1 |
2094
|
jones
|
/**
|
2 |
|
|
* '$RCSfile$'
|
3 |
|
|
* Copyright: 2004 University of New Mexico and the
|
4 |
|
|
* Regents of the University of California
|
5 |
2022
|
costa
|
*
|
6 |
2094
|
jones
|
* '$Author$'
|
7 |
|
|
* '$Date$'
|
8 |
|
|
* '$Revision$'
|
9 |
|
|
*
|
10 |
|
|
* This program is free software; you can redistribute it and/or modify
|
11 |
|
|
* it under the terms of the GNU General Public License as published by
|
12 |
|
|
* the Free Software Foundation; either version 2 of the License, or
|
13 |
|
|
* (at your option) any later version.
|
14 |
|
|
*
|
15 |
|
|
* This program is distributed in the hope that it will be useful,
|
16 |
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
17 |
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
18 |
|
|
* GNU General Public License for more details.
|
19 |
|
|
*
|
20 |
|
|
* You should have received a copy of the GNU General Public License
|
21 |
|
|
* along with this program; if not, write to the Free Software
|
22 |
|
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
23 |
2022
|
costa
|
*/
|
24 |
|
|
|
25 |
|
|
package edu.ucsb.nceas.metacat.harvesterClient;
|
26 |
|
|
|
27 |
|
|
import java.io.InputStream;
|
28 |
|
|
import java.io.InputStreamReader;
|
29 |
|
|
import java.io.IOException;
|
30 |
2086
|
costa
|
import java.io.PrintStream;
|
31 |
2022
|
costa
|
import java.io.StringReader;
|
32 |
|
|
import java.net.MalformedURLException;
|
33 |
|
|
import java.net.URL;
|
34 |
2139
|
costa
|
import java.sql.Connection;
|
35 |
2036
|
costa
|
import java.sql.ResultSet;
|
36 |
|
|
import java.sql.SQLException;
|
37 |
|
|
import java.sql.Statement;
|
38 |
2022
|
costa
|
|
39 |
2031
|
costa
|
import edu.ucsb.nceas.metacat.client.InsufficientKarmaException;
|
40 |
|
|
import edu.ucsb.nceas.metacat.client.Metacat;
|
41 |
|
|
import edu.ucsb.nceas.metacat.client.MetacatException;
|
42 |
|
|
import edu.ucsb.nceas.metacat.client.MetacatInaccessibleException;
|
43 |
2022
|
costa
|
import edu.ucsb.nceas.utilities.IOUtil;
|
44 |
|
|
|
45 |
|
|
|
46 |
|
|
/**
|
47 |
|
|
* HarvestDocument manages operations and data for a single document to be
|
48 |
|
|
* harvested.
|
49 |
|
|
*
|
50 |
|
|
* @author costa
|
51 |
|
|
*/
|
52 |
|
|
public class HarvestDocument {
|
53 |
2036
|
costa
|
|
54 |
|
|
|
55 |
|
|
private String docid; // scope + identifier
|
56 |
|
|
private String docidFull; // scope + identifier + revision
|
57 |
2031
|
costa
|
String documentType;
|
58 |
|
|
String documentURL;
|
59 |
2022
|
costa
|
private Harvester harvester;
|
60 |
|
|
private HarvestSiteSchedule harvestSiteSchedule;
|
61 |
2031
|
costa
|
int identifier;
|
62 |
|
|
int revision;
|
63 |
|
|
String scope;
|
64 |
2155
|
costa
|
|
65 |
|
|
/* These booleans keep track of status information. They are used when
|
66 |
|
|
* generating email reports.
|
67 |
|
|
*/
|
68 |
|
|
boolean accessError = false;
|
69 |
|
|
boolean inserted = false;
|
70 |
|
|
boolean metacatHasIt = false;
|
71 |
|
|
boolean updated = false;
|
72 |
|
|
boolean uploadError = false;
|
73 |
2022
|
costa
|
|
74 |
|
|
|
75 |
|
|
/**
|
76 |
|
|
* Creates a new instance of HarvestDocument. Initialized with the data
|
77 |
|
|
* that was read from a single <document> element in site document list.
|
78 |
|
|
*
|
79 |
|
|
* @param harvester the parent Harvester object
|
80 |
|
|
* @param harvestSiteSchedule the parent HarvestSiteSchedule object
|
81 |
|
|
* @param scope the value of the <scope> element
|
82 |
|
|
* @param identifier the value of the <identifier> element
|
83 |
|
|
* @param revision the value of the <revision> element
|
84 |
|
|
* @param documentType the value of the <documentType> element
|
85 |
|
|
* @param documentURL the value of the <documentURL> element
|
86 |
|
|
*/
|
87 |
2139
|
costa
|
public HarvestDocument (
|
88 |
2022
|
costa
|
Harvester harvester,
|
89 |
|
|
HarvestSiteSchedule harvestSiteSchedule,
|
90 |
|
|
String scope,
|
91 |
|
|
int identifier,
|
92 |
|
|
int revision,
|
93 |
|
|
String documentType,
|
94 |
|
|
String documentURL
|
95 |
|
|
) {
|
96 |
|
|
this.harvester = harvester;
|
97 |
|
|
this.harvestSiteSchedule = harvestSiteSchedule;
|
98 |
|
|
this.documentType = documentType;
|
99 |
|
|
this.documentURL = documentURL;
|
100 |
|
|
this.scope = scope;
|
101 |
|
|
this.identifier = identifier;
|
102 |
|
|
this.revision = revision;
|
103 |
|
|
|
104 |
2036
|
costa
|
this.docid = scope + "." + identifier;
|
105 |
|
|
this.docidFull = this.docid + "." + revision;
|
106 |
2022
|
costa
|
}
|
107 |
|
|
|
108 |
|
|
|
109 |
|
|
/**
|
110 |
|
|
* Retrieve the document from the site using its <documentURL> value.
|
111 |
|
|
*
|
112 |
|
|
* @return A StringReader containing the document string.
|
113 |
|
|
*/
|
114 |
2139
|
costa
|
public StringReader getSiteDocument() {
|
115 |
2022
|
costa
|
String documentString;
|
116 |
|
|
InputStream inputStream;
|
117 |
|
|
InputStreamReader inputStreamReader;
|
118 |
|
|
StringReader stringReader = null;
|
119 |
|
|
URL url;
|
120 |
|
|
|
121 |
|
|
try {
|
122 |
|
|
url = new URL(documentURL);
|
123 |
|
|
inputStream = url.openStream();
|
124 |
|
|
inputStreamReader = new InputStreamReader(inputStream);
|
125 |
|
|
documentString = IOUtil.getAsString(inputStreamReader, true);
|
126 |
|
|
stringReader = new StringReader(documentString);
|
127 |
2036
|
costa
|
harvester.addLogEntry(0,
|
128 |
|
|
"Retrieved: " + documentURL,
|
129 |
4175
|
daigle
|
"harvester.GetDocSuccess",
|
130 |
2036
|
costa
|
harvestSiteSchedule.siteScheduleID,
|
131 |
|
|
null,
|
132 |
|
|
"");
|
133 |
2022
|
costa
|
}
|
134 |
|
|
catch (MalformedURLException e) {
|
135 |
2155
|
costa
|
accessError = true;
|
136 |
4175
|
daigle
|
harvester.addLogEntry(1, "MalformedURLException", "harvester.GetDocError",
|
137 |
2031
|
costa
|
harvestSiteSchedule.siteScheduleID, this,
|
138 |
|
|
"MalformedURLException: " + e.getMessage());
|
139 |
2022
|
costa
|
}
|
140 |
|
|
catch (IOException e) {
|
141 |
2155
|
costa
|
accessError = true;
|
142 |
4175
|
daigle
|
harvester.addLogEntry(1, "IOException", "harvester.GetDocError",
|
143 |
2031
|
costa
|
harvestSiteSchedule.siteScheduleID, this,
|
144 |
|
|
"IOException: " + e.getMessage());
|
145 |
2022
|
costa
|
}
|
146 |
|
|
|
147 |
|
|
return stringReader;
|
148 |
|
|
}
|
149 |
|
|
|
150 |
|
|
|
151 |
|
|
/**
|
152 |
|
|
* Harvest the document from the site. Unless Metacat already has the
|
153 |
|
|
* document, retrieve the document from the site and put (insert or
|
154 |
|
|
* update) it to Metacat. If Metacat already has the document, determine
|
155 |
|
|
* the highest revision stored in Metacat so that this can be reported
|
156 |
|
|
* back to the user.
|
157 |
|
|
*/
|
158 |
|
|
public void harvestDocument() {
|
159 |
|
|
int highestRevision;
|
160 |
2036
|
costa
|
boolean insert = false;
|
161 |
2022
|
costa
|
String metacatReturnString;
|
162 |
|
|
StringReader stringReader;
|
163 |
2036
|
costa
|
boolean update = false;
|
164 |
2022
|
costa
|
|
165 |
|
|
/* If metacat already has this document, determine the highest revision in
|
166 |
|
|
* metacat and report it to the user; else, insert or delete the document
|
167 |
|
|
* into metacat.
|
168 |
|
|
*/
|
169 |
2036
|
costa
|
highestRevision = metacatHighestRevision();
|
170 |
|
|
|
171 |
|
|
if (highestRevision == -1) {
|
172 |
|
|
insert = true;
|
173 |
|
|
}
|
174 |
|
|
else if (revision > highestRevision) {
|
175 |
|
|
update = true;
|
176 |
|
|
}
|
177 |
|
|
else {
|
178 |
2155
|
costa
|
metacatHasIt = true;
|
179 |
2031
|
costa
|
harvester.addLogEntry(0,
|
180 |
2036
|
costa
|
"Attempting to update " + docid + " to revision " +
|
181 |
|
|
revision + ". Metacat has document revision " +
|
182 |
|
|
highestRevision + ".",
|
183 |
4175
|
daigle
|
"harvester.MetacatHasDoc",
|
184 |
2036
|
costa
|
harvestSiteSchedule.siteScheduleID,
|
185 |
|
|
null,
|
186 |
|
|
"");
|
187 |
2022
|
costa
|
}
|
188 |
2036
|
costa
|
|
189 |
|
|
if (insert || update) {
|
190 |
2022
|
costa
|
stringReader = getSiteDocument();
|
191 |
|
|
if (stringReader != null) {
|
192 |
2036
|
costa
|
if (validateDocument()) {
|
193 |
|
|
putMetacatDocument(insert, update, stringReader);
|
194 |
2022
|
costa
|
}
|
195 |
|
|
}
|
196 |
|
|
}
|
197 |
|
|
}
|
198 |
2031
|
costa
|
|
199 |
2022
|
costa
|
|
200 |
|
|
/**
|
201 |
5169
|
costa
|
* Boolean to determine whether the string returned by the Metacat client for
|
202 |
|
|
* an insert or update operation indicates that the operation succeeded.
|
203 |
|
|
*
|
204 |
|
|
* @param metacatReturnString The string returned by the Metacat client.
|
205 |
|
|
* @return true if the return string indicates success, else false
|
206 |
|
|
*/
|
207 |
|
|
private boolean isMetacatSuccessString(String metacatReturnString) {
|
208 |
|
|
boolean isMetacatSuccessString = false;
|
209 |
|
|
|
210 |
|
|
if ((metacatReturnString != null) &&
|
211 |
|
|
(metacatReturnString.contains("<success>"))
|
212 |
|
|
) {
|
213 |
|
|
isMetacatSuccessString = true;
|
214 |
|
|
}
|
215 |
|
|
|
216 |
|
|
return isMetacatSuccessString;
|
217 |
|
|
}
|
218 |
|
|
|
219 |
|
|
|
220 |
|
|
/**
|
221 |
2031
|
costa
|
* Logs a metacat document error to the harvest detail log.
|
222 |
|
|
*
|
223 |
|
|
* @param insert true if insert operation, false is update
|
224 |
|
|
* @param metacatReturnString string returned from the insert or update
|
225 |
|
|
* @param exceptionName name of the exception class
|
226 |
|
|
* @param e the exception object
|
227 |
|
|
*/
|
228 |
|
|
private void logMetacatError (boolean insert,
|
229 |
|
|
String metacatReturnString,
|
230 |
|
|
String exceptionName,
|
231 |
|
|
Exception e
|
232 |
|
|
) {
|
233 |
2155
|
costa
|
uploadError = true;
|
234 |
|
|
|
235 |
2031
|
costa
|
if (insert) {
|
236 |
2036
|
costa
|
harvester.addLogEntry(1,
|
237 |
|
|
metacatReturnString,
|
238 |
4175
|
daigle
|
"harvester.InsertDocError",
|
239 |
2036
|
costa
|
harvestSiteSchedule.siteScheduleID,
|
240 |
|
|
this,
|
241 |
|
|
exceptionName + ": " + e.getMessage());
|
242 |
2031
|
costa
|
}
|
243 |
|
|
else {
|
244 |
2036
|
costa
|
harvester.addLogEntry(1,
|
245 |
|
|
metacatReturnString,
|
246 |
4175
|
daigle
|
"harvester.UpdateDocError",
|
247 |
2036
|
costa
|
harvestSiteSchedule.siteScheduleID,
|
248 |
|
|
this,
|
249 |
|
|
exceptionName + ": " + e.getMessage());
|
250 |
2031
|
costa
|
}
|
251 |
|
|
}
|
252 |
|
|
|
253 |
|
|
|
254 |
|
|
/**
|
255 |
2022
|
costa
|
* Determines the highest revision that Metacat has for this document.
|
256 |
|
|
*
|
257 |
2036
|
costa
|
* @return int representing the highest revision for this document in
|
258 |
|
|
* Metacat. Returns -1 if Metacat does not currently hold the
|
259 |
|
|
* document.
|
260 |
2022
|
costa
|
*/
|
261 |
2139
|
costa
|
public int metacatHighestRevision() {
|
262 |
|
|
Connection conn = harvester.getConnection();
|
263 |
2036
|
costa
|
int highestRevision = -1;
|
264 |
|
|
String query = "SELECT REV FROM XML_DOCUMENTS WHERE DOCID = " +
|
265 |
|
|
"'" + docid + "'";
|
266 |
|
|
Statement stmt;
|
267 |
2022
|
costa
|
|
268 |
2036
|
costa
|
try {
|
269 |
2139
|
costa
|
stmt = conn.createStatement();
|
270 |
2036
|
costa
|
ResultSet rs = stmt.executeQuery(query);
|
271 |
|
|
|
272 |
|
|
while (rs.next()) {
|
273 |
|
|
highestRevision = rs.getInt("REV");
|
274 |
|
|
}
|
275 |
|
|
|
276 |
|
|
stmt.close();
|
277 |
|
|
}
|
278 |
|
|
catch(SQLException e) {
|
279 |
|
|
System.out.println("SQLException: " + e.getMessage());
|
280 |
2031
|
costa
|
}
|
281 |
2036
|
costa
|
|
282 |
|
|
return highestRevision;
|
283 |
2022
|
costa
|
}
|
284 |
|
|
|
285 |
|
|
|
286 |
|
|
/**
|
287 |
|
|
* Print the data fields and values in this HarvestDocument object.
|
288 |
2086
|
costa
|
*
|
289 |
|
|
* @param out the PrintStream to write to
|
290 |
2022
|
costa
|
*/
|
291 |
2139
|
costa
|
public void printOutput(PrintStream out) {
|
292 |
2086
|
costa
|
out.println("* scope: " + scope);
|
293 |
|
|
out.println("* identifier: " + identifier);
|
294 |
|
|
out.println("* revision: " + revision);
|
295 |
|
|
out.println("* documentType: " + documentType);
|
296 |
|
|
out.println("* documentURL: " + documentURL);
|
297 |
2022
|
costa
|
}
|
298 |
|
|
|
299 |
|
|
|
300 |
|
|
/**
|
301 |
2155
|
costa
|
* Print the document URL following by its scope.identifier.revision.
|
302 |
|
|
* Used for report generation.
|
303 |
|
|
*
|
304 |
|
|
* @param out the PrintStream to write to
|
305 |
|
|
*/
|
306 |
|
|
public void prettyPrint(PrintStream out) {
|
307 |
|
|
out.println("* " + docidFull + " (" + documentURL + ")");
|
308 |
|
|
}
|
309 |
|
|
|
310 |
|
|
|
311 |
|
|
/**
|
312 |
2022
|
costa
|
* Insert or update this document to Metacat. If revision equals 1, do an
|
313 |
|
|
* insert; otherwise, do an update.
|
314 |
2036
|
costa
|
*
|
315 |
|
|
* @param insert true if this is an insert operation
|
316 |
|
|
* @param update true if this is an update operation
|
317 |
|
|
* @param stringReader the StringReader object holding the document text
|
318 |
2022
|
costa
|
*/
|
319 |
2036
|
costa
|
private void putMetacatDocument(boolean insert,
|
320 |
|
|
boolean update,
|
321 |
|
|
StringReader stringReader) {
|
322 |
2022
|
costa
|
Metacat metacat = harvester.metacat;
|
323 |
|
|
String metacatReturnString = "";
|
324 |
2031
|
costa
|
|
325 |
2022
|
costa
|
if (harvester.connectToMetacat()) {
|
326 |
|
|
try {
|
327 |
5169
|
costa
|
String harvestOperationCode = "";
|
328 |
|
|
|
329 |
2031
|
costa
|
if (insert) {
|
330 |
5169
|
costa
|
harvestOperationCode = "harvester.InsertDocSuccess";
|
331 |
2036
|
costa
|
metacatReturnString = metacat.insert(docidFull, stringReader, null);
|
332 |
5169
|
costa
|
this.inserted = true;
|
333 |
2022
|
costa
|
}
|
334 |
2036
|
costa
|
else if (update) {
|
335 |
5169
|
costa
|
harvestOperationCode = "harvester.UpdateDocSuccess";
|
336 |
2036
|
costa
|
metacatReturnString = metacat.update(docidFull, stringReader, null);
|
337 |
5169
|
costa
|
this.updated = true;
|
338 |
2022
|
costa
|
}
|
339 |
5169
|
costa
|
|
340 |
|
|
if (isMetacatSuccessString(metacatReturnString)) {
|
341 |
|
|
String message = docidFull + " : " + metacatReturnString;
|
342 |
|
|
harvester.addLogEntry(0, message, harvestOperationCode,
|
343 |
|
|
harvestSiteSchedule.siteScheduleID, null, "");
|
344 |
|
|
}
|
345 |
|
|
else {
|
346 |
|
|
this.inserted = false;
|
347 |
|
|
this.updated = false;
|
348 |
|
|
final String exceptionName = "UnreportedMetacatException";
|
349 |
|
|
final String exceptionMessage =
|
350 |
|
|
"Metacat insert/update failed without reporting an exception";
|
351 |
|
|
Exception e = new Exception(exceptionMessage);
|
352 |
|
|
logMetacatError(insert, metacatReturnString, exceptionName, e);
|
353 |
|
|
}
|
354 |
2022
|
costa
|
}
|
355 |
|
|
catch (MetacatInaccessibleException e) {
|
356 |
2031
|
costa
|
logMetacatError(insert, metacatReturnString,
|
357 |
|
|
"MetacatInaccessibleException", e);
|
358 |
2022
|
costa
|
}
|
359 |
|
|
catch (InsufficientKarmaException e) {
|
360 |
2031
|
costa
|
logMetacatError(insert, metacatReturnString,
|
361 |
|
|
"InsufficientKarmaException", e);
|
362 |
2022
|
costa
|
}
|
363 |
|
|
catch (MetacatException e) {
|
364 |
2031
|
costa
|
logMetacatError(insert, metacatReturnString, "MetacatException", e);
|
365 |
2022
|
costa
|
}
|
366 |
|
|
catch (IOException e) {
|
367 |
2031
|
costa
|
logMetacatError(insert, metacatReturnString, "IOException", e);
|
368 |
2022
|
costa
|
}
|
369 |
5169
|
costa
|
catch (Exception e) {
|
370 |
|
|
logMetacatError(insert, metacatReturnString, "Exception", e);
|
371 |
|
|
}
|
372 |
2022
|
costa
|
}
|
373 |
|
|
}
|
374 |
2036
|
costa
|
|
375 |
|
|
|
376 |
|
|
/**
|
377 |
|
|
* Validate the document to determine whether it is valid EML prior to
|
378 |
|
|
* inserting or updating it to Metacat. This is QA/QC measure.
|
379 |
|
|
* Not yet implemented.
|
380 |
|
|
*
|
381 |
|
|
* @return true if the document is valid EML, otherwise false
|
382 |
|
|
*/
|
383 |
|
|
private boolean validateDocument () {
|
384 |
|
|
boolean success = true;
|
385 |
|
|
|
386 |
|
|
/*if (success) {
|
387 |
|
|
harvester.addLogEntry(0,
|
388 |
|
|
"Validated: " + documentURL,
|
389 |
4175
|
daigle
|
"harvester.ValidateDocSuccess",
|
390 |
2036
|
costa
|
harvestSiteSchedule.siteScheduleID,
|
391 |
|
|
null,
|
392 |
|
|
"");
|
393 |
|
|
}
|
394 |
|
|
else {
|
395 |
4175
|
daigle
|
harvester.addLogEntry(1, "Error validating document", "harvester.ValidateDocError",
|
396 |
2036
|
costa
|
harvestSiteSchedule.siteScheduleID, this, "");
|
397 |
|
|
}*/
|
398 |
|
|
|
399 |
|
|
return success;
|
400 |
|
|
}
|
401 |
|
|
|
402 |
2022
|
costa
|
}
|