MDEV-5357 REGEXP word boundaries don't work

Applied a patch from Philip Hazel implementing the non-standard
syntax for word boundaries in PCRE, for compatibility with the
old Henry Spencer's regex library.

11 files changed, 192 insertions, 69 deletions

diff --git a/mysql-test/include/ctype_regex_utf8.inc b/mysql-test/include/ctype_regex_utf8.inc
new file mode 100644
index 00000000000..d389cb214f7
--- /dev/null
+++ b/mysql-test/include/ctype_regex_utf8.inc
@@ -0,0 +1,24 @@
+#
+# Bug #3928 regexp [[:>:]] and UTF-8
+#
+SELECT @@character_set_client, @@collation_connection;
+
+# This should return TRUE
+select  'вася'  rlike '\\bвася\\b';
+select  'вася ' rlike '\\bвася\\b';
+select ' вася'  rlike '\\bвася\\b';
+select ' вася ' rlike '\\bвася\\b';
+
+select  'вася'  rlike '[[:<:]]вася[[:>:]]';
+select  'вася ' rlike '[[:<:]]вася[[:>:]]';
+select ' вася'  rlike '[[:<:]]вася[[:>:]]';
+select ' вася ' rlike '[[:<:]]вася[[:>:]]';
+
+# This should return FALSE
+select  'васяz' rlike '\\bвася\\b';
+select 'zвася'  rlike '\\bвася\\b';
+select 'zвасяz' rlike '\\bвася\\b';
+
+select  'васяz' rlike '[[:<:]]вася[[:>:]]';
+select 'zвася'  rlike '[[:<:]]вася[[:>:]]';
+select 'zвасяz' rlike '[[:<:]]вася[[:>:]]';
diff --git a/mysql-test/include/ctype_utf8mb4.inc b/mysql-test/include/ctype_utf8mb4.inc
index af3a4564026..9ee2414e142 100644
--- a/mysql-test/include/ctype_utf8mb4.inc
+++ b/mysql-test/include/ctype_utf8mb4.inc
@@ -224,25 +224,9 @@ drop table t1;
 #
 # Testing regexp
 #
-set collation_connection=utf8mb4_general_ci;
---source include/ctype_regex.inc
-set names utf8mb4;
-
-#
-# Bug #3928 regexp [[:>:]] and UTF-8
-#
 set names utf8mb4;
-
-# This should return TRUE
-select  'вася'  rlike '\\bвася\\b';
-select  'вася ' rlike '\\bвася\\b';
-select ' вася'  rlike '\\bвася\\b';
-select ' вася ' rlike '\\bвася\\b';
-
-# This should return FALSE
-select  'васяz' rlike '\\bвася\\b';
-select 'zвася'  rlike '\\bвася\\b';
-select 'zвасяz' rlike '\\bвася\\b';
+--source include/ctype_regex.inc
+--source include/ctype_regex_utf8.inc
 
 #
 # Bug #4555
diff --git a/mysql-test/r/ctype_utf8.result b/mysql-test/r/ctype_utf8.result
index 7342dffa1e0..f98fe649f80 100644
--- a/mysql-test/r/ctype_utf8.result
+++ b/mysql-test/r/ctype_utf8.result
@@ -270,7 +270,7 @@ b
 select * from t1 where a = 'b' and a != 'b';
 a
 drop table t1;
-set collation_connection=utf8_general_ci;
+set names utf8;
 drop table if exists t1;
 create table t1 as
 select repeat(' ', 64) as s1, repeat(' ',64) as s2
@@ -314,8 +314,9 @@ NULL
 NULL
 NULL
 drop table t1;
-set names utf8;
-set names utf8;
+SELECT @@character_set_client, @@collation_connection;
+@@character_set_client	@@collation_connection
+utf8	utf8_general_ci
 select  'вася'  rlike '\\bвася\\b';
 'вася'  rlike '\\bвася\\b'
 1
@@ -328,6 +329,18 @@ select ' вася'  rlike '\\bвася\\b';
 select ' вася ' rlike '\\bвася\\b';
 ' вася ' rlike '\\bвася\\b'
 1
+select  'вася'  rlike '[[:<:]]вася[[:>:]]';
+'вася'  rlike '[[:<:]]вася[[:>:]]'
+1
+select  'вася ' rlike '[[:<:]]вася[[:>:]]';
+'вася ' rlike '[[:<:]]вася[[:>:]]'
+1
+select ' вася'  rlike '[[:<:]]вася[[:>:]]';
+' вася'  rlike '[[:<:]]вася[[:>:]]'
+1
+select ' вася ' rlike '[[:<:]]вася[[:>:]]';
+' вася ' rlike '[[:<:]]вася[[:>:]]'
+1
 select  'васяz' rlike '\\bвася\\b';
 'васяz' rlike '\\bвася\\b'
 0
@@ -337,6 +350,15 @@ select 'zвася'  rlike '\\bвася\\b';
 select 'zвасяz' rlike '\\bвася\\b';
 'zвасяz' rlike '\\bвася\\b'
 0
+select  'васяz' rlike '[[:<:]]вася[[:>:]]';
+'васяz' rlike '[[:<:]]вася[[:>:]]'
+0
+select 'zвася'  rlike '[[:<:]]вася[[:>:]]';
+'zвася'  rlike '[[:<:]]вася[[:>:]]'
+0
+select 'zвасяz' rlike '[[:<:]]вася[[:>:]]';
+'zвасяz' rlike '[[:<:]]вася[[:>:]]'
+0
 CREATE TABLE t1 (a enum ('Y', 'N') DEFAULT 'N' COLLATE utf8_unicode_ci);
 ALTER TABLE t1 ADD COLUMN b CHAR(20);
 DROP TABLE t1;
diff --git a/mysql-test/r/ctype_utf8mb4.result b/mysql-test/r/ctype_utf8mb4.result
index e9608188e9f..4580d90c5bc 100644
--- a/mysql-test/r/ctype_utf8mb4.result
+++ b/mysql-test/r/ctype_utf8mb4.result
@@ -270,7 +270,7 @@ b
 select * from t1 where a = 'b' and a != 'b';
 a
 drop table t1;
-set collation_connection=utf8mb4_general_ci;
+set names utf8mb4;
 drop table if exists t1;
 create table t1 as
 select repeat(' ', 64) as s1, repeat(' ',64) as s2
@@ -314,8 +314,9 @@ NULL
 NULL
 NULL
 drop table t1;
-set names utf8mb4;
-set names utf8mb4;
+SELECT @@character_set_client, @@collation_connection;
+@@character_set_client	@@collation_connection
+utf8mb4	utf8mb4_general_ci
 select  'вася'  rlike '\\bвася\\b';
 'вася'  rlike '\\bвася\\b'
 1
@@ -328,6 +329,18 @@ select ' вася'  rlike '\\bвася\\b';
 select ' вася ' rlike '\\bвася\\b';
 ' вася ' rlike '\\bвася\\b'
 1
+select  'вася'  rlike '[[:<:]]вася[[:>:]]';
+'вася'  rlike '[[:<:]]вася[[:>:]]'
+1
+select  'вася ' rlike '[[:<:]]вася[[:>:]]';
+'вася ' rlike '[[:<:]]вася[[:>:]]'
+1
+select ' вася'  rlike '[[:<:]]вася[[:>:]]';
+' вася'  rlike '[[:<:]]вася[[:>:]]'
+1
+select ' вася ' rlike '[[:<:]]вася[[:>:]]';
+' вася ' rlike '[[:<:]]вася[[:>:]]'
+1
 select  'васяz' rlike '\\bвася\\b';
 'васяz' rlike '\\bвася\\b'
 0
@@ -337,6 +350,15 @@ select 'zвася'  rlike '\\bвася\\b';
 select 'zвасяz' rlike '\\bвася\\b';
 'zвасяz' rlike '\\bвася\\b'
 0
+select  'васяz' rlike '[[:<:]]вася[[:>:]]';
+'васяz' rlike '[[:<:]]вася[[:>:]]'
+0
+select 'zвася'  rlike '[[:<:]]вася[[:>:]]';
+'zвася'  rlike '[[:<:]]вася[[:>:]]'
+0
+select 'zвасяz' rlike '[[:<:]]вася[[:>:]]';
+'zвасяz' rlike '[[:<:]]вася[[:>:]]'
+0
 CREATE TABLE t1 (a enum ('Y', 'N') DEFAULT 'N' COLLATE utf8mb4_unicode_ci);
 ALTER TABLE t1 ADD COLUMN b CHAR(20);
 DROP TABLE t1;
diff --git a/mysql-test/r/ctype_utf8mb4_heap.result b/mysql-test/r/ctype_utf8mb4_heap.result
index 0ffe26b5f25..bcacdd4e16e 100644
--- a/mysql-test/r/ctype_utf8mb4_heap.result
+++ b/mysql-test/r/ctype_utf8mb4_heap.result
@@ -260,7 +260,7 @@ b
 select * from t1 where a = 'b' and a != 'b';
 a
 drop table t1;
-set collation_connection=utf8mb4_general_ci;
+set names utf8mb4;
 drop table if exists t1;
 create table t1 as
 select repeat(' ', 64) as s1, repeat(' ',64) as s2
@@ -304,8 +304,9 @@ NULL
 NULL
 NULL
 drop table t1;
-set names utf8mb4;
-set names utf8mb4;
+SELECT @@character_set_client, @@collation_connection;
+@@character_set_client	@@collation_connection
+utf8mb4	utf8mb4_general_ci
 select  'вася'  rlike '\\bвася\\b';
 'вася'  rlike '\\bвася\\b'
 1
@@ -318,6 +319,18 @@ select ' вася'  rlike '\\bвася\\b';
 select ' вася ' rlike '\\bвася\\b';
 ' вася ' rlike '\\bвася\\b'
 1
+select  'вася'  rlike '[[:<:]]вася[[:>:]]';
+'вася'  rlike '[[:<:]]вася[[:>:]]'
+1
+select  'вася ' rlike '[[:<:]]вася[[:>:]]';
+'вася ' rlike '[[:<:]]вася[[:>:]]'
+1
+select ' вася'  rlike '[[:<:]]вася[[:>:]]';
+' вася'  rlike '[[:<:]]вася[[:>:]]'
+1
+select ' вася ' rlike '[[:<:]]вася[[:>:]]';
+' вася ' rlike '[[:<:]]вася[[:>:]]'
+1
 select  'васяz' rlike '\\bвася\\b';
 'васяz' rlike '\\bвася\\b'
 0
@@ -327,6 +340,15 @@ select 'zвася'  rlike '\\bвася\\b';
 select 'zвасяz' rlike '\\bвася\\b';
 'zвасяz' rlike '\\bвася\\b'
 0
+select  'васяz' rlike '[[:<:]]вася[[:>:]]';
+'васяz' rlike '[[:<:]]вася[[:>:]]'
+0
+select 'zвася'  rlike '[[:<:]]вася[[:>:]]';
+'zвася'  rlike '[[:<:]]вася[[:>:]]'
+0
+select 'zвасяz' rlike '[[:<:]]вася[[:>:]]';
+'zвасяz' rlike '[[:<:]]вася[[:>:]]'
+0
 CREATE TABLE t1 (a enum ('Y', 'N') DEFAULT 'N' COLLATE utf8mb4_unicode_ci) ENGINE heap;
 ALTER TABLE t1 ADD COLUMN b CHAR(20);
 DROP TABLE t1;
diff --git a/mysql-test/r/ctype_utf8mb4_innodb.result b/mysql-test/r/ctype_utf8mb4_innodb.result
index 3e1554cd0ae..2375ca3bb92 100644
--- a/mysql-test/r/ctype_utf8mb4_innodb.result
+++ b/mysql-test/r/ctype_utf8mb4_innodb.result
@@ -270,7 +270,7 @@ b
 select * from t1 where a = 'b' and a != 'b';
 a
 drop table t1;
-set collation_connection=utf8mb4_general_ci;
+set names utf8mb4;
 drop table if exists t1;
 create table t1 as
 select repeat(' ', 64) as s1, repeat(' ',64) as s2
@@ -314,8 +314,9 @@ NULL
 NULL
 NULL
 drop table t1;
-set names utf8mb4;
-set names utf8mb4;
+SELECT @@character_set_client, @@collation_connection;
+@@character_set_client	@@collation_connection
+utf8mb4	utf8mb4_general_ci
 select  'вася'  rlike '\\bвася\\b';
 'вася'  rlike '\\bвася\\b'
 1
@@ -328,6 +329,18 @@ select ' вася'  rlike '\\bвася\\b';
 select ' вася ' rlike '\\bвася\\b';
 ' вася ' rlike '\\bвася\\b'
 1
+select  'вася'  rlike '[[:<:]]вася[[:>:]]';
+'вася'  rlike '[[:<:]]вася[[:>:]]'
+1
+select  'вася ' rlike '[[:<:]]вася[[:>:]]';
+'вася ' rlike '[[:<:]]вася[[:>:]]'
+1
+select ' вася'  rlike '[[:<:]]вася[[:>:]]';
+' вася'  rlike '[[:<:]]вася[[:>:]]'
+1
+select ' вася ' rlike '[[:<:]]вася[[:>:]]';
+' вася ' rlike '[[:<:]]вася[[:>:]]'
+1
 select  'васяz' rlike '\\bвася\\b';
 'васяz' rlike '\\bвася\\b'
 0
@@ -337,6 +350,15 @@ select 'zвася'  rlike '\\bвася\\b';
 select 'zвасяz' rlike '\\bвася\\b';
 'zвасяz' rlike '\\bвася\\b'
 0
+select  'васяz' rlike '[[:<:]]вася[[:>:]]';
+'васяz' rlike '[[:<:]]вася[[:>:]]'
+0
+select 'zвася'  rlike '[[:<:]]вася[[:>:]]';
+'zвася'  rlike '[[:<:]]вася[[:>:]]'
+0
+select 'zвасяz' rlike '[[:<:]]вася[[:>:]]';
+'zвасяz' rlike '[[:<:]]вася[[:>:]]'
+0
 CREATE TABLE t1 (a enum ('Y', 'N') DEFAULT 'N' COLLATE utf8mb4_unicode_ci) ENGINE InnoDB;
 ALTER TABLE t1 ADD COLUMN b CHAR(20);
 DROP TABLE t1;
diff --git a/mysql-test/r/ctype_utf8mb4_myisam.result b/mysql-test/r/ctype_utf8mb4_myisam.result
index 4efcedf708f..b467f07ece9 100644
--- a/mysql-test/r/ctype_utf8mb4_myisam.result
+++ b/mysql-test/r/ctype_utf8mb4_myisam.result
@@ -270,7 +270,7 @@ b
 select * from t1 where a = 'b' and a != 'b';
 a
 drop table t1;
-set collation_connection=utf8mb4_general_ci;
+set names utf8mb4;
 drop table if exists t1;
 create table t1 as
 select repeat(' ', 64) as s1, repeat(' ',64) as s2
@@ -314,8 +314,9 @@ NULL
 NULL
 NULL
 drop table t1;
-set names utf8mb4;
-set names utf8mb4;
+SELECT @@character_set_client, @@collation_connection;
+@@character_set_client	@@collation_connection
+utf8mb4	utf8mb4_general_ci
 select  'вася'  rlike '\\bвася\\b';
 'вася'  rlike '\\bвася\\b'
 1
@@ -328,6 +329,18 @@ select ' вася'  rlike '\\bвася\\b';
 select ' вася ' rlike '\\bвася\\b';
 ' вася ' rlike '\\bвася\\b'
 1
+select  'вася'  rlike '[[:<:]]вася[[:>:]]';
+'вася'  rlike '[[:<:]]вася[[:>:]]'
+1
+select  'вася ' rlike '[[:<:]]вася[[:>:]]';
+'вася ' rlike '[[:<:]]вася[[:>:]]'
+1
+select ' вася'  rlike '[[:<:]]вася[[:>:]]';
+' вася'  rlike '[[:<:]]вася[[:>:]]'
+1
+select ' вася ' rlike '[[:<:]]вася[[:>:]]';
+' вася ' rlike '[[:<:]]вася[[:>:]]'
+1
 select  'васяz' rlike '\\bвася\\b';
 'васяz' rlike '\\bвася\\b'
 0
@@ -337,6 +350,15 @@ select 'zвася'  rlike '\\bвася\\b';
 select 'zвасяz' rlike '\\bвася\\b';
 'zвасяz' rlike '\\bвася\\b'
 0
+select  'васяz' rlike '[[:<:]]вася[[:>:]]';
+'васяz' rlike '[[:<:]]вася[[:>:]]'
+0
+select 'zвася'  rlike '[[:<:]]вася[[:>:]]';
+'zвася'  rlike '[[:<:]]вася[[:>:]]'
+0
+select 'zвасяz' rlike '[[:<:]]вася[[:>:]]';
+'zвасяz' rlike '[[:<:]]вася[[:>:]]'
+0
 CREATE TABLE t1 (a enum ('Y', 'N') DEFAULT 'N' COLLATE utf8mb4_unicode_ci) ENGINE MyISAM;
 ALTER TABLE t1 ADD COLUMN b CHAR(20);
 DROP TABLE t1;
diff --git a/mysql-test/t/ctype_utf8.test b/mysql-test/t/ctype_utf8.test
index 6f2222b8e45..468804130f4 100644
--- a/mysql-test/t/ctype_utf8.test
+++ b/mysql-test/t/ctype_utf8.test
@@ -199,25 +199,9 @@ drop table t1;
 #
 # Testing regexp
 #
-set collation_connection=utf8_general_ci;
---source include/ctype_regex.inc
-set names utf8;
-
-#
-# Bug #3928 regexp [[:>:]] and UTF-8
-#
 set names utf8;
-
-# This should return TRUE
-select  'вася'  rlike '\\bвася\\b';
-select  'вася ' rlike '\\bвася\\b';
-select ' вася'  rlike '\\bвася\\b';
-select ' вася ' rlike '\\bвася\\b';
-
-# This should return FALSE
-select  'васяz' rlike '\\bвася\\b';
-select 'zвася'  rlike '\\bвася\\b';
-select 'zвасяz' rlike '\\bвася\\b';
+--source include/ctype_regex.inc
+--source include/ctype_regex_utf8.inc
 
 #
 # Bug #4555
diff --git a/mysql-test/t/ctype_utf8mb4.test b/mysql-test/t/ctype_utf8mb4.test
index 934adb50cca..7a3c67bb417 100644
--- a/mysql-test/t/ctype_utf8mb4.test
+++ b/mysql-test/t/ctype_utf8mb4.test
@@ -197,25 +197,9 @@ drop table t1;
 #
 # Testing regexp
 #
-set collation_connection=utf8mb4_general_ci;
---source include/ctype_regex.inc
-set names utf8mb4;
-
-#
-# Bug #3928 regexp [[:>:]] and UTF-8
-#
 set names utf8mb4;
-
-# This should return TRUE
-select  'вася'  rlike '\\bвася\\b';
-select  'вася ' rlike '\\bвася\\b';
-select ' вася'  rlike '\\bвася\\b';
-select ' вася ' rlike '\\bвася\\b';
-
-# This should return FALSE
-select  'васяz' rlike '\\bвася\\b';
-select 'zвася'  rlike '\\bвася\\b';
-select 'zвасяz' rlike '\\bвася\\b';
+--source include/ctype_regex.inc
+--source include/ctype_regex_utf8.inc
 
 #
 # Bug #4555
diff --git a/pcre/pcre_compile.c b/pcre/pcre_compile.c
index 0ebb3f168f1..a307372fbfe 100644
--- a/pcre/pcre_compile.c
+++ b/pcre/pcre_compile.c
@@ -253,6 +253,19 @@ static const verbitem verbs[] = {
 static const int verbcount = sizeof(verbs)/sizeof(verbitem);
 
 
+/* Substitutes for [[:<:]] and [[:>:]], which mean start and end of word in 
+another regex library. */
+
+static const pcre_uchar sub_start_of_word[] = {
+  CHAR_BACKSLASH, CHAR_b, CHAR_LEFT_PARENTHESIS, CHAR_QUESTION_MARK,
+  CHAR_EQUALS_SIGN, CHAR_BACKSLASH, CHAR_w, CHAR_RIGHT_PARENTHESIS, '\0' }; 
+
+static const pcre_uchar sub_end_of_word[] = {
+  CHAR_BACKSLASH, CHAR_b, CHAR_LEFT_PARENTHESIS, CHAR_QUESTION_MARK,
+  CHAR_LESS_THAN_SIGN, CHAR_EQUALS_SIGN, CHAR_BACKSLASH, CHAR_w,
+  CHAR_RIGHT_PARENTHESIS, '\0' }; 
+
+
 /* Tables of names of POSIX character classes and their lengths. The names are
 now all in a single string, to reduce the number of relocations when a shared
 library is dynamically loaded. The list of lengths is terminated by a zero
@@ -4036,8 +4049,30 @@ for (;; ptr++)
       goto FAILED;
       }
     goto NORMAL_CHAR;
+    
+    /* In another (POSIX) regex library, the ugly syntax [[:<:]] and [[:>:]] is 
+    used for "start of word" and "end of word". As these are otherwise illegal
+    sequences, we don't break anything by recognizing them. They are replaced
+    by \b(?=\w) and \b(?<=\w) respectively. Sequences like [a[:<:]] are
+    erroneous and are handled by the normal code below. */
 
     case CHAR_LEFT_SQUARE_BRACKET:
+    if (STRNCMP_UC_C8(ptr+1, STRING_WEIRD_STARTWORD, 6) == 0)
+      {
+      nestptr = ptr + 7;
+      ptr = sub_start_of_word - 1;
+      continue;  
+      }  
+
+    if (STRNCMP_UC_C8(ptr+1, STRING_WEIRD_ENDWORD, 6) == 0)
+      {
+      nestptr = ptr + 7;
+      ptr = sub_end_of_word - 1;
+      continue;  
+      }  
+
+    /* Handle a real character class. */
+ 
     previous = code;
 
     /* PCRE supports POSIX class stuff inside a class. Perl gives an error if
diff --git a/pcre/pcre_internal.h b/pcre/pcre_internal.h
index 307069ca9d6..cd6ef3ed83e 100644
--- a/pcre/pcre_internal.h
+++ b/pcre/pcre_internal.h
@@ -1794,6 +1794,8 @@ only. */
 #define STRING_xdigit               STR_x STR_d STR_i STR_g STR_i STR_t
 
 #define STRING_DEFINE               STR_D STR_E STR_F STR_I STR_N STR_E
+#define STRING_WEIRD_STARTWORD      STR_LEFT_SQUARE_BRACKET STR_COLON STR_LESS_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET
+#define STRING_WEIRD_ENDWORD        STR_LEFT_SQUARE_BRACKET STR_COLON STR_GREATER_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET
 
 #define STRING_CR_RIGHTPAR             STR_C STR_R STR_RIGHT_PARENTHESIS
 #define STRING_LF_RIGHTPAR             STR_L STR_F STR_RIGHT_PARENTHESIS