ABAP Keyword Documentation → ABAP - Reference → Processing Internal Data → Character String and Byte String Processing → Expressions and Functions for String Processing → Regular Expressions → Syntax of Regular Expressions
Character String Patterns
Character strings are represented by chainings or operators.
Other versions: 7.31 | 7.40 | 7.54
Chainings
Chainings are valid regular expressions that are written after each other. If r and s are regular expressions, the chaining rs matches all character strings that can be formed from the chaining of character strings that match r and s.
Examples
The following table shows some results from a test.
| Pattern | Text | match |
|---|---|---|
| H[aeu]llo | Hallo | X |
| H[aeu]llo | Hello | X |
| H[aeu]llo | Hullo | X |
| H[aeu]llo | Hollo | - |
H[aeu]llo is the chaining of five regular expressions for single characters.
Operators for Character Strings
These operators are made up of the special characters {, }, *, +, ?, |, (, ), and \. The special characters can be made into literal characters using the prefix \ or by enclosing with \Q ... \E.
Chaining Operators
The operators {n}, {n,m}, *, +, and ? (where n and m are natural numbers, including zero) can be written directly after a regular expression r, and thus create chainings rrr... of the regular expression:
- The regular expression r{n} is equivalent to an n-fold chaining of r. The regular expression r{0} matches an empty character string, and therefore also the offset before the first character of a character string, the spaces between the characters in character strings, and the offset after the last character in a character string.
- The regular expression r{n,m} is equivalent to at least n and a maximum of m chainings of r. The value of n must be less than or equal to the value of m. The expression r{n} is equivalent to at least an n-fold chaining of r.
- The regular expression r? is equivalent to r{0,1}, which means the expression r or the empty character string.
- The regular expression r* is equivalent to r{0,}, in other words a chaining of r of any length, including the empty character string. When using subgroups (see below), and in a text search, r* matches the longest possible substring (greedy behavior).
- The regular expression r+ is equivalent to r{1,}, in other words a chaining of any length of r excluding the empty character string. When using subgroups, and in a text search, r+ matches the longest possible substring (greedy behavior).
- The regular expressions r{n,m}?, r*? and r+? are reserved for future language enhancements (non-greedy behavior) and currently raise the exception CX_SY_INVALID_REGEX.
Note
For a regular expression with chaining operators, the first rule is that where possible the entire expression must match. This rule restricts the length of character strings that match the chaining with the operators *and + and therefore their greedy behavior.
Examples
The following table shows some results from a test.
| pattern | Text | match |
|---|---|---|
| Hel | Hello | X |
| H. | Hello | X |
| . | Hello | - |
| . | Hello | X |
| .+H.+e.+l.+l.+o.+ | Hello | - |
| xHxexlxlxox | Hello | X |
| l+ | ll | X |
Example
The first partial expression a+ is compared with the first five characters
"aaaaa" from text and the last character
"a" from text is reserved for the second partial expression a.
DATA TEXT type STRING.
DATA result_tab TYPE match_result_tab.
text = 'aaaaaa'.
FIND ALL OCCURRENCES OF REGEX '(a+)(a)'
IN text RESULTS result_tab.
Alternatives
The operator | can be written between two regular expressions r and s, and thereby creates one single regular expression r|s, that matches both r as well as s.
Note
Chainings and other operators form a stronger bond than |, in other words r|st and r|s+ are equivalent to r|(?:st) or r|(?:s+) but not to (?:r|s)t or (?:r|s)+.
Examples
The following table shows some results from a test.
| Pattern | Text | match |
|---|---|---|
| H(e | a | u)llo |
| H(e | a | u)llo |
| He | a | ullo |
| He | a | ullo |
Subgroups
The operators ( ... ) and (?: ... ) group chainings of regular expressions, forming a unit. This affects the scope of other operators such as * or |.Here, the regular expressions (r) and (?:r) correspond to the regular expression r. The difference between ( ... ) and (?: ... ) is that the operator ( ... ) saves the substrings found in registers and (?: ... ) does not.
Note
The greedy behavior of chaining operators mentioned above also applies to subgroups (from left to right). This does not break the rule that generally requires the entire regular expression to match.
Examples
The following table shows some results from a test.
| Pattern | Text | match |
|---|---|---|
| Tral+a | Tralala | - |
| Tr(al)+a | Tralala | X |
| Tr(?:al)+a | Tralala | X |
In the first expression, the chaining with the operator + affects the literal character l, and in the second and third expression it affects the subgroup al.
Subgroups with Registration
Besides its effect with respect to creating subgroups, the operator ( ... ) also saves the substrings (when matching the regular expression to a character string) in the correct order in tabs. The subgroups ( ... ) of the expression match these substrings. In this process, an operator \1, \2, \3, ... is assigned to each subgroup, which can be specified within the expression after its subgroup, and thus acts as a placeholder for the character string stored in the corresponding register. In text replacements, the special characters $1, $2, $3, ... can be used to access the last assignment to the register.
The number of subgroups and registers is only limited by the capacity of the platform.
Notes
The addition SUBMATCHES of the statements
FIND and REPLACE and
the identically named column of the results table filled using the addition RESULTS
can be used to access the content of all registers for a found location. The class CL_ABAP_MATCHER contains the method GET_SUBMATCH for this purpose.
If only grouping is required and no substrings need to be saved in registers, the operator (?: ... ) can be used instead of ( ... ). With respect to the formation of subgroups, both operators have the same effect. However (?: ... ) does not save anything in registers.
Examples
The following table shows some results from a test.
| Pattern | Text | match |
|---|---|---|
| (["']).+\1 | " Hello " | X |
| (["']).+\1 | "Hello' | - |
| (["']).+\1 | 'Hello' | X |
The chaining (["']).+\1 matches all text strings of which the first character is " or ' and the last character is the same as the first. In the two successful checks, the register is given the values " or '.
Example
The example demonstrates the greedy behavior of the operator + in subgroups and its relation to the primary rule that the entire regular expression must match where possible. The first subgroup takes up as many characters as possible "a". It is assigned the first four characters "aaaa". One character "a" remains for each of the other two subgroups.
DATA text TYPE string.
DATA result_tab TYPE match_result_tab.
text = 'aaaaaa'.
FIND ALL OCCURRENCES OF REGEX '(a+)(a+)(a+)'
IN text RESULTS result_tab.
Literal Characters
The operators \Q ... \E form a character string of literal characters from all enclosed characters. Special characters are ignored in this character string.
The following table shows some results from a test.
| pattern | Text | match |
|---|---|---|
| .+\w\d | Special: \w\d | - |
| .+\w\d | Special: \w\d | X |
| .+\Q\w\d\E | Special: \w\d | X |
Reserved Enhancements
The character string (? ... ) is generally reserved for later language enhancements. Apart from the operators already supported (?:... ), (?=...), (?!...), and (?> ... ), this string raises the exception CX_SY_INVALID_REGEX.