Android-builders-club/OsmAnd
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

Process wiki index

Browse source
This commit is contained in:
Victor Shcherb 2012-01-09 00:17:53 +01:00
parent 933dbf694a
commit 77f9064258
1 changed files with 26 additions and 18 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

42
DataExtractionOSM/src/net/osmand/data/index/WikiIndexer.java
View file

@ -299,7 +299,9 @@ public class WikiIndexer {
} }
} else if (name.equals("text")) { } else if (name.equals("text")) {
if(parseText) { if(parseText) {
log.debug("Article accepted " + cid + " " + title.toString()); if(id % 50 == 0) {
log.debug("Article accepted " + cid + " " + title.toString());
}
analyzeTextForGeoInfoNew(); analyzeTextForGeoInfoNew();
} }
ctext = null; ctext = null;
@ -478,32 +480,38 @@ public class WikiIndexer {
private void analyzeTextForGeoInfoNew() throws XMLStreamException { private void analyzeTextForGeoInfoNew() throws XMLStreamException {
StringBuilder description = new StringBuilder(); StringBuilder description = new StringBuilder();
int beg = 0; try {
int h = findOpenBrackets(beg);
int beg = 0;
int h = findOpenBrackets(beg);
// 1. Find main header section {{ ... lat, lon }}
while (h != -1 && text.substring(beg, h).trim().length() == 0 ) {
beg = findClosedBrackets(h);
if(beg == -1){
return;
}
beg += 2;
h = findOpenBrackets(beg);
// 1. Find main header section {{ ... lat, lon }}
while (h != -1 && text.substring(beg, h).trim().length() == 0 ) {
beg = findClosedBrackets(h);
if(beg == -1){
return;
} }
beg += 2;
h = findOpenBrackets(beg);
} // 3. Parse main subcategory name
// 3. Parse main subcategory name
// for (int j = h + 2; j < e; j++) { // for (int j = h + 2; j < e; j++) {
// if (Character.isWhitespace(text.charAt(j)) || text.charAt(j) == '|') { // if (Character.isWhitespace(text.charAt(j)) || text.charAt(j) == '|') {
// subcategory = text.substring(h + 2, j).trim(); // subcategory = text.substring(h + 2, j).trim();
// break; // break;
// } // }
// } // }
// Special case // Special case
// 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;
} }