From 22955b5b2368b17aab97d032024f2c5e4f55c625 Mon Sep 17 00:00:00 2001
From: William D C M SILVA <colen@apache.org>
Date: Mon, 2 Jan 2017 23:18:11 -0200
Subject: [PATCH] Adds cache to Chunker

See issue OPENNLP-274
---
 .../opennlp/tools/chunker/ChunkerFactory.java |  4 ++
 .../java/opennlp/tools/chunker/ChunkerME.java |  6 +--
 .../DefaultChunkerContextGenerator.java       | 51 +++++++++++++++++++
 3 files changed, 58 insertions(+), 3 deletions(-)

diff --git a/opennlp-tools/src/main/java/opennlp/tools/chunker/ChunkerFactory.java b/opennlp-tools/src/main/java/opennlp/tools/chunker/ChunkerFactory.java
index 1cb772fc1..132328444 100644
--- a/opennlp-tools/src/main/java/opennlp/tools/chunker/ChunkerFactory.java
+++ b/opennlp-tools/src/main/java/opennlp/tools/chunker/ChunkerFactory.java
@@ -60,4 +60,8 @@ public SequenceValidator<String> getSequenceValidator() {
   public ChunkerContextGenerator getContextGenerator() {
     return new DefaultChunkerContextGenerator();
   }
+
+  public ChunkerContextGenerator getContextGenerator(int cache) {
+    return new DefaultChunkerContextGenerator(cache);
+  }
 }
diff --git a/opennlp-tools/src/main/java/opennlp/tools/chunker/ChunkerME.java b/opennlp-tools/src/main/java/opennlp/tools/chunker/ChunkerME.java
index 3ed4f9c2c..3e0272ff0 100644
--- a/opennlp-tools/src/main/java/opennlp/tools/chunker/ChunkerME.java
+++ b/opennlp-tools/src/main/java/opennlp/tools/chunker/ChunkerME.java
@@ -94,7 +94,7 @@ private ChunkerME(ChunkerModel model, int beamSize, SequenceValidator<String> se
   @Deprecated
   private ChunkerME(ChunkerModel model, int beamSize) {
 
-   contextGenerator = model.getFactory().getContextGenerator();
+   contextGenerator = model.getFactory().getContextGenerator(beamSize);
    sequenceValidator = model.getFactory().getSequenceValidator();
 
     if (model.getChunkerSequenceModel() != null) {
@@ -177,7 +177,7 @@ public static ChunkerModel train(String lang, ObjectStream<ChunkSample> in,
     SequenceClassificationModel<String> seqChunkerModel = null;
 
     if (TrainerType.EVENT_MODEL_TRAINER.equals(trainerType)) {
-      ObjectStream<Event> es = new ChunkerEventStream(in, factory.getContextGenerator());
+      ObjectStream<Event> es = new ChunkerEventStream(in, factory.getContextGenerator(beamSize));
       EventTrainer trainer = TrainerFactory.getEventTrainer(mlParams.getSettings(),
           manifestInfoEntries);
       chunkerModel = trainer.train(es);
@@ -188,7 +188,7 @@ else if (TrainerType.SEQUENCE_TRAINER.equals(trainerType)) {
 
       // TODO: This will probably cause issue, since the feature generator uses the outcomes array
 
-      ChunkSampleSequenceStream ss = new ChunkSampleSequenceStream(in, factory.getContextGenerator());
+      ChunkSampleSequenceStream ss = new ChunkSampleSequenceStream(in, factory.getContextGenerator(beamSize));
       seqChunkerModel = trainer.train(ss);
     }
     else {
diff --git a/opennlp-tools/src/main/java/opennlp/tools/chunker/DefaultChunkerContextGenerator.java b/opennlp-tools/src/main/java/opennlp/tools/chunker/DefaultChunkerContextGenerator.java
index 3bf4ba452..728049c6a 100644
--- a/opennlp-tools/src/main/java/opennlp/tools/chunker/DefaultChunkerContextGenerator.java
+++ b/opennlp-tools/src/main/java/opennlp/tools/chunker/DefaultChunkerContextGenerator.java
@@ -18,16 +18,38 @@
 
 package opennlp.tools.chunker;
 
+import java.util.Objects;
+
+import opennlp.tools.util.Cache;
+
 /** Features based on chunking model described in Fei Sha and Fernando Pereira. Shallow
  *  parsing with conditional random fields. In Proceedings of HLT-NAACL 2003. Association
  *  for Computational Linguistics, 2003.
  */
 public class DefaultChunkerContextGenerator implements ChunkerContextGenerator {
 
+  private Cache<String, String[]> contextsCache;
+  private Object wordsKey;
+  
+  protected static final String SB = "*SB*";
+  
   /**
    * Creates the default context generator a chunker.
    */
   public DefaultChunkerContextGenerator() {
+    this(0);
+  }
+  
+
+  /**
+   * Creates the default context generator a chunker.
+   * 
+   * @param cacheSize
+   */
+  public DefaultChunkerContextGenerator(int cacheSize) {
+    if (cacheSize > 0) {
+      contextsCache = new Cache<>(cacheSize);
+    }
   }
 
   public String[] getContext(int index, String[] sequence, String[] priorDecisions, Object[] additionalContext) {
@@ -35,6 +57,8 @@ public String[] getContext(int index, String[] sequence, String[] priorDecisions
   }
 
   public String[] getContext(int i, String[] toks, String[] tags, String[] preds) {
+    String predprev = null, predprevprev = null;
+    
 	// Words in a 5-word window
     String w_2, w_1, w0, w1, w2;
 
@@ -43,6 +67,30 @@ public String[] getContext(int i, String[] toks, String[] tags, String[] preds)
 
     // Previous predictions
     String p_2, p_1;
+    
+    String cacheKey = null;
+    if (contextsCache != null) {
+      if (i - 1 >= 0) {
+        predprev = preds[i - 1];
+
+        if (i - 2 >= 0) {
+          predprevprev = preds[i - 2];
+        }
+      }
+
+      cacheKey = i + predprev + predprevprev;
+      if (Objects.equals(wordsKey, toks)) {
+        // same sentence
+        String[] cachedContexts = contextsCache.get(cacheKey);
+        if (cachedContexts != null) {
+          return cachedContexts;
+        }
+      } else {
+        // new sentence
+        contextsCache.clear();
+        wordsKey = toks;
+      }
+    }
 
     if (i < 2) {
       w_2 = "w_2=bos";
@@ -140,6 +188,9 @@ public String[] getContext(int i, String[] toks, String[] tags, String[] preds)
         p_1 + w0 + w1
     };
 
+    if (contextsCache != null) {
+      contextsCache.put(cacheKey,features);
+    }
     return features;
   }
 }