Wicri:Dilib source, module Utf8, commande Utf8FromHexEntity

Bibliothèque Dilib (ressources numériques) Module Utf8

Cette page introduit le fichier source pour générer la commande de conversion Utf8FromHexEntity.

Liste des modules Dilib Noyau Unix... CharSet, Buffer, , RegExp, StrDict, Utf8 Modules SRI, infométrie Explor, Hfd, Infometrics, Index, Tei Modules Sxml SxmlNode, SxmlProc, SxPath, SgmlFast Flux de données ArXiv, Atilf, BnF, Ccsd, Getty, Inist, Inra, Istex, Loria, Marc, Nlm, Wicri Interfaces avec logiciels MediaWiki Installation, versions... Dilib V0.6.3, DilibMake, installation et compilation   Module Utf8 Fonctions fonctions de conversion fonctions sur Buffer Commandes Utf8ToLc Dilib, Module Utf8, commande Utf8SplitWordUsual Dilib, Module Utf8, commande Utf8LangBuildProfile

Code source

/********************************************************************************
ATTENTION : la version de référence de ce source est sur le wiki Wicri/Outils
WARNING:    reference source code is on Wicri/Outils (fr)  

http://ticri.univ-lorraine.fr/wicri-outils.fr/index.php/Wicri:Dilib_source,_module_Utf8,_commande_Utf8FromHexEntity
================================================================================

     Module   : Utf8
     Command  : Utf8FromHexEntity
     File     : Utf8FromHexEntity.lex

 ********************************************************************************/
%{
#include "stdio.h"
#include "string.h"
#include <stdlib.h> 
#include "Buffer.h"

int Hexa2bin(char *hexa)
{
  int i,l,res,w;
  l=strlen(hexa);
  i=l;
  res=0;
  w=1;
  while (i--)
    {
      if (isdigit(hexa[i]))res+=(hexa[i]&0x0F)*w;
      else res+=((hexa[i]&0x07)+9)*w;
      w=16*w;
    }
  return res;
}

Buffer *Utf8CharBuffer=NULL;

char *Utf8CharFromHexa(char *hexa)
{
  int charNumber;
  if(!Utf8CharBuffer)Utf8CharBuffer=BufferCreate(10,10);
  BufferReset(Utf8CharBuffer);
  charNumber=Hexa2bin(hexa);
  if (charNumber<33)                           /*  ASCII control  char */
    {
      BufferStrcpy(Utf8CharBuffer, "&#x");
      BufferStrcat(Utf8CharBuffer, hexa);
      BufferStrcat(Utf8CharBuffer, ";");
      return (BufferString(Utf8CharBuffer));
    }
  if (charNumber<128)                            /* ASCII with XML exceptions */
    {
      switch (charNumber)
	{
	case 38: return ("&amp");
	case 60: return ("&lt;");
	case 62: return ("&gt;");
	default: 
	  BufferCharCat(Utf8CharBuffer, charNumber);
	  return (BufferString(Utf8CharBuffer));
	}
    }
  if  (charNumber<2048)                          /* sth like &#x999;  */
    {
      int n1, n2;
      n2=(charNumber & 0x3F) | 0x80;
      n1=((charNumber & 0x7C0)/0x40) | 0xC0;
      BufferCharCat(Utf8CharBuffer,n1);
      BufferCharCat(Utf8CharBuffer,n2);
      return (BufferString(Utf8CharBuffer));
    }
  if (charNumber < 0x10000)                       /* sth like &#x9999;  */
    {
      int n1, n2, n3;
      n3=(charNumber & 0x3F) | 0x80;
      n2=((charNumber & 0xFC0)/0x40) | 0x80;
      n1=((charNumber & 0xF000)/0x1000) | 0xE0;
      BufferCharCat(Utf8CharBuffer,n1);
      BufferCharCat(Utf8CharBuffer,n2);
      BufferCharCat(Utf8CharBuffer,n3);
      return (BufferString(Utf8CharBuffer));
    }
  BufferStrcpy(Utf8CharBuffer, "&#x");             /* to be improved */
  BufferStrcat(Utf8CharBuffer, hexa);
  BufferStrcat(Utf8CharBuffer, ";");
  return (BufferString(Utf8CharBuffer));
}
%}

%START UTF8

%%
"&#x"                 BEGIN UTF8;
<UTF8>[A-Fa-f0-9]+    printf("%s",Utf8CharFromHexa(yytext));
<UTF8>";"             BEGIN 0;
%%

main()
{
  yylex();
  exit(EXIT_SUCCESS);
}

Voir aussi

• Dilib, module Utf8, commande Utf8FromHexEntity