Revision 47109
Added by Michele Artini over 6 years ago
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/msro/workflows/nodes/GeneratePersonReportJobNode.java | ||
|---|---|---|
| 3 | 3 |
import java.io.PrintWriter; |
| 4 | 4 |
import java.io.StringReader; |
| 5 | 5 |
import java.io.StringWriter; |
| 6 |
import java.nio.charset.StandardCharsets; |
|
| 6 | 7 |
import java.nio.file.Files; |
| 7 | 8 |
import java.nio.file.Paths; |
| 8 | 9 |
import java.text.SimpleDateFormat; |
| ... | ... | |
| 43 | 44 |
|
| 44 | 45 |
protected class SinglePersonReport {
|
| 45 | 46 |
|
| 47 |
private final String id; |
|
| 46 | 48 |
private final String title; |
| 47 | 49 |
private final List<String> oldCreators = new ArrayList<>(); |
| 48 | 50 |
private final List<String> newCreators = new ArrayList<>(); |
| 49 | 51 |
|
| 50 |
public SinglePersonReport(final String title) {
|
|
| 52 |
public SinglePersonReport(final String id, final String title) {
|
|
| 53 |
this.id = id; |
|
| 51 | 54 |
this.title = title; |
| 52 | 55 |
} |
| 53 | 56 |
|
| 57 |
public String getId() {
|
|
| 58 |
return id; |
|
| 59 |
} |
|
| 60 |
|
|
| 54 | 61 |
public String getTitle() {
|
| 55 | 62 |
return title; |
| 56 | 63 |
} |
| ... | ... | |
| 80 | 87 |
} |
| 81 | 88 |
|
| 82 | 89 |
public boolean hasCorrections() {
|
| 83 |
/* |
|
| 84 |
* final int size = newCreators.size(); |
|
| 85 |
* |
|
| 86 |
* if (oldCreators.size() != size) { return true; }
|
|
| 87 |
* |
|
| 88 |
* final Set<String> set = new HashSet<>(); set.addAll(oldCreators); set.addAll(newCreators); |
|
| 89 |
* |
|
| 90 |
* return set.size() != size; |
|
| 91 |
*/ |
|
| 92 |
|
|
| 93 | 90 |
return (oldCreators.size() > 0) || (newCreators.size() > 0); |
| 94 | 91 |
} |
| 95 | 92 |
|
| ... | ... | |
| 97 | 94 |
public String toString() {
|
| 98 | 95 |
final StringWriter sw = new StringWriter(); |
| 99 | 96 |
sw.write("** ");
|
| 97 |
sw.write(id); |
|
| 98 |
sw.write(" - ");
|
|
| 100 | 99 |
sw.write(title); |
| 101 | 100 |
sw.write(" **\n");
|
| 102 |
sw.write("ORIGINAL:\n");
|
|
| 103 |
for (int i = 0; i < oldCreators.size(); i++) {
|
|
| 104 |
sw.write(String.format("%5d) %s\n", i + 1, oldCreators.get(i)));
|
|
| 101 |
|
|
| 102 |
if (oldCreators.size() > 0) {
|
|
| 103 |
sw.write("ORIGINAL:\n");
|
|
| 104 |
for (int i = 0; i < oldCreators.size(); i++) {
|
|
| 105 |
final String s = oldCreators.get(i); |
|
| 106 |
sw.write(String.format("%5d) %s\n", i + 1, StringUtils.isNotBlank(s) ? s : "[empty]"));
|
|
| 107 |
} |
|
| 105 | 108 |
} |
| 106 |
sw.write("CORRECTED:\n");
|
|
| 107 |
for (int i = 0; i < newCreators.size(); i++) {
|
|
| 108 |
sw.write(String.format("%5d) %s\n", i + 1, newCreators.get(i)));
|
|
| 109 |
if (newCreators.size() > 0) {
|
|
| 110 |
sw.write("CORRECTED:\n");
|
|
| 111 |
for (int i = 0; i < newCreators.size(); i++) {
|
|
| 112 |
final String s = newCreators.get(i); |
|
| 113 |
sw.write(String.format("%5d) %s\n", i + 1, StringUtils.isNotBlank(s) ? s : "[empty]"));
|
|
| 114 |
} |
|
| 109 | 115 |
} |
| 110 | 116 |
return sw.toString(); |
| 111 | 117 |
} |
| ... | ... | |
| 114 | 120 |
|
| 115 | 121 |
@Override |
| 116 | 122 |
protected String execute(final Env env) throws Exception {
|
| 123 |
|
|
| 117 | 124 |
@SuppressWarnings("unchecked")
|
| 125 |
final ResultSet<String> rsIn = env.getAttribute(inputEprParam, ResultSet.class); |
|
| 118 | 126 |
|
| 119 |
final ResultSet<String> rsIn = env.getAttribute(inputEprParam, ResultSet.class); |
|
| 120 | 127 |
final SimpleDataTransformer f = dataTransformerFactory.createTransformer(ruleId); |
| 121 | 128 |
final SAXReader reader = new SAXReader(); |
| 122 | 129 |
|
| 123 | 130 |
final String fileName = "/tmp/report_" + (new SimpleDateFormat("yyyyMMdd_HHmmss_S")).format(new Date()) + ".txt";
|
| 124 | 131 |
|
| 125 |
try (PrintWriter pw = new PrintWriter(Files.newBufferedWriter(Paths.get(fileName)))) {
|
|
| 132 |
try (PrintWriter pw = new PrintWriter(Files.newBufferedWriter(Paths.get(fileName), StandardCharsets.UTF_8))) {
|
|
| 126 | 133 |
StreamSupport.stream(resultSetClient.iter(rsIn, String.class).spliterator(), false).map( |
| 127 | 134 |
xml -> {
|
| 128 | 135 |
try {
|
| 129 | 136 |
final Document docOld = reader.read(new StringReader(xml)); |
| 130 | 137 |
final Document docNew = reader.read(new StringReader(f.apply(xml))); |
| 131 | 138 |
|
| 132 |
final SinglePersonReport report = new SinglePersonReport(docNew.valueOf("//*[local-name() = 'title']"));
|
|
| 139 |
final String id = docNew.valueOf("//*[local-name() = 'recordIdentifier']");
|
|
| 140 |
final String title = docNew.valueOf("//*[local-name() = 'title']");
|
|
| 141 |
|
|
| 142 |
final SinglePersonReport report = new SinglePersonReport(id, title); |
|
| 133 | 143 |
for (final Object o : docOld.selectNodes("//*[local-name() = 'creator']")) {
|
| 134 | 144 |
final String p1 = ((Node) o).valueOf("./*[local-name() = 'surname']").trim();
|
| 135 | 145 |
final String p2 = ((Node) o).valueOf("./*[local-name() = 'name']").trim();
|
Also available in: Unified diff
wf to generate a report of corrected authors