Improved error handling in TextScoring.cpp. Better constructors in Te…

…xtScoring.h.
ad-freiburg · Feb 27, 2025 · 4c09e59 · 4c09e59
1 parent 563ce80
commit 4c09e59
Show file tree

Hide file tree

Showing 2 changed files with 32 additions and 39 deletions.
diff --git a/src/index/TextScoring.cpp b/src/index/TextScoring.cpp
@@ -71,14 +71,17 @@ void ScoreData::addDocumentOrLiteralToScoreDataInvertedIndex(
 
 // ____________________________________________________________________________
 float ScoreData::getScore(WordIndex wordIndex, TextRecordIndex contextId) {
+  AD_CORRECTNESS_CHECK(!(scoringMetric_ == TextScoringMetric::EXPLICIT),
+                       "This method shouldn't be called for explicit scores.");
   // Retrieve inner map
-  if (!invertedIndex_.contains(wordIndex)) {
+  auto it = invertedIndex_.find(wordIndex);
+  if (it == invertedIndex_.end()) {
     LOG(DEBUG) << "Didn't find word in Inverted Scoring Index. WordId: "
                << wordIndex << std::endl;
     return 0;
   }
   calculateAVDL();
-  InnerMap& innerMap = invertedIndex_.find(wordIndex)->second;
+  InnerMap& innerMap = it->second;
   size_t df = innerMap.size();
   float idf = std::log2f(nofDocuments_ / df);
 
@@ -92,18 +95,17 @@ float ScoreData::getScore(WordIndex wordIndex, TextRecordIndex contextId) {
     docId = DocumentIndex::make(contextId.get());
   } else {
     auto it = docIdSet_.upper_bound(convertedContextId);
-    if (it == docIdSet_.end()) {
-      if (docIdSet_.empty()) {
-        AD_THROW("docIdSet is empty and shouldn't be");
-      }
-      LOG(DEBUG) << "Requesting a contextId that is bigger than the largest "
-                    "docId. contextId: "
-                 << contextId.get() << " Largest docId: " << *docIdSet_.rbegin()
-                 << std::endl;
-      return 0;
-    } else {
-      docId = *it;
-    }
+    AD_CORRECTNESS_CHECK(!docIdSet_.empty(),
+                         "docIdSet is empty and shouldn't be.");
+    AD_CORRECTNESS_CHECK(
+        !(it == docIdSet_.end()),
+        absl::StrCat("Requesting a contextId that is bigger than the largest "
+                     "docId. Requested contextId: ",
+                     contextId.get(),
+                     " Largest docId: ", docIdSet_.rbegin()->get(),
+                     " This hints on faulty input data for wordsfile.tsv and "
+                     "or docsfile.tsv"));
+    docId = *it;
   }
   auto ret1 = innerMap.find(docId);
   if (ret1 == innerMap.end()) {
@@ -118,12 +120,13 @@ float ScoreData::getScore(WordIndex wordIndex, TextRecordIndex contextId) {
   }
 
   auto ret2 = docLengthMap_.find(docId);
-  if (ret2 == docLengthMap_.end()) {
-    LOG(DEBUG)
-        << "The calculated docId doesn't exist in the dochLengthMap. docId: "
-        << docId << std::endl;
-    return 0;
-  }
+  AD_CORRECTNESS_CHECK(
+      !(ret2 == docLengthMap_.end()),
+      absl::StrCat("The calculated docId doesn't exist in the docLengthMap. "
+                   "The requested contextId was: ",
+                   contextId.get(), " The calculated docId was: ", docId.get(),
+                   " This hints on faulty input data for wordsfile.tsv and or "
+                   "docsfile.tsv"));
   size_t dl = ret2->second;
   float alpha = (1 - b_ + b_ * (dl / averageDocumentLength_));
   float tf_star = (tf * (k_ + 1)) / (k_ * alpha + tf);

diff --git a/src/index/TextScoring.h b/src/index/TextScoring.h
@@ -9,20 +9,12 @@
 
 class ScoreData {
  public:
-  ScoreData()
-      : scoringMetric_(TextScoringMetric::EXPLICIT), b_(0.75), k_(1.75){};
+  ScoreData() = default;
 
-  ScoreData(LocaleManager localeManager)
-      : scoringMetric_(TextScoringMetric::EXPLICIT),
-        b_(0.75),
-        k_(1.75),
-        localeManager_(localeManager){};
+  ScoreData(LocaleManager localeManager) : localeManager_(localeManager){};
 
   ScoreData(LocaleManager localeManager, TextScoringMetric scoringMetric)
-      : scoringMetric_(scoringMetric),
-        b_(0.75),
-        k_(1.75),
-        localeManager_(localeManager){};
+      : scoringMetric_(scoringMetric), localeManager_(localeManager){};
 
   ScoreData(LocaleManager localeManager, TextScoringMetric scoringMetric,
             std::pair<float, float> bAndKParam)
@@ -31,21 +23,19 @@ class ScoreData {
         k_(bAndKParam.second),
         localeManager_(localeManager){};
 
-  // Getters
-  TextScoringMetric getScoringMetric() { return scoringMetric_; }
+  TextScoringMetric getScoringMetric() const { return scoringMetric_; }
+
+  float getScore(WordIndex wordIndex, TextRecordIndex contextId);
 
-  // Functions
   void calculateScoreData(const string& docsFileName, bool addWordsFromLiterals,
                           const Index::TextVocab& textVocab,
                           const Index::Vocab& vocab);
 
-  float getScore(WordIndex wordIndex, TextRecordIndex contextId);
-
  private:
   //
-  TextScoringMetric scoringMetric_;
-  float b_;
-  float k_;
+  TextScoringMetric scoringMetric_ = TextScoringMetric::EXPLICIT;
+  float b_ = 0.75;
+  float k_ = 1.75;
 
   //
   LocaleManager localeManager_;