Revision 2036
Added by Duane Costa over 20 years ago
| HarvestDocument.java | ||
|---|---|---|
| 12 | 12 |
import java.io.StringReader; |
| 13 | 13 |
import java.net.MalformedURLException; |
| 14 | 14 |
import java.net.URL; |
| 15 |
import java.sql.ResultSet; |
|
| 16 |
import java.sql.SQLException; |
|
| 17 |
import java.sql.Statement; |
|
| 15 | 18 |
|
| 16 | 19 |
import edu.ucsb.nceas.metacat.client.InsufficientKarmaException; |
| 17 | 20 |
import edu.ucsb.nceas.metacat.client.Metacat; |
| ... | ... | |
| 27 | 30 |
* @author costa |
| 28 | 31 |
*/ |
| 29 | 32 |
public class HarvestDocument {
|
| 30 |
|
|
| 31 |
private String documentName; |
|
| 33 |
|
|
| 34 |
|
|
| 35 |
private String docid; // scope + identifier |
|
| 36 |
private String docidFull; // scope + identifier + revision |
|
| 32 | 37 |
String documentType; |
| 33 | 38 |
String documentURL; |
| 34 | 39 |
private Harvester harvester; |
| ... | ... | |
| 67 | 72 |
this.identifier = identifier; |
| 68 | 73 |
this.revision = revision; |
| 69 | 74 |
|
| 70 |
this.documentName = scope + "." + identifier; |
|
| 75 |
this.docid = scope + "." + identifier; |
|
| 76 |
this.docidFull = this.docid + "." + revision; |
|
| 71 | 77 |
} |
| 72 | 78 |
|
| 73 | 79 |
|
| ... | ... | |
| 89 | 95 |
inputStreamReader = new InputStreamReader(inputStream); |
| 90 | 96 |
documentString = IOUtil.getAsString(inputStreamReader, true); |
| 91 | 97 |
stringReader = new StringReader(documentString); |
| 92 |
harvester.addLogEntry(0, "", "GetDocSuccess", |
|
| 93 |
harvestSiteSchedule.siteScheduleID, null, ""); |
|
| 98 |
harvester.addLogEntry(0, |
|
| 99 |
"Retrieved: " + documentURL, |
|
| 100 |
"GetDocSuccess", |
|
| 101 |
harvestSiteSchedule.siteScheduleID, |
|
| 102 |
null, |
|
| 103 |
""); |
|
| 94 | 104 |
} |
| 95 | 105 |
catch (MalformedURLException e) {
|
| 96 | 106 |
harvester.addLogEntry(1, "MalformedURLException", "GetDocError", |
| ... | ... | |
| 116 | 126 |
*/ |
| 117 | 127 |
public void harvestDocument() {
|
| 118 | 128 |
int highestRevision; |
| 129 |
boolean insert = false; |
|
| 119 | 130 |
String metacatReturnString; |
| 120 | 131 |
StringReader stringReader; |
| 132 |
boolean update = false; |
|
| 121 | 133 |
|
| 122 | 134 |
/* If metacat already has this document, determine the highest revision in |
| 123 | 135 |
* metacat and report it to the user; else, insert or delete the document |
| 124 | 136 |
* into metacat. |
| 125 | 137 |
*/ |
| 126 |
if (metacatHasDocument()) {
|
|
| 127 |
highestRevision = metacatHighestRevision(); |
|
| 138 |
highestRevision = metacatHighestRevision(); |
|
| 139 |
|
|
| 140 |
if (highestRevision == -1) {
|
|
| 141 |
insert = true; |
|
| 142 |
} |
|
| 143 |
else if (revision > highestRevision) {
|
|
| 144 |
update = true; |
|
| 145 |
} |
|
| 146 |
else {
|
|
| 128 | 147 |
harvester.addLogEntry(0, |
| 129 |
"Metacat has document: " + documentName + |
|
| 130 |
", highest revision: " + highestRevision, |
|
| 148 |
"Attempting to update " + docid + " to revision " + |
|
| 149 |
revision + ". Metacat has document revision " + |
|
| 150 |
highestRevision + ".", |
|
| 131 | 151 |
"MetacatHasDoc", |
| 132 |
harvestSiteSchedule.siteScheduleID, null, ""); |
|
| 152 |
harvestSiteSchedule.siteScheduleID, |
|
| 153 |
null, |
|
| 154 |
""); |
|
| 133 | 155 |
} |
| 134 |
else {
|
|
| 156 |
|
|
| 157 |
if (insert || update) {
|
|
| 135 | 158 |
stringReader = getSiteDocument(); |
| 136 | 159 |
if (stringReader != null) {
|
| 137 |
if (parseDocument()) {
|
|
| 138 |
putMetacatDocument(stringReader); |
|
| 160 |
if (validateDocument()) {
|
|
| 161 |
putMetacatDocument(insert, update, stringReader);
|
|
| 139 | 162 |
} |
| 140 | 163 |
} |
| 141 | 164 |
} |
| ... | ... | |
| 156 | 179 |
Exception e |
| 157 | 180 |
) {
|
| 158 | 181 |
if (insert) {
|
| 159 |
harvester.addLogEntry(1, metacatReturnString, "InsertDocError", |
|
| 160 |
harvestSiteSchedule.siteScheduleID, |
|
| 161 |
this, exceptionName + ": " + e.getMessage()); |
|
| 182 |
harvester.addLogEntry(1, |
|
| 183 |
metacatReturnString, |
|
| 184 |
"InsertDocError", |
|
| 185 |
harvestSiteSchedule.siteScheduleID, |
|
| 186 |
this, |
|
| 187 |
exceptionName + ": " + e.getMessage()); |
|
| 162 | 188 |
} |
| 163 | 189 |
else {
|
| 164 |
harvester.addLogEntry(1, metacatReturnString, "UpdateDocError", |
|
| 165 |
harvestSiteSchedule.siteScheduleID, |
|
| 166 |
this, exceptionName + ": " + e.getMessage()); |
|
| 190 |
harvester.addLogEntry(1, |
|
| 191 |
metacatReturnString, |
|
| 192 |
"UpdateDocError", |
|
| 193 |
harvestSiteSchedule.siteScheduleID, |
|
| 194 |
this, |
|
| 195 |
exceptionName + ": " + e.getMessage()); |
|
| 167 | 196 |
} |
| 168 | 197 |
} |
| 169 | 198 |
|
| 170 | 199 |
|
| 171 | 200 |
/** |
| 172 |
* Boolean to determine whether Metacat already has this document. |
|
| 173 |
* |
|
| 174 |
* @return true if Metacat has the document, otherwise false |
|
| 175 |
*/ |
|
| 176 |
private boolean metacatHasDocument() {
|
|
| 177 |
boolean hasDocument = false; |
|
| 178 |
|
|
| 179 |
return hasDocument; |
|
| 180 |
} |
|
| 181 |
|
|
| 182 |
|
|
| 183 |
/** |
|
| 184 | 201 |
* Determines the highest revision that Metacat has for this document. |
| 185 | 202 |
* |
| 186 |
* @return int representing the highest revision for this document in Metacat |
|
| 203 |
* @return int representing the highest revision for this document in |
|
| 204 |
* Metacat. Returns -1 if Metacat does not currently hold the |
|
| 205 |
* document. |
|
| 187 | 206 |
*/ |
| 188 | 207 |
private int metacatHighestRevision() {
|
| 189 |
int highestRevision = 0; |
|
| 190 |
|
|
| 191 |
return highestRevision; |
|
| 192 |
} |
|
| 193 |
|
|
| 194 |
|
|
| 195 |
/** |
|
| 196 |
* Parse the document to determine whether it is valid EML prior to inserting |
|
| 197 |
* or updating it to Metacat. This is QA/QC measure. Currently unimplemented. |
|
| 198 |
* |
|
| 199 |
* @return true if the document is valid EML, otherwise false |
|
| 200 |
*/ |
|
| 201 |
private boolean parseDocument () {
|
|
| 202 |
boolean success = true; |
|
| 208 |
int highestRevision = -1; |
|
| 209 |
String query = "SELECT REV FROM XML_DOCUMENTS WHERE DOCID = " + |
|
| 210 |
"'" + docid + "'"; |
|
| 211 |
Statement stmt; |
|
| 203 | 212 |
|
| 204 |
if (success) {
|
|
| 205 |
harvester.addLogEntry(0, "", "ValidateDocSuccess", |
|
| 206 |
harvestSiteSchedule.siteScheduleID, null, ""); |
|
| 213 |
try {
|
|
| 214 |
stmt = harvester.conn.createStatement(); |
|
| 215 |
ResultSet rs = stmt.executeQuery(query); |
|
| 216 |
|
|
| 217 |
while (rs.next()) {
|
|
| 218 |
highestRevision = rs.getInt("REV");
|
|
| 219 |
} |
|
| 220 |
|
|
| 221 |
stmt.close(); |
|
| 222 |
} |
|
| 223 |
catch(SQLException e) {
|
|
| 224 |
System.out.println("SQLException: " + e.getMessage());
|
|
| 207 | 225 |
} |
| 208 |
else {
|
|
| 209 |
harvester.addLogEntry(1, "Error validating document", "ValidateDocError", |
|
| 210 |
harvestSiteSchedule.siteScheduleID, this, ""); |
|
| 211 |
} |
|
| 212 |
|
|
| 213 |
return success; |
|
| 226 |
|
|
| 227 |
return highestRevision; |
|
| 214 | 228 |
} |
| 215 | 229 |
|
| 216 | 230 |
|
| ... | ... | |
| 218 | 232 |
* Print the data fields and values in this HarvestDocument object. |
| 219 | 233 |
*/ |
| 220 | 234 |
void printOutput() {
|
| 221 |
System.out.println("scope: " + scope);
|
|
| 222 |
System.out.println("identifier: " + identifier);
|
|
| 223 |
System.out.println("revision: " + revision);
|
|
| 224 |
System.out.println("documentType: " + documentType);
|
|
| 225 |
System.out.println("documentURL: " + documentURL);
|
|
| 226 |
System.out.println("documentName: " + documentName);
|
|
| 235 |
System.out.println("* scope: " + scope);
|
|
| 236 |
System.out.println("* identifier: " + identifier);
|
|
| 237 |
System.out.println("* revision: " + revision);
|
|
| 238 |
System.out.println("* documentType: " + documentType);
|
|
| 239 |
System.out.println("* documentURL: " + documentURL);
|
|
| 227 | 240 |
} |
| 228 | 241 |
|
| 229 | 242 |
|
| 230 | 243 |
/** |
| 231 | 244 |
* Insert or update this document to Metacat. If revision equals 1, do an |
| 232 | 245 |
* insert; otherwise, do an update. |
| 246 |
* |
|
| 247 |
* @param insert true if this is an insert operation |
|
| 248 |
* @param update true if this is an update operation |
|
| 249 |
* @param stringReader the StringReader object holding the document text |
|
| 233 | 250 |
*/ |
| 234 |
private void putMetacatDocument(StringReader stringReader) {
|
|
| 235 |
String docid = scope + "." + identifier + "." + revision;
|
|
| 236 |
boolean insert = (revision == 1);
|
|
| 251 |
private void putMetacatDocument(boolean insert,
|
|
| 252 |
boolean update,
|
|
| 253 |
StringReader stringReader) {
|
|
| 237 | 254 |
Metacat metacat = harvester.metacat; |
| 238 | 255 |
String metacatReturnString = ""; |
| 239 | 256 |
|
| 240 | 257 |
if (harvester.connectToMetacat()) {
|
| 241 | 258 |
try {
|
| 242 | 259 |
if (insert) {
|
| 243 |
metacatReturnString = metacat.insert(docid, stringReader, null); |
|
| 244 |
harvester.addLogEntry(0, docid + " : " + metacatReturnString, |
|
| 260 |
metacatReturnString = metacat.insert(docidFull, stringReader, null);
|
|
| 261 |
harvester.addLogEntry(0, docidFull + " : " + metacatReturnString,
|
|
| 245 | 262 |
"InsertDocSuccess", |
| 246 | 263 |
harvestSiteSchedule.siteScheduleID, |
| 247 | 264 |
null, ""); |
| 248 | 265 |
} |
| 249 |
else {
|
|
| 250 |
metacatReturnString = metacat.update(docid, stringReader, null); |
|
| 251 |
harvester.addLogEntry(0, docid + " : " + metacatReturnString, |
|
| 266 |
else if (update) {
|
|
| 267 |
metacatReturnString = metacat.update(docidFull, stringReader, null);
|
|
| 268 |
harvester.addLogEntry(0, docidFull + " : " + metacatReturnString,
|
|
| 252 | 269 |
"UpdateDocSuccess", |
| 253 | 270 |
harvestSiteSchedule.siteScheduleID, |
| 254 | 271 |
null, ""); |
| ... | ... | |
| 270 | 287 |
} |
| 271 | 288 |
} |
| 272 | 289 |
} |
| 290 |
|
|
| 291 |
|
|
| 292 |
/** |
|
| 293 |
* Validate the document to determine whether it is valid EML prior to |
|
| 294 |
* inserting or updating it to Metacat. This is QA/QC measure. |
|
| 295 |
* Not yet implemented. |
|
| 296 |
* |
|
| 297 |
* @return true if the document is valid EML, otherwise false |
|
| 298 |
*/ |
|
| 299 |
private boolean validateDocument () {
|
|
| 300 |
boolean success = true; |
|
| 301 |
|
|
| 302 |
/*if (success) {
|
|
| 303 |
harvester.addLogEntry(0, |
|
| 304 |
"Validated: " + documentURL, |
|
| 305 |
"ValidateDocSuccess", |
|
| 306 |
harvestSiteSchedule.siteScheduleID, |
|
| 307 |
null, |
|
| 308 |
""); |
|
| 309 |
} |
|
| 310 |
else {
|
|
| 311 |
harvester.addLogEntry(1, "Error validating document", "ValidateDocError", |
|
| 312 |
harvestSiteSchedule.siteScheduleID, this, ""); |
|
| 313 |
}*/ |
|
| 314 |
|
|
| 315 |
return success; |
|
| 316 |
} |
|
| 317 |
|
|
| 273 | 318 |
} |
Also available in: Unified diff
Additional Harvester development