Groonga - fulltext search engine.

[Groonga-commit] groonga/grnxx at 96ad811 [master] Update Column::contains/find_one() to use an index if available. (#125)

susumu.yata null+****@clear*****
Thu Dec 4 12:26:20 JST 2014

• Previous message: [Groonga-commit] groonga/groonga.org at a24bfbc [gh-pages] ja: add a case example of Rroonga at CrowdWorks
• Next message: [Groonga-commit] groonga/grnxx at 033f33a [master] Reduce the scan range of Column<Int>::contains/find_one().
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

susumu.yata	2014-12-04 12:26:20 +0900 (Thu, 04 Dec 2014)

  New Revision: 96ad811aa1fc8272ae778b43d3879c0b1e617788
  https://github.com/groonga/grnxx/commit/96ad811aa1fc8272ae778b43d3879c0b1e617788

  Message:
    Update Column::contains/find_one() to use an index if available. (#125)

  Modified files:
    lib/grnxx/impl/column/scalar/float.cpp
    lib/grnxx/impl/column/scalar/int.cpp
    lib/grnxx/impl/column/scalar/text.cpp

  Modified: lib/grnxx/impl/column/scalar/float.cpp (+9 -3)
===================================================================
--- lib/grnxx/impl/column/scalar/float.cpp    2014-12-04 11:24:48 +0900 (3ecb23c)
+++ lib/grnxx/impl/column/scalar/float.cpp    2014-12-04 12:26:20 +0900 (ff62a2a)
@@ -1,7 +1,7 @@
 #include "grnxx/impl/column/scalar/float.hpp"
 
 #include "grnxx/impl/table.hpp"
-//#include "grnxx/index.hpp"
+#include "grnxx/impl/index.hpp"
 
 namespace grnxx {
 namespace impl {
@@ -59,7 +59,10 @@ void Column<Float>::get(Int row_id, Datum *datum) const {
 }
 
 bool Column<Float>::contains(const Datum &datum) const {
-  // TODO: Use an index if exists.
+  // TODO: Choose the best index.
+  if (!indexes_.is_empty()) {
+    return indexes_[0]->contains(datum);
+  }
   Float value = parse_datum(datum);
   size_t valid_size = get_valid_size();
   if (value.is_na()) {
@@ -79,7 +82,10 @@ bool Column<Float>::contains(const Datum &datum) const {
 }
 
 Int Column<Float>::find_one(const Datum &datum) const {
-  // TODO: Use an index if exists.
+  // TODO: Choose the best index.
+  if (!indexes_.is_empty()) {
+    return indexes_[0]->find_one(datum);
+  }
   Float value = parse_datum(datum);
   size_t valid_size = get_valid_size();
   if (value.is_na()) {

  Modified: lib/grnxx/impl/column/scalar/int.cpp (+9 -36)
===================================================================
--- lib/grnxx/impl/column/scalar/int.cpp    2014-12-04 11:24:48 +0900 (a000dbd)
+++ lib/grnxx/impl/column/scalar/int.cpp    2014-12-04 12:26:20 +0900 (1d88056)
@@ -2,7 +2,7 @@
 
 #include "grnxx/impl/db.hpp"
 #include "grnxx/impl/table.hpp"
-//#include "grnxx/impl/index.hpp"
+#include "grnxx/impl/index.hpp"
 
 #include <unordered_set>
 
@@ -82,7 +82,10 @@ void Column<Int>::get(Int row_id, Datum *datum) const {
 }
 
 bool Column<Int>::contains(const Datum &datum) const {
-  // TODO: Use an index if exists.
+  // TODO: Choose the best index.
+  if (!indexes_.is_empty()) {
+    return indexes_[0]->contains(datum);
+  }
   Int value = parse_datum(datum);
   if (value.is_na()) {
     for (size_t i = 0; i < values_.size(); ++i) {
@@ -101,7 +104,10 @@ bool Column<Int>::contains(const Datum &datum) const {
 }
 
 Int Column<Int>::find_one(const Datum &datum) const {
-  // TODO: Use an index if exists.
+  // TODO: Choose the best index.
+  if (!indexes_.is_empty()) {
+    return indexes_[0]->find_one(datum);
+  }
   Int value = parse_datum(datum);
   if (value.is_na()) {
     for (size_t i = 0; i < values_.size(); ++i) {
@@ -117,39 +123,6 @@ Int Column<Int>::find_one(const Datum &datum) const {
     }
   }
   return Int::na();
-
-//  // TODO: Cursor should not be used because it takes time.
-//  //       Also, cursor operations can fail due to memory allocation.
-//  Int value = datum.force_int();
-//  if (indexes_.size() != 0) {
-//    return indexes_[0]->find_one(datum);
-//  } else {
-//    // TODO: A full scan takes time.
-//    //       An index should be required for a key column.
-
-//    // TODO: Functor-based inline callback may be better in this case,
-//    //       because it does not require memory allocation.
-
-//    // Scan the column to find "value".
-//    auto cursor = table_->create_cursor(nullptr);
-//    if (!cursor) {
-//      return NULL_ROW_ID;
-//    }
-//    Array<Record> records;
-//    for ( ; ; ) {
-//      auto result = cursor->read(nullptr, 1024, &records);
-//      if (!result.is_ok || result.count == 0) {
-//        return NULL_ROW_ID;
-//      }
-//      for (Int i = 0; i < result.count; ++i) {
-//        if (values_[records.get_row_id(i)] == value) {
-//          return records.get_row_id(i);
-//        }
-//      }
-//      records.clear();
-//    }
-//  }
-//  return NULL_ROW_ID;
 }
 
 void Column<Int>::set_key_attribute() {

  Modified: lib/grnxx/impl/column/scalar/text.cpp (+9 -45)
===================================================================
--- lib/grnxx/impl/column/scalar/text.cpp    2014-12-04 11:24:48 +0900 (e69995c)
+++ lib/grnxx/impl/column/scalar/text.cpp    2014-12-04 12:26:20 +0900 (281429c)
@@ -3,9 +3,8 @@
 #include <cstring>
 #include <set>
 
-#include "grnxx/impl/db.hpp"
 #include "grnxx/impl/table.hpp"
-//#include "grnxx/impl/index.hpp"
+#include "grnxx/impl/index.hpp"
 
 namespace grnxx {
 namespace impl {
@@ -144,7 +143,10 @@ void Column<Text>::get(Int row_id, Datum *datum) const {
 }
 
 bool Column<Text>::contains(const Datum &datum) const {
-  // TODO: Use an index if exists.
+  // TODO: Choose the best index.
+  if (!indexes_.is_empty()) {
+    return indexes_[0]->contains(datum);
+  }
   Text value = parse_datum(datum);
   size_t valid_size = get_valid_size();
   if (value.is_na()) {
@@ -165,7 +167,10 @@ bool Column<Text>::contains(const Datum &datum) const {
 }
 
 Int Column<Text>::find_one(const Datum &datum) const {
-  // TODO: Use an index if exists.
+  // TODO: Choose the best index.
+  if (!indexes_.is_empty()) {
+    return indexes_[0]->find_one(datum);
+  }
   Text value = parse_datum(datum);
   size_t valid_size = get_valid_size();
   if (value.is_na()) {
@@ -185,47 +190,6 @@ Int Column<Text>::find_one(const Datum &datum) const {
   return Int::na();
 }
 
-//Int Column<Text>::find_one(const Datum &datum) const {
-//  // TODO: Cursor should not be used because it takes time.
-//  // Also, cursor operations can fail due to memory allocation.
-//  Text value = datum.force_text();
-//  if (indexes_.size() != 0) {
-//    auto cursor = indexes_[0]->find(nullptr, value);
-//    Array<Record> records;
-//    auto result = cursor->read(nullptr, 1, &records);
-//    if (!result.is_ok || (result.count == 0)) {
-//      return NULL_ROW_ID;
-//    }
-//    return true;
-//  } else {
-//    // TODO: A full scan takes time.
-//    // An index should be required for a key column.
-
-//    // TODO: Functor-based inline callback may be better in this case,
-//    // because it does not require memory allocation.
-
-//    // Scan the column to find "value".
-//    auto cursor = table_->create_cursor(nullptr);
-//    if (!cursor) {
-//      return NULL_ROW_ID;
-//    }
-//    Array<Record> records;
-//    for ( ; ; ) {
-//      auto result = cursor->read(nullptr, 1024, &records);
-//      if (!result.is_ok || result.count == 0) {
-//        return NULL_ROW_ID;
-//      }
-//      for (Int i = 0; i < result.count; ++i) {
-//        if (get(records.get_row_id(i)) == value) {
-//          return records.get_row_id(i);
-//        }
-//      }
-//      records.clear();
-//    }
-//  }
-//  return NULL_ROW_ID;
-//}
-
 void Column<Text>::set_key_attribute() {
   if (is_key_) {
     throw "Key column";  // TODO
-------------- next part --------------
HTML����������������������������...
Download 

• Previous message: [Groonga-commit] groonga/groonga.org at a24bfbc [gh-pages] ja: add a case example of Rroonga at CrowdWorks
• Next message: [Groonga-commit] groonga/grnxx at 033f33a [master] Reduce the scan range of Column<Int>::contains/find_one().
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]