longer regex error messages

extra/comp_err.c

@@ -31,6 +31,7 @@
 #include <m_string.h>
 #include <my_getopt.h>
 #include <my_dir.h>
+#include <ctype.h>
 
 #define MAX_ROWS  3000
 #define ERRORS_PER_RANGE 1000
@@ -749,18 +750,19 @@ static struct message *find_message(struct errors *err, const char *lang,
                    for the format specifiers
 
   RETURN VALUE
-    Returns the checksum for all the characters of the
+    Returns the checksum for all letters of the
     format specifiers
 
     Ex.
-     "text '%-64.s' text part 2 %d'"
-            ^^^^^^              ^^
+     "text '%-.64s' text part 2 %zu'"
+                 ^               ^^
             characters will be xored to form checksum
 
+    Non-letters are skipped, because they do not change the type
+    of the argument.
+
     NOTE:
-      Does not support format specifiers with positional args
-      like "%2$s" but that is not yet supported by my_vsnprintf
-      either.
+      Does not support format specifiers with positional args like "%2$s"
 */
 
 static ha_checksum checksum_format_specifier(const char* msg)
@@ -777,20 +779,17 @@ static ha_checksum checksum_format_specifier(const char* msg)
       start= p+1; /* Entering format specifier */
       num_format_specifiers++;
     }
-    else if (start)
+    else if (start && isalpha(*p))
     {
+      chksum= my_checksum(chksum, p, 1);
       switch(*p) {
       case 'd':
       case 'u':
       case 'x':
       case 's':
       case 'M':
-        chksum= my_checksum(chksum, (uchar*) start, (uint) (p + 1 - start));
         start= 0; /* Not in format specifier anymore */
         break;
-
-      default:
-        break;
       }
     }
 

mysql-test/main/func_regexp_pcre.result

@@ -793,7 +793,7 @@ SELECT 'a\nb' RLIKE '(?-s)a.b';
 0
 SET default_regex_flags=DEFAULT;
 SELECT REGEXP_SUBSTR('Monday Mon','^((?<DN>Mon|Fri|Sun)day|(?<DN>Tue)sday).*(?P=DN)$');
-ERROR 42000: Got error 'two named subpatterns have the same name (PCRE2_DUPNAMES not set' from regexp
+ERROR 42000: Regex error 'two named subpatterns have the same name (PCRE2_DUPNAMES not set) at offset 30'
 SET default_regex_flags='DUPNAMES';
 SELECT REGEXP_SUBSTR('Monday Mon','^((?<DN>Mon|Fri|Sun)day|(?<DN>Tue)sday).*(?P=DN)$');
 REGEXP_SUBSTR('Monday Mon','^((?<DN>Mon|Fri|Sun)day|(?<DN>Tue)sday).*(?P=DN)$')
@@ -817,10 +817,10 @@ SELECT 'AB' RLIKE 'A# this is a comment\nB';
 1
 SET default_regex_flags=DEFAULT;
 SELECT 'Aq' RLIKE 'A\\q';
-ERROR 42000: Got error 'unrecognized character follows \ at offset 2' from regexp
+ERROR 42000: Regex error 'unrecognized character follows \ at offset 2'
 SET default_regex_flags='EXTRA';
 SELECT 'Aq' RLIKE 'A\\q';
-ERROR 42000: Got error 'unrecognized character follows \ at offset 2' from regexp
+ERROR 42000: Regex error 'unrecognized character follows \ at offset 2'
 SET default_regex_flags=DEFAULT;
 SELECT 'a\nb\nc' RLIKE '^b$';
 'a\nb\nc' RLIKE '^b$'
@@ -860,7 +860,7 @@ SELECT 0xE001 REGEXP @regCheck;
 0xE001 REGEXP @regCheck
 0
 Warnings:
-Warning	1139	Got error 'UTF-8 error: 1 byte missing at end' from regexp
+Warning	1139	Regex error 'UTF-8 error: 1 byte missing at end'
 # Testing workaround N1: This makes the pattern to be a binary string:
 SET NAMES latin1;
 SET @regCheck= X'E001';

mysql-test/main/func_regexp_pcre_debug.result

@@ -3,7 +3,7 @@ SELECT 'a' RLIKE 'a';
 'a' RLIKE 'a'
 0
 Warnings:
-Warning	1139	Got error 'pcre_exec: Internal error (-123)' from regexp
+Warning	1139	Regex error 'pcre_exec: Internal error (-123)'
 SET debug_dbug='';
 SELECT 'a' RLIKE 'a';
 'a' RLIKE 'a'

sql/share/errmsg-utf8.txt

@@ -3186,13 +3186,13 @@ ER_INVALID_USE_OF_NULL 22004
         swe "Felaktig använding av NULL"
         ukr "Хибне використання значення NULL"
 ER_REGEXP_ERROR 42000 
-        cze "Regulární výraz vrátil chybu '%-.64s'"
+        cze "Regulární výraz vrátil chybu: %s"
         dan "Fik fejl '%-.64s' fra regexp"
         nla "Fout '%-.64s' ontvangen van regexp"
-        eng "Got error '%-.64s' from regexp"
-        est "regexp tagastas vea '%-.64s'"
+        eng "Regex error '%s'"
+        est "regexp tagastas vea: %s"
         fre "Erreur '%-.64s' provenant de regexp"
-        ger "regexp lieferte Fehler '%-.64s'"
+        ger "Regexp Fehler %s"
         hindi "regexp में '%-.64s' त्रुटि हुई"
         hun "'%-.64s' hiba a regularis kifejezes hasznalata soran (regexp)"
         ita "Errore '%-.64s' da regexp"
@@ -3200,11 +3200,11 @@ ER_REGEXP_ERROR 42000
         kor "regexp에서 '%-.64s'가 났습니다."
         por "Obteve erro '%-.64s' em regexp"
         rum "Eroarea '%-.64s' obtinuta din expresia regulara (regexp)"
-        rus "Получена ошибка '%-.64s' от регулярного выражения"
-        serbian "Funkcija regexp je vratila grešku '%-.64s'"
+        rus "Ошибка регулярного выражения: %s"
+        serbian "Funkcija regexp je vratila grešku: %s"
         spa "Obtenido error '%-.64s' de regexp"
         swe "Fick fel '%-.64s' från REGEXP"
-        ukr "Отримано помилку '%-.64s' від регулярного виразу"
+        ukr "Помилка регулярного виразу: %s"
 ER_MIX_OF_GROUP_FUNC_AND_FIELDS 42000 
         cze "Pokud není žádná GROUP BY klauzule, není dovoleno současné použití GROUP položek (MIN(),MAX(),COUNT()...) s ne GROUP položkami"
         dan "Sammenblanding af GROUP kolonner (MIN(),MAX(),COUNT()...) uden GROUP kolonner er ikke tilladt, hvis der ikke er noget GROUP BY prædikat"