[q-lang-cvs] q/src w3centities.h,NONE,1.1 qbase.c,1.25,1.26

[Albert G. <ag...@us...>]

Update of /cvsroot/q-lang/q/src
In directory sc8-pr-cvs16.sourceforge.net:/tmp/cvs-serv28296

Modified Files:
	qbase.c 
Added Files:
	w3centities.h 
Log Message:
add xml entity character escapes, as suggested by Rob Hubbard and John Cowan

Index: qbase.c
===================================================================
RCS file: /cvsroot/q-lang/q/src/qbase.c,v
retrieving revision 1.25
retrieving revision 1.26
diff -C2 -d -r1.25 -r1.26
*** qbase.c	24 Sep 2007 12:20:59 -0000	1.25
--- qbase.c	16 Jan 2008 08:28:15 -0000	1.26
***************
*** 1019,1022 ****
--- 1019,1051 ----
  }
  
+ #include "w3centities.h"
+ 
+ static int entcmp(const void *_x, const void *_y)
+ {
+   const char *x = (const char*)_x;
+   const Entity *y = (const Entity*)_y;
+   return strcmp(x, y->name);
+ }
+ 
+ static inline long entity(const char *name)
+ {
+   Entity *x = bsearch(name, entities, sizeof(entities) / sizeof(Entity),
+ 		      sizeof(Entity), entcmp);
+   return x?x->c:-1;
+ }
+ 
+ static long parse_entity(char *s, char **t)
+ {
+   long c;
+   char *p = strchr(s, ';');
+   if (!p) return -1;
+   *p = 0;
+   c = entity(s);
+   *p = ';';
+   if (c < 0) return c;
+   *t = p+1;
+   return c;
+ }
+ 
  static char *
  scanchar(char *t, char **s, char **p)
***************
*** 1049,1052 ****
--- 1078,1099 ----
        *t++ = c;
        break;
+     case '&': {
+       char *r;
+       long c = parse_entity(*s, s);
+       if (c >= 0) {
+ #ifdef HAVE_UNICODE
+ 	if (c >= 0x110000)
+ 	  c %=  0x110000;
+ 	u8encode(t, c);
+ 	t += strlen(t);
+ #else
+ 	*t++ = (char)c;
+ #endif
+       } else {
+ 	*p = *s-1;
+ 	*t++ = c;
+       }
+       break;
+     }
      case '(': {
        if ('0' <= **s && **s <= '9') {

--- NEW FILE: w3centities.h ---
/* generated from w3centities.ent Wed Jan 16 08:17:24 2008 */

typedef struct ENTITY { char *name; long c; } Entity;

static Entity entities[] = {
  { "AElig", 0x000C6 },
  { "Aacgr", 0x00386 },
  { "Aacute", 0x000C1 },
  { "Abreve", 0x00102 },
  { "Acirc", 0x000C2 },
  { "Acy", 0x00410 },
  { "Afr", 0x1D504 },
  { "Agr", 0x00391 },
  { "Agrave", 0x000C0 },
  { "Alpha", 0x00391 },
  { "Amacr", 0x00100 },
  { "And", 0x02A53 },
  { "Aogon", 0x00104 },
  { "Aopf", 0x1D538 },
[...2081 lines suppressed...]
  { "yicy", 0x00457 },
  { "yopf", 0x1D56A },
  { "yscr", 0x1D4CE },
  { "yucy", 0x0044E },
  { "yuml", 0x000FF },
  { "zacute", 0x0017A },
  { "zcaron", 0x0017E },
  { "zcy", 0x00437 },
  { "zdot", 0x0017C },
  { "zeetrf", 0x02128 },
  { "zeta", 0x003B6 },
  { "zfr", 0x1D537 },
  { "zgr", 0x003B6 },
  { "zhcy", 0x00436 },
  { "zigrarr", 0x021DD },
  { "zopf", 0x1D56B },
  { "zscr", 0x1D4CF },
  { "zwj", 0x0200D },
  { "zwnj", 0x0200C },
};