Merge pull request #10283 from rouault/sqlite_stddev

SQLite/GPKG: add STDDEV_POP() and STDDEV_SAMP() aggregate functions
OSGeo · Jul 6, 2024 · 75ad908 · 75ad908
2 parents 8c0fcc3 + b012d9a
commit 75ad908
Show file tree

Hide file tree

Showing 3 changed files with 92 additions and 0 deletions.
diff --git a/autotest/ogr/ogr_sqlite.py b/autotest/ogr/ogr_sqlite.py
@@ -4095,3 +4095,15 @@ def test_ogr_sql_ST_Area_on_ellipsoid(tmp_vsimem, require_spatialite):
     with ds.ExecuteSQL("SELECT ST_Area(null, 1) FROM my_layer") as sql_lyr:
         f = sql_lyr.GetNextFeature()
         assert f[0] is None
+
+
+def test_ogr_sqlite_stddev():
+    """Test STDDEV_POP() and STDDEV_SAMP"""
+
+    ds = ogr.Open(":memory:", update=1)
+    ds.ExecuteSQL("CREATE TABLE test(v REAL)")
+    ds.ExecuteSQL("INSERT INTO test VALUES (4),(NULL),('invalid'),(5)")
+    with ds.ExecuteSQL("SELECT STDDEV_POP(v), STDDEV_SAMP(v) FROM test") as sql_lyr:
+        f = sql_lyr.GetNextFeature()
+        assert f.GetField(0) == pytest.approx(0.5, rel=1e-15)
+        assert f.GetField(1) == pytest.approx(0.5**0.5, rel=1e-15)
diff --git a/doc/source/user/sql_sqlite_dialect.rst b/doc/source/user/sql_sqlite_dialect.rst
@@ -201,6 +201,15 @@ For example we can select the annotation features as:
 
     SELECT * FROM nation WHERE OGR_STYLE LIKE 'LABEL%'
 
+Statistics functions
+++++++++++++++++++++
+
+In addition to standard COUNT(), SUM(), AVG(), MIN(), MAX(), the following
+aggregate functions are available:
+
+- STDDEV_POP: (GDAL >= 3.10) numerical population standard deviation.
+- STDDEV_SAMP: (GDAL >= 3.10) numerical `sample standard deviation <https://en.wikipedia.org/wiki/Standard_deviation#Sample_standard_deviation>`__
+
 Spatialite SQL functions
 ++++++++++++++++++++++++
 

diff --git a/ogr/ogrsf_frmts/sqlite/ogrsqlitesqlfunctionscommon.cpp b/ogr/ogrsf_frmts/sqlite/ogrsqlitesqlfunctionscommon.cpp
@@ -336,6 +336,69 @@ static void OGRSQLITE_LIKE(sqlite3_context *pContext, int argc,
                                                insensitive, bUTF8Strings));
 }
 
+/************************************************************************/
+/*                       OGRSQLITE_STDDEV_Step()                        */
+/************************************************************************/
+
+// Welford's online algorithm for variance:
+// https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Welford's_online_algorithm
+struct OGRSQLITE_STDDEV_Context
+{
+    int64_t nValues;
+    double dfMean;
+    double dfM2;  // Accumulator for squared distance from the mean
+};
+
+static void OGRSQLITE_STDDEV_Step(sqlite3_context *pContext, int /* argc*/,
+                                  sqlite3_value **argv)
+{
+    auto pAggCtxt =
+        static_cast<OGRSQLITE_STDDEV_Context *>(sqlite3_aggregate_context(
+            pContext, static_cast<int>(sizeof(OGRSQLITE_STDDEV_Context))));
+    const auto eType = sqlite3_value_type(argv[0]);
+    if (eType != SQLITE_INTEGER && eType != SQLITE_FLOAT)
+        return;
+
+    const double dfValue = sqlite3_value_double(argv[0]);
+    pAggCtxt->nValues++;
+    const double dfDelta = dfValue - pAggCtxt->dfMean;
+    pAggCtxt->dfMean += dfDelta / pAggCtxt->nValues;
+    const double dfDelta2 = dfValue - pAggCtxt->dfMean;
+    pAggCtxt->dfM2 += dfDelta * dfDelta2;
+}
+
+/************************************************************************/
+/*                    OGRSQLITE_STDDEV_POP_Finalize()                   */
+/************************************************************************/
+
+static void OGRSQLITE_STDDEV_POP_Finalize(sqlite3_context *pContext)
+{
+    auto pAggCtxt =
+        static_cast<OGRSQLITE_STDDEV_Context *>(sqlite3_aggregate_context(
+            pContext, static_cast<int>(sizeof(OGRSQLITE_STDDEV_Context))));
+    if (pAggCtxt->nValues > 0)
+    {
+        sqlite3_result_double(pContext,
+                              sqrt(pAggCtxt->dfM2 / pAggCtxt->nValues));
+    }
+}
+
+/************************************************************************/
+/*                    OGRSQLITE_STDDEV_SAMP_Finalize()                  */
+/************************************************************************/
+
+static void OGRSQLITE_STDDEV_SAMP_Finalize(sqlite3_context *pContext)
+{
+    auto pAggCtxt =
+        static_cast<OGRSQLITE_STDDEV_Context *>(sqlite3_aggregate_context(
+            pContext, static_cast<int>(sizeof(OGRSQLITE_STDDEV_Context))));
+    if (pAggCtxt->nValues > 1)
+    {
+        sqlite3_result_double(pContext,
+                              sqrt(pAggCtxt->dfM2 / (pAggCtxt->nValues - 1)));
+    }
+}
+
 /************************************************************************/
 /*                 OGRSQLiteRegisterSQLFunctionsCommon()                */
 /************************************************************************/
@@ -365,6 +428,14 @@ static OGRSQLiteExtensionData *OGRSQLiteRegisterSQLFunctionsCommon(sqlite3 *hDB)
                                 OGRSQLITE_LIKE, nullptr, nullptr);
     }
 
+    sqlite3_create_function(hDB, "STDDEV_POP", 1, UTF8_INNOCUOUS, nullptr,
+                            nullptr, OGRSQLITE_STDDEV_Step,
+                            OGRSQLITE_STDDEV_POP_Finalize);
+
+    sqlite3_create_function(hDB, "STDDEV_SAMP", 1, UTF8_INNOCUOUS, nullptr,
+                            nullptr, OGRSQLITE_STDDEV_Step,
+                            OGRSQLITE_STDDEV_SAMP_Finalize);
+
     pData->SetRegExpCache(OGRSQLiteRegisterRegExpFunction(hDB));
 
     return pData;