Fix a bug where attributes would have the wrong freq (B0)

The reading code is accepting 0..15 and this outputs
0..31.

Change-Id: Ia638950dc2ce0b8743cbb6a71fbad866e9208786
This commit is contained in:
Jean Chalard 2011-12-26 20:12:25 +09:00
parent 10862aa4d0
commit d8120ff72e

View File

@ -123,6 +123,12 @@ public class XmlDictInputOutput {
private final static String BIGRAM_W2_ATTRIBUTE = "w2";
private final static String BIGRAM_FREQ_ATTRIBUTE = "p";
// In this version of the XML file, the bigram frequency is given as an int 0..XML_MAX
private final static int XML_MAX = 256;
// In memory and in the binary dictionary the bigram frequency is 0..MEMORY_MAX
private final static int MEMORY_MAX = 16;
private final static int XML_TO_MEMORY_RATIO = XML_MAX / MEMORY_MAX;
String mW1;
final HashMap<String, ArrayList<WeightedString>> mBigramsMap;
@ -138,7 +144,7 @@ public class XmlDictInputOutput {
} else if (BIGRAM_W2_TAG.equals(localName)) {
String w2 = attrs.getValue(uri, BIGRAM_W2_ATTRIBUTE);
int freq = Integer.parseInt(attrs.getValue(uri, BIGRAM_FREQ_ATTRIBUTE));
WeightedString bigram = new WeightedString(w2, freq / 8);
WeightedString bigram = new WeightedString(w2, freq / XML_TO_MEMORY_RATIO);
ArrayList<WeightedString> bigramList = mBigramsMap.get(mW1);
if (null == bigramList) bigramList = new ArrayList<WeightedString>();
bigramList.add(bigram);