add StripeStatistic api and test

apache · Feb 25, 2025 · f31ebb9 · f31ebb9
1 parent da6cfc1
commit f31ebb9
Show file tree

Hide file tree

Showing 4 changed files with 49 additions and 1 deletion.
diff --git a/c++/include/orc/Reader.hh b/c++/include/orc/Reader.hh
@@ -495,10 +495,17 @@ namespace orc {
      */
     virtual uint64_t getNumberOfStripeStatistics() const = 0;
 
+    /**
+     * Get the statistics about a stripe .
+     * @param stripeIndex the index of the stripe (0 to N-1) to get statistics about
+     * @return the statistics about that stripe without reading row group index statistics
+     */
+    virtual std::unique_ptr<Statistics> getStripeStatisticsOnly(uint64_t stripeIndex) const = 0;
+
     /**
      * Get the statistics about a stripe.
      * @param stripeIndex the index of the stripe (0 to N-1) to get statistics about
-     * @return the statistics about that stripe
+     * @return the statistics about that stripe and row group index statistics
      */
     virtual std::unique_ptr<StripeStatistics> getStripeStatistics(uint64_t stripeIndex) const = 0;
 

diff --git a/c++/src/Reader.cc b/c++/src/Reader.cc
@@ -751,6 +751,20 @@ namespace orc {
     return *(contents_->schema.get());
   }
 
+  std::unique_ptr<Statistics>
+  ReaderImpl::getStripeStatisticsOnly(uint64_t stripeIndex) const {
+    if (!isMetadataLoaded_) {
+      readMetadata();
+    }
+    if (contents_->metadata == nullptr) {
+      throw std::logic_error("No stripe statistics in file");
+    }
+    StatContext statContext(hasCorrectStatistics());
+    return std::unique_ptr<Statistics>(new StatisticsImpl(
+        contents_->metadata->stripestats(static_cast<int>(stripeIndex)),
+        statContext));
+  }
+
   std::unique_ptr<StripeStatistics> ReaderImpl::getStripeStatistics(uint64_t stripeIndex) const {
     if (!isMetadataLoaded_) {
       readMetadata();

diff --git a/c++/src/Reader.hh b/c++/src/Reader.hh
@@ -346,6 +346,8 @@ namespace orc {
 
     std::unique_ptr<ColumnStatistics> getColumnStatistics(uint32_t columnId) const override;
 
+    std::unique_ptr<Statistics> getStripeStatisticsOnly(uint64_t stripeIndex) const override;
+
     std::string getSerializedFileTail() const override;
 
     const Type& getType() const override;

diff --git a/c++/test/TestStripeIndexStatistics.cc b/c++/test/TestStripeIndexStatistics.cc
@@ -83,6 +83,31 @@ namespace orc {
         "length: "
         "8000\n",
         stringColStats->toString());
+
+    std::unique_ptr<orc::Statistics> stripeLevelStats = reader->getStripeStatisticsOnly(0);
+    const orc::IntegerColumnStatistics* stripeLevelIntColStats;
+    stripeLevelIntColStats = reinterpret_cast<const orc::IntegerColumnStatistics*>(
+        stripeLevelStats->getColumnStatistics(1));
+    EXPECT_EQ(
+        "Data type: Integer\nValues: 6000\nHas null: no\nMinimum: 1\nMaximum: 6000\nSum: "
+        "18003000\n",
+        stripeLevelIntColStats->toString());
+
+    const orc::StringColumnStatistics* stripeLevelStringColStats;
+    stripeLevelStringColStats = reinterpret_cast<const orc::StringColumnStatistics*>(
+        stripeLevelStats->getColumnStatistics(2));
+    EXPECT_EQ(
+        "Data type: String\nValues: 6000\nHas null: no\nMinimum: 1000\nMaximum: 9a\nTotal length: "
+        "23892\n",
+        stripeLevelStringColStats->toString());
+
+    intColStats =
+        reinterpret_cast<const orc::IntegerColumnStatistics*>(stripeStats->getColumnStatistics(1));
+    stringColStats =
+        reinterpret_cast<const orc::StringColumnStatistics*>(stripeStats->getColumnStatistics(2));
+
+    EXPECT_EQ(intColStats->toString(), stripeLevelStringColStats->toString());
+    EXPECT_EQ(stringColStats->toString(), stripeLevelStringColStats->toString());
   }
 
 }  // namespace orc