From bf4d0dcfe2d30b42861293af995a3b8d5c4acfb1 Mon Sep 17 00:00:00 2001
From: Michael Okoko <okokomichaels@outlook.com>
Date: Fri, 30 Jul 2021 06:55:17 +0100
Subject: [PATCH] implement parse and serialize for histogram json

---
 sql/sql_statistics.cc | 36 ++++++++++++++++++++++++++++--------
 sql/sql_statistics.h  |  6 ++++--
 2 files changed, 32 insertions(+), 10 deletions(-)

diff --git a/sql/sql_statistics.cc b/sql/sql_statistics.cc
index 9e22c82d920..8bd181e1939 100644
--- a/sql/sql_statistics.cc
+++ b/sql/sql_statistics.cc
@@ -1261,12 +1261,6 @@ bool Histogram_binary::parse(MEM_ROOT *mem_root, Histogram_type type_arg, const
 */
 void Histogram_binary::serialize(Field *field)
 {
-  if (get_type() == JSON)
-  {
-    field->store((char*)get_values(), strlen((char*)get_values()), 
-                 &my_charset_bin);
-  }
-  else
     field->store((char*)get_values(), get_width(), &my_charset_bin);
 }
 
@@ -1287,6 +1281,32 @@ void Histogram_json::init_for_collection(MEM_ROOT *mem_root, Histogram_type htyp
   size = (uint8) size_arg;
 }
 
+bool Histogram_json::parse(MEM_ROOT *mem_root, Histogram_type type_arg, const uchar *ptr, uint size_arg)
+{
+  size = (uint8) size_arg;
+  type = type_arg;
+  // I think we could use memcpy here, but not sure about how to get the right size
+  // since we can't depend on size_arg (it's zero for json histograms)
+  // also, does it make sense to cast here? or we can modify json_get_array_items
+  // to accept uchar*
+  const char *json = (char *)ptr;
+  int vt;
+  bool result = json_get_array_items(json, json + strlen(json), &vt, hist_buckets);
+  fprintf(stderr,"==============\n");
+  fprintf(stderr,"histogram: %s\n", json);
+  fprintf(stderr, "json_get_array_items() returned %s\n", result ? "true" : "false");
+  fprintf(stderr, "value type after json_get_array_items() is %d\n", vt);
+  fprintf(stderr, " JSV_BAD_JSON=%d, JSON_VALUE_ARRAY=%d\n", (int)JSV_BAD_JSON, (int)JSON_VALUE_ARRAY);
+  fprintf(stderr, "hist_buckets.size()=%zu\n", hist_buckets.size());
+  return false;
+}
+
+void Histogram_json::serialize(Field *field)
+{
+  field->store((char*)get_values(), strlen((char*)get_values()),
+               &my_charset_bin);
+}
+
 /*
   An object of the class Index_stat is created to read statistical
   data on tables from the statistical table table_stat, to update
@@ -1987,9 +2007,9 @@ public:
     @brief
     Get the pointer to the histogram built for table_field
   */
-  Histogram_binary *get_histogram()
+  Histogram_base *get_histogram()
   {
-    return dynamic_cast<Histogram_binary *>(table_field->collected_stats->histogram_);
+    return table_field->collected_stats->histogram_;
   }
 
 };
diff --git a/sql/sql_statistics.h b/sql/sql_statistics.h
index 76ab8694789..3e4e917dc46 100644
--- a/sql/sql_statistics.h
+++ b/sql/sql_statistics.h
@@ -16,6 +16,7 @@
 #ifndef SQL_STATISTICS_H
 #define SQL_STATISTICS_H
 
+#include <vector>
 /*
   For COMPLEMENTARY_FOR_QUERIES and PREFERABLY_FOR_QUERIES they are
   similar to the COMPLEMENTARY and PREFERABLY respectively except that
@@ -341,11 +342,12 @@ private:
   Histogram_type type;
   uint8 size; /* Number of elements in the histogram*/
   uchar *values;
+  std::vector<std::string> hist_buckets;
 
 public:
-  bool parse(MEM_ROOT *mem_root, Histogram_type type_arg, const uchar *ptr, uint size) override {return false;}
+  bool parse(MEM_ROOT *mem_root, Histogram_type type_arg, const uchar *ptr, uint size) override;
 
-  void serialize(Field *to_field) override{}
+  void serialize(Field *field) override;
 
   // returns number of buckets in the histogram
   uint get_width() override