Reading dictionary containing timestamps in Java Side.

Just skipping historical information fields.

Bug: 11281877
Change-Id: I43d2adaa576b7da11ed3ca54990265dbb6f53b08

6 files changed, 48 insertions, 52 deletions

diff --git a/java/src/com/android/inputmethod/latin/makedict/AbstractDictDecoder.java b/java/src/com/android/inputmethod/latin/makedict/AbstractDictDecoder.java
index f8fa68f45..370782b33 100644
--- a/java/src/com/android/inputmethod/latin/makedict/AbstractDictDecoder.java
+++ b/java/src/com/android/inputmethod/latin/makedict/AbstractDictDecoder.java
@@ -48,7 +48,7 @@ public abstract class AbstractDictDecoder implements DictDecoder {
           throw new UnsupportedFormatException("Unsupported version : " + version);
         }
         // TODO: Remove this field.
-        final int optionsFlags = HeaderReader.readOptionFlags(headerBuffer);
+        HeaderReader.readOptionFlags(headerBuffer);
         final int headerSize = HeaderReader.readHeaderSize(headerBuffer);
         if (headerSize < 0) {
             throw new UnsupportedFormatException("header size can't be negative.");
@@ -59,8 +59,8 @@ public abstract class AbstractDictDecoder implements DictDecoder {
 
         final FileHeader header = new FileHeader(headerSize,
                 new FusionDictionary.DictionaryOptions(attributes),
-                new FormatOptions(version,
-                        0 != (optionsFlags & FormatSpec.CONTAINS_TIMESTAMP_FLAG)));
+                new FormatOptions(version, FileHeader.ATTRIBUTE_VALUE_TRUE.equals(
+                        attributes.get(FileHeader.HAS_HISTORICAL_INFO_ATTRIBUTE))));
         return header;
     }
 
diff --git a/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java b/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java
index 9a24c47af..31747155e 100644
--- a/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java
+++ b/java/src/com/android/inputmethod/latin/makedict/BinaryDictDecoderUtils.java
@@ -499,7 +499,6 @@ public final class BinaryDictDecoderUtils {
         final int nodeArrayOriginPos = dictDecoder.getPosition();
 
         do { // Scan the linked-list node.
-            final int nodeArrayHeadPos = dictDecoder.getPosition();
             final int count = dictDecoder.readPtNodeCount();
             int groupPos = dictDecoder.getPosition();
             for (int i = count; i > 0; --i) { // Scan the array of PtNode.
diff --git a/java/src/com/android/inputmethod/latin/makedict/BinaryDictEncoderUtils.java b/java/src/com/android/inputmethod/latin/makedict/BinaryDictEncoderUtils.java
index bb40e0dd5..eff8fc375 100644
--- a/java/src/com/android/inputmethod/latin/makedict/BinaryDictEncoderUtils.java
+++ b/java/src/com/android/inputmethod/latin/makedict/BinaryDictEncoderUtils.java
@@ -756,14 +756,6 @@ public class BinaryDictEncoderUtils {
     }
 
     /**
-     * Makes the 2-byte value for options flags. Unused at the moment, and always 0.
-     */
-    private static final int makeOptionsValue(final FormatOptions formatOptions) {
-        // TODO: why doesn't this handle CONTAINS_TIMESTAMP_FLAG?
-        return 0;
-    }
-
-    /**
      * Makes the flag value for a shortcut.
      *
      * @param more whether there are more attributes after this one.
@@ -949,7 +941,8 @@ public class BinaryDictEncoderUtils {
         headerBuffer.write((byte) (0xFF & version));
 
         // Options flags
-        final int options = makeOptionsValue(formatOptions);
+        // TODO: Remove this field.
+        final int options = 0;
         headerBuffer.write((byte) (0xFF & (options >> 8)));
         headerBuffer.write((byte) (0xFF & options));
         final int headerSizeOffset = headerBuffer.size();
diff --git a/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java b/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java
index 437fa942b..61c17fc46 100644
--- a/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java
+++ b/java/src/com/android/inputmethod/latin/makedict/FormatSpec.java
@@ -192,10 +192,6 @@ public final class FormatSpec {
     static final int MINIMUM_SUPPORTED_VERSION = VERSION2;
     static final int MAXIMUM_SUPPORTED_VERSION = VERSION4;
 
-    // These options need to be the same numeric values as the one in the native reading code.
-    // TODO: Make the native reading code read this variable.
-    static final int CONTAINS_TIMESTAMP_FLAG = 0x10;
-
     // TODO: Make this value adaptative to content data, store it in the header, and
     // use it in the reading code.
     static final int MAX_WORD_LENGTH = Constants.DICTIONARY_MAX_WORD_LENGTH;
@@ -249,26 +245,26 @@ public final class FormatSpec {
     static final String TRIE_FILE_EXTENSION = ".trie";
     public static final String HEADER_FILE_EXTENSION = ".header";
     static final String FREQ_FILE_EXTENSION = ".freq";
-    static final String UNIGRAM_TIMESTAMP_FILE_EXTENSION = ".timestamp";
     // tat = Terminal Address Table
     static final String TERMINAL_ADDRESS_TABLE_FILE_EXTENSION = ".tat";
     static final String BIGRAM_FILE_EXTENSION = ".bigram";
     static final String SHORTCUT_FILE_EXTENSION = ".shortcut";
     static final String LOOKUP_TABLE_FILE_SUFFIX = "_lookup";
     static final String CONTENT_TABLE_FILE_SUFFIX = "_index";
+    static final int FLAGS_IN_FREQ_FILE_SIZE = 1;
     static final int FREQUENCY_AND_FLAGS_SIZE = 2;
     static final int TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE = 3;
     static final int UNIGRAM_TIMESTAMP_SIZE = 4;
+    static final int UNIGRAM_COUNTER_SIZE = 1;
+    static final int UNIGRAM_LEVEL_SIZE = 1;
 
     // With the English main dictionary as of October 2013, the size of bigram address table is
     // is 345KB with the block size being 16.
     // This is 54% of that of full address table.
     static final int BIGRAM_ADDRESS_TABLE_BLOCK_SIZE = 16;
-    static final int BIGRAM_CONTENT_COUNT = 2;
+    static final int BIGRAM_CONTENT_COUNT = 1;
     static final int BIGRAM_FREQ_CONTENT_INDEX = 0;
-    static final int BIGRAM_TIMESTAMP_CONTENT_INDEX = 1;
     static final String BIGRAM_FREQ_CONTENT_ID = "_freq";
-    static final String BIGRAM_TIMESTAMP_CONTENT_ID = "_timestamp";
     static final int BIGRAM_TIMESTAMP_SIZE = 4;
     static final int BIGRAM_COUNTER_SIZE = 1;
     static final int BIGRAM_LEVEL_SIZE = 1;
diff --git a/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java b/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java
index fdf2ae7b5..5b0e8399a 100644
--- a/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java
+++ b/java/src/com/android/inputmethod/latin/makedict/FusionDictionary.java
@@ -61,6 +61,7 @@ public final class FusionDictionary implements Iterable<Word> {
             mData = new ArrayList<PtNode>();
         }
         public PtNodeArray(ArrayList<PtNode> data) {
+            Collections.sort(data, PTNODE_COMPARATOR);
             mData = data;
         }
     }
diff --git a/java/src/com/android/inputmethod/latin/makedict/Ver4DictDecoder.java b/java/src/com/android/inputmethod/latin/makedict/Ver4DictDecoder.java
index 7071893d2..f23022992 100644
--- a/java/src/com/android/inputmethod/latin/makedict/Ver4DictDecoder.java
+++ b/java/src/com/android/inputmethod/latin/makedict/Ver4DictDecoder.java
@@ -143,7 +143,7 @@ public class Ver4DictDecoder extends AbstractDictDecoder {
         mTerminalAddressTableBuffer = mBufferFactory.getDictionaryBuffer(
                 getFile(FILETYPE_TERMINAL_ADDRESS_TABLE));
         mBigramReader = new BigramContentReader(mDictDirectory.getName(),
-                mDictDirectory, mBufferFactory, false);
+                mDictDirectory, mBufferFactory);
         mBigramReader.openBuffers();
         mShortcutReader = new ShortcutContentReader(mDictDirectory.getName(), mDictDirectory,
                 mBufferFactory);
@@ -184,39 +184,24 @@ public class Ver4DictDecoder extends AbstractDictDecoder {
      */
     protected static class BigramContentReader extends SparseTableContentReader {
         public BigramContentReader(final String name, final File baseDir,
-                final DictionaryBufferFactory factory, final boolean hasTimestamp) {
+                final DictionaryBufferFactory factory) {
             super(name + FormatSpec.BIGRAM_FILE_EXTENSION,
                     FormatSpec.BIGRAM_ADDRESS_TABLE_BLOCK_SIZE, baseDir,
-                    getContentFilenames(name, hasTimestamp), getContentIds(hasTimestamp), factory);
+                    getContentFilenames(name), getContentIds(), factory);
         }
 
         // TODO: Consolidate this method and BigramContentWriter.getContentFilenames.
-        protected static String[] getContentFilenames(final String name,
-                final boolean hasTimestamp) {
-            final String[] contentFilenames;
-            if (hasTimestamp) {
-                contentFilenames = new String[] { name + FormatSpec.BIGRAM_FILE_EXTENSION,
-                        name + FormatSpec.BIGRAM_FILE_EXTENSION };
-            } else {
-                contentFilenames = new String[] { name + FormatSpec.BIGRAM_FILE_EXTENSION };
-            }
-            return contentFilenames;
+        protected static String[] getContentFilenames(final String name) {
+            return new String[] { name + FormatSpec.BIGRAM_FILE_EXTENSION };
         }
 
         // TODO: Consolidate this method and BigramContentWriter.getContentIds.
-        protected static String[] getContentIds(final boolean hasTimestamp) {
-            final String[] contentIds;
-            if (hasTimestamp) {
-                contentIds = new String[] { FormatSpec.BIGRAM_FREQ_CONTENT_ID,
-                        FormatSpec.BIGRAM_TIMESTAMP_CONTENT_ID };
-            } else {
-                contentIds = new String[] { FormatSpec.BIGRAM_FREQ_CONTENT_ID };
-            }
-            return contentIds;
+        protected static String[] getContentIds() {
+            return new String[] { FormatSpec.BIGRAM_FREQ_CONTENT_ID };
         }
 
         public ArrayList<PendingAttribute> readTargetsAndFrequencies(final int terminalId,
-                final DictBuffer terminalAddressTableBuffer) {
+                final DictBuffer terminalAddressTableBuffer, final FormatOptions options) {
             final ArrayList<PendingAttribute> bigrams = CollectionUtils.newArrayList();
             read(FormatSpec.BIGRAM_FREQ_CONTENT_INDEX, terminalId,
                     new SparseTableContentReaderInterface() {
@@ -226,14 +211,25 @@ public class Ver4DictDecoder extends AbstractDictDecoder {
                                 // If bigrams.size() reaches FormatSpec.MAX_BIGRAMS_IN_A_PTNODE,
                                 // remaining bigram entries are ignored.
                                 final int bigramFlags = buffer.readUnsignedByte();
+                                final int probability;
+
+                                if (options.mHasTimestamp) {
+                                    probability = buffer.readUnsignedByte();
+                                    final int pos = buffer.position();
+                                    // Skip historical info.
+                                    buffer.position(pos + FormatSpec.BIGRAM_TIMESTAMP_SIZE
+                                            + FormatSpec.BIGRAM_LEVEL_SIZE
+                                            + FormatSpec.BIGRAM_COUNTER_SIZE);
+                                } else {
+                                    probability = bigramFlags
+                                            & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY;
+                                }
                                 final int targetTerminalId = buffer.readUnsignedInt24();
                                 terminalAddressTableBuffer.position(targetTerminalId
                                         * FormatSpec.TERMINAL_ADDRESS_TABLE_ADDRESS_SIZE);
                                 final int targetAddress =
                                         terminalAddressTableBuffer.readUnsignedInt24();
-                                bigrams.add(new PendingAttribute(bigramFlags
-                                        & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_FREQUENCY,
-                                        targetAddress));
+                                bigrams.add(new PendingAttribute(probability, targetAddress));
                                 if (0 == (bigramFlags
                                         & FormatSpec.FLAG_BIGRAM_SHORTCUT_ATTR_HAS_NEXT)) {
                                     break;
@@ -286,8 +282,19 @@ public class Ver4DictDecoder extends AbstractDictDecoder {
     }
 
     protected static class PtNodeReader extends AbstractDictDecoder.PtNodeReader {
-        protected static int readFrequency(final DictBuffer frequencyBuffer, final int terminalId) {
-            frequencyBuffer.position(terminalId * FormatSpec.FREQUENCY_AND_FLAGS_SIZE + 1);
+        protected static int readFrequency(final DictBuffer frequencyBuffer, final int terminalId,
+                final FormatOptions formatOptions) {
+            final int readingPos;
+            if (formatOptions.mHasTimestamp) {
+                final int entrySize = FormatSpec.FREQUENCY_AND_FLAGS_SIZE
+                        + FormatSpec.UNIGRAM_TIMESTAMP_SIZE + FormatSpec.UNIGRAM_LEVEL_SIZE
+                        + FormatSpec.UNIGRAM_COUNTER_SIZE;
+                readingPos = terminalId * entrySize + FormatSpec.FLAGS_IN_FREQ_FILE_SIZE;
+            } else {
+                readingPos = terminalId * FormatSpec.FREQUENCY_AND_FLAGS_SIZE
+                        + FormatSpec.FLAGS_IN_FREQ_FILE_SIZE;
+            }
+            frequencyBuffer.position(readingPos);
             return frequencyBuffer.readUnsignedByte();
         }
 
@@ -354,12 +361,12 @@ public class Ver4DictDecoder extends AbstractDictDecoder {
     }
 
     @Override
-    public PtNodeInfo readPtNode(int ptNodePos, FormatOptions options) {
+    public PtNodeInfo readPtNode(final int ptNodePos, final FormatOptions options) {
         final Ver4PtNodeInfo nodeInfo = readVer4PtNodeInfo(ptNodePos, options);
 
         final int frequency;
         if (0 != (FormatSpec.FLAG_IS_TERMINAL & nodeInfo.mFlags)) {
-            frequency = PtNodeReader.readFrequency(mFrequencyBuffer, nodeInfo.mTerminalId);
+            frequency = PtNodeReader.readFrequency(mFrequencyBuffer, nodeInfo.mTerminalId, options);
         } else {
             frequency = PtNode.NOT_A_TERMINAL;
         }
@@ -367,7 +374,7 @@ public class Ver4DictDecoder extends AbstractDictDecoder {
         final ArrayList<WeightedString> shortcutTargets = mShortcutReader.readShortcuts(
                 nodeInfo.mTerminalId);
         final ArrayList<PendingAttribute> bigrams = mBigramReader.readTargetsAndFrequencies(
-                nodeInfo.mTerminalId, mTerminalAddressTableBuffer);
+                nodeInfo.mTerminalId, mTerminalAddressTableBuffer, options);
 
         return new PtNodeInfo(ptNodePos, ptNodePos + nodeInfo.mNodeSize, nodeInfo.mFlags,
                 nodeInfo.mCharacters, frequency, nodeInfo.mParentPos, nodeInfo.mChildrenPos,