Groonga - fulltext search engine.

[Groonga-commit] groonga/grnxx at f6bb500 [master] Fix a bug that Column<Vector<T>> cannot find N/A. (#135)

susumu.yata null+****@clear*****
Fri Dec 26 12:08:11 JST 2014

• Previous message: [Groonga-commit] droonga/droonga-engine at 78b0c93 [buffered-forward] Initialize live-nodes-list.json certainly
• Next message: [Groonga-commit] Still Failing: droonga/droonga-engine#995 (master - 753da13)
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

susumu.yata	2014-12-26 12:08:11 +0900 (Fri, 26 Dec 2014)

  New Revision: f6bb500131e2d55a25a8e97069e5304a39318e64
  https://github.com/groonga/grnxx/commit/f6bb500131e2d55a25a8e97069e5304a39318e64

  Message:
    Fix a bug that Column<Vector<T>> cannot find N/A. (#135)

  Modified files:
    lib/grnxx/impl/column/vector/bool.cpp
    lib/grnxx/impl/column/vector/bool.hpp
    lib/grnxx/impl/column/vector/float.cpp
    lib/grnxx/impl/column/vector/float.hpp
    lib/grnxx/impl/column/vector/geo_point.cpp
    lib/grnxx/impl/column/vector/geo_point.hpp
    lib/grnxx/impl/column/vector/int.cpp
    lib/grnxx/impl/column/vector/int.hpp
    lib/grnxx/impl/column/vector/text.cpp
    lib/grnxx/impl/column/vector/text.hpp

  Modified: lib/grnxx/impl/column/vector/bool.cpp (+27 -32)
===================================================================
--- lib/grnxx/impl/column/vector/bool.cpp    2014-12-26 11:55:16 +0900 (578bedc)
+++ lib/grnxx/impl/column/vector/bool.cpp    2014-12-26 12:08:11 +0900 (75ba880)
@@ -84,38 +84,44 @@ void Column<Vector<Bool>>::get(Int row_id, Datum *datum) const {
 
 bool Column<Vector<Bool>>::contains(const Datum &datum) const {
   // TODO: Use an index if exists.
-  Vector<Bool> value = parse_datum(datum);
-  size_t valid_size = get_valid_size();
-  if (value.is_na()) {
-    for (size_t i = 0; i < valid_size; ++i) {
-      if (headers_[i] == na_header()) {
-        return true;
-      }
-    }
-  } else {
-    for (size_t i = 0; i < valid_size; ++i) {
-      // TODO: Improve this.
-      if (get(Int(i)).match(value)) {
-        return true;
-      }
-    }
-  }
-  return false;
+  return !scan(parse_datum(datum)).is_na();
 }
 
 Int Column<Vector<Bool>>::find_one(const Datum &datum) const {
   // TODO: Use an index if exists.
-  Vector<Bool> value = parse_datum(datum);
-  size_t valid_size = get_valid_size();
+  return scan(parse_datum(datum));
+}
+
+void Column<Vector<Bool>>::unset(Int row_id) {
+  Vector<Bool> value = get(row_id);
+  if (!value.is_na()) {
+    // TODO: Update indexes if exist.
+//    for (size_t i = 0; i < num_indexes(); ++i) {
+//      indexes_[i]->remove(row_id, value);
+//    }
+    headers_[row_id.raw()] = na_header();
+  }
+}
+
+Int Column<Vector<Bool>>::scan(const Vector<Bool> &value) const {
+  if (table_->max_row_id().is_na()) {
+    return Int::na();
+  }
+  size_t table_size = table_->max_row_id().raw() + 1;
+  size_t valid_size =
+      (headers_.size() < table_size) ? headers_.size() : table_size;
   if (value.is_na()) {
+    if (headers_.size() < table_size) {
+      return table_->max_row_id();
+    }
     for (size_t i = 0; i < valid_size; ++i) {
-      if (headers_[i] == na_header()) {
+      if (headers_[i] == na_header() && table_->_test_row(i)) {
         return Int(i);
       }
     }
   } else {
     for (size_t i = 0; i < valid_size; ++i) {
-      // TODO: Improve this.
+      // TODO: Improve this (get() checks the range of its argument).
       if (get(Int(i)).match(value)) {
         return Int(i);
       }
@@ -124,17 +130,6 @@ Int Column<Vector<Bool>>::find_one(const Datum &datum) const {
   return Int::na();
 }
 
-void Column<Vector<Bool>>::unset(Int row_id) {
-  Vector<Bool> value = get(row_id);
-  if (!value.is_na()) {
-    // TODO: Update indexes if exist.
-//    for (size_t i = 0; i < num_indexes(); ++i) {
-//      indexes_[i]->remove(row_id, value);
-//    }
-    headers_[row_id.raw()] = na_header();
-  }
-}
-
 size_t Column<Vector<Bool>>::get_valid_size() const {
   if (table_->max_row_id().is_na()) {
     return 0;

  Modified: lib/grnxx/impl/column/vector/bool.hpp (+6 -0)
===================================================================
--- lib/grnxx/impl/column/vector/bool.hpp    2014-12-26 11:55:16 +0900 (ab58e7c)
+++ lib/grnxx/impl/column/vector/bool.hpp    2014-12-26 12:08:11 +0900 (3e8067c)
@@ -71,6 +71,12 @@ class Column<Vector<Bool>> : public ColumnBase {
     return std::numeric_limits<uint64_t>::max();
   }
 
+  // Scan the column to find "value".
+  //
+  // If found, returns the row ID.
+  // If not found, returns N/A.
+  Int scan(const Vector<Bool> &value) const;
+
   // Return the active column size.
   size_t get_valid_size() const;
 

  Modified: lib/grnxx/impl/column/vector/float.cpp (+29 -34)
===================================================================
--- lib/grnxx/impl/column/vector/float.cpp    2014-12-26 11:55:16 +0900 (3f4eeee)
+++ lib/grnxx/impl/column/vector/float.cpp    2014-12-26 12:08:11 +0900 (8f2488b)
@@ -84,44 +84,12 @@ void Column<Vector<Float>>::get(Int row_id, Datum *datum) const {
 
 bool Column<Vector<Float>>::contains(const Datum &datum) const {
   // TODO: Use an index if exists.
-  Vector<Float> value = parse_datum(datum);
-  size_t valid_size = get_valid_size();
-  if (value.is_na()) {
-    for (size_t i = 0; i < valid_size; ++i) {
-      if (headers_[i] == na_header()) {
-        return true;
-      }
-    }
-  } else {
-    for (size_t i = 0; i < valid_size; ++i) {
-      // TODO: Improve this.
-      if (get(Int(i)).match(value)) {
-        return true;
-      }
-    }
-  }
-  return false;
+  return !scan(parse_datum(datum)).is_na();
 }
 
 Int Column<Vector<Float>>::find_one(const Datum &datum) const {
   // TODO: Use an index if exists.
-  Vector<Float> value = parse_datum(datum);
-  size_t valid_size = get_valid_size();
-  if (value.is_na()) {
-    for (size_t i = 0; i < valid_size; ++i) {
-      if (headers_[i] == na_header()) {
-        return Int(i);
-      }
-    }
-  } else {
-    for (size_t i = 0; i < valid_size; ++i) {
-      // TODO: Improve this.
-      if (get(Int(i)).match(value)) {
-        return Int(i);
-      }
-    }
-  }
-  return Int::na();
+  return scan(parse_datum(datum));
 }
 
 void Column<Vector<Float>>::unset(Int row_id) {
@@ -145,6 +113,33 @@ void Column<Vector<Float>>::read(ArrayCRef<Record> records,
   }
 }
 
+Int Column<Vector<Float>>::scan(const Vector<Float> &value) const {
+  if (table_->max_row_id().is_na()) {
+    return Int::na();
+  }
+  size_t table_size = table_->max_row_id().raw() + 1;
+  size_t valid_size =
+      (headers_.size() < table_size) ? headers_.size() : table_size;
+  if (value.is_na()) {
+    if (headers_.size() < table_size) {
+      return table_->max_row_id();
+    }
+    for (size_t i = 0; i < valid_size; ++i) {
+      if (headers_[i] == na_header() && table_->_test_row(i)) {
+        return Int(i);
+      }
+    }
+  } else {
+    for (size_t i = 0; i < valid_size; ++i) {
+      // TODO: Improve this (get() checks the range of its argument).
+      if (get(Int(i)).match(value)) {
+        return Int(i);
+      }
+    }
+  }
+  return Int::na();
+}
+
 size_t Column<Vector<Float>>::get_valid_size() const {
   if (table_->max_row_id().is_na()) {
     return 0;

  Modified: lib/grnxx/impl/column/vector/float.hpp (+6 -0)
===================================================================
--- lib/grnxx/impl/column/vector/float.hpp    2014-12-26 11:55:16 +0900 (473f319)
+++ lib/grnxx/impl/column/vector/float.hpp    2014-12-26 12:08:11 +0900 (55f66fe)
@@ -71,6 +71,12 @@ class Column<Vector<Float>> : public ColumnBase {
     return std::numeric_limits<uint64_t>::max();
   }
 
+  // Scan the column to find "value".
+  //
+  // If found, returns the row ID.
+  // If not found, returns N/A.
+  Int scan(const Vector<Float> &value) const;
+
   // Return the active column size.
   size_t get_valid_size() const;
 

  Modified: lib/grnxx/impl/column/vector/geo_point.cpp (+29 -34)
===================================================================
--- lib/grnxx/impl/column/vector/geo_point.cpp    2014-12-26 11:55:16 +0900 (e57bd27)
+++ lib/grnxx/impl/column/vector/geo_point.cpp    2014-12-26 12:08:11 +0900 (c8190fd)
@@ -85,44 +85,12 @@ void Column<Vector<GeoPoint>>::get(Int row_id, Datum *datum) const {
 
 bool Column<Vector<GeoPoint>>::contains(const Datum &datum) const {
   // TODO: Use an index if exists.
-  Vector<GeoPoint> value = parse_datum(datum);
-  size_t valid_size = get_valid_size();
-  if (value.is_na()) {
-    for (size_t i = 0; i < valid_size; ++i) {
-      if (headers_[i] == na_header()) {
-        return true;
-      }
-    }
-  } else {
-    for (size_t i = 0; i < valid_size; ++i) {
-      // TODO: Improve this.
-      if (get(Int(i)).match(value)) {
-        return true;
-      }
-    }
-  }
-  return false;
+  return !scan(parse_datum(datum)).is_na();
 }
 
 Int Column<Vector<GeoPoint>>::find_one(const Datum &datum) const {
   // TODO: Use an index if exists.
-  Vector<GeoPoint> value = parse_datum(datum);
-  size_t valid_size = get_valid_size();
-  if (value.is_na()) {
-    for (size_t i = 0; i < valid_size; ++i) {
-      if (headers_[i] == na_header()) {
-        return Int(i);
-      }
-    }
-  } else {
-    for (size_t i = 0; i < valid_size; ++i) {
-      // TODO: Improve this.
-      if (get(Int(i)).match(value)) {
-        return Int(i);
-      }
-    }
-  }
-  return Int::na();
+  return scan(parse_datum(datum));
 }
 
 void Column<Vector<GeoPoint>>::unset(Int row_id) {
@@ -146,6 +114,33 @@ void Column<Vector<GeoPoint>>::read(ArrayCRef<Record> records,
   }
 }
 
+Int Column<Vector<GeoPoint>>::scan(const Vector<GeoPoint> &value) const {
+  if (table_->max_row_id().is_na()) {
+    return Int::na();
+  }
+  size_t table_size = table_->max_row_id().raw() + 1;
+  size_t valid_size =
+      (headers_.size() < table_size) ? headers_.size() : table_size;
+  if (value.is_na()) {
+    if (headers_.size() < table_size) {
+      return table_->max_row_id();
+    }
+    for (size_t i = 0; i < valid_size; ++i) {
+      if (headers_[i] == na_header() && table_->_test_row(i)) {
+        return Int(i);
+      }
+    }
+  } else {
+    for (size_t i = 0; i < valid_size; ++i) {
+      // TODO: Improve this (get() checks the range of its argument).
+      if (get(Int(i)).match(value)) {
+        return Int(i);
+      }
+    }
+  }
+  return Int::na();
+}
+
 size_t Column<Vector<GeoPoint>>::get_valid_size() const {
   if (table_->max_row_id().is_na()) {
     return 0;

  Modified: lib/grnxx/impl/column/vector/geo_point.hpp (+6 -0)
===================================================================
--- lib/grnxx/impl/column/vector/geo_point.hpp    2014-12-26 11:55:16 +0900 (f1fb553)
+++ lib/grnxx/impl/column/vector/geo_point.hpp    2014-12-26 12:08:11 +0900 (46dc6e8)
@@ -72,6 +72,12 @@ class Column<Vector<GeoPoint>> : public ColumnBase {
     return std::numeric_limits<uint64_t>::max();
   }
 
+  // Scan the column to find "value".
+  //
+  // If found, returns the row ID.
+  // If not found, returns N/A.
+  Int scan(const Vector<GeoPoint> &value) const;
+
   // Return the active column size.
   size_t get_valid_size() const;
 

  Modified: lib/grnxx/impl/column/vector/int.cpp (+29 -34)
===================================================================
--- lib/grnxx/impl/column/vector/int.cpp    2014-12-26 11:55:16 +0900 (047672f)
+++ lib/grnxx/impl/column/vector/int.cpp    2014-12-26 12:08:11 +0900 (a529efa)
@@ -99,44 +99,12 @@ void Column<Vector<Int>>::get(Int row_id, Datum *datum) const {
 
 bool Column<Vector<Int>>::contains(const Datum &datum) const {
   // TODO: Use an index if exists.
-  Vector<Int> value = parse_datum(datum);
-  size_t valid_size = get_valid_size();
-  if (value.is_na()) {
-    for (size_t i = 0; i < valid_size; ++i) {
-      if (headers_[i] == na_header()) {
-        return true;
-      }
-    }
-  } else {
-    for (size_t i = 0; i < valid_size; ++i) {
-      // TODO: Improve this.
-      if (get(Int(i)).match(value)) {
-        return true;
-      }
-    }
-  }
-  return false;
+  return !scan(parse_datum(datum)).is_na();
 }
 
 Int Column<Vector<Int>>::find_one(const Datum &datum) const {
   // TODO: Use an index if exists.
-  Vector<Int> value = parse_datum(datum);
-  size_t valid_size = get_valid_size();
-  if (value.is_na()) {
-    for (size_t i = 0; i < valid_size; ++i) {
-      if (headers_[i] == na_header()) {
-        return Int(i);
-      }
-    }
-  } else {
-    for (size_t i = 0; i < valid_size; ++i) {
-      // TODO: Improve this.
-      if (get(Int(i)).match(value)) {
-        return Int(i);
-      }
-    }
-  }
-  return Int::na();
+  return scan(parse_datum(datum));
 }
 
 void Column<Vector<Int>>::unset(Int row_id) {
@@ -160,6 +128,33 @@ void Column<Vector<Int>>::read(ArrayCRef<Record> records,
   }
 }
 
+Int Column<Vector<Int>>::scan(const Vector<Int> &value) const {
+  if (table_->max_row_id().is_na()) {
+    return Int::na();
+  }
+  size_t table_size = table_->max_row_id().raw() + 1;
+  size_t valid_size =
+      (headers_.size() < table_size) ? headers_.size() : table_size;
+  if (value.is_na()) {
+    if (headers_.size() < table_size) {
+      return table_->max_row_id();
+    }
+    for (size_t i = 0; i < valid_size; ++i) {
+      if (headers_[i] == na_header() && table_->_test_row(i)) {
+        return Int(i);
+      }
+    }
+  } else {
+    for (size_t i = 0; i < valid_size; ++i) {
+      // TODO: Improve this (get() checks the range of its argument).
+      if (get(Int(i)).match(value)) {
+        return Int(i);
+      }
+    }
+  }
+  return Int::na();
+}
+
 size_t Column<Vector<Int>>::get_valid_size() const {
   if (table_->max_row_id().is_na()) {
     return 0;

  Modified: lib/grnxx/impl/column/vector/int.hpp (+6 -0)
===================================================================
--- lib/grnxx/impl/column/vector/int.hpp    2014-12-26 11:55:16 +0900 (0e8e352)
+++ lib/grnxx/impl/column/vector/int.hpp    2014-12-26 12:08:11 +0900 (27fb975)
@@ -71,6 +71,12 @@ class Column<Vector<Int>> : public ColumnBase {
     return std::numeric_limits<uint64_t>::max();
   }
 
+  // Scan the column to find "value".
+  //
+  // If found, returns the row ID.
+  // If not found, returns N/A.
+  Int scan(const Vector<Int> &value) const;
+
   // Return the active column size.
   size_t get_valid_size() const;
 

  Modified: lib/grnxx/impl/column/vector/text.cpp (+29 -34)
===================================================================
--- lib/grnxx/impl/column/vector/text.cpp    2014-12-26 11:55:16 +0900 (89a89ad)
+++ lib/grnxx/impl/column/vector/text.cpp    2014-12-26 12:08:11 +0900 (facc8c7)
@@ -86,44 +86,12 @@ void Column<Vector<Text>>::get(Int row_id, Datum *datum) const {
 
 bool Column<Vector<Text>>::contains(const Datum &datum) const {
   // TODO: Use an index if exists.
-  Vector<Text> value = parse_datum(datum);
-  size_t valid_size = get_valid_size();
-  if (value.is_na()) {
-    for (size_t i = 0; i < valid_size; ++i) {
-      if (headers_[i].size.is_na()) {
-        return true;
-      }
-    }
-  } else {
-    for (size_t i = 0; i < valid_size; ++i) {
-      // TODO: Improve this.
-      if (get(Int(i)).match(value)) {
-        return true;
-      }
-    }
-  }
-  return false;
+  return !scan(parse_datum(datum)).is_na();
 }
 
 Int Column<Vector<Text>>::find_one(const Datum &datum) const {
   // TODO: Use an index if exists.
-  Vector<Text> value = parse_datum(datum);
-  size_t valid_size = get_valid_size();
-  if (value.is_na()) {
-    for (size_t i = 0; i < valid_size; ++i) {
-      if (headers_[i].size.is_na()) {
-        return Int(i);
-      }
-    }
-  } else {
-    for (size_t i = 0; i < valid_size; ++i) {
-      // TODO: Improve this.
-      if (get(Int(i)).match(value)) {
-        return Int(i);
-      }
-    }
-  }
-  return Int::na();
+  return scan(parse_datum(datum));
 }
 
 void Column<Vector<Text>>::unset(Int row_id) {
@@ -147,6 +115,33 @@ void Column<Vector<Text>>::read(ArrayCRef<Record> records,
   }
 }
 
+Int Column<Vector<Text>>::scan(const Vector<Text> &value) const {
+  if (table_->max_row_id().is_na()) {
+    return Int::na();
+  }
+  size_t table_size = table_->max_row_id().raw() + 1;
+  size_t valid_size =
+      (headers_.size() < table_size) ? headers_.size() : table_size;
+  if (value.is_na()) {
+    if (headers_.size() < table_size) {
+      return table_->max_row_id();
+    }
+    for (size_t i = 0; i < valid_size; ++i) {
+      if (headers_[i].size.is_na() && table_->_test_row(i)) {
+        return Int(i);
+      }
+    }
+  } else {
+    for (size_t i = 0; i < valid_size; ++i) {
+      // TODO: Improve this (get() checks the range of its argument).
+      if (get(Int(i)).match(value)) {
+        return Int(i);
+      }
+    }
+  }
+  return Int::na();
+}
+
 size_t Column<Vector<Text>>::get_valid_size() const {
   if (table_->max_row_id().is_na()) {
     return 0;

  Modified: lib/grnxx/impl/column/vector/text.hpp (+6 -0)
===================================================================
--- lib/grnxx/impl/column/vector/text.hpp    2014-12-26 11:55:16 +0900 (2e0e75b)
+++ lib/grnxx/impl/column/vector/text.hpp    2014-12-26 12:08:11 +0900 (b791839)
@@ -63,6 +63,12 @@ class Column<Vector<Text>> : public ColumnBase {
     return Header{ 0, Int::na() };
   }
 
+  // Scan the column to find "value".
+  //
+  // If found, returns the row ID.
+  // If not found, returns N/A.
+  Int scan(const Vector<Text> &value) const;
+
   // Return the active column size.
   size_t get_valid_size() const;
 
-------------- next part --------------
HTML����������������������������...
다운로드 

• Previous message: [Groonga-commit] droonga/droonga-engine at 78b0c93 [buffered-forward] Initialize live-nodes-list.json certainly
• Next message: [Groonga-commit] Still Failing: droonga/droonga-engine#995 (master - 753da13)
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]