database/mariadb
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2025-08-07 00:04:31 +03:00
Code Activity

MDEV-26764: JSON_HB Histograms: handle BINARY and unassigned characters

Browse Source 
Encode such characters in hex.
This commit is contained in:
Sergei Petrunia
2021-12-03 20:13:43 +03:00
parent 748b293c14
commit d8d57d2c27
4 changed files with 130 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
sql/opt_histogram_json.h
View File

@@ -32,12 +32,18 @@
"histogram_hb": [
{ "start": "value", "size":nnn.nn, "ndv": nnn },
...
{ "start": "value", "size":nnn.nn, "ndv": nnn, "end": "value"}
// Optionally, start and/or end can be replaced with _hex variant
{ "start_hex: "value", "size":nnn.nn, "ndv":nnn},
...
{ "start": "value", "size":nnn.nn, "ndv": nnn, "end": "value"},
]
}
The histogram is an object with single member named Histogram_json_hb::
JSON_NAME. The value of that member is an array of buckets.
Each bucket is an object with these members:
"start" - the first value in the bucket.
"size" - fraction of table rows that is contained in the bucket.
@@ -51,6 +57,11 @@
The exception is single-point buckets where last value is the same as the
first value.
start/end can be replaced with start_hex/end_hex. In _hex variant, the
constant is encoded in hex. This encoding is used to handle so called
"unassigned characters": some non-UTF8 charsets have byte combinations that
are not mapped to any UTF8 character.
*/
class Histogram_json_hb : public Histogram_base