ut_string.cpp File Reference

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <algorithm>
#include <math.h>
#include <ctype.h>
#include "ut_types.h"
#include "ut_misc.h"
#include "ut_assert.h"
#include "ut_string.h"
#include "ut_debugmsg.h"
#include "ut_growbuf.h"
#include <fribidi.h>
#include "ut_mbtowc.h"
#include "ut_wctomb.h"
#include "ut_string_class.h"
#include "xap_EncodingManager.h"
#include "ut_case.h"

Defines
#define	UT_STRING_CPP
Functions
bool	UT_XML_cloneNoAmpersands (gchar *&rszDest, const gchar *szSource)
bool	UT_XML_cloneConvAmpersands (gchar *&rszDest, const gchar *szSource)
const gchar *	UT_XML_transNoAmpersands (const gchar *szSource)
bool	UT_ensureValidXML (std::string &str)
	XML cannot contain any control characters except , , , see bug 8565 (http://www.w3.org/TR/REC-xml/#charsets).
bool	UT_isValidXML (const char *pString)
bool	UT_validXML (char *pString)
void	UT_decodeUTF8string (const gchar *pString, UT_uint32 len, UT_GrowBuf *pResult)
UT_uint32	UT_UCS2_strlen (const UT_UCS2Char *string)
bool	UT_isSmartQuotableCharacter (UT_UCSChar c)
bool	UT_isSmartQuotedCharacter (UT_UCSChar c)
bool	UT_UCS4_isupper (UT_UCS4Char c)
bool	UT_UCS4_islower (UT_UCS4Char c)
bool	UT_UCS4_isspace (UT_UCS4Char c)
bool	UT_UCS4_isalpha (UT_UCS4Char c)
bool	UT_UCS4_isSentenceSeparator (UT_UCS4Char c)
bool	UT_UCS4_isdigit (UT_UCS4Char c)
UT_UCS4Char *	UT_UCS4_strncpy (UT_UCS4Char *dest, const UT_UCS4Char *src, UT_uint32 n)
UT_UCS4Char *	UT_UCS4_strnrev (UT_UCS4Char *src, UT_uint32 n)
UT_UCS4Char *	UT_UCS4_strstr (const UT_UCS4Char *phaystack, const UT_UCS4Char *pneedle)
UT_sint32	UT_UCS4_strcmp (const UT_UCS4Char *left, const UT_UCS4Char *right)
UT_UCS4Char	UT_UCS4_toupper (UT_UCS4Char c)
	Convert a given character to uppercase.
UT_UCS4Char	UT_UCS4_tolower (UT_UCS4Char c)
UT_UCS4Char *	UT_UCS4_stristr (const UT_UCS4Char *phaystack, const UT_UCS4Char *pneedle)
UT_uint32	UT_UCS4_strlen (const UT_UCS4Char *string)
UT_uint32	UT_UCS4_strlen_as_char (const UT_UCS4Char *string)
UT_UCS4Char *	UT_UCS4_strcpy (UT_UCS4Char *dest, const UT_UCS4Char *src)
UT_UCS4Char *	UT_UCS4_strcpy_char (UT_UCS4Char *dest, const char *src)
UT_UCS4Char *	UT_UCS4_strncpy_char (UT_UCS4Char *dest, const char *src, int n)
UT_UCS4Char *	UT_UCS4_strcpy_utf8_char (UT_UCS4Char *dest, const char *src)
char *	UT_UCS4_strcpy_to_char (char *dest, const UT_UCS4Char *src)
char *	UT_UCS4_strncpy_to_char (char *dest, const UT_UCS4Char *src, int n)
bool	UT_UCS4_cloneString (UT_UCS4Char **dest, const UT_UCS4Char *src)
bool	UT_UCS4_cloneString_char (UT_UCS4Char **dest, const char *src)
static const char *	s_pass_name (const char *&csstr, char end)
static const char *	s_pass_value (const char *&csstr)
static const char *	s_pass_string (const char *&csstr_ptr)
static void	s_pass_whitespace (const char *&csstr)
void	UT_parse_attributes (const char *attributes, std::map< std::string, std::string > &map)
void	UT_parse_properties (const char *properties, std::map< std::string, std::string > &map)
const char *	std_size_string (float f)
UT_BidiCharType	UT_bidiGetCharType (UT_UCS4Char c)
bool	UT_bidiReorderString (const UT_UCS4Char *pStrIn, UT_uint32 len, UT_BidiCharType baseDir, UT_UCS4Char *pStrOut)
bool	UT_bidiMapLog2Vis (const UT_UCS4Char *pStrIn, UT_uint32 len, UT_BidiCharType baseDir, UT_uint32 *pL2V, UT_uint32 *pV2L, UT_Byte *pEmbed)
bool	UT_bidiGetMirrorChar (UT_UCS4Char c, UT_UCS4Char &mc)

---

Define Documentation

#define UT_STRING_CPP

---

Function Documentation

static const char* s_pass_name	(	const char *&	csstr,
		char	end
	)			[static]

References UT_UTF8Stringbuf::charCode(), and UT_UCS4_isspace().

Referenced by UT_parse_attributes(), and UT_parse_properties().

static const char* s_pass_string

(

const char *&

csstr_ptr

)

[static]

Referenced by UT_parse_attributes().

static const char* s_pass_value

(

const char *&

csstr

)

[static]

References UT_UTF8Stringbuf::charCode(), and UT_UCS4_isspace().

Referenced by UT_parse_properties().

static void s_pass_whitespace

(

const char *&

csstr

)

[static]

References UT_UTF8Stringbuf::charCode(), and UT_UCS4_isspace().

Referenced by UT_parse_attributes(), and UT_parse_properties().

const char* std_size_string

(

float

f

)

References int().

Referenced by IE_Imp_RTF::ApplyParagraphAttributes(), IE_Imp_RTF::buildAllProps(), IE_Imp_RTF::buildCharacterProps(), Defun_EV_GetToolbarItemState_Fn(), XAP_UnixDialog_FontChooser::runModal(), XAP_CocoaDialog_FontChooserController::selectSize:, and streamRead().

UT_BidiCharType UT_bidiGetCharType

(

UT_UCS4Char

c

)

bool UT_bidiGetMirrorChar	(	UT_UCS4Char	c,
		UT_UCS4Char &	mc
	)

bool UT_bidiMapLog2Vis	(	const UT_UCS4Char *	pStrIn,
		UT_uint32	len,
		UT_BidiCharType	baseDir,
		UT_uint32 *	pL2V,
		UT_uint32 *	pV2L,
		UT_Byte *	pEmbed
	)

References UT_ASSERT_HARMLESS, and UT_return_val_if_fail.

bool UT_bidiReorderString	(	const UT_UCS4Char *	pStrIn,
		UT_uint32	len,
		UT_BidiCharType	baseDir,
		UT_UCS4Char *	pStrOut
	)

pStrOut needs to contain space for len characters + terminating 0

References UT_ASSERT_HARMLESS, UT_return_val_if_fail, and UT_UCS4_strncpy().

void UT_decodeUTF8string	(	const gchar *	pString,
		UT_uint32	len,
		UT_GrowBuf *	pResult
	)

References UT_GrowBuf::append(), gchar, and UT_ASSERT.

Referenced by XAP_Prefs::savePrefsFile(), XAP_DiskStringSet::setValue(), and AP_DiskStringSet::setValue().

bool UT_ensureValidXML

(

std::string &

str

)

XML cannot contain any control characters except ,
, , see bug 8565 (http://www.w3.org/TR/REC-xml/#charsets).

Ensure a key or property is valid XML.

This function removes any illegal characters and invalid utf-8 sequences.

Parameters:

str

the string to modify in place

Return values:

%true

if the string was valid before. false if it needed changes.

References gchar, s, UT_ASSERT_HARMLESS, UT_NOT_IMPLEMENTED, and xxx_UT_DEBUGMSG.

Referenced by PP_AttrProp::setAttribute(), PP_AttrProp::setProperty(), and TFTEST_MAIN().

bool UT_isSmartQuotableCharacter

(

UT_UCSChar

c

)

Referenced by FL_DocLayout::considerSmartQuoteCandidateAt(), and fl_BlockLayout::doclistener_insertSpan().

bool UT_isSmartQuotedCharacter

(

UT_UCSChar

c

)

References UCS_LDBLQUOTE, UCS_LQUOTE, UCS_RDBLQUOTE, and UCS_RQUOTE.

Referenced by fp_TextRun::isOneItem().

bool UT_isValidXML

(

const char *

s

)

Parameters:

s

The string of characters which is to be checked for XML-validity.

Return values:

TRUE

if the characters are all valid for XML, FALSE if any one of them is not.

NB: this function also checks that the string is valid utf-8

References s.

Referenced by TFTEST_MAIN().

void UT_parse_attributes	(	const char *	attributes,
		std::map< std::string, std::string > &	map
	)

References s_pass_name(), s_pass_string(), and s_pass_whitespace().

void UT_parse_properties	(	const char *	properties,
		std::map< std::string, std::string > &	map
	)

References s_pass_name(), s_pass_value(), and s_pass_whitespace().

Referenced by AP_Convert::print(), IE_Imp::setProps(), and IE_Exp::setProps().

UT_uint32 UT_UCS2_strlen

(

const UT_UCS2Char *

string

)

Referenced by IE_Imp_MsWord_97::_getBookmarkName().

bool UT_UCS4_cloneString	(	UT_UCS4Char **	dest,
		const UT_UCS4Char *	src
	)

References UT_calloc(), and UT_UCS4_strlen().

Referenced by AP_UnixDialog_Spell::_convertFromMB(), AP_Dialog_Paragraph::_createPreviewFromGC(), AP_Dialog_Replace::_manageList(), FV_View::findGetFindString(), FV_View::findGetReplaceString(), FV_View::findSetFindString(), FV_View::findSetReplaceString(), XAP_Dialog_FontChooser::setDrawString(), XAP_Preview_Zoom::setString(), AP_Preview_Paragraph_Block::setText(), XAP_Dictionary::suggestWord(), BarbarismChecker::suggestWord(), and utf8_to_utf32().

bool UT_UCS4_cloneString_char	(	UT_UCS4Char **	dest,
		const char *	src
	)

References UT_calloc(), and UT_UCS4_strcpy_char().

Referenced by fp_Line::_calculateWidthOfRun(), fp_ForcedPageBreakRun::_draw(), fp_ForcedColumnBreakRun::_draw(), AP_Preview_PageNumbers::AP_Preview_PageNumbers(), XAP_Dialog_FontChooser::event_previewExposed(), ODe_Style_Style::ParagraphProps::fetchAttributesFromAbiProps(), FV_View::findGetFindString(), FV_View::findGetReplaceString(), AP_Dialog_Replace::getFindString(), AP_Dialog_Replace::getReplaceString(), s_doInsertDateTime(), s_getUCSText(), XAP_Dialog_FontChooser::setDrawString(), XAP_Preview_Zoom::setString(), XAP_FontPreview::setText(), XAP_CocoaDialog_FontChooser::updatePreview, XAP_Dialog_FontChooser::XAP_Dialog_FontChooser(), and XAP_FontPreview::XAP_FontPreview().

bool UT_UCS4_isalpha

(

UT_UCS4Char

c

)

References UT_BIDI_IS_LETTER, and UT_bidiGetCharType().

Referenced by FV_View::toggleCase(), UT_isWordDelimiter(), and whatKindOfChar().

bool UT_UCS4_isdigit

(

UT_UCS4Char

c

)

References digits_table, and s_cmp_digits().

Referenced by AP_Win32Dialog_Goto::_onCommand(), IE_Exp_RTF::_write_rtf_header(), and RTF_msword97_level::ParseLevelText().

bool UT_UCS4_islower

(

UT_UCS4Char

c

)

References case_table, s_cmp_case(), and case_entry::type.

Referenced by _toggleToggle(), BarbarismChecker::suggestWord(), and FV_View::toggleCase().

bool UT_UCS4_isSentenceSeparator

(

UT_UCS4Char

c

)

Referenced by fl_BlockLayout::isSentenceSeparator(), and FV_View::toggleCase().

bool UT_UCS4_isspace

(

UT_UCS4Char

c

)

References whitespace_table.

Referenced by FV_View::_isSpaceBefore(), _toggleFirstCapital(), AbiGoogle_invoke(), fp_TextRun::canMergeWithNext(), IE_Imp_Applix::s_getTagName(), s_pass_name(), s_pass_value(), s_pass_whitespace(), FV_View::toggleCase(), UT_UCS4String::UT_UCS4String(), UT_UTF8_Base64Decode(), and whatKindOfChar().

bool UT_UCS4_isupper

(

UT_UCS4Char

c

)

References case_table, s_cmp_case(), and case_entry::type.

Referenced by BarbarismChecker::suggestWord().

UT_sint32 UT_UCS4_strcmp	(	const UT_UCS4Char *	left,
		const UT_UCS4Char *	right
	)

References UT_ASSERT.

Referenced by AP_Dialog_Replace::_manageList(), fp_FieldRun::_setValue(), operator<(), operator==(), IE_Imp_RTF::ReadRevisionTable(), and AP_Dialog_Replace::setFindString().

UT_UCS4Char* UT_UCS4_strcpy	(	UT_UCS4Char *	dest,
		const UT_UCS4Char *	src
	)

References UT_ASSERT.

Referenced by fp_FieldRun::_setValue(), AP_Dialog_Spell::addChangeAll(), and UT_UCS4_strcpy_utf8_char().

UT_UCS4Char* UT_UCS4_strcpy_char	(	UT_UCS4Char *	dest,
		const char *	src
	)

References XAP_EncodingManager::get_instance(), UT_UCS4_mbtowc::mbtowc(), and UT_ASSERT.

Referenced by AP_TopRuler::_drawTickMark(), AP_Dialog_Styles::_populateAbiPreview(), XAP_Win32Dialog_Insert_Symbol::_setFontFromCombo(), XAP_Dictionary::addWord(), fp_FieldShortFileNameRun::calculateValue(), fp_FieldFileNameRun::calculateValue(), fp_FieldDateRun::calculateValue(), fp_FieldTimeRun::calculateValue(), fp_FieldEndnoteRefRun::calculateValue(), fp_FieldEndnoteAnchorRun::calculateValue(), fp_FieldFootnoteAnchorRun::calculateValue(), fp_FieldFootnoteRefRun::calculateValue(), fp_FieldTimeZoneRun::calculateValue(), fp_FieldDateTimeCustomRun::calculateValue(), fp_FieldAMPMRun::calculateValue(), fp_FieldMilTimeRun::calculateValue(), fp_FieldDOYRun::calculateValue(), fp_FieldWkdayRun::calculateValue(), fp_FieldDefaultDateNoTimeRun::calculateValue(), fp_FieldDefaultDateRun::calculateValue(), fp_FieldMthDayYearRun::calculateValue(), fp_FieldMonthDayYearRun::calculateValue(), fp_FieldDDMMYYRun::calculateValue(), fp_FieldMMDDYYRun::calculateValue(), AP_LeftRuler::drawLU(), AbiCommand::insertText(), UT_UCS4String::operator+=(), AbiCommand::parseTokens(), AbiCommand::replaceAll(), AbiCommand::replaceNext(), AP_Win32App::s_fromWinLocaleToUCS4(), fd_Field::update(), and UT_UCS4_cloneString_char().

char* UT_UCS4_strcpy_to_char	(	char *	dest,
		const UT_UCS4Char *	src
	)

References XAP_EncodingManager::get_instance(), UT_ASSERT, and UT_Wctomb::wctomb_or_fallback().

Referenced by AP_CocoaDialog_Replace::_populateWindowData, XAP_Dictionary::addWord(), AP_Dialog_MarkRevisions::getComment1(), AP_Dialog_ListRevisions::getNthItemText(), and AP_Dialog_InsertHyperlink::setDoc().

UT_UCS4Char* UT_UCS4_strcpy_utf8_char	(	UT_UCS4Char *	dest,
		const char *	src
	)

References UT_UCS4String::ucs4_str(), UT_ASSERT, and UT_UCS4_strcpy().

Referenced by EV_Win32Toolbar::_ComboWndProc(), AP_Dialog_Styles::_createCharPreviewFromGC(), AP_Dialog_Styles::_populateAbiPreview(), and AP_Win32Dialog_Goto::GoTo().

UT_UCS4Char* UT_UCS4_stristr	(	const UT_UCS4Char *	phaystack,
		const UT_UCS4Char *	pneedle
	)

References UT_UCS4_tolower().

UT_uint32 UT_UCS4_strlen

(

const UT_UCS4Char *

string

)

Referenced by AP_Win32Dialog_Spell::_change(), AP_Win32Dialog_Spell::_changeAll(), EV_Win32Toolbar::_ComboWndProc(), FV_View::_computeFindPrefix(), fp_FieldRun::_defaultDraw(), fp_EndOfParagraphRun::_draw(), fp_ForcedLineBreakRun::_draw(), fp_Run::_drawTextLine(), FV_View::_findNext(), FV_View::_findPrev(), FV_View::_findReplace(), FV_View::_findReplaceReverse(), s_AbiWord_1_Listener::_handleRevisions(), fp_EndOfParagraphRun::_lookupProperties(), fp_ForcedLineBreakRun::_lookupProperties(), AP_Win32Dialog_Replace::_onCommand(), AP_Dialog_Styles::_populateAbiPreview(), AP_CocoaDialog_Replace::_populateWindowData, fp_FieldRun::_recalcWidth(), fp_FieldRun::_setValue(), AP_Win32Dialog_Spell::_showMisspelledWord(), _ucs4ToLatin1(), _ucsToAscii(), _ucsToUTF8(), s_RTF_ListenerWriteDoc::_writeFieldTrailer(), AP_Dialog_Spell::addChangeAll(), AP_Dialog_MarkRevisions::addRevision(), AiksaurusABI_ucsToAscii(), fp_FieldListLabelRun::calculateValue(), AP_Dialog_Spell::changeWordWith(), XAP_Dictionary::countCommonChars(), XAP_Preview_Zoom::draw(), XAP_Preview_FontPreview::draw(), AP_Preview_PageNumbers::draw(), AP_Lists_preview::draw(), XAP_Dialog_FontChooser::event_previewExposed(), UT_Win32LocaleString::fromUCS4(), AP_Dialog_MarkRevisions::getComment1(), AP_Dialog_Lists::getListLabel(), AP_Dialog_ListRevisions::getNthItemText(), FV_View::gotoTarget(), AP_UnixDialog_Spell::onChangeAllClicked(), AP_UnixDialog_Spell::onChangeClicked(), UT_UCS4String::operator+=(), UT_UCS4String::operator=(), AP_Dialog_Lists::PopulateDialogData(), s_doInsertDateTime(), XAP_Dictionary::save(), XAP_Dialog_FontChooser::setDrawString(), AP_Preview_Paragraph_Block::setText(), BarbarismChecker::suggestExactWord(), XAP_Dictionary::suggestWord(), fd_Field::update(), UT_UCS4_cloneString(), and IE_Exp_OpenXML::writeText().

UT_uint32 UT_UCS4_strlen_as_char

(

const UT_UCS4Char *

string

)

References XAP_EncodingManager::get_instance(), and UT_Wctomb::wctomb_or_fallback().

Referenced by AP_Dialog_InsertHyperlink::setDoc().

UT_UCS4Char* UT_UCS4_strncpy	(	UT_UCS4Char *	dest,
		const UT_UCS4Char *	src,
		UT_uint32	n
	)

References UT_ASSERT.

Referenced by AD_Document::addRevision(), GR_XPRenderInfo::append(), PD_Document::changeDocPropeties(), GR_XPRenderInfo::cut(), GR_XPRenderInfo::split(), and UT_bidiReorderString().

UT_UCS4Char* UT_UCS4_strncpy_char	(	UT_UCS4Char *	dest,
		const char *	src,
		int	n
	)

References XAP_EncodingManager::get_instance(), UT_UCS4_mbtowc::mbtowc(), and UT_ASSERT.

char* UT_UCS4_strncpy_to_char	(	char *	dest,
		const UT_UCS4Char *	src,
		int	n
	)

References XAP_EncodingManager::get_instance(), UT_ASSERT, and UT_Wctomb::wctomb_or_fallback().

Referenced by AP_Dialog_Spell::addChangeAll(), and AP_Dialog_Spell::inChangeAll().

UT_UCS4Char* UT_UCS4_strnrev	(	UT_UCS4Char *	src,
		UT_uint32	n
	)

Referenced by fp_TextRun::_refreshDrawBuffer().

UT_UCS4Char* UT_UCS4_strstr	(	const UT_UCS4Char *	phaystack,
		const UT_UCS4Char *	pneedle
	)

Referenced by XAP_Dictionary::countCommonChars().

UT_UCS4Char UT_UCS4_tolower

(

UT_UCS4Char

c

)

References case_table, XAP_EncodingManager::get_instance(), case_entry::other, s_cmp_case(), and case_entry::type.

Referenced by IE_Imp_MsWord_97::_appendChar(), FV_View::_computeFindPrefix(), FV_View::_findNext(), FV_View::_findPrev(), _toggleFirstCapital(), _toggleLower(), _toggleToggle(), ev_Win32Keyboard::onKeyDown(), BarbarismChecker::suggestWord(), and UT_UCS4_stristr().

UT_UCS4Char UT_UCS4_toupper

(

UT_UCS4Char

c

)

Convert a given character to uppercase.

References case_table, XAP_EncodingManager::get_instance(), case_entry::other, s_cmp_case(), and case_entry::type.

Referenced by _toggleFirstCapital(), _toggleToggle(), _toggleUpper(), BarbarismChecker::suggestWord(), and FV_View::toggleCase().

bool UT_validXML

(

char *

pString

)

XML cannot contain any control characters except ,
, , see bug 8565 (http://www.w3.org/TR/REC-xml/#charsets)

This function removes any illegal characters and invalid utf-8 sequences.

The return value of true indicates that the string was modified

References UT_String::c_str(), gchar, UT_String::length(), UT_String::reserve(), s, UT_ASSERT, UT_ASSERT_HARMLESS, and UT_NOT_IMPLEMENTED.

Referenced by TFTEST_MAIN().

bool UT_XML_cloneConvAmpersands	(	gchar *&	rszDest,
		const gchar *	szSource
	)

References gchar, and UT_calloc().

Referenced by localizeMenuItem().

bool UT_XML_cloneNoAmpersands	(	gchar *&	rszDest,
		const gchar *	szSource
	)

References gchar, and UT_calloc().

const gchar* UT_XML_transNoAmpersands

(

const gchar *

szSource

)

References gchar, and UT_calloc().