Rechercher dans le manuel MySQL
12.5.2 Regular Expressions
Table 12.9 Regular Expression Functions and Operators
Name | Description |
---|---|
NOT REGEXP |
Negation of REGEXP |
REGEXP |
Whether string matches regular expression |
REGEXP_INSTR() |
Starting index of substring matching regular expression |
REGEXP_LIKE() |
Whether string matches regular expression |
REGEXP_REPLACE() |
Replace substrings matching regular expression |
REGEXP_SUBSTR() |
Return substring matching regular expression |
RLIKE |
Whether string matches regular expression |
A regular expression is a powerful way of specifying a pattern for a complex search. This section discusses the functions and operators available for regular expression matching and illustrates, with examples, some of the special characters and constructs that can be used for regular expression operations. See also Section 3.3.4.7, “Pattern Matching”.
MySQL implements regular expression support using International Components for Unicode (ICU), which provides full Unicode support and is multibyte safe. (Prior to MySQL 8.0.4, MySQL used Henry Spencer's implementation of regular expressions, which operates in byte-wise fashion and is not multibyte safe. For information about ways in which applications that use regular expressions may be affected by the implementation change, see Regular Expression Compatibility Considerations.)
Regular Expression Functions and Operators
,expr
NOT REGEXPpat
expr
NOT RLIKEpat
This is the same as
NOT (
.expr
REGEXPpat
)
,expr
REGEXPpat
expr
RLIKEpat
Returns 1 if the string
expr
matches the regular expression specified by the patternpat
, 0 otherwise. Ifexpr
orpat
isNULL
, the return value isNULL
.REGEXP
andRLIKE
are synonyms forREGEXP_LIKE()
.For additional information about how matching occurs, see the description for
REGEXP_LIKE()
.- +------------------------+
- +------------------------+
- | 1 |
- +------------------------+
- +---------------------------------------+
- +---------------------------------------+
- | 0 |
- +---------------------------------------+
- +---------------------+
- +---------------------+
- | 1 |
- +---------------------+
- +----------------+-----------------------+
- +----------------+-----------------------+
- | 1 | 0 |
- +----------------+-----------------------+
REGEXP_INSTR(
expr
,pat
[,pos
[,occurrence
[,return_option
[,match_type
]]]])Returns the starting index of the substring of the string
expr
that matches the regular expression specified by the patternpat
, 0 if there is no match. Ifexpr
orpat
isNULL
, the return value isNULL
. Character indexes begin at 1.REGEXP_INSTR()
takes these optional arguments:pos
: The position inexpr
at which to start the search. If omitted, the default is 1.occurrence
: Which occurrence of a match to search for. If omitted, the default is 1.return_option
: Which type of position to return. If this value is 0,REGEXP_INSTR()
returns the position of the matched substring's first character. If this value is 1,REGEXP_INSTR()
returns the position following the matched substring. If omitted, the default is 0.match_type
: A string that specifies how to perform matching. The meaning is as described forREGEXP_LIKE()
.
For additional information about how matching occurs, see the description for
REGEXP_LIKE()
.- +------------------------------------+
- | REGEXP_INSTR('dog cat dog', 'dog') |
- +------------------------------------+
- | 1 |
- +------------------------------------+
- +---------------------------------------+
- | REGEXP_INSTR('dog cat dog', 'dog', 2) |
- +---------------------------------------+
- | 9 |
- +---------------------------------------+
- +-------------------------------------+
- | REGEXP_INSTR('aa aaa aaaa', 'a{2}') |
- +-------------------------------------+
- | 1 |
- +-------------------------------------+
- +-------------------------------------+
- | REGEXP_INSTR('aa aaa aaaa', 'a{4}') |
- +-------------------------------------+
- | 8 |
- +-------------------------------------+
REGEXP_LIKE(
expr
,pat
[,match_type
])Returns 1 if the string
expr
matches the regular expression specified by the patternpat
, 0 otherwise. Ifexpr
orpat
isNULL
, the return value isNULL
.The pattern can be an extended regular expression, the syntax for which is discussed in Regular Expression Syntax. The pattern need not be a literal string. For example, it can be specified as a string expression or table column.
The optional
match_type
argument is a string that may contain any or all the following characters specifying how to perform matching:c
: Case sensitive matching.i
: Case insensitive matching.m
: Multiple-line mode. Recognize line terminators within the string. The default behavior is to match line terminators only at the start and end of the string expression.n
: The.
character matches line terminators. The default is for.
matching to stop at the end of a line.u
: Unix-only line endings. Only the newline character is recognized as a line ending by the.
,^
, and$
match operators.
If characters specifying contradictory options are specified within
match_type
, the rightmost one takes precedence.By default, regular expression operations use the character set and collation of the
expr
andpat
arguments when deciding the type of a character and performing the comparison. If the arguments have different character sets or collations, coercibility rules apply as described in Section 10.8.4, “Collation Coercibility in Expressions”. Arguments may be specified with explicit collation indicators to change comparison behavior.- +---------------------------------------+
- | REGEXP_LIKE('CamelCase', 'CAMELCASE') |
- +---------------------------------------+
- | 1 |
- +---------------------------------------+
- +------------------------------------------------------------------+
- +------------------------------------------------------------------+
- | 0 |
- +------------------------------------------------------------------+
match_type
may be specified with thec
ori
characters to override the default case sensitivity. Exception: If either argument is a binary string, the arguments are handled in case-sensitive fashion as binary strings, even ifmatch_type
contains thei
character.NoteBecause MySQL uses the C escape syntax in strings (for example,
\n
to represent the newline character), you must double any\
that you use in yourexpr
andpat
arguments.- +-------------------------------+
- | REGEXP_LIKE('Michael!', '.*') |
- +-------------------------------+
- | 1 |
- +-------------------------------+
- +----------------------------------------------+
- | REGEXP_LIKE('new*\n*line', 'new\\*.\\*line') |
- +----------------------------------------------+
- | 0 |
- +----------------------------------------------+
- +----------------------------+
- | REGEXP_LIKE('a', '^[a-d]') |
- +----------------------------+
- | 1 |
- +----------------------------+
- +-----------------------+------------------------------+
- +-----------------------+------------------------------+
- | 1 | 0 |
- +-----------------------+------------------------------+
- +---------------------------+
- | REGEXP_LIKE('abc', 'ABC') |
- +---------------------------+
- | 1 |
- +---------------------------+
- +--------------------------------+
- | REGEXP_LIKE('abc', 'ABC', 'c') |
- +--------------------------------+
- | 0 |
- +--------------------------------+
REGEXP_REPLACE(
expr
,pat
,repl
[,pos
[,occurrence
[,match_type
]]])Replaces occurrences in the string
expr
that match the regular expression specified by the patternpat
with the replacement stringrepl
, and returns the resulting string. Ifexpr
,pat
, orrepl
isNULL
, the return value isNULL
.REGEXP_REPLACE()
takes these optional arguments:pos
: The position inexpr
at which to start the search. If omitted, the default is 1.occurrence
: Which occurrence of a match to replace. If omitted, the default is 0 (which means “replace all occurrences”).match_type
: A string that specifies how to perform matching. The meaning is as described forREGEXP_LIKE()
.
Prior to MySQL 8.0.17, the result returned by this function used the
UTF-16
character set; in MySQL 8.0.17 and later, the character set and collation of the expression searched for matches is used. (Bug #94203, Bug #29308212)For additional information about how matching occurs, see the description for
REGEXP_LIKE()
.- +-----------------------------------+
- | REGEXP_REPLACE('a b c', 'b', 'X') |
- +-----------------------------------+
- +-----------------------------------+
- +----------------------------------------------------+
- | REGEXP_REPLACE('abc def ghi', '[a-z]+', 'X', 1, 3) |
- +----------------------------------------------------+
- +----------------------------------------------------+
REGEXP_SUBSTR(
expr
,pat
[,pos
[,occurrence
[,match_type
]]])Returns the substring of the string
expr
that matches the regular expression specified by the patternpat
,NULL
if there is no match. Ifexpr
orpat
isNULL
, the return value isNULL
.REGEXP_SUBSTR()
takes these optional arguments:pos
: The position inexpr
at which to start the search. If omitted, the default is 1.occurrence
: Which occurrence of a match to search for. If omitted, the default is 1.match_type
: A string that specifies how to perform matching. The meaning is as described forREGEXP_LIKE()
.
Prior to MySQL 8.0.17, the result returned by this function used the
UTF-16
character set; in MySQL 8.0.17 and later, the character set and collation of the expression searched for matches is used. (Bug #94203, Bug #29308212)For additional information about how matching occurs, see the description for
REGEXP_LIKE()
.- +----------------------------------------+
- | REGEXP_SUBSTR('abc def ghi', '[a-z]+') |
- +----------------------------------------+
- | abc |
- +----------------------------------------+
- +----------------------------------------------+
- | REGEXP_SUBSTR('abc def ghi', '[a-z]+', 1, 3) |
- +----------------------------------------------+
- | ghi |
- +----------------------------------------------+
A regular expression describes a set of strings. The simplest
regular expression is one that has no special characters in
it. For example, the regular expression
hello
matches hello
and
nothing else.
Nontrivial regular expressions use certain special constructs
so that they can match more than one string. For example, the
regular expression hello|world
contains the
|
alternation operator and matches either
the hello
or world
.
As a more complex example, the regular expression
B[an]*s
matches any of the strings
Bananas
, Baaaaas
,
Bs
, and any other string starting with a
B
, ending with an s
, and
containing any number of a
or
n
characters in between.
The following list covers some of the basic special characters and constructs that can be used in regular expressions. For information about the full regular expression syntax supported by the ICU library used to implement regular expression support, visit the International Components for Unicode website.
^
Match the beginning of a string.
$
Match the end of a string.
.
Match any character (including carriage return and newline, although to match these in the middle of a string, the
m
(multiple line) match-control character or the(?m)
within-pattern modifier must be given).a*
Match any sequence of zero or more
a
characters.a+
Match any sequence of one or more
a
characters.a?
Match either zero or one
a
character.de|abc
Alternation; match either of the sequences
de
orabc
.(abc)*
Match zero or more instances of the sequence
abc
.{1}
,{2,3}
Repetition;
{
andn
}{
notation provide a more general way of writing regular expressions that match many occurrences of the previous atom (or “piece”) of the pattern.m
,n
}m
andn
are integers.a*
Can be written as
a{0,}
.a+
Can be written as
a{1,}
.a?
Can be written as
a{0,1}
.
To be more precise,
a{
matches exactlyn
}n
instances ofa
.a{
matchesn
,}n
or more instances ofa
.a{
matchesm
,n
}m
throughn
instances ofa
, inclusive. If bothm
andn
are given,m
must be less than or equal ton
.[a-dX]
,[^a-dX]
Matches any character that is (or is not, if
^
is used) eithera
,b
,c
,d
orX
. A-
character between two other characters forms a range that matches all characters from the first character to the second. For example,[0-9]
matches any decimal digit. To include a literal]
character, it must immediately follow the opening bracket[
. To include a literal-
character, it must be written first or last. Any character that does not have a defined special meaning inside a[]
pair matches only itself.[=character_class=]
Within a bracket expression (written using
[
and]
),[=character_class=]
represents an equivalence class. It matches all characters with the same collation value, including itself. For example, ifo
and(+)
are the members of an equivalence class,[[=o=]]
,[[=(+)=]]
, and[o(+)]
are all synonymous. An equivalence class may not be used as an endpoint of a range.[:character_class:]
Within a bracket expression (written using
[
and]
),[:character_class:]
represents a character class that matches all characters belonging to that class. The following table lists the standard class names. These names stand for the character classes defined in thectype(3)
manual page. A particular locale may provide other class names. A character class may not be used as an endpoint of a range.Character Class Name Meaning alnum
Alphanumeric characters alpha
Alphabetic characters blank
Whitespace characters cntrl
Control characters digit
Digit characters graph
Graphic characters lower
Lowercase alphabetic characters print
Graphic or space characters punct
Punctuation characters space
Space, tab, newline, and carriage return upper
Uppercase alphabetic characters xdigit
Hexadecimal digit characters
To use a literal instance of a special character in a regular
expression, precede it by two backslash (\) characters. The
MySQL parser interprets one of the backslashes, and the
regular expression library interprets the other. For example,
to match the string 1+2
that contains the
special +
character, only the last of the
following regular expressions is the correct one:
REGEXP_LIKE()
and similar
functions use resources that can be controlled by setting
system variables:
The match engine uses memory for its internal stack. To control the maximum available memory for the stack in bytes, set the
regexp_stack_limit
system variable.The match engine operates in steps. To control the maximum number of steps performed by the engine (and thus indirectly the execution time), set the
regexp_time_limit
system variable. Because this limit is expressed as number of steps, it affects execution time only indirectly. Typically, it is on the order of milliseconds.
Prior to MySQL 8.0.4, MySQL used the Henry Spencer regular expression library to support regular expression operations, rather than International Components for Unicode (ICU). The following discussion describes differences between the Spencer and ICU libraries that may affect applications:
With the Spencer library, the
REGEXP
andRLIKE
operators work in byte-wise fashion, so they are not multibyte safe and may produce unexpected results with multibyte character sets. In addition, these operators compare characters by their byte values and accented characters may not compare as equal even if a given collation treats them as equal.ICU has full Unicode support and is multibyte safe. Its regular expression functions treat all strings as
UTF-16
. You should keep in mind that positional indexes are based on 16-bit chunks and not on code points. This means that, when passed to such functions, characters using more than one chunk may produce unanticipated results, such as those shown here:- +--------------------------+
- | REGEXP_INSTR('??b', 'b') |
- +--------------------------+
- | 5 |
- +--------------------------+
- +--------------------------------+
- | REGEXP_INSTR('??bxxx', 'b', 4) |
- +--------------------------------+
- | 5 |
- +--------------------------------+
Characters within the Unicode Basic Multilingual Plane, which includes characters used by most modern languages, are safe in this regard:
- +----------------------------+
- | REGEXP_INSTR('бжb', 'b') |
- +----------------------------+
- | 3 |
- +----------------------------+
- +----------------------------+
- | REGEXP_INSTR('עבb', 'b') |
- +----------------------------+
- | 3 |
- +----------------------------+
- +------------------------------------+
- | REGEXP_INSTR('µå周çб', '周') |
- +------------------------------------+
- | 3 |
- +------------------------------------+
Emoji, such as the “sushi” character
🍣
(U+1F363) used in the first two examples, are not included in the Basic Multilingual Plane, but rather in Unicode's Supplementary Multilingual Plane. Another issue can arise with emoji and other 4-byte characters whenREGEXP_SUBSTR()
or a similar function begins searching in the middle of a character. Each of the two statements in the following example starts from the second 2-byte position in the first argument. The first statement works on a string consisting solely of 2-byte (BMP) characters. The second statement contains 4-byte characters which are incorrectly interpreted in the result because the first two bytes are stripped off and so the remainder of the character data is misaligned.- +----------------------------------------+
- | REGEXP_SUBSTR('周周周周', '.*', 2) |
- +----------------------------------------+
- | 周周周 |
- +----------------------------------------+
- +--------------------------------+
- | REGEXP_SUBSTR('????', '.*', 2) |
- +--------------------------------+
- | ?㳟揘㳟揘㳟揘 |
- +--------------------------------+
For the
.
operator, the Spencer library matches line-terminator characters (carriage return, newline) anywhere in string expressions, including in the middle. To match line terminator characters in the middle of strings with ICU, specify them
match-control character.The Spencer library supports word-beginning and word-end boundary markers (
[[:<:]]
and[[:>:]]
notation). ICU does not. For ICU, you can use\b
to match word boundaries; double the backslash because MySQL interprets it as the escape character within strings.The Spencer library supports collating element bracket expressions (
[.characters.]
notation). ICU does not.For repetition counts (
{n}
and{m,n}
notation), the Spencer library has a maximum of 255. ICU has no such limit, although the maximum number of match engine steps can be limited by setting theregexp_time_limit
system variable.ICU interprets parentheses as metacharacters. To specify a literal open or close parenthesis
(
in a regular expression, it must be escaped:- +-------------------------+
- | REGEXP_LIKE('(', '\\(') |
- +-------------------------+
- | 1 |
- +-------------------------+
- +-------------------------+
- | REGEXP_LIKE(')', '\\)') |
- +-------------------------+
- | 1 |
- +-------------------------+
ICU also interprets square brackets as metacharacters, but only the opening square bracket need be escaped to be used as a literal character:
- unclosed bracket expression.
- +-------------------------+
- | REGEXP_LIKE('[', '\\[') |
- +-------------------------+
- | 1 |
- +-------------------------+
- +-----------------------+
- | REGEXP_LIKE(']', ']') |
- +-----------------------+
- | 1 |
- +-----------------------+
Deutsche Übersetzung
Sie haben gebeten, diese Seite auf Deutsch zu besuchen. Momentan ist nur die Oberfläche übersetzt, aber noch nicht der gesamte Inhalt.Wenn Sie mir bei Übersetzungen helfen wollen, ist Ihr Beitrag willkommen. Alles, was Sie tun müssen, ist, sich auf der Website zu registrieren und mir eine Nachricht zu schicken, in der Sie gebeten werden, Sie der Gruppe der Übersetzer hinzuzufügen, die Ihnen die Möglichkeit gibt, die gewünschten Seiten zu übersetzen. Ein Link am Ende jeder übersetzten Seite zeigt an, dass Sie der Übersetzer sind und einen Link zu Ihrem Profil haben.
Vielen Dank im Voraus.
Dokument erstellt 26/06/2006, zuletzt geändert 26/10/2018
Quelle des gedruckten Dokuments:https://www.gaudry.be/de/mysql-rf-regexp.html
Die Infobro ist eine persönliche Seite, deren Inhalt in meiner alleinigen Verantwortung liegt. Der Text ist unter der CreativeCommons-Lizenz (BY-NC-SA) verfügbar. Weitere Informationen auf die Nutzungsbedingungen und dem Autor.
Referenzen
Diese Verweise und Links verweisen auf Dokumente, die während des Schreibens dieser Seite konsultiert wurden, oder die zusätzliche Informationen liefern können, aber die Autoren dieser Quellen können nicht für den Inhalt dieser Seite verantwortlich gemacht werden.
Der Autor Diese Website ist allein dafür verantwortlich, wie die verschiedenen Konzepte und Freiheiten, die mit den Nachschlagewerken gemacht werden, hier dargestellt werden. Denken Sie daran, dass Sie mehrere Quellinformationen austauschen müssen, um das Risiko von Fehlern zu reduzieren.