From 59d3cd598e0c54464005f7ff19ad9a32ea936117 Mon Sep 17 00:00:00 2001
From: Peter Wu <peter@lekensteyn.nl>
Date: Sat, 4 Apr 2015 13:09:26 +0200
Subject: Debugging tweaks, be less verbose unless requested

---
 spellchecker/run.sh                  |  4 ++++
 spellchecker/src/CorpusReader.java   |  2 ++
 spellchecker/src/SpellCorrector.java | 28 ++++++++++++++++++----------
 3 files changed, 24 insertions(+), 10 deletions(-)

diff --git a/spellchecker/run.sh b/spellchecker/run.sh
index 956ce67..e2df2d4 100755
--- a/spellchecker/run.sh
+++ b/spellchecker/run.sh
@@ -2,6 +2,10 @@
 classpath="${CLASSPATH:+$CLASSPATH:}"
 classpath+=build/classes
 
+# Possible env vars:
+#DEBUG_SCORE=1  # Enable debugging of the score
+#NO_PEACH=1     # Print less debug information, use peach I/O formats
+
 ant -q -S compile >&2 &&
 NO_PEACH=1 \
 java -ea -cp "$classpath" SpellChecker
diff --git a/spellchecker/src/CorpusReader.java b/spellchecker/src/CorpusReader.java
index 2e0a855..371c916 100644
--- a/spellchecker/src/CorpusReader.java
+++ b/spellchecker/src/CorpusReader.java
@@ -159,6 +159,8 @@ public class CorpusReader {
             p = (a + .001) / (b + 1);
         }
 
+        //System.err.println("P(" + word + "|" + ngram + ") = " + p);
+
         return p;
     }
 
diff --git a/spellchecker/src/SpellCorrector.java b/spellchecker/src/SpellCorrector.java
index fbbfbef..a015ac2 100644
--- a/spellchecker/src/SpellCorrector.java
+++ b/spellchecker/src/SpellCorrector.java
@@ -33,6 +33,13 @@ public class SpellCorrector {
         this.cmr = cmr;
     }
 
+    private void debugPrint(String str) {
+        // print debugging information if NO_PEACH is set.
+        if (System.getenv("NO_PEACH") != null) {
+            System.err.println(str);
+        }
+    }
+
     /**
      * Converts a set of word indices which were modified to an array of length
      * {@code n} of which elements are set to true when those should not be
@@ -70,6 +77,7 @@ public class SpellCorrector {
 
             // try to find a better suggestion for this word.
             Map<String, Double> candidates = getCandidateWords(old_word);
+            debugPrint("\n\n LOOKING AT " + i + " " + old_word);
             candidates.forEach((word, channel_probability) -> {
                 rater.tryWord(word_index, word, channel_probability);
             });
@@ -88,19 +96,19 @@ public class SpellCorrector {
                     if (subResult != null) {
                         // Yes it did. Is this the best one so far?
                         if (subResult.getBestScore() > rater.getBestScore()) {
-                            System.err.println("Subresult is better!");
+                            debugPrint("Subresult is better!");
                             bestResult = subResult;
                         } else {
-                            System.err.println("Subresult is not better!");
+                            debugPrint("Subresult is not better!");
                         }
                     } else {
-                        System.err.println("no subresult found");
+                        debugPrint("no subresult found");
                     }
                 }
                 // make the context to be editable again
                 modifiedIndices.remove(i);
             } else {
-                System.err.println("No suggestion found for " + old_word + ".");
+                debugPrint("No suggestion found for " + old_word + ".");
             }
         }
         return bestResult;
@@ -272,7 +280,9 @@ public class SpellCorrector {
                         LM_PROBABILITY_UNMODIFIED);
             }
             sentence_probability = combineProbabilities(word_likelihoods);
-            debugScore();
+            if (DEBUG_SCORE) {
+                debugScore();
+            }
         }
 
         /**
@@ -306,10 +316,7 @@ public class SpellCorrector {
 
             assert channel_probability > 0.0;
             String debug_word = null;
-            if (DEBUG_SCORE
-                    && (word.equals("he")
-                    || word.equals("hme")
-                    || word.equals("home"))) {
+            if (DEBUG_SCORE) {
                 debug_word = "";
             }
 
@@ -351,7 +358,7 @@ public class SpellCorrector {
             p = prior * channel_probability;
 
             if (debug_word != null) {
-                System.err.println("# " + word + " p=" + p
+                debugPrint("# " + word + " p=" + p
                         + " chan=" + channel_probability
                         + " prior=" + prior + debug_word);
             }
@@ -413,6 +420,7 @@ public class SpellCorrector {
             // better than the original sentence.
             if ((best_modification != null && effect.probability > best_modification.probability)
                     || (best_modification == null && effect.probability > sentence_probability)) {
+                debugPrint("Found better word! " + word + " p=" + effect.probability);
                 best_modification = effect;
             }
         }
-- 
cgit v1.2.1