am d64b8c97: Merge "Read the group count correctly from the binary format"
* commit 'd64b8c97fea0ab5dd6d16feef329127dc7ecc833': Read the group count correctly from the binary formatmain
commit
082d3f6255
|
@ -264,20 +264,29 @@ public class BinaryDictInputOutput {
|
||||||
return size;
|
return size;
|
||||||
}
|
}
|
||||||
|
|
||||||
|
/**
|
||||||
|
* Compute the binary size of the group count
|
||||||
|
* @param count the group count
|
||||||
|
* @return the size of the group count, either 1 or 2 bytes.
|
||||||
|
*/
|
||||||
|
private static int getGroupCountSize(final int count) {
|
||||||
|
if (MAX_CHARGROUPS_FOR_ONE_BYTE_CHARGROUP_COUNT >= count) {
|
||||||
|
return 1;
|
||||||
|
} else if (MAX_CHARGROUPS_IN_A_NODE >= count) {
|
||||||
|
return 2;
|
||||||
|
} else {
|
||||||
|
throw new RuntimeException("Can't have more than " + MAX_CHARGROUPS_IN_A_NODE
|
||||||
|
+ " groups in a node (found " + count +")");
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
/**
|
/**
|
||||||
* Compute the binary size of the group count for a node
|
* Compute the binary size of the group count for a node
|
||||||
* @param node the node
|
* @param node the node
|
||||||
* @return the size of the group count, either 1 or 2 bytes.
|
* @return the size of the group count, either 1 or 2 bytes.
|
||||||
*/
|
*/
|
||||||
private static int getGroupCountSize(final Node node) {
|
private static int getGroupCountSize(final Node node) {
|
||||||
if (MAX_CHARGROUPS_FOR_ONE_BYTE_CHARGROUP_COUNT >= node.mData.size()) {
|
return getGroupCountSize(node.mData.size());
|
||||||
return 1;
|
|
||||||
} else if (MAX_CHARGROUPS_IN_A_NODE >= node.mData.size()) {
|
|
||||||
return 2;
|
|
||||||
} else {
|
|
||||||
throw new RuntimeException("Can't have more than " + MAX_CHARGROUPS_IN_A_NODE
|
|
||||||
+ " groups in a node (found " + node.mData.size() +")");
|
|
||||||
}
|
|
||||||
}
|
}
|
||||||
|
|
||||||
/**
|
/**
|
||||||
|
@ -913,7 +922,7 @@ public class BinaryDictInputOutput {
|
||||||
addressPointer += 3;
|
addressPointer += 3;
|
||||||
break;
|
break;
|
||||||
default:
|
default:
|
||||||
throw new RuntimeException("Has attribute with no address");
|
throw new RuntimeException("Has shortcut targets with no address");
|
||||||
}
|
}
|
||||||
shortcutTargets.add(new PendingAttribute(targetFlags & FLAG_ATTRIBUTE_FREQUENCY,
|
shortcutTargets.add(new PendingAttribute(targetFlags & FLAG_ATTRIBUTE_FREQUENCY,
|
||||||
targetAddress));
|
targetAddress));
|
||||||
|
@ -944,7 +953,7 @@ public class BinaryDictInputOutput {
|
||||||
addressPointer += 3;
|
addressPointer += 3;
|
||||||
break;
|
break;
|
||||||
default:
|
default:
|
||||||
throw new RuntimeException("Has attribute with no address");
|
throw new RuntimeException("Has bigrams with no address");
|
||||||
}
|
}
|
||||||
bigrams.add(new PendingAttribute(bigramFlags & FLAG_ATTRIBUTE_FREQUENCY,
|
bigrams.add(new PendingAttribute(bigramFlags & FLAG_ATTRIBUTE_FREQUENCY,
|
||||||
bigramAddress));
|
bigramAddress));
|
||||||
|
@ -955,6 +964,19 @@ public class BinaryDictInputOutput {
|
||||||
childrenAddress, shortcutTargets, bigrams);
|
childrenAddress, shortcutTargets, bigrams);
|
||||||
}
|
}
|
||||||
|
|
||||||
|
/**
|
||||||
|
* Reads and returns the char group count out of a file and forwards the pointer.
|
||||||
|
*/
|
||||||
|
private static int readCharGroupCount(RandomAccessFile source) throws IOException {
|
||||||
|
final int msb = source.readUnsignedByte();
|
||||||
|
if (MAX_CHARGROUPS_FOR_ONE_BYTE_CHARGROUP_COUNT >= msb) {
|
||||||
|
return msb;
|
||||||
|
} else {
|
||||||
|
return ((MAX_CHARGROUPS_FOR_ONE_BYTE_CHARGROUP_COUNT & msb) << 8)
|
||||||
|
+ source.readUnsignedByte();
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
/**
|
/**
|
||||||
* Finds, as a string, the word at the address passed as an argument.
|
* Finds, as a string, the word at the address passed as an argument.
|
||||||
*
|
*
|
||||||
|
@ -968,8 +990,8 @@ public class BinaryDictInputOutput {
|
||||||
int address) throws IOException {
|
int address) throws IOException {
|
||||||
final long originalPointer = source.getFilePointer();
|
final long originalPointer = source.getFilePointer();
|
||||||
source.seek(headerSize);
|
source.seek(headerSize);
|
||||||
final int count = source.readUnsignedByte();
|
final int count = readCharGroupCount(source);
|
||||||
int groupOffset = 1; // 1 for the group count
|
int groupOffset = getGroupCountSize(count);
|
||||||
final StringBuilder builder = new StringBuilder();
|
final StringBuilder builder = new StringBuilder();
|
||||||
String result = null;
|
String result = null;
|
||||||
|
|
||||||
|
@ -1025,9 +1047,9 @@ public class BinaryDictInputOutput {
|
||||||
Map<Integer, Node> reverseNodeMap, Map<Integer, CharGroup> reverseGroupMap)
|
Map<Integer, Node> reverseNodeMap, Map<Integer, CharGroup> reverseGroupMap)
|
||||||
throws IOException {
|
throws IOException {
|
||||||
final int nodeOrigin = (int)(source.getFilePointer() - headerSize);
|
final int nodeOrigin = (int)(source.getFilePointer() - headerSize);
|
||||||
final int count = source.readUnsignedByte();
|
final int count = readCharGroupCount(source);
|
||||||
final ArrayList<CharGroup> nodeContents = new ArrayList<CharGroup>();
|
final ArrayList<CharGroup> nodeContents = new ArrayList<CharGroup>();
|
||||||
int groupOffset = nodeOrigin + 1; // 1 byte for the group count
|
int groupOffset = nodeOrigin + getGroupCountSize(count);
|
||||||
for (int i = count; i > 0; --i) {
|
for (int i = count; i > 0; --i) {
|
||||||
CharGroupInfo info = readCharGroup(source, groupOffset);
|
CharGroupInfo info = readCharGroup(source, groupOffset);
|
||||||
ArrayList<WeightedString> shortcutTargets = null;
|
ArrayList<WeightedString> shortcutTargets = null;
|
||||||
|
|
Loading…
Reference in New Issue