UTF8 String tools class. More...

#include <minitscript/utilities/UTF8StringTools.h>

Collaboration diagram for UTF8StringTools:

Static Public Member Functions
static const bool	startsWith (const string &str, const string &prefix)
	Checks if string starts with prefix. More...

static const bool	viewStartsWith (const string_view &str, const string_view &prefix)
	Checks if string starts with prefix. More...

static const bool	endsWith (const string &str, const string &suffix)
	Checks if string ends with suffix. More...

static const bool	viewEndsWith (const string_view &str, const string_view &suffix)
	Checks if string ends with suffix. More...

static const string	replace (const string &str, const string &what, const string &by, int64_t beginIndex=0, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Replace string with another string. More...

static int64_t	indexOf (const string &str, const string &what, int64_t beginIndex=0, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Finds first index of given string. More...

static int64_t	firstIndexOf (const string &str, const string &what, int64_t beginIndex=0, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Finds first index of given string. More...

static int64_t	lastIndexOf (const string &str, const string &what, int64_t endIndex=string::npos, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Finds last index of given string. More...

static int64_t	firstIndexOfChars (const string &str, const string &what, int64_t beginIndex=0, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache srcCache=nullptr, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache whatCache=nullptr)
	Finds first index of characters provided within given string. More...

static int64_t	lastIndexOfChars (const string &str, const string &what, int64_t endIndex=string::npos, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache srcCache=nullptr, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache whatCache=nullptr)
	Finds last index of characters provided within given string. More...

static const string	substring (const string &str, int64_t beginIndex, int64_t endIndex=string::npos, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache srcCache=nullptr, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache cache=nullptr)
	Returns substring of given string from begin index to end index. More...

static const string_view	viewSubstring (const string_view &str, int64_t beginIndex, int64_t endIndex, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Returns substring of given string from begin index to end index. More...

static bool	equalsIgnoreCase (const string &string1, const string &string2, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache string1Cache=nullptr, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache string2Cache=nullptr)
	Checks if strings equal ignoring case. More...

static const string	trim (const string &str, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Trim string. More...

static const string_view	viewTrim (const string_view &str, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Trim string. More...

static const string	toLowerCase (const string &str, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Transform string to lower case. More...

static const string	toUpperCase (const string &str, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Transform string to upper case. More...

static bool	regexMatch (const string &str, const string &pattern, smatch *matches=nullptr)
	Check if pattern matches whole string. More...

static bool	regexSearch (const string &str, const string &pattern, smatch *matches=nullptr)
	Do regex pattern search. More...

static const string	regexReplace (const string &str, const string &pattern, const string &by)
	Replace regex pattern with given string. More...

static const vector< string >	tokenize (const string &str, const string &delimiters, bool emptyTokens=false)
	Tokenize. More...

static const string	padLeft (const string &str, const string &by, int64_t toLength, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Pad a string left. More...

static const string	padRight (const string &str, const string &by, int64_t toLength, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Pad a string right. More...

static const string	indent (const string &str, const string &with, int64_t count)
	Indent a string. More...

static const string	generate (const string &what, int64_t count=1)
	Generate a string. More...

static int64_t	getLength (const string &str, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Get UTF8 string length. More...

static const string	getCharAt (const string &str, int64_t index, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Get UTF8 character at given index. More...

static uint32_t	getCodePointAt (const string &str, int64_t index, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Get UTF8 code point at given index. More...

static int64_t	getUTF8BinaryIndex (const string &str, int64_t charIdx, ::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *cache=nullptr)
	Get UTF8 binary buffer index. More...

Detailed Description

UTF8 String tools class.

Author: Andreas Drewke

Definition at line 21 of file UTF8StringTools.h.

Member Function Documentation

◆ startsWith()

static const bool startsWith	(	const string &	str,
		const string &	prefix
	)

inlinestatic

Checks if string starts with prefix.

Parameters

str	string
prefix	prefix string

Returns: if string starts with prefix

Definition at line 30 of file UTF8StringTools.h.

◆ viewStartsWith()

static const bool viewStartsWith	(	const string_view &	str,
		const string_view &	prefix
	)

inlinestatic

Checks if string starts with prefix.

Parameters

str	string
prefix	prefix string

Returns: if string starts with prefix

Definition at line 40 of file UTF8StringTools.h.

◆ endsWith()

static const bool endsWith	(	const string &	str,
		const string &	suffix
	)

inlinestatic

Checks if string ends with suffix.

Parameters

str	string
suffix	suffix string

Returns: if string ends with suffix

Definition at line 50 of file UTF8StringTools.h.

◆ viewEndsWith()

static const bool viewEndsWith	(	const string_view &	str,
		const string_view &	suffix
	)

inlinestatic

Checks if string ends with suffix.

Parameters

str	string
suffix	suffix string

Returns: if string ends with suffix

Definition at line 62 of file UTF8StringTools.h.

◆ replace()

const string replace	(	const string &	str,
		const string &	what,
		const string &	by,
		int64_t	beginIndex = `0`,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

static

Replace string with another string.

Parameters

str	string
what	what to replace
by	to replace by
beginIndex	index to begin with
cache	str UTF8 position cache

Returns: replace result

Definition at line 26 of file UTF8StringTools.cpp.

◆ indexOf()

static int64_t indexOf	(	const string &	str,
		const string &	what,
		int64_t	beginIndex = `0`,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

inlinestatic

Finds first index of given string.

Parameters

str	string
what	what
beginIndex	begin index
cache	str UTF8 position cache

Returns: index or string::npos if not found

Definition at line 93 of file UTF8StringTools.h.

◆ firstIndexOf()

int64_t firstIndexOf	(	const string &	str,
		const string &	what,
		int64_t	beginIndex = `0`,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

static

Finds first index of given string.

Parameters

str	string
what	what
beginIndex	begin index
cache	str UTF8 position cache

Returns: index or string::npos if not found

Definition at line 44 of file UTF8StringTools.cpp.

◆ lastIndexOf()

int64_t lastIndexOf	(	const string &	str,
		const string &	what,
		int64_t	endIndex = `string::npos`,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

static

Finds last index of given string.

Parameters

str	string
what	what
endIndex	end index or string::npos
cache	str UTF8 position cache

Returns: index or string::npos if not found

Definition at line 63 of file UTF8StringTools.cpp.

◆ firstIndexOfChars()

int64_t firstIndexOfChars	(	const string &	str,
		const string &	what,
		int64_t	beginIndex = `0`,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	srcCache = `nullptr`,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	whatCache = `nullptr`
	)

static

Finds first index of characters provided within given string.

Parameters

str	string
what	what
beginIndex	begin index
srcCache	str UTF8 position cache
whatCache	what UTF8 position cache

Returns: index or string::npos if not found

Definition at line 93 of file UTF8StringTools.cpp.

◆ lastIndexOfChars()

int64_t lastIndexOfChars	(	const string &	str,
		const string &	what,
		int64_t	endIndex = `string::npos`,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	srcCache = `nullptr`,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	whatCache = `nullptr`
	)

static

Finds last index of characters provided within given string.

Parameters

str	string
what	what
endIndex	end index or string::npos
srcCache	str UTF8 position cache
whatCache	what UTF8 position cache

Returns: index or string::npos if not found

Definition at line 111 of file UTF8StringTools.cpp.

◆ substring()

static const string substring	(	const string &	str,
		int64_t	beginIndex,
		int64_t	endIndex = `string::npos`,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	srcCache = `nullptr`,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

inlinestatic

Returns substring of given string from begin index to end index.

Parameters

str	string
beginIndex	begin index
endIndex	end index or string::npos
cache	str UTF8 position cache

Returns: substring result

Definition at line 162 of file UTF8StringTools.h.

◆ viewSubstring()

const string_view viewSubstring	(	const string_view &	str,
		int64_t	beginIndex,
		int64_t	endIndex,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

static

Returns substring of given string from begin index to end index.

Parameters

str	string
beginIndex	begin index
endIndex	end index or string::npos
cache	str UTF8 position cache

Returns: substring result

Definition at line 140 of file UTF8StringTools.cpp.

◆ equalsIgnoreCase()

bool equalsIgnoreCase	(	const string &	string1,
		const string &	string2,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	string1Cache = `nullptr`,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	string2Cache = `nullptr`
	)

static

Checks if strings equal ignoring case.

Parameters

string1	string 1
string2	string 2
string1Cache	string1 UTF8 position cache
string2Cache	string2 UTF8 position cache

Returns: equality

Definition at line 156 of file UTF8StringTools.cpp.

◆ trim()

const string trim	(	const string &	str,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

static

Trim string.

Parameters

str	string
cache	UTF8 position cache

Returns: trimmed string

Definition at line 180 of file UTF8StringTools.cpp.

◆ viewTrim()

const string_view viewTrim	(	const string_view &	str,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

static

Trim string.

Parameters

str	string
cache	UTF8 position cache

Returns: trimmed string

Definition at line 185 of file UTF8StringTools.cpp.

◆ toLowerCase()

const string toLowerCase	(	const string &	str,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

static

Transform string to lower case.

Parameters

str	string
cache	UTF8 position cache

Returns: lowercase string

Definition at line 204 of file UTF8StringTools.cpp.

◆ toUpperCase()

const string toUpperCase	(	const string &	str,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

static

Transform string to upper case.

Parameters

str	string
cache	UTF8 position cache

Returns: uppercase string

Definition at line 214 of file UTF8StringTools.cpp.

◆ regexMatch()

bool regexMatch	(	const string &	str,
		const string &	pattern,
		smatch *	matches = `nullptr`
	)

static

Check if pattern matches whole string.

Parameters

str	string
pattern	pattern
matches	matches

Returns: if pattern matches whole string

Definition at line 224 of file UTF8StringTools.cpp.

◆ regexSearch()

bool regexSearch	(	const string &	str,
		const string &	pattern,
		smatch *	matches = `nullptr`
	)

static

Do regex pattern search.

Parameters

str	string
pattern	pattern
matches	matches

Returns: if search was successful

Definition at line 232 of file UTF8StringTools.cpp.

◆ regexReplace()

const string regexReplace	(	const string &	str,
		const string &	pattern,
		const string &	by
	)

static

Replace regex pattern with given string.

Parameters

str	string
pattern	pattern
by	replace string

Returns: replace result

Definition at line 240 of file UTF8StringTools.cpp.

◆ tokenize()

const vector< string > tokenize	(	const string &	str,
		const string &	delimiters,
		bool	emptyTokens = `false`
	)

static

Tokenize.

Parameters

str	string
delimiters	delimiters
emptyTokens	include empty tokens

Returns: tokens

Definition at line 244 of file UTF8StringTools.cpp.

◆ padLeft()

const string padLeft	(	const string &	str,
		const string &	by,
		int64_t	toLength,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

static

Pad a string left.

Parameters

str	string
by	by
toLength	to length
cache	str UTF8 position cache

Returns: padded string

Definition at line 250 of file UTF8StringTools.cpp.

◆ padRight()

const string padRight	(	const string &	str,
		const string &	by,
		int64_t	toLength,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

static

Pad a string right.

Parameters

str	string
by	by
toLength	to length
cache	str UTF8 position cache

Returns: padded string

Definition at line 256 of file UTF8StringTools.cpp.

◆ indent()

static const string indent	(	const string &	str,
		const string &	with,
		int64_t	count
	)

inlinestatic

Indent a string.

Parameters

str	string
with	with
count	count

Returns: resulting string

Definition at line 287 of file UTF8StringTools.h.

◆ generate()

static const string generate	(	const string &	what,
		int64_t	count = `1`
	)

inlinestatic

Generate a string.

Parameters

what	what
count	count

Returns: resulting string

Definition at line 299 of file UTF8StringTools.h.

◆ getLength()

int64_t getLength	(	const string &	str,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

static

Get UTF8 string length.

Parameters

str	string
cache	UTF8 position cache

Returns: UTF8 string length

Definition at line 264 of file UTF8StringTools.cpp.

◆ getCharAt()

const string getCharAt	(	const string &	str,
		int64_t	index,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

static

Get UTF8 character at given index.

Parameters

str	string
index	index
cache	UTF8 position cache

Returns: character as string

Definition at line 270 of file UTF8StringTools.cpp.

◆ getCodePointAt()

uint32_t getCodePointAt	(	const string &	str,
		int64_t	index,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

static

Get UTF8 code point at given index.

Parameters

str	string
index	index
cache	UTF8 position cache

Returns: character as codepoint

Definition at line 278 of file UTF8StringTools.cpp.

◆ getUTF8BinaryIndex()

int64_t getUTF8BinaryIndex	(	const string &	str,
		int64_t	charIdx,
		::minitscript::utilities::UTF8CharacterIterator::UTF8PositionCache *	cache = `nullptr`
	)

static

Get UTF8 binary buffer index.

Parameters

str	string
charIdx	character index
cache	UTF8 position cache

Returns: UTF binary buffer position from given character/code point index

Definition at line 286 of file UTF8StringTools.cpp.

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

/home/andreas/Development/drewke.net/minitscript/src/minitscript/utilities/UTF8StringTools.h
/home/andreas/Development/drewke.net/minitscript/src/minitscript/utilities/UTF8StringTools.cpp

Static Public Member Functions

Detailed Description

Member Function Documentation

◆ startsWith()

◆ viewStartsWith()

◆ endsWith()

◆ viewEndsWith()

◆ replace()

◆ indexOf()

◆ firstIndexOf()

◆ lastIndexOf()

◆ firstIndexOfChars()

◆ lastIndexOfChars()

◆ substring()

◆ viewSubstring()

◆ equalsIgnoreCase()

◆ trim()

◆ viewTrim()

◆ toLowerCase()

◆ toUpperCase()

◆ regexMatch()

◆ regexSearch()

◆ regexReplace()

◆ tokenize()

◆ padLeft()

◆ padRight()

◆ indent()

◆ generate()

◆ getLength()

◆ getCharAt()

◆ getCodePointAt()

◆ getUTF8BinaryIndex()