From: drh <>
Date: Tue, 23 Jul 2024 16:23:46 +0000 (+0000)
Subject: Enhance the percentile() extension function to include the median()
X-Git-Tag: version-3.47.0~280
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=171c944345ac9f11711679f3c68e3d792cb529cd;p=thirdparty%2Fsqlite.git

Enhance the percentile() extension function to include the median()
variant.  Update the implementation to implement its own sorting
algorithm, so that the extension no longer depends on qsort().

FossilOrigin-Name: 6e31b1bab1f014933c671a12a5930c1e88d611cfe68d389e9ce888bd8842addd
---

diff --git a/ext/misc/percentile.c b/ext/misc/percentile.c
index d83bc5b838..9a586eafbc 100644
--- a/ext/misc/percentile.c
+++ b/ext/misc/percentile.c
@@ -57,6 +57,8 @@
 **  (12)  The percentile(Y,P) is implemented as a single C99 source-code
 **        file that compiles into a shared-library or DLL that can be loaded
 **        into SQLite using the sqlite3_load_extension() interface.
+**
+**  (13)  A separate median(Y) function is the equivalent percentile(Y,50).
 */
 #include "sqlite3ext.h"
 SQLITE_EXTENSION_INIT1
@@ -103,16 +105,21 @@ static void percentStep(sqlite3_context *pCtx, int argc, sqlite3_value **argv){
   double rPct;
   int eType;
   double y;
-  assert( argc==2 );
-
-  /* Requirement 3:  P must be a number between 0 and 100 */
-  eType = sqlite3_value_numeric_type(argv[1]);
-  rPct = sqlite3_value_double(argv[1]);
-  if( (eType!=SQLITE_INTEGER && eType!=SQLITE_FLOAT)
-   || rPct<0.0 || rPct>100.0 ){
-    sqlite3_result_error(pCtx, "2nd argument to percentile() is not "
-                         "a number between 0.0 and 100.0", -1);
-    return;
+  assert( argc==2 || argc==1 );
+
+  if( argc==1 ){
+    /* Requirement 13:  median(Y) is the same as percentile(Y,50). */
+    rPct = 50.0;
+  }else{
+    /* Requirement 3:  P must be a number between 0 and 100 */
+    eType = sqlite3_value_numeric_type(argv[1]);
+    rPct = sqlite3_value_double(argv[1]);
+    if( (eType!=SQLITE_INTEGER && eType!=SQLITE_FLOAT)
+     || rPct<0.0 || rPct>100.0 ){
+       sqlite3_result_error(pCtx, "2nd argument to percentile() is not "
+                           "a number between 0.0 and 100.0", -1);
+      return;
+    }
   }
 
   /* Allocate the session context. */
@@ -165,14 +172,52 @@ static void percentStep(sqlite3_context *pCtx, int argc, sqlite3_value **argv){
 }
 
 /*
-** Compare to doubles for sorting using qsort()
+** Sort an array of doubles.
 */
-static int SQLITE_CDECL doubleCmp(const void *pA, const void *pB){
-  double a = *(double*)pA;
-  double b = *(double*)pB;
-  if( a==b ) return 0;
-  if( a<b ) return -1;
-  return +1;
+static void sortDoubles(double *a, int n){
+  int iLt;       /* Entries with index less than iLt are less than rPivot */
+  int iGt;       /* Entries with index iGt or more are greater than rPivot */
+  int i;         /* Loop counter */
+  double rPivot; /* The pivot value */
+  double rTmp;   /* Temporary used to swap two values */
+
+  if( n<2 ) return;
+  if( n>5 ){
+    rPivot = (a[0] + a[n/2] + a[n-1])/3.0;
+  }else{
+    rPivot = a[n/2];
+  }
+  iLt = i = 0;
+  iGt = n;
+  while( i<iGt ){
+    if( a[i]<rPivot ){
+      if( i>iLt ){
+        rTmp = a[i];
+        a[i] = a[iLt];
+        a[iLt] = rTmp;
+      }
+      iLt++;
+      i++;
+    }else if( a[i]>rPivot ){
+      do{
+        iGt--;
+      }while( iGt>i && a[iGt]>rPivot );
+      rTmp = a[i];
+      a[i] = a[iGt];
+      a[iGt] = rTmp;
+    }else{
+      i++;
+    }
+  }
+  if( iLt>=2 ) sortDoubles(a, iLt);
+  if( n-iGt>=2 ) sortDoubles(a+iGt, n-iGt);
+
+/* Uncomment for testing */
+#if 0
+  for(i=0; i<n-1; i++){
+    assert( a[i]<=a[i+1] );
+  }
+#endif
 }
 
 /*
@@ -188,7 +233,7 @@ static void percentFinal(sqlite3_context *pCtx){
   if( p==0 ) return;
   if( p->a==0 ) return;
   if( p->nUsed ){
-    qsort(p->a, p->nUsed, sizeof(double), doubleCmp);
+    sortDoubles(p->a, p->nUsed);
     ix = (p->rPct-1.0)*(p->nUsed-1)*0.01;
     i1 = (unsigned)ix;
     i2 = ix==(double)i1 || i1==p->nUsed-1 ? i1 : i1+1;
@@ -216,5 +261,10 @@ int sqlite3_percentile_init(
   rc = sqlite3_create_function(db, "percentile", 2, 
                                SQLITE_UTF8|SQLITE_INNOCUOUS, 0,
                                0, percentStep, percentFinal);
+  if( rc==SQLITE_OK ){
+    rc = sqlite3_create_function(db, "median", 1, 
+                                 SQLITE_UTF8|SQLITE_INNOCUOUS, 0,
+                                 0, percentStep, percentFinal);
+  }
   return rc;
 }
diff --git a/manifest b/manifest
index d00f89cfcc..7ec408ed08 100644
--- a/manifest
+++ b/manifest
@@ -1,5 +1,5 @@
-C Doc\stypo\sfix.\sNo\scode\schanges.
-D 2024-07-22T21:46:55.078
+C Enhance\sthe\spercentile()\sextension\sfunction\sto\sinclude\sthe\smedian()\nvariant.\s\sUpdate\sthe\simplementation\sto\simplement\sits\sown\ssorting\nalgorithm,\sso\sthat\sthe\sextension\sno\slonger\sdepends\son\sqsort().
+D 2024-07-23T16:23:46.925
 F .fossil-settings/empty-dirs dbb81e8fc0401ac46a1491ab34a7f2c7c0452f2f06b54ebb845d024ca8283ef1
 F .fossil-settings/ignore-glob 35175cdfcf539b2318cb04a9901442804be81cd677d8b889fcc9149c21f239ea
 F LICENSE.md df5091916dbb40e6e9686186587125e1b2ff51f022cc334e886c19a0e9982724
@@ -404,7 +404,7 @@ F ext/misc/nextchar.c 7877914c2a80c2f181dd04c3dbef550dfb54c93495dc03da2403b5dd58
 F ext/misc/noop.c f1a21cc9b7a4e667e5c8458d80ba680b8bd4315a003f256006046879f679c5a0
 F ext/misc/normalize.c bd84355c118e297522aba74de34a4fd286fc775524e0499b14473918d09ea61f
 F ext/misc/pcachetrace.c f4227ce03fb16aa8d6f321b72dd051097419d7a028a9853af048bee7645cb405
-F ext/misc/percentile.c b9086e223d583bdaf8cb73c98a6539d501a2fc4282654adbfea576453d82e691
+F ext/misc/percentile.c f15eb0fb0c2a95a482382b41415b9d1ae66eabf857ecac8e1e0eb3aa6b44e4d8
 F ext/misc/prefixes.c 82645f79229877afab08c8b08ca1e7fa31921280906b90a61c294e4f540cd2a6
 F ext/misc/qpvtab.c fc189e127f68f791af90a487f4460ec91539a716daf45a0c357e963fd47cc06c
 F ext/misc/randomjson.c ef835fc64289e76ac4873b85fe12f9463a036168d7683cf2b773e36e6262c4ed
@@ -1503,7 +1503,7 @@ F test/parser1.test 6ccdf5e459a5dc4673d3273dc311a7e9742ca952dd0551a6a6320d27035c
 F test/pcache.test c8acbedd3b6fd0f9a7ca887a83b11d24a007972b
 F test/pcache2.test af7f3deb1a819f77a6d0d81534e97d1cf62cd442
 F test/pendingrace.test 6aa33756b950c4529f79c4f3817a9a1e4025bd0d9961571a05c0279bd183d9c6
-F test/percentile.test 4243af26b8f3f4555abe166f723715a1f74c77ff
+F test/percentile.test 16fa70efabac459223d4886604abc629cacfb5f00ca7c3160da47fbeac870674
 F test/permutations.test 405542f1d659942994a6b38a9e024cf5cfd23eaa68c806aeb24a72d7c9186e80
 F test/pg_common.tcl 3b27542224db1e713ae387459b5d117c836a5f6e328846922993b6d2b7640d9f
 F test/pragma.test 11cb9310c42f921918f7f563e3c0b6e70f9f9c3a6a1cf12af8fccb6c574f3882
@@ -2195,8 +2195,8 @@ F vsixtest/vsixtest.tcl 6a9a6ab600c25a91a7acc6293828957a386a8a93
 F vsixtest/vsixtest.vcxproj.data 2ed517e100c66dc455b492e1a33350c1b20fbcdc
 F vsixtest/vsixtest.vcxproj.filters 37e51ffedcdb064aad6ff33b6148725226cd608e
 F vsixtest/vsixtest_TemporaryKey.pfx e5b1b036facdb453873e7084e1cae9102ccc67a0
-P 6cd9f55a975b5237efee8776efce7e7836b41905ca750f82be3b90aa04c1dff2
-R 61b15192d04959d281e94594e02bce6a
-U stephan
-Z aae0cc74fee2d9ed676d7939f647c2b8
+P 8d558ad25bfbdea04de87616d4e3f664b5749a7d23643d1a0238e991b4bb337e
+R 6164638135694f2a60de1effceddb452
+U drh
+Z a2b6c7a02bd489d715da4964c630a442
 # Remove this line to create a well-formed Fossil manifest.
diff --git a/manifest.uuid b/manifest.uuid
index f3de54cfb0..e1c8d42a87 100644
--- a/manifest.uuid
+++ b/manifest.uuid
@@ -1 +1 @@
-8d558ad25bfbdea04de87616d4e3f664b5749a7d23643d1a0238e991b4bb337e
+6e31b1bab1f014933c671a12a5930c1e88d611cfe68d389e9ce888bd8842addd
diff --git a/test/percentile.test b/test/percentile.test
index b2bd061e86..a54248e6da 100644
--- a/test/percentile.test
+++ b/test/percentile.test
@@ -38,6 +38,9 @@ foreach {in out} {
     execsql {SELECT percentile(x,$in) FROM t1}
   } $out
 }
+do_execsql_test percentile-1.1.median {
+  SELECT median(x) FROM t1;
+} 8.0
 
 # Add some NULL values.
 #