1
|
/*
|
2
|
* HarvestDocument.java
|
3
|
*
|
4
|
* Created on January 14, 2004, 4:37 PM
|
5
|
*/
|
6
|
|
7
|
package edu.ucsb.nceas.metacat.harvesterClient;
|
8
|
|
9
|
import java.io.InputStream;
|
10
|
import java.io.InputStreamReader;
|
11
|
import java.io.IOException;
|
12
|
import java.io.StringReader;
|
13
|
import java.net.MalformedURLException;
|
14
|
import java.net.URL;
|
15
|
|
16
|
import edu.ucsb.nceas.metacat.client.InsufficientKarmaException;
|
17
|
import edu.ucsb.nceas.metacat.client.Metacat;
|
18
|
import edu.ucsb.nceas.metacat.client.MetacatException;
|
19
|
import edu.ucsb.nceas.metacat.client.MetacatInaccessibleException;
|
20
|
import edu.ucsb.nceas.utilities.IOUtil;
|
21
|
|
22
|
|
23
|
/**
|
24
|
* HarvestDocument manages operations and data for a single document to be
|
25
|
* harvested.
|
26
|
*
|
27
|
* @author costa
|
28
|
*/
|
29
|
public class HarvestDocument {
|
30
|
|
31
|
private String documentName;
|
32
|
String documentType;
|
33
|
String documentURL;
|
34
|
private Harvester harvester;
|
35
|
private HarvestSiteSchedule harvestSiteSchedule;
|
36
|
int identifier;
|
37
|
int revision;
|
38
|
String scope;
|
39
|
|
40
|
|
41
|
/**
|
42
|
* Creates a new instance of HarvestDocument. Initialized with the data
|
43
|
* that was read from a single <document> element in site document list.
|
44
|
*
|
45
|
* @param harvester the parent Harvester object
|
46
|
* @param harvestSiteSchedule the parent HarvestSiteSchedule object
|
47
|
* @param scope the value of the <scope> element
|
48
|
* @param identifier the value of the <identifier> element
|
49
|
* @param revision the value of the <revision> element
|
50
|
* @param documentType the value of the <documentType> element
|
51
|
* @param documentURL the value of the <documentURL> element
|
52
|
*/
|
53
|
public HarvestDocument(
|
54
|
Harvester harvester,
|
55
|
HarvestSiteSchedule harvestSiteSchedule,
|
56
|
String scope,
|
57
|
int identifier,
|
58
|
int revision,
|
59
|
String documentType,
|
60
|
String documentURL
|
61
|
) {
|
62
|
this.harvester = harvester;
|
63
|
this.harvestSiteSchedule = harvestSiteSchedule;
|
64
|
this.documentType = documentType;
|
65
|
this.documentURL = documentURL;
|
66
|
this.scope = scope;
|
67
|
this.identifier = identifier;
|
68
|
this.revision = revision;
|
69
|
|
70
|
this.documentName = scope + "." + identifier;
|
71
|
}
|
72
|
|
73
|
|
74
|
/**
|
75
|
* Retrieve the document from the site using its <documentURL> value.
|
76
|
*
|
77
|
* @return A StringReader containing the document string.
|
78
|
*/
|
79
|
private StringReader getSiteDocument() {
|
80
|
String documentString;
|
81
|
InputStream inputStream;
|
82
|
InputStreamReader inputStreamReader;
|
83
|
StringReader stringReader = null;
|
84
|
URL url;
|
85
|
|
86
|
try {
|
87
|
url = new URL(documentURL);
|
88
|
inputStream = url.openStream();
|
89
|
inputStreamReader = new InputStreamReader(inputStream);
|
90
|
documentString = IOUtil.getAsString(inputStreamReader, true);
|
91
|
stringReader = new StringReader(documentString);
|
92
|
harvester.addLogEntry(0, "", "GetDocSuccess",
|
93
|
harvestSiteSchedule.siteScheduleID, null, "");
|
94
|
}
|
95
|
catch (MalformedURLException e) {
|
96
|
harvester.addLogEntry(1, "MalformedURLException", "GetDocError",
|
97
|
harvestSiteSchedule.siteScheduleID, this,
|
98
|
"MalformedURLException: " + e.getMessage());
|
99
|
}
|
100
|
catch (IOException e) {
|
101
|
harvester.addLogEntry(1, "IOException", "GetDocError",
|
102
|
harvestSiteSchedule.siteScheduleID, this,
|
103
|
"IOException: " + e.getMessage());
|
104
|
}
|
105
|
|
106
|
return stringReader;
|
107
|
}
|
108
|
|
109
|
|
110
|
/**
|
111
|
* Harvest the document from the site. Unless Metacat already has the
|
112
|
* document, retrieve the document from the site and put (insert or
|
113
|
* update) it to Metacat. If Metacat already has the document, determine
|
114
|
* the highest revision stored in Metacat so that this can be reported
|
115
|
* back to the user.
|
116
|
*/
|
117
|
public void harvestDocument() {
|
118
|
int highestRevision;
|
119
|
String metacatReturnString;
|
120
|
StringReader stringReader;
|
121
|
|
122
|
/* If metacat already has this document, determine the highest revision in
|
123
|
* metacat and report it to the user; else, insert or delete the document
|
124
|
* into metacat.
|
125
|
*/
|
126
|
if (metacatHasDocument()) {
|
127
|
highestRevision = metacatHighestRevision();
|
128
|
harvester.addLogEntry(0,
|
129
|
"Metacat has document: " + documentName +
|
130
|
", highest revision: " + highestRevision,
|
131
|
"MetacatHasDoc",
|
132
|
harvestSiteSchedule.siteScheduleID, null, "");
|
133
|
}
|
134
|
else {
|
135
|
stringReader = getSiteDocument();
|
136
|
if (stringReader != null) {
|
137
|
if (parseDocument()) {
|
138
|
putMetacatDocument(stringReader);
|
139
|
}
|
140
|
}
|
141
|
}
|
142
|
}
|
143
|
|
144
|
|
145
|
/**
|
146
|
* Logs a metacat document error to the harvest detail log.
|
147
|
*
|
148
|
* @param insert true if insert operation, false is update
|
149
|
* @param metacatReturnString string returned from the insert or update
|
150
|
* @param exceptionName name of the exception class
|
151
|
* @param e the exception object
|
152
|
*/
|
153
|
private void logMetacatError (boolean insert,
|
154
|
String metacatReturnString,
|
155
|
String exceptionName,
|
156
|
Exception e
|
157
|
) {
|
158
|
if (insert) {
|
159
|
harvester.addLogEntry(1, metacatReturnString, "InsertDocError",
|
160
|
harvestSiteSchedule.siteScheduleID,
|
161
|
this, exceptionName + ": " + e.getMessage());
|
162
|
}
|
163
|
else {
|
164
|
harvester.addLogEntry(1, metacatReturnString, "UpdateDocError",
|
165
|
harvestSiteSchedule.siteScheduleID,
|
166
|
this, exceptionName + ": " + e.getMessage());
|
167
|
}
|
168
|
}
|
169
|
|
170
|
|
171
|
/**
|
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
|
* Determines the highest revision that Metacat has for this document.
|
185
|
*
|
186
|
* @return int representing the highest revision for this document in Metacat
|
187
|
*/
|
188
|
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;
|
203
|
|
204
|
if (success) {
|
205
|
harvester.addLogEntry(0, "", "ValidateDocSuccess",
|
206
|
harvestSiteSchedule.siteScheduleID, null, "");
|
207
|
}
|
208
|
else {
|
209
|
harvester.addLogEntry(1, "Error validating document", "ValidateDocError",
|
210
|
harvestSiteSchedule.siteScheduleID, this, "");
|
211
|
}
|
212
|
|
213
|
return success;
|
214
|
}
|
215
|
|
216
|
|
217
|
/**
|
218
|
* Print the data fields and values in this HarvestDocument object.
|
219
|
*/
|
220
|
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);
|
227
|
}
|
228
|
|
229
|
|
230
|
/**
|
231
|
* Insert or update this document to Metacat. If revision equals 1, do an
|
232
|
* insert; otherwise, do an update.
|
233
|
*/
|
234
|
private void putMetacatDocument(StringReader stringReader) {
|
235
|
String docid = scope + "." + identifier + "." + revision;
|
236
|
boolean insert = (revision == 1);
|
237
|
Metacat metacat = harvester.metacat;
|
238
|
String metacatReturnString = "";
|
239
|
|
240
|
if (harvester.connectToMetacat()) {
|
241
|
try {
|
242
|
if (insert) {
|
243
|
metacatReturnString = metacat.insert(docid, stringReader, null);
|
244
|
harvester.addLogEntry(0, docid + " : " + metacatReturnString,
|
245
|
"InsertDocSuccess",
|
246
|
harvestSiteSchedule.siteScheduleID,
|
247
|
null, "");
|
248
|
}
|
249
|
else {
|
250
|
metacatReturnString = metacat.update(docid, stringReader, null);
|
251
|
harvester.addLogEntry(0, docid + " : " + metacatReturnString,
|
252
|
"UpdateDocSuccess",
|
253
|
harvestSiteSchedule.siteScheduleID,
|
254
|
null, "");
|
255
|
}
|
256
|
}
|
257
|
catch (MetacatInaccessibleException e) {
|
258
|
logMetacatError(insert, metacatReturnString,
|
259
|
"MetacatInaccessibleException", e);
|
260
|
}
|
261
|
catch (InsufficientKarmaException e) {
|
262
|
logMetacatError(insert, metacatReturnString,
|
263
|
"InsufficientKarmaException", e);
|
264
|
}
|
265
|
catch (MetacatException e) {
|
266
|
logMetacatError(insert, metacatReturnString, "MetacatException", e);
|
267
|
}
|
268
|
catch (IOException e) {
|
269
|
logMetacatError(insert, metacatReturnString, "IOException", e);
|
270
|
}
|
271
|
}
|
272
|
}
|
273
|
}
|