContentTypeInEML2(eml2Docid); |
|
| 127 | 147 |
|
| 128 | 148 |
} |
| 129 | 149 |
|
| ... | ... | |
| 135 | 155 |
return contentType; |
| 136 | 156 |
}//getContentType |
| 137 | 157 |
|
| 158 |
/* Method to find content type base on data format*/ |
|
| 159 |
private void findContentTypeInEML2(String eml2DocId) |
|
| 160 |
{
|
|
| 161 |
DocumentImpl xmlDoc = null; |
|
| 162 |
String xmlString = null; |
|
| 163 |
StringReader read = null; |
|
| 164 |
InputSource in = null; |
|
| 165 |
DocumentBuilderFactory dfactory = null; |
|
| 166 |
Document doc = null; |
|
| 167 |
// create xml document |
|
| 168 |
try |
|
| 169 |
{
|
|
| 170 |
xmlDoc = new DocumentImpl(eml2DocId); |
|
| 171 |
xmlString = xmlDoc.toString(); |
|
| 172 |
// create dom tree |
|
| 173 |
read = new StringReader(xmlString); |
|
| 174 |
in = new InputSource(read); |
|
| 175 |
dfactory = DocumentBuilderFactory.newInstance(); |
|
| 176 |
dfactory.setNamespaceAware(false); |
|
| 177 |
doc = dfactory.newDocumentBuilder().parse(in); |
|
| 178 |
} |
|
| 179 |
catch (Exception e) |
|
| 180 |
{
|
|
| 181 |
// if faild, set default value |
|
| 182 |
contentType = DEFAULTCONTENTTYPE; |
|
| 183 |
MetaCatUtil.debugMessage("Error in ContentTypeProvider." +
|
|
| 184 |
"findContentTypeInEML2()" + e.getMessage(), 30); |
|
| 185 |
return; |
|
| 186 |
} |
|
| 187 |
Node dataFormatNode = findDataFormatNodeInEML2(doc, DATAFILEPATH, |
|
| 188 |
dataFileId); |
|
| 189 |
NodeList childList = dataFormatNode.getChildNodes(); |
|
| 190 |
// go through childList |
|
| 191 |
for (int i = 0; i<childList.getLength(); i++) |
|
| 192 |
{
|
|
| 193 |
Node child = childList.item(i); |
|
| 194 |
|
|
| 195 |
// if has text format child set to text/plain |
|
| 196 |
if (child.getNodeName() != null && child.getNodeName().equals(TEXTFORMAT)) |
|
| 197 |
{
|
|
| 198 |
MetaCatUtil.debugMessage("in text format", 35);
|
|
| 199 |
contentType = TEXTYPE; |
|
| 200 |
} |
|
| 201 |
|
|
| 202 |
//external format |
|
| 203 |
if (child.getNodeName() != null && child.getNodeName().equals(EXTENALFORMAT)) |
|
| 204 |
{
|
|
| 205 |
MetaCatUtil.debugMessage("in external format ", 35);
|
|
| 206 |
String format = getTextValueForGivenChildTag(child, FORMATNAME); |
|
| 207 |
MetaCatUtil.debugMessage("The format is: "+format, 35);
|
|
| 208 |
// if we can find the format in the contentTypeHash table |
|
| 209 |
contentType = (String)lookUpContentType(format); |
|
| 210 |
if (contentType == null) |
|
| 211 |
{
|
|
| 212 |
contentType = BINARYTYPE; |
|
| 213 |
} |
|
| 214 |
} |
|
| 215 |
|
|
| 216 |
// binaryRasterFormat |
|
| 217 |
if (child.getNodeName() != null && child.getNodeName(). |
|
| 218 |
equals(BINARYRASTERFORMAT)) |
|
| 219 |
{
|
|
| 220 |
contentType = BINARYTYPE; |
|
| 221 |
}//if |
|
| 222 |
}//for |
|
| 223 |
//if contentype still be null, set default value |
|
| 224 |
if (contentType == null) |
|
| 225 |
{
|
|
| 226 |
contentType = DEFAULTCONTENTTYPE; |
|
| 227 |
} |
|
| 228 |
} |
|
| 229 |
|
|
| 230 |
/* Method get text value of given child tagname*/ |
|
| 231 |
private String getTextValueForGivenChildTag(Node parentNode, |
|
| 232 |
String childTagName) |
|
| 233 |
{
|
|
| 234 |
String textValue = null; |
|
| 235 |
NodeList childList = parentNode.getChildNodes(); |
|
| 236 |
for (int i= 0; i<childList.getLength();i++) |
|
| 237 |
{
|
|
| 238 |
Node child = childList.item(i); |
|
| 239 |
if (child.getNodeName() != null && child.getNodeName().equals(childTagName)) |
|
| 240 |
{
|
|
| 241 |
MetaCatUtil.debugMessage("Find child node: " + childTagName, 35);
|
|
| 242 |
Node textNode = child.getFirstChild(); |
|
| 243 |
if (textNode.getNodeType() == Node.TEXT_NODE) |
|
| 244 |
{
|
|
| 245 |
textValue = textNode.getNodeValue(); |
|
| 246 |
}//if |
|
| 247 |
}//if |
|
| 248 |
}//for |
|
| 249 |
MetaCatUtil.debugMessage("The text value for element- " + childTagName +
|
|
| 250 |
" is " + textValue, 30); |
|
| 251 |
return textValue; |
|
| 252 |
}//getTExtValueForGivenChildTag |
|
| 253 |
|
|
| 254 |
/* Find the data format node in eml2 document */ |
|
| 255 |
private Node findDataFormatNodeInEML2(Document xml, String xPath, |
|
| 256 |
String targetDocId) |
|
| 257 |
{
|
|
| 258 |
Node targetNode = null; |
|
| 259 |
Node node = findDataFileNodeInEML2(xml, xPath, targetDocId); |
|
| 260 |
// get the phycial the prent is online, grandparent is distribution |
|
| 261 |
// the grand'parent is physical |
|
| 262 |
Node phyicalNode = node.getParentNode().getParentNode().getParentNode(); |
|
| 263 |
NodeList list = phyicalNode.getChildNodes(); |
|
| 264 |
for (int i = 0; i<list.getLength(); i++) |
|
| 265 |
{
|
|
| 266 |
Node kid = list.item(i); |
|
| 267 |
// find dataFormat node |
|
| 268 |
if (kid.getNodeType() == node.ELEMENT_NODE && |
|
| 269 |
kid.getNodeName().equals(DATAFORMAT)) |
|
| 270 |
{
|
|
| 271 |
targetNode = kid; |
|
| 272 |
break; |
|
| 273 |
}//if |
|
| 274 |
}//for |
|
| 275 |
MetaCatUtil.debugMessage("dataFormat node'name: "+
|
|
| 276 |
targetNode.getNodeName(), 35); |
|
| 277 |
return targetNode; |
|
| 278 |
} |
|
| 279 |
/* Find the datafile node */ |
|
| 280 |
private Node findDataFileNodeInEML2(Document xml, String xPath, |
|
| 281 |
String targetDocId) |
|
| 282 |
{
|
|
| 283 |
Node dataFileNode = null; |
|
| 284 |
NodeList list = null; |
|
| 285 |
try |
|
| 286 |
{
|
|
| 287 |
list = XPathAPI.selectNodeList(xml, xPath); |
|
| 288 |
} |
|
| 289 |
catch (Exception e) |
|
| 290 |
{
|
|
| 291 |
// catch an error and return null |
|
| 292 |
MetaCatUtil.debugMessage("Error in findDataFileNode: "+e.getMessage(), 30);
|
|
| 293 |
return dataFileNode; |
|
| 294 |
} |
|
| 295 |
// go through the list and find target docid in online/url |
|
| 296 |
for (int i = 0; i<list.getLength(); i++) |
|
| 297 |
{
|
|
| 298 |
Node node = list.item(i); |
|
| 299 |
Node textNode = node.getFirstChild(); |
|
| 300 |
if (textNode.getNodeType() == node.TEXT_NODE) |
|
| 301 |
{
|
|
| 302 |
String URLData = textNode.getNodeValue(); |
|
| 303 |
MetaCatUtil.debugMessage("online/url text data: " + URLData, 30);
|
|
| 304 |
//Only handle data file in local metacat server |
|
| 305 |
if (URLData.indexOf(MetaCatUtil.getOption("httpserver")) != -1 ||
|
|
| 306 |
URLData.indexOf(MetaCatUtil.getOption("server")) != -1)
|
|
| 307 |
{
|
|
| 308 |
// Get docid from url |
|
| 309 |
String docId =MetaCatUtil.getDocIdWithRevFromOnlineURL(URLData); |
|
| 310 |
// Get rid of revision |
|
| 311 |
docId = MetaCatUtil.getDocIdFromString(docId); |
|
| 312 |
MetaCatUtil.debugMessage("docid from url element in xml is: "+
|
|
| 313 |
docId, 30); |
|
| 314 |
//if this docid equals target one, we find it |
|
| 315 |
if (docId != null && docId.equals(targetDocId)) |
|
| 316 |
{
|
|
| 317 |
MetaCatUtil.debugMessage("Find target docid in online/url: "+
|
|
| 318 |
docId, 30); |
|
| 319 |
dataFileNode = node; |
|
| 320 |
break; |
|
| 321 |
} |
|
| 322 |
}//if |
|
| 323 |
|
|
| 324 |
}//if |
|
| 325 |
}//for |
|
| 326 |
MetaCatUtil.debugMessage("online/url node's name: " +
|
|
| 327 |
dataFileNode.getNodeName(), 35); |
|
| 328 |
return dataFileNode; |
|
| 329 |
}//findDataFileNode |
|
| 330 |
|
|
| 138 | 331 |
/* Get relative docid list and packagetype */ |
| 139 | 332 |
private Vector getRelativeDocIdList(String id) |
| 140 | 333 |
{
|
| ... | ... | |
| 292 | 485 |
DBConnectionPool.returnDBConnection(conn, serialNumber); |
| 293 | 486 |
} |
| 294 | 487 |
}//finally |
| 295 |
MetaCatUtil.debugMessage("!!!!!!!!!target docid is: "+ docId + " "+
|
|
| 488 |
MetaCatUtil.debugMessage("target docid is: "+ docId + " "+
|
|
| 296 | 489 |
"for target doctype: "+targetType, 25); |
| 297 | 490 |
return docId; |
| 298 | 491 |
} |
| ... | ... | |
| 382 | 575 |
contentTypeHash.put(XML, XMLTYPE); |
| 383 | 576 |
contentTypeHash.put(HTML,HTMLTYPE); |
| 384 | 577 |
contentTypeHash.put(GIF, GIFTYPE); |
| 578 |
contentTypeHash.put(JPEG, JPEGTYPE); |
|
| 385 | 579 |
contentTypeHash.put(BMP, BMPTYPE); |
| 386 | 580 |
contentTypeHash.put(TAR, TARTYPE); |
| 387 | 581 |
contentTypeHash.put(ZIP, ZIPTYPE); |
| ... | ... | |
| 389 | 583 |
|
| 390 | 584 |
}//constructrContentHashTable(); |
| 391 | 585 |
|
| 586 |
|
|
| 587 |
|
|
| 392 | 588 |
public static void main(String[] argus) |
| 393 | 589 |
{
|
| 394 | 590 |
try |
| 395 | 591 |
{
|
| 396 | 592 |
DBConnectionPool pool = DBConnectionPool.getInstance(); |
| 397 |
ContentTypeProvider provider = new ContentTypeProvider("tao.9830");
|
|
| 593 |
//ContentTypeProvider provider = new ContentTypeProvider("tao.9830");
|
|
| 594 |
ContentTypeProvider provider = new ContentTypeProvider("tao.0001");
|
|
| 398 | 595 |
String str = provider.getContentType(); |
| 399 | 596 |
MetaCatUtil.debugMessage("content type is : " + str, 20);
|
| 400 |
|
|
| 401 | 597 |
} |
| 402 | 598 |
catch(Exception e) |
| 403 | 599 |
{
|
Also available in: Unified diff
Add code to handle eml2 contentype.