Process wiki index
This commit is contained in:
parent
933dbf694a
commit
77f9064258
1 changed files with 26 additions and 18 deletions
|
@ -299,7 +299,9 @@ public class WikiIndexer {
|
||||||
}
|
}
|
||||||
} else if (name.equals("text")) {
|
} else if (name.equals("text")) {
|
||||||
if(parseText) {
|
if(parseText) {
|
||||||
|
if(id % 50 == 0) {
|
||||||
log.debug("Article accepted " + cid + " " + title.toString());
|
log.debug("Article accepted " + cid + " " + title.toString());
|
||||||
|
}
|
||||||
analyzeTextForGeoInfoNew();
|
analyzeTextForGeoInfoNew();
|
||||||
}
|
}
|
||||||
ctext = null;
|
ctext = null;
|
||||||
|
@ -478,6 +480,8 @@ public class WikiIndexer {
|
||||||
|
|
||||||
private void analyzeTextForGeoInfoNew() throws XMLStreamException {
|
private void analyzeTextForGeoInfoNew() throws XMLStreamException {
|
||||||
StringBuilder description = new StringBuilder();
|
StringBuilder description = new StringBuilder();
|
||||||
|
try {
|
||||||
|
|
||||||
int beg = 0;
|
int beg = 0;
|
||||||
int h = findOpenBrackets(beg);
|
int h = findOpenBrackets(beg);
|
||||||
|
|
||||||
|
@ -504,6 +508,10 @@ public class WikiIndexer {
|
||||||
// 4. Parse main subcategory name
|
// 4. Parse main subcategory name
|
||||||
processDescription(description, beg);
|
processDescription(description, beg);
|
||||||
|
|
||||||
|
} catch (RuntimeException e) {
|
||||||
|
description.setLength(0);
|
||||||
|
log.error(e.getMessage(), e);
|
||||||
|
}
|
||||||
if (description.length() > 0) {
|
if (description.length() > 0) {
|
||||||
writeNode(clat, clon, subcategory, description);
|
writeNode(clat, clon, subcategory, description);
|
||||||
}
|
}
|
||||||
|
@ -513,7 +521,7 @@ public class WikiIndexer {
|
||||||
if (text.charAt(i) != start.charAt(0)) {
|
if (text.charAt(i) != start.charAt(0)) {
|
||||||
return -1;
|
return -1;
|
||||||
}
|
}
|
||||||
for (int j = 1; j < start.length(); j++) {
|
for (int j = 1; j < start.length() && i + j < text.length(); j++) {
|
||||||
if (text.charAt(i + j) != start.charAt(j)) {
|
if (text.charAt(i + j) != start.charAt(j)) {
|
||||||
return -1;
|
return -1;
|
||||||
}
|
}
|
||||||
|
|
Loading…
Reference in a new issue