Singleton helper class to map html entities to other formats. More...

#include <src/htmlentity.h>

Collaboration diagram for HtmlEntityMapper:

Classes
struct	PerlSymb

Public Types
enum	SymType { Sym_Unknown = -1 , Sym_nbsp , Sym_iexcl , Sym_cent , Sym_pound , Sym_curren , Sym_yen , Sym_brvbar , Sym_sect , Sym_uml , Sym_copy , Sym_ordf , Sym_laquo , Sym_not , Sym_shy , Sym_reg , Sym_macr , Sym_deg , Sym_plusmn , Sym_sup2 , Sym_sup3 , Sym_acute , Sym_micro , Sym_para , Sym_middot , Sym_cedil , Sym_sup1 , Sym_ordm , Sym_raquo , Sym_frac14 , Sym_frac12 , Sym_frac34 , Sym_iquest , Sym_Agrave , Sym_Aacute , Sym_Acirc , Sym_Atilde , Sym_Auml , Sym_Aring , Sym_AElig , Sym_Ccedil , Sym_Egrave , Sym_Eacute , Sym_Ecirc , Sym_Euml , Sym_Igrave , Sym_Iacute , Sym_Icirc , Sym_Iuml , Sym_ETH , Sym_Ntilde , Sym_Ograve , Sym_Oacute , Sym_Ocirc , Sym_Otilde , Sym_Ouml , Sym_times , Sym_Oslash , Sym_Ugrave , Sym_Uacute , Sym_Ucirc , Sym_Uuml , Sym_Yacute , Sym_THORN , Sym_szlig , Sym_agrave , Sym_aacute , Sym_acirc , Sym_atilde , Sym_auml , Sym_aring , Sym_aelig , Sym_ccedil , Sym_egrave , Sym_eacute , Sym_ecirc , Sym_euml , Sym_igrave , Sym_iacute , Sym_icirc , Sym_iuml , Sym_eth , Sym_ntilde , Sym_ograve , Sym_oacute , Sym_ocirc , Sym_otilde , Sym_ouml , Sym_divide , Sym_oslash , Sym_ugrave , Sym_uacute , Sym_ucirc , Sym_uuml , Sym_yacute , Sym_thorn , Sym_yuml , Sym_fnof , Sym_Alpha , Sym_Beta , Sym_Gamma , Sym_Delta , Sym_Epsilon , Sym_Zeta , Sym_Eta , Sym_Theta , Sym_Iota , Sym_Kappa , Sym_Lambda , Sym_Mu , Sym_Nu , Sym_Xi , Sym_Omicron , Sym_Pi , Sym_Rho , Sym_Sigma , Sym_Tau , Sym_Upsilon , Sym_Phi , Sym_Chi , Sym_Psi , Sym_Omega , Sym_alpha , Sym_beta , Sym_gamma , Sym_delta , Sym_epsilon , Sym_zeta , Sym_eta , Sym_theta , Sym_iota , Sym_kappa , Sym_lambda , Sym_mu , Sym_nu , Sym_xi , Sym_omicron , Sym_pi , Sym_rho , Sym_sigmaf , Sym_sigma , Sym_tau , Sym_upsilon , Sym_phi , Sym_chi , Sym_psi , Sym_omega , Sym_thetasym , Sym_upsih , Sym_piv , Sym_bull , Sym_hellip , Sym_prime , Sym_Prime , Sym_oline , Sym_frasl , Sym_weierp , Sym_image , Sym_real , Sym_trade , Sym_alefsym , Sym_larr , Sym_uarr , Sym_rarr , Sym_darr , Sym_harr , Sym_crarr , Sym_lArr , Sym_uArr , Sym_rArr , Sym_dArr , Sym_hArr , Sym_forall , Sym_part , Sym_exist , Sym_empty , Sym_nabla , Sym_isin , Sym_notin , Sym_ni , Sym_prod , Sym_sum , Sym_minus , Sym_lowast , Sym_radic , Sym_prop , Sym_infin , Sym_ang , Sym_and , Sym_or , Sym_cap , Sym_cup , Sym_int , Sym_there4 , Sym_sim , Sym_cong , Sym_asymp , Sym_ne , Sym_equiv , Sym_le , Sym_ge , Sym_sub , Sym_sup , Sym_nsub , Sym_sube , Sym_supe , Sym_oplus , Sym_otimes , Sym_perp , Sym_sdot , Sym_lceil , Sym_rceil , Sym_lfloor , Sym_rfloor , Sym_lang , Sym_rang , Sym_loz , Sym_spades , Sym_clubs , Sym_hearts , Sym_diams , Sym_quot , Sym_amp , Sym_lt , Sym_gt , Sym_OElig , Sym_oelig , Sym_Scaron , Sym_scaron , Sym_Yuml , Sym_circ , Sym_tilde , Sym_ensp , Sym_emsp , Sym_thinsp , Sym_zwnj , Sym_zwj , Sym_lrm , Sym_rlm , Sym_ndash , Sym_mdash , Sym_lsquo , Sym_rsquo , Sym_sbquo , Sym_ldquo , Sym_rdquo , Sym_bdquo , Sym_dagger , Sym_Dagger , Sym_permil , Sym_lsaquo , Sym_rsaquo , Sym_euro , Sym_tm , Sym_apos , Sym_BSlash , Sym_At , Sym_Less , Sym_Greater , Sym_Amp , Sym_Dollar , Sym_Hash , Sym_DoubleColon , Sym_Percent , Sym_Pipe , Sym_Quot , Sym_Minus , Sym_Plus , Sym_Dot , Sym_Colon , Sym_Equal , Sym_Exclam , Sym_Quest }
enum	PerlType { Perl_unknown = 0 , Perl_string , Perl_char , Perl_symbol , Perl_umlaut , Perl_acute , Perl_grave , Perl_circ , Perl_slash , Perl_tilde , Perl_cedilla , Perl_ring }
using	HtmlEntityMapperFunc = std::function<DString(SymType)>

Public Member Functions
SymType	name2sym (const DString &symName) const
	Give code of the requested HTML entity name.
const char *	utf8 (SymType symb, bool useInPrintf=false) const
	Access routine to the UTF8 code of the HTML entity.
const char *	html (SymType symb, bool useInPrintf=false) const
	Access routine to the html code of the HTML entity.
const char *	xml (SymType symb) const
	Access routine to the XML code of the HTML entity.
const char *	docbook (SymType symb) const
	Access routine to the docbook code of the HTML entity.
const char *	latex (SymType symb) const
	Access routine to the LaTeX code of the HTML entity.
const char *	man (SymType symb) const
	Access routine to the man code of the HTML entity.
const char *	rtf (SymType symb) const
	Access routine to the RTF code of the HTML entity.
const PerlSymb *	perl (SymType symb) const
	Access routine to the perl struct with the perl code of the HTML entity.
void	writeXMLSchema (TextStream &t)
DString	convertCharEntitiesToUTF8 (const DString &s) const
template<class T>
const char *	writeHtmlEntity (T &result, const char s, HtmlEntityMapperFunc &&mapper, const char fallback)

Static Public Member Functions
static HtmlEntityMapper &	instance ()
	Returns the one and only instance of the HTML entity mapper.

Private Member Functions
void	validate ()
	Routine to check if the entries of the html_entities are numbered correctly.
	HtmlEntityMapper ()
	~HtmlEntityMapper ()

Private Attributes
std::unordered_map< std::string, SymType >	m_name2sym

Static Private Attributes
static HtmlEntityMapper *	s_instance

Detailed Description

Singleton helper class to map html entities to other formats.

Definition at line 29 of file htmlentity.h.

Member Typedef Documentation

◆ HtmlEntityMapperFunc

using HtmlEntityMapper::HtmlEntityMapperFunc = std::function<DString(SymType)>

Definition at line 116 of file htmlentity.h.

Member Enumeration Documentation

◆ PerlType

enum HtmlEntityMapper::PerlType

Enumerator
Perl_unknown
Perl_string
Perl_char
Perl_symbol
Perl_umlaut
Perl_acute
Perl_grave
Perl_circ
Perl_slash
Perl_tilde
Perl_cedilla
Perl_ring

Definition at line 94 of file htmlentity.h.

                  { Perl_unknown = 0, Perl_string, Perl_char, Perl_symbol, Perl_umlaut,
                    Perl_acute, Perl_grave, Perl_circ, Perl_slash, Perl_tilde,
                    Perl_cedilla, Perl_ring
                  };

◆ SymType

enum HtmlEntityMapper::SymType

Enumerator
Sym_Unknown
Sym_nbsp
Sym_iexcl
Sym_cent
Sym_pound
Sym_curren
Sym_yen
Sym_brvbar
Sym_sect
Sym_uml
Sym_copy
Sym_ordf
Sym_laquo
Sym_not
Sym_shy
Sym_reg
Sym_macr
Sym_deg
Sym_plusmn
Sym_sup2
Sym_sup3
Sym_acute
Sym_micro
Sym_para
Sym_middot
Sym_cedil
Sym_sup1
Sym_ordm
Sym_raquo
Sym_frac14
Sym_frac12
Sym_frac34
Sym_iquest
Sym_Agrave
Sym_Aacute
Sym_Acirc
Sym_Atilde
Sym_Auml
Sym_Aring
Sym_AElig
Sym_Ccedil
Sym_Egrave
Sym_Eacute
Sym_Ecirc
Sym_Euml
Sym_Igrave
Sym_Iacute
Sym_Icirc
Sym_Iuml
Sym_ETH
Sym_Ntilde
Sym_Ograve
Sym_Oacute
Sym_Ocirc
Sym_Otilde
Sym_Ouml
Sym_times
Sym_Oslash
Sym_Ugrave
Sym_Uacute
Sym_Ucirc
Sym_Uuml
Sym_Yacute
Sym_THORN
Sym_szlig
Sym_agrave
Sym_aacute
Sym_acirc
Sym_atilde
Sym_auml
Sym_aring
Sym_aelig
Sym_ccedil
Sym_egrave
Sym_eacute
Sym_ecirc
Sym_euml
Sym_igrave
Sym_iacute
Sym_icirc
Sym_iuml
Sym_eth
Sym_ntilde
Sym_ograve
Sym_oacute
Sym_ocirc
Sym_otilde
Sym_ouml
Sym_divide
Sym_oslash
Sym_ugrave
Sym_uacute
Sym_ucirc
Sym_uuml
Sym_yacute
Sym_thorn
Sym_yuml
Sym_fnof
Sym_Alpha
Sym_Beta
Sym_Gamma
Sym_Delta
Sym_Epsilon
Sym_Zeta
Sym_Eta
Sym_Theta
Sym_Iota
Sym_Kappa
Sym_Lambda
Sym_Mu
Sym_Nu
Sym_Xi
Sym_Omicron
Sym_Pi
Sym_Rho
Sym_Sigma
Sym_Tau
Sym_Upsilon
Sym_Phi
Sym_Chi
Sym_Psi
Sym_Omega
Sym_alpha
Sym_beta
Sym_gamma
Sym_delta
Sym_epsilon
Sym_zeta
Sym_eta
Sym_theta
Sym_iota
Sym_kappa
Sym_lambda
Sym_mu
Sym_nu
Sym_xi
Sym_omicron
Sym_pi
Sym_rho
Sym_sigmaf
Sym_sigma
Sym_tau
Sym_upsilon
Sym_phi
Sym_chi
Sym_psi
Sym_omega
Sym_thetasym
Sym_upsih
Sym_piv
Sym_bull
Sym_hellip
Sym_prime
Sym_Prime
Sym_oline
Sym_frasl
Sym_weierp
Sym_image
Sym_real
Sym_trade
Sym_alefsym
Sym_larr
Sym_uarr
Sym_rarr
Sym_darr
Sym_harr
Sym_crarr
Sym_lArr
Sym_uArr
Sym_rArr
Sym_dArr
Sym_hArr
Sym_forall
Sym_part
Sym_exist
Sym_empty
Sym_nabla
Sym_isin
Sym_notin
Sym_ni
Sym_prod
Sym_sum
Sym_minus
Sym_lowast
Sym_radic
Sym_prop
Sym_infin
Sym_ang
Sym_and
Sym_or
Sym_cap
Sym_cup
Sym_int
Sym_there4
Sym_sim
Sym_cong
Sym_asymp
Sym_ne
Sym_equiv
Sym_le
Sym_ge
Sym_sub
Sym_sup
Sym_nsub
Sym_sube
Sym_supe
Sym_oplus
Sym_otimes
Sym_perp
Sym_sdot
Sym_lceil
Sym_rceil
Sym_lfloor
Sym_rfloor
Sym_lang
Sym_rang
Sym_loz
Sym_spades
Sym_clubs
Sym_hearts
Sym_diams
Sym_quot
Sym_amp
Sym_lt
Sym_gt
Sym_OElig
Sym_oelig
Sym_Scaron
Sym_scaron
Sym_Yuml
Sym_circ
Sym_tilde
Sym_ensp
Sym_emsp
Sym_thinsp
Sym_zwnj
Sym_zwj
Sym_lrm
Sym_rlm
Sym_ndash
Sym_mdash
Sym_lsquo
Sym_rsquo
Sym_sbquo
Sym_ldquo
Sym_rdquo
Sym_bdquo
Sym_dagger
Sym_Dagger
Sym_permil
Sym_lsaquo
Sym_rsaquo
Sym_euro
Sym_tm
Sym_apos
Sym_BSlash
Sym_At
Sym_Less
Sym_Greater
Sym_Amp
Sym_Dollar
Sym_Hash
Sym_DoubleColon
Sym_Percent
Sym_Pipe
Sym_Quot
Sym_Minus
Sym_Plus
Sym_Dot
Sym_Colon
Sym_Equal
Sym_Exclam
Sym_Quest

Definition at line 32 of file htmlentity.h.

                 { Sym_Unknown = -1,
                   Sym_nbsp, Sym_iexcl, Sym_cent, Sym_pound, Sym_curren,
                   Sym_yen, Sym_brvbar, Sym_sect, Sym_uml, Sym_copy,
                   Sym_ordf, Sym_laquo, Sym_not, Sym_shy, Sym_reg,
                   Sym_macr, Sym_deg, Sym_plusmn, Sym_sup2, Sym_sup3,
                   Sym_acute, Sym_micro, Sym_para, Sym_middot, Sym_cedil,
                   Sym_sup1, Sym_ordm, Sym_raquo, Sym_frac14, Sym_frac12,
                   Sym_frac34, Sym_iquest, Sym_Agrave, Sym_Aacute, Sym_Acirc,
                   Sym_Atilde, Sym_Auml, Sym_Aring, Sym_AElig, Sym_Ccedil,
                   Sym_Egrave, Sym_Eacute, Sym_Ecirc, Sym_Euml, Sym_Igrave,
                   Sym_Iacute, Sym_Icirc, Sym_Iuml, Sym_ETH, Sym_Ntilde,
                   Sym_Ograve, Sym_Oacute, Sym_Ocirc, Sym_Otilde, Sym_Ouml,
                   Sym_times, Sym_Oslash, Sym_Ugrave, Sym_Uacute, Sym_Ucirc,
                   Sym_Uuml, Sym_Yacute, Sym_THORN, Sym_szlig, Sym_agrave,
                   Sym_aacute, Sym_acirc, Sym_atilde, Sym_auml, Sym_aring,
                   Sym_aelig, Sym_ccedil, Sym_egrave, Sym_eacute, Sym_ecirc,
                   Sym_euml, Sym_igrave, Sym_iacute, Sym_icirc, Sym_iuml,
                   Sym_eth, Sym_ntilde, Sym_ograve, Sym_oacute, Sym_ocirc,
                   Sym_otilde, Sym_ouml, Sym_divide, Sym_oslash, Sym_ugrave,
                   Sym_uacute, Sym_ucirc, Sym_uuml, Sym_yacute, Sym_thorn,
                   Sym_yuml, Sym_fnof, Sym_Alpha, Sym_Beta, Sym_Gamma,
                   Sym_Delta, Sym_Epsilon, Sym_Zeta, Sym_Eta, Sym_Theta,
                   Sym_Iota, Sym_Kappa, Sym_Lambda, Sym_Mu, Sym_Nu,
                   Sym_Xi, Sym_Omicron, Sym_Pi, Sym_Rho, Sym_Sigma,
                   Sym_Tau, Sym_Upsilon, Sym_Phi, Sym_Chi, Sym_Psi,
                   Sym_Omega, Sym_alpha, Sym_beta, Sym_gamma, Sym_delta,
                   Sym_epsilon, Sym_zeta, Sym_eta, Sym_theta, Sym_iota,
                   Sym_kappa, Sym_lambda, Sym_mu, Sym_nu, Sym_xi,
                   Sym_omicron, Sym_pi, Sym_rho, Sym_sigmaf, Sym_sigma,
                   Sym_tau, Sym_upsilon, Sym_phi, Sym_chi, Sym_psi,
                   Sym_omega, Sym_thetasym, Sym_upsih, Sym_piv, Sym_bull,
                   Sym_hellip, Sym_prime, Sym_Prime, Sym_oline, Sym_frasl,
                   Sym_weierp, Sym_image, Sym_real, Sym_trade, Sym_alefsym,
                   Sym_larr, Sym_uarr, Sym_rarr, Sym_darr, Sym_harr,
                   Sym_crarr, Sym_lArr, Sym_uArr, Sym_rArr, Sym_dArr,
                   Sym_hArr, Sym_forall, Sym_part, Sym_exist, Sym_empty,
                   Sym_nabla, Sym_isin, Sym_notin, Sym_ni, Sym_prod,
                   Sym_sum, Sym_minus, Sym_lowast, Sym_radic, Sym_prop,
                   Sym_infin, Sym_ang, Sym_and, Sym_or, Sym_cap,
                   Sym_cup, Sym_int, Sym_there4, Sym_sim, Sym_cong,
                   Sym_asymp, Sym_ne, Sym_equiv, Sym_le, Sym_ge,
                   Sym_sub, Sym_sup, Sym_nsub, Sym_sube, Sym_supe,
                   Sym_oplus, Sym_otimes, Sym_perp, Sym_sdot, Sym_lceil,
                   Sym_rceil, Sym_lfloor, Sym_rfloor, Sym_lang, Sym_rang,
                   Sym_loz, Sym_spades, Sym_clubs, Sym_hearts, Sym_diams,
                   Sym_quot, Sym_amp, Sym_lt, Sym_gt, Sym_OElig,
                   Sym_oelig, Sym_Scaron, Sym_scaron, Sym_Yuml, Sym_circ,
                   Sym_tilde, Sym_ensp, Sym_emsp, Sym_thinsp, Sym_zwnj,
                   Sym_zwj, Sym_lrm, Sym_rlm, Sym_ndash, Sym_mdash,
                   Sym_lsquo, Sym_rsquo, Sym_sbquo, Sym_ldquo, Sym_rdquo,
                   Sym_bdquo, Sym_dagger, Sym_Dagger, Sym_permil, Sym_lsaquo,
                   Sym_rsaquo, Sym_euro,
 
                   /* doxygen extensions */
                   Sym_tm, Sym_apos,
 
                   /* doxygen commands mapped */
                   Sym_BSlash, Sym_At, Sym_Less, Sym_Greater, Sym_Amp,
                   Sym_Dollar, Sym_Hash, Sym_DoubleColon, Sym_Percent, Sym_Pipe,
                   Sym_Quot, Sym_Minus, Sym_Plus, Sym_Dot, Sym_Colon, Sym_Equal,
                   Sym_Exclam, Sym_Quest
                 };

Constructor & Destructor Documentation

◆ HtmlEntityMapper()

HtmlEntityMapper::HtmlEntityMapper ( )

private

Definition at line 326 of file htmlentity.cpp.

{
  for (const auto &entity : g_htmlEntities)
  {
    m_name2sym.emplace(entity.item,entity.symb);
  }
  validate();
}

References g_htmlEntities, m_name2sym, and validate().

Referenced by instance().

◆ ~HtmlEntityMapper()

HtmlEntityMapper::~HtmlEntityMapper ( )

private

Definition at line 335 of file htmlentity.cpp.

336{

337}

Member Function Documentation

◆ convertCharEntitiesToUTF8()

DString HtmlEntityMapper::convertCharEntitiesToUTF8 ( const DString & s ) const

Definition at line 494 of file htmlentity.cpp.

{
  if (str.empty()) return DString();
 
  std::string s = str.data();
  static const reg::Ex re(R"(&\a\w*;)");
  reg::Iterator it(s,re);
  reg::Iterator end;
 
  DString result;
  result.reserve(str.length()+32);
  size_t p=0, i=0, l=0;
  for (; it!=end ; ++it)
  {
    const auto &match = *it;
    p = match.position();
    l = match.length();
    if (p>i)
    {
      result+=s.substr(i,p-i);
    }
    DString entity(match.str());
    SymType symType = name2sym(entity);
    const char *code=nullptr;
    if (symType!=Sym_Unknown && (code=utf8(symType)))
    {
      result+=code;
    }
    else
    {
      result+=entity;
    }
    i=p+l;
  }
  result+=s.substr(i);
  //printf("convertCharEntitiesToUTF8(%s)->%s\n",qPrint(s),qPrint(result));
  return result;
}

References DString::data(), DString::DString(), DString::empty(), end(), DString::length(), name2sym(), DString::reserve(), DString::substr(), Sym_Unknown, and utf8().

Referenced by convertToJSString(), getSQLDocBlock(), and parseCommentAsText().

◆ docbook()

const char * HtmlEntityMapper::docbook ( HtmlEntityMapper::SymType symb ) const

Access routine to the docbook code of the HTML entity.

Parameters

symb	Code of the requested HTML entity

Returns: the docbook code of the HTML entity, in case the docbook code is unknown nullptr is returned.

Definition at line 403 of file htmlentity.cpp.

{
  return g_htmlEntities[symb].docbook;
}

References g_htmlEntities.

Referenced by DocbookGenerator::convertToDocbook(), and DocbookDocVisitor::operator()().

◆ html()

const char * HtmlEntityMapper::html	(	HtmlEntityMapper::SymType	symb,
		bool	useInPrintf = false ) const

Access routine to the html code of the HTML entity.

Parameters

symb	Code of the requested HTML entity
useInPrintf	If true the result will be escaped such that it can be used in a printf string pattern

Returns: the html representation of the HTML entity, in case the html code is unknown nullptr is returned.

Definition at line 374 of file htmlentity.cpp.

{
  if (useInPrintf && symb==HtmlEntityMapper::Sym_Percent)
  {
    return "%%"; // escape for printf
  }
  else
  {
    return g_htmlEntities[symb].html;
  }
}

References g_htmlEntities, and Sym_Percent.

Referenced by HtmlDocVisitor::filter(), HtmlDocVisitor::operator()(), and TextDocVisitor::operator()().

◆ instance()

HtmlEntityMapper & HtmlEntityMapper::instance ( )

static

Returns the one and only instance of the HTML entity mapper.

Definition at line 340 of file htmlentity.cpp.

{
  static HtmlEntityMapper inst;
  return inst;
}

References HtmlEntityMapper().

Referenced by DocbookGenerator::convertToDocbook(), convertToJSString(), convertToXML(), DocSymbol::decodeSymbol(), HtmlDocVisitor::filter(), ManDocVisitor::filter(), RTFDocVisitor::filter(), filterLatexString(), generateSqlite3ForPage(), generateXML(), generateXMLForPage(), generateXMLForRequirement(), generateXMLForRequirements(), getSQLDocBlock(), DocbookDocVisitor::operator()(), HtmlDocVisitor::operator()(), LatexDocVisitor::operator()(), ManDocVisitor::operator()(), PerlModDocVisitor::operator()(), PrintDocVisitor::operator()(), RTFDocVisitor::operator()(), TextDocVisitor::operator()(), XmlDocVisitor::operator()(), and parseCommentAsText().

◆ latex()

const char * HtmlEntityMapper::latex ( HtmlEntityMapper::SymType symb ) const

Access routine to the LaTeX code of the HTML entity.

Parameters

symb	Code of the requested HTML entity

Returns: the LaTeX code of the HTML entity, in case the LaTeX code is unknown nullptr is returned.

Definition at line 414 of file htmlentity.cpp.

{
  return g_htmlEntities[symb].latex;
}

References g_htmlEntities.

Referenced by filterLatexString(), and LatexDocVisitor::operator()().

◆ man()

const char * HtmlEntityMapper::man ( HtmlEntityMapper::SymType symb ) const

Access routine to the man code of the HTML entity.

Parameters

symb	Code of the requested HTML entity

Returns: the man of the HTML entity, in case the man code is unknown nullptr is returned.

Definition at line 425 of file htmlentity.cpp.

{
  return g_htmlEntities[symb].man;
}

References g_htmlEntities.

Referenced by ManDocVisitor::filter(), and ManDocVisitor::operator()().

◆ name2sym()

HtmlEntityMapper::SymType HtmlEntityMapper::name2sym ( const DString & symName ) const

Give code of the requested HTML entity name.

Parameters

symName HTML entity name without & and ;

Returns: the code for the requested HTML entity name, in case the requested HTML item does not exist HtmlEntityMapper::Sym_unknown is returned.

Definition at line 459 of file htmlentity.cpp.

{
  auto it = m_name2sym.find(symName.str());
  return it!=m_name2sym.end() ? it->second : HtmlEntityMapper::Sym_Unknown;
}

References m_name2sym, DString::str(), and Sym_Unknown.

Referenced by convertCharEntitiesToUTF8(), DocSymbol::decodeSymbol(), and writeHtmlEntity().

◆ perl()

const HtmlEntityMapper::PerlSymb * HtmlEntityMapper::perl ( HtmlEntityMapper::SymType symb ) const

Access routine to the perl struct with the perl code of the HTML entity.

Parameters

symb	Code of the requested HTML entity

Returns: the pointer to perl struct with the perl code of the HTML entity, in case the perl code does not exists the nullptr pointer is entered in the symb field and in the HtmlEntityMapper::Perl_unknown in the type field.

Definition at line 448 of file htmlentity.cpp.

{
  return &g_htmlEntities[symb].perl;
}

References g_htmlEntities.

Referenced by PerlModDocVisitor::operator()().

◆ rtf()

const char * HtmlEntityMapper::rtf ( HtmlEntityMapper::SymType symb ) const

Access routine to the RTF code of the HTML entity.

Parameters

symb	Code of the requested HTML entity

Returns: the RTF of the HTML entity, in case the RTF code is unknown nullptr is returned.

Definition at line 436 of file htmlentity.cpp.

{
  return g_htmlEntities[symb].rtf;
}

References g_htmlEntities.

Referenced by RTFDocVisitor::filter(), and RTFDocVisitor::operator()().

◆ utf8()

const char * HtmlEntityMapper::utf8	(	HtmlEntityMapper::SymType	symb,
		bool	useInPrintf = false ) const

Access routine to the UTF8 code of the HTML entity.

Parameters

symb	Code of the requested HTML entity
useInPrintf	If true the result will be escaped such that it can be used in a printf string pattern

Returns: the UTF8 code of the HTML entity, in case the UTF code is unknown nullptr is returned.

Definition at line 354 of file htmlentity.cpp.

{
  if (useInPrintf && symb==HtmlEntityMapper::Sym_Percent)
  {
    return "%%"; // escape for printf
  }
  else
  {
    return g_htmlEntities[symb].UTF8;
  }
}

References g_htmlEntities, and Sym_Percent.

Referenced by convertCharEntitiesToUTF8(), and PrintDocVisitor::operator()().

◆ validate()

void HtmlEntityMapper::validate ( )

private

Routine to check if the entries of the html_entities are numbered correctly.

in case of a mismatch a warning message is given.

Definition at line 481 of file htmlentity.cpp.

{
  int i=0;
  for (const auto &entity : g_htmlEntities)
  {
    if (i != entity.symb)
    {
      warn_uncond("Internal inconsistency, htmlentries code {} (item={})\n",i,entity.item);
    }
    i++;
  }
}

References g_htmlEntities, and warn_uncond.

Referenced by HtmlEntityMapper().

◆ writeHtmlEntity()

template<class T>

const char * HtmlEntityMapper::writeHtmlEntity	(	T &	result,
		const char *	s,
		HtmlEntityMapperFunc &&	mapper,
		const char *	fallback )

inline

Writes an HTML entity for the current symbol and advances the input pointer.

Template Parameters

T	Type of the output sink used to write encoded output.

Parameters

result	Output target receiving the encoded entity or fallback text.
s	Pointer to the start of a potential HTML entity in the input text.
mapper	Callback that maps a entity symbol type to its HTML entity string.
fallback	Fallback string written when no entity mapping is available.

Returns: Pointer to the position after the processed HTML entity in the input text.

Definition at line 127 of file htmlentity.h.

    {
      assert(s!=nullptr);
      assert(s[0]=='&');
      const char *q = s+1;
      size_t cnt = 2; // we have to count & and ; as well
      while ((*q >= 'a' && *q <= 'z') || (*q >= 'A' && *q <= 'Z') || (*q >= '0' && *q <= '9'))
      {
        cnt++;
        q++;
      }
      if (*q == ';') // valid entity name
      {
        SymType res = name2sym(DString(s).left(cnt));
        if (res!=Sym_Unknown)
        {
          result += mapper(res);
          return q+1;
        }
      }
      result += fallback;
      return s+1;
    }

References DString::DString(), name2sym(), and Sym_Unknown.

Referenced by DocbookGenerator::convertToDocbook(), convertToXML(), HtmlDocVisitor::filter(), ManDocVisitor::filter(), RTFDocVisitor::filter(), and filterLatexString().

◆ writeXMLSchema()

void HtmlEntityMapper::writeXMLSchema ( TextStream & t )

Definition at line 465 of file htmlentity.cpp.

{
  for (size_t i=0;i<g_htmlEntities.size();i++)
  {
    DString bareName = g_htmlEntities[i].xml;
    if (!bareName.empty() && bareName.at(0)=='<' && bareName.endsWith("/>"))
    {
      bareName = bareName.mid(1,bareName.length()-3); // strip < and />
      t << "      <xsd:element name=\"" << bareName << "\" type=\"docEmptyType\" />\n";
    }
  }
}

References DString::at(), DString::empty(), DString::endsWith(), g_htmlEntities, DString::length(), and DString::mid().

Referenced by generateXML().

◆ xml()

const char * HtmlEntityMapper::xml ( HtmlEntityMapper::SymType symb ) const

Access routine to the XML code of the HTML entity.

Parameters

symb	Code of the requested HTML entity

Returns: the XML code of the HTML entity, in case the XML code is unknown nullptr is returned.

Definition at line 392 of file htmlentity.cpp.

{
  return g_htmlEntities[symb].xml;
}

References g_htmlEntities.

Referenced by convertToXML(), and XmlDocVisitor::operator()().

Member Data Documentation

◆ m_name2sym

std::unordered_map<std::string,SymType> HtmlEntityMapper::m_name2sym

private

Definition at line 157 of file htmlentity.h.

Referenced by HtmlEntityMapper(), and name2sym().

◆ s_instance

HtmlEntityMapper* HtmlEntityMapper::s_instance

staticprivate

Definition at line 156 of file htmlentity.h.

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

src/htmlentity.h
src/htmlentity.cpp

Classes

Public Types

Public Member Functions

Static Public Member Functions

Private Member Functions

Private Attributes

Static Private Attributes

Detailed Description

Member Typedef Documentation

◆ HtmlEntityMapperFunc

Member Enumeration Documentation

◆ PerlType

◆ SymType

Constructor & Destructor Documentation

◆ HtmlEntityMapper()

◆ ~HtmlEntityMapper()

Member Function Documentation

◆ convertCharEntitiesToUTF8()

◆ docbook()

◆ html()

◆ instance()

◆ latex()

◆ man()

◆ name2sym()

◆ perl()

◆ rtf()

◆ utf8()

◆ validate()

◆ writeHtmlEntity()

◆ writeXMLSchema()

◆ xml()

Member Data Documentation

◆ m_name2sym

◆ s_instance