Improve tests and test results to cover larger cases

Signed-off-by: Michael Okoko <okokomichaels@outlook.com>

mysql-test/main/statistics_fetch.test

@@ -3,38 +3,47 @@
 --echo # todo: should be merged with statistics_json.test
 --echo #
 
+set @save_histogram_type=@@histogram_type;
+set @save_histogram_size=@@histogram_size;
+
 create table ten(a int primary key);
 insert into ten values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9);
 
-create table t1 (a varchar(255));
-insert into t1 select concat('a-', a) from ten;
-
-analyze table t1 persistent for all;
-select * from mysql.column_stats where table_name='t1';
-explain extended select * from t1 where a between 'a-3a' and 'zzzzzzzzz';
-
-create table t2(a int);
-insert into t2 select a*10 from ten;
-analyze table t2 persistent for all;
-explain extended select * from t2 where a between '44' and '55';
+create table t1_bin (a varchar(255));
+insert into t1_bin select concat('a-', a) from ten;
 
-drop table t1;
-drop table t2;
+set histogram_size=10;
+analyze table t1_bin persistent for all;
+select hex(histogram) from mysql.column_stats where table_name='t1_bin';
+explain extended select * from t1_bin where a between 'a-3a' and 'zzzzzzzzz';
+analyze select * from t1_bin where a between 'a-3a' and 'zzzzzzzzz';
 
-create table t1 (a varchar(255));
-insert into t1 select concat('a-', a) from ten;
+create table t1_json (a varchar(255));
+insert into t1_json select concat('a-', a) from ten;
 set histogram_type=json;
+analyze table t1_json persistent for all;
+select * from mysql.column_stats where table_name='t1_json';
+explain extended select * from t1_json where a between 'a-3a' and 'zzzzzzzzz';
+analyze select * from t1_json where a between 'a-3a' and 'zzzzzzzzz';
 
-analyze table t1 persistent for all;
-select * from mysql.column_stats where table_name='t1';
-explain extended select * from t1 where a between 'a-3a' and 'zzzzzzzzz';
 
-create table t2(a int);
-insert into t2 select a*10 from ten;
-set histogram_type=json;
-analyze table t2 persistent for all;
-explain extended select * from t2 where a between '44' and '55';
+create table t2_bin(a int);
+insert into t2_bin select a*10 from ten;
+set histogram_type=@save_histogram_type;
+analyze table t2_bin persistent for all;
+explain extended select * from t2_bin where a between '44' and '55';
+analyze select * from t2_bin where a between '44' and '55';
 
-drop table t1;
-drop table t2;
+create table t2_json(a int);
+insert into t2_json select a*10 from ten;
+set histogram_type=json;
+analyze table t2_json persistent for all;
+select * from mysql.column_stats where table_name='t2_json';
+explain extended select * from t2_json where a between '44' and '55';
+analyze select * from t2_json where a between '44' and '55';
+
+drop table t1_bin;
+drop table t1_json;
+drop table t2_bin;
+drop table t2_json;
 

mysql-test/main/statistics_json.test

@@ -12,34 +12,80 @@ set @save_histogram_type=@@histogram_type;
 set @save_histogram_size=@@histogram_size;
 
 CREATE TABLE t1 (
-  a int,
-  b varchar(32),
-  c char(2),
-  d double
+    a int NOT NULL PRIMARY KEY,
+    b varchar(32),
+    c char(16),
+    d date,
+    e double,
+    f bit(3),
+    INDEX idx1 (b, e),
+    INDEX idx2 (c, d),
+    INDEX idx3 (d),
+    INDEX idx4 (e, b, d)
 );
 
 --disable_result_log
-INSERT INTO t1 SELECT seq, seq, seq, seq from seq_1_to_25;
+
+INSERT INTO t1 VALUES
+  (0, NULL, NULL, NULL, NULL, NULL),
+  (7, 'xxxxxxxxxxxxxxxxxxxxxxxxxx', 'dddddddd', '1990-05-15', 0.1, b'100'),
+  (17, 'vvvvvvvvvvvvv', 'aaaa', '1989-03-12', 0.01, b'101'),
+  (1, 'vvvvvvvvvvvvv', NULL, '1989-03-12', 0.01, b'100'),
+  (12, 'wwwwwwwwwwwwwwwwwwwwwwwwwwww', 'dddddddd', '1999-07-23', 0.112, b'001'),
+  (23, 'vvvvvvvvvvvvv', 'dddddddd', '1999-07-23', 0.1, b'100'),
+  (8, 'vvvvvvvvvvvvv', 'aaaa', '1999-07-23', 0.1, b'100'),
+  (22, 'xxxxxxxxxxxxxxxxxxxxxxxxxx', 'aaaa', '1989-03-12', 0.112, b'001'),
+  (31, 'wwwwwwwwwwwwwwwwwwwwwwwwwwww', 'aaaa', '1999-07-23', 0.01, b'001'),
+  (10, NULL, 'aaaa', NULL, 0.01, b'010'),
+  (5, 'wwwwwwwwwwwwwwwwwwwwwwwwwwww', 'dddddddd', '1999-07-23', 0.1, b'100'),
+  (15, 'vvvvvvvvvvvvv', 'ccccccccc', '1990-05-15', 0.1, b'010'),
+  (30, NULL, 'bbbbbb', NULL, NULL, b'100'),
+  (38, 'zzzzzzzzzzzzzzzzzz', 'bbbbbb', NULL, NULL, NULL),
+  (18, 'zzzzzzzzzzzzzzzzzz', 'ccccccccc', '1990-05-15', 0.01, b'010'),
+  (9, 'yyy', 'bbbbbb', '1998-08-28', 0.01, NULL),
+  (29, 'vvvvvvvvvvvvv', 'dddddddd', '1999-07-23', 0.012, b'010'),
+  (3, 'yyy', 'dddddddd',  '1990-05-15', 0.112, b'010'),
+  (39, 'zzzzzzzzzzzzzzzzzz', 'bbbbbb', NULL, 0.01, b'100'),
+  (14, 'xxxxxxxxxxxxxxxxxxxxxxxxxx', 'ccccccccc', '1990-05-15', 0.1, b'100'),
+  (40, 'zzzzzzzzzzzzzzzzzz', 'bbbbbb', '1989-03-12', NULL, NULL),
+  (44, NULL, 'aaaa', '1989-03-12', NULL, b'010'),
+  (19, 'vvvvvvvvvvvvv', 'ccccccccc', '1990-05-15', 0.012, b'011'),
+  (21, 'zzzzzzzzzzzzzzzzzz', 'dddddddd', '1989-03-12', 0.112, b'100'),
+  (45, NULL, NULL, '1989-03-12', NULL, b'011'),
+  (2, 'wwwwwwwwwwwwwwwwwwwwwwwwwwww', 'ccccccccc', '1990-05-15', 0.1, b'001'),
+  (35, 'yyy', 'aaaa', '1990-05-15', 0.05, b'011'),
+  (4, 'vvvvvvvvvvvvv', 'dddddddd', '1999-07-23', 0.01, b'101'),
+  (47, NULL, 'aaaa', '1990-05-15', 0.05, b'010'),
+  (42, NULL, 'ccccccccc', '1989-03-12', 0.01, b'010'),
+  (32, NULL, 'bbbbbb', '1990-05-15', 0.01, b'011'),
+  (49, 'wwwwwwwwwwwwwwwwwwwwwwwwwwww' , 'aaaa', '1990-05-15', NULL, NULL),
+  (43, 'wwwwwwwwwwwwwwwwwwwwwwwwwwww' , 'bbbbbb', '1990-05-15', NULL, b'100'),
+  (37, 'yyy', NULL, '1989-03-12', 0.05, b'011'),
+  (41, 'xxxxxxxxxxxxxxxxxxxxxxxxxx', 'ccccccccc', '1990-05-15', 0.05, NULL),
+  (34, 'yyy', NULL, NULL, NULL, NULL),
+  (33, 'zzzzzzzzzzzzzzzzzz', 'dddddddd', '1989-03-12', 0.05, b'011'),
+  (24,  'wwwwwwwwwwwwwwwwwwwwwwwwwwww', 'dddddddd', '1990-05-15', 0.01,  b'101'),
+  (11, 'yyy', 'ccccccccc', '1999-07-23', 0.1, NULL),
+  (25, 'zzzzzzzzzzzzzzzzzz', 'bbb', '1989-03-12', 0.01,  b'101');
 --enable_result_log
 
 SET histogram_type='JSON';
-# set histogram size to be < row count (25 in this case) to see how histogram behaves
-set histogram_size=10;
+set histogram_size=25;
 
-ANALYZE TABLE t1 PERSISTENT FOR ALL;
-SELECT * FROM mysql.column_stats WHERE table_name='t1';
-explain extended select * from t1 where b between '20' and '70';
+ANALYZE TABLE t1 persistent for all;
+SELECT * FROM mysql.table_stats;
+SELECT * FROM mysql.column_stats;
+SELECT * FROM mysql.index_stats;
+SELECT COUNT(*) FROM t1;
 
-# We then test different valid JSON strings that are invalid histograms.
-UPDATE mysql.column_stats SET histogram='["1", {"a": "b"}, "2"]' WHERE table_name='t1';
-FLUSH TABLES;
---error ER_JSON_HISTOGRAM_PARSE_FAILED
-SELECT * FROM t1;
+explain extended select * from t1 where a between '20' and '70';
+analyze select * from t1 where a between '20' and '70';
 
-UPDATE mysql.column_stats SET histogram='{}' WHERE table_name='t1';
-FLUSH TABLES;
---error ER_JSON_HISTOGRAM_PARSE_FAILED
-SELECT * FROM t1;
+# todo: test different valid JSON strings that are invalid histograms.
+# UPDATE mysql.column_stats SET histogram='["1", {"a": "b"}, "2"]' WHERE table_name='t1';
+# FLUSH TABLES;
+# --error ER_JSON_HISTOGRAM_PARSE_FAILED
+# explain extended select * from t1 where a between '20' and '70';
 
 DELETE FROM mysql.column_stats;
 DROP TABLE t1;
@@ -56,12 +102,14 @@ use world;
 --enable_query_log
 
 set histogram_type='JSON';
-set histogram_size=25;
+set histogram_size=50;
 --disable_result_log
-ANALYZE TABLE Country PERSISTENT FOR ALL;
+ANALYZE TABLE Country, City, CountryLanguage persistent for all;
 --enable_result_log
 
 SELECT column_name, min_value, max_value, hist_size, hist_type, histogram FROM mysql.column_stats;
+explain extended select * from Country where 'Code' between 'BBC' and 'GGG';
+analyze select * from Country where 'Code' between 'BBC' and 'GGG';
 
 set histogram_type=@save_histogram_type;
 set histogram_size=@save_histogram_size;

sql/sql_statistics.cc

@@ -1564,9 +1564,9 @@ int Histogram_json::find_bucket(Field *field, const uchar *endpoint)
     mid_val = histogram_bounds[mid];
 
     int res = field->key_cmp((uchar*) mid_val.data(), endpoint);
-    min_bucket_index = mid;
     if (res < 0) {
       low = mid + 1;
+      min_bucket_index = mid;
     } else if (res > 0) {
       high = mid - 1;
     } else {
@@ -1574,6 +1574,9 @@ int Histogram_json::find_bucket(Field *field, const uchar *endpoint)
       break;
     }
   }
+
+  if (min_bucket_index == -1)
+    min_bucket_index = high;
   return min_bucket_index;
 }