This class handles. More...

#include <regexp_facade.h>

Public Member Functions
bool	SetPattern (Item *pattern_expr, uint32_t flags)
	Sets the pattern if called for the first time or the pattern_expr is non-constant. More...

std::optional< bool >	Matches (Item *subject_expr, int start, int occurrence)
	Tries to match the subject against the compiled regular expression. More...

std::optional< int >	Find (Item *subject_expr, int start, int occurrence, bool after_match)
	Searches the subject for a match of the compiled regular expression and returns a position. More...

String *	Replace (Item subject_expr, Item replacement_expr, int start, int occurrence, String *result)

String *	Substr (Item subject_expr, int start, int occurrence, String result)

void	cleanup ()
	Delete the "engine" data structure after execution. More...

bool	EngineHasWarning () const
	Did any operation return a warning? For unit testing. More...

Private Member Functions
bool	Reset (Item *subject_expr, int start=1)
	Resets the compiled regular expression with a new string. More...

bool	SetupEngine (Item *pattern_expr, uint flags)
	Actually compiles the regular expression. More...

int	ConvertCodePointToLibPosition (int position) const
	Converts a string position in m_current_subject. More...

int	ConvertLibPositionToCodePoint (int position) const
	Converts a string position in m_current_subject. More...

String *	AssignResult (const char str, size_t length, String result)
	Helper function for setting the result from SQL regular expression functions that return a string value. More...

Private Attributes
unique_ptr_destroy_only< Regexp_engine >	m_engine
	Used for all the actual regular expression matching, search-and-replace, and positional and string information. More...

std::u16string	m_current_subject
	ICU does not copy the subject string, so we keep the subject buffer here. More...

Detailed Description

This class handles.

Conversion to the regexp library's character set, and buffers the converted strings during matching.
Re-compilation of the regular expression in case the pattern is a field reference or otherwise non-constant.
NULL handling.
Conversion between indexing conventions. Clients of this class can use one-based indexing, while the classes used by this class use zero-based indexing.

Member Function Documentation

◆ AssignResult()

String * regexp::Regexp_facade::AssignResult	(	const char *	str,
		size_t	length,
		String *	result
	)

private

Helper function for setting the result from SQL regular expression functions that return a string value.

Depending on character sets used by arguments and result, this function may copy, convert or just set the result. In particular, it handles the special case of the BINARY character set being interpreted as CP-1252.

Parameters

	str	The result string from the regexp function.
	length	Length in bytes.
[out]	result	The result string.

Returns: A pointer to the same string as the argument, or nullptr in case of failure.

◆ cleanup()

void regexp::Regexp_facade::cleanup ( )

inline

Delete the "engine" data structure after execution.

◆ ConvertCodePointToLibPosition()

int regexp::Regexp_facade::ConvertCodePointToLibPosition ( int position ) const

private

Converts a string position in m_current_subject.

Parameters

position One-based code point position.

Returns: Zero-based byte position.

◆ ConvertLibPositionToCodePoint()

int regexp::Regexp_facade::ConvertLibPositionToCodePoint ( int position ) const

private

Converts a string position in m_current_subject.

Parameters

position Zero-based UTF-16 position.

Returns: Zero-based code point position.

◆ EngineHasWarning()

bool regexp::Regexp_facade::EngineHasWarning ( ) const

inline

Did any operation return a warning? For unit testing.

◆ Find()

std::optional< int > regexp::Regexp_facade::Find	(	Item *	subject_expr,
		int	start,
		int	occurrence,
		bool	after_match
	)

Searches the subject for a match of the compiled regular expression and returns a position.

Parameters

subject_expr	The string to search.
start	Start position, 1-based.
occurrence	Which occurrence of the pattern should be searched for.
after_match	If true, the position following the end of the match is returned. If false, the position before the match is returned.

Returns: The first character of the match, or a null value if not found.

◆ Matches()

std::optional< bool > regexp::Regexp_facade::Matches	(	Item *	subject_expr,
		int	start,
		int	occurrence
	)

Tries to match the subject against the compiled regular expression.

Parameters

subject_expr	Is evaluated into a string to search.
start	Start position, 1-based.
occurrence	Which occurrence of the pattern should be searched for.

Return values

true	A match was found.
false	A match was not found.
nullptr	Either the engine was not compiled, or subject_expr evaluates to NULL. This is useful for the Item_func_regexp object, since it doesn't have to make a special case for when the regular expression is NULL. Instead, the case is handled here in the facade.

◆ Replace()

String * regexp::Regexp_facade::Replace	(	Item *	subject_expr,
		Item *	replacement_expr,
		int	start,
		int	occurrence,
		String *	result
	)

Parameters

	subject_expr	The string to search.
	replacement_expr	The string to replace the match with.
	start	Start position, 1-based.
	occurrence	Which occurrence of the pattern should be searched for.
[in,out]	result	Holds the buffer for writing the result.

◆ Reset()

bool regexp::Regexp_facade::Reset	(	Item *	subject_expr,
		int	start = `1`
	)

private

Resets the compiled regular expression with a new string.

Parameters

subject_expr	The new string to search.
start	If present, start on this code point.

Return values

false	OK.
true	Either there is no compiled regular expression, or the expression evaluated to `NULL`.

◆ SetPattern()

bool regexp::Regexp_facade::SetPattern	(	Item *	pattern_expr,
		uint32_t	flags
	)

Sets the pattern if called for the first time or the pattern_expr is non-constant.

This function is meant to be called for every row in a command such as

SELECT regexp_like( column, 'a+' ) FROM table;

In this case, the client of this class may call SetPattern() for every row without paying any penalty, as this becomes a no-op for all consecutive calls. In cases such as

SELECT regexp_like( column, regexp_column ) FROM table;

The regexp_column expression is non-constant and hence we have to recompile the regular expression for each row.

◆ SetupEngine()

bool regexp::Regexp_facade::SetupEngine	(	Item *	pattern_expr,
		uint	flags
	)

private

Actually compiles the regular expression.

◆ Substr()

String * regexp::Regexp_facade::Substr	(	Item *	subject_expr,
		int	start,
		int	occurrence,
		String *	result
	)

Member Data Documentation

◆ m_current_subject

std::u16string regexp::Regexp_facade::m_current_subject

private

ICU does not copy the subject string, so we keep the subject buffer here.

A call to Reset() causes it to be overwritten.

See also: Regexp_engine::reset()

◆ m_engine

unique_ptr_destroy_only<Regexp_engine> regexp::Regexp_facade::m_engine

private

Used for all the actual regular expression matching, search-and-replace, and positional and string information.

If either the regular expression pattern or the subject is NULL, this pointer is empty.

The documentation for this class was generated from the following files:

sql/regexp/regexp_facade.h
sql/regexp/regexp_facade.cc

Public Member Functions

Private Member Functions

Private Attributes

Detailed Description

Member Function Documentation

◆ AssignResult()

◆ cleanup()

◆ ConvertCodePointToLibPosition()

◆ ConvertLibPositionToCodePoint()

◆ EngineHasWarning()

◆ Find()

◆ Matches()

◆ Replace()

◆ Reset()

◆ SetPattern()

◆ SetupEngine()

◆ Substr()

Member Data Documentation

◆ m_current_subject

◆ m_engine