--- poppler/GfxFont.cc	2006/11/14 22:07:55	1.1
+++ poppler/GfxFont.cc	2006/11/15 08:27:39
@@ -773,42 +773,52 @@ Gfx8BitFont::Gfx8BitFont(XRef *xref, cha
     }
   }
 
+  // construct the char code -> Unicode mapping object
+  ctu = CharCodeToUnicode::make8BitToUnicode(toUnicode);
+
   // pass 2: try to fill in the missing chars, looking for names of
   // the form 'Axx', 'xx', 'Ann', 'ABnn', or 'nn', where 'A' and 'B'
   // are any letters, 'xx' is two hex digits, and 'nn' is 2-4
   // decimal digits
-  if (missing && globalParams->getMapNumericCharNames()) {
+  if (missing) {
     for (code = 0; code < 256; ++code) {
       if ((charName = enc[code]) && !toUnicode[code] &&
 	  strcmp(charName, ".notdef")) {
-	n = strlen(charName);
-	code2 = -1;
-	if (hex && n == 3 && isalpha(charName[0]) &&
-	    isxdigit(charName[1]) && isxdigit(charName[2])) {
-	  sscanf(charName+1, "%x", &code2);
-	} else if (hex && n == 2 &&
-		   isxdigit(charName[0]) && isxdigit(charName[1])) {
-	  sscanf(charName, "%x", &code2);
-	} else if (!hex && n >= 2 && n <= 4 &&
-		   isdigit(charName[0]) && isdigit(charName[1])) {
-	  code2 = atoi(charName);
-	} else if (n >= 3 && n <= 5 &&
-		   isdigit(charName[1]) && isdigit(charName[2])) {
-	  code2 = atoi(charName+1);
-	} else if (n >= 4 && n <= 6 &&
-		   isdigit(charName[2]) && isdigit(charName[3])) {
-	  code2 = atoi(charName+2);
-	}
-	if (code2 >= 0 && code2 <= 0xff) {
-	  toUnicode[code] = (Unicode)code2;
+	m = sizeof(uBuf)/sizeof(Unicode);
+	if (globalParams->getMapNumericCharNames() &&
+	    (n = parseNumericCharCode(charName, hex, uBuf, m))) {
+	  ctu->setMapping((CharCode)code, uBuf, n);
+	} else if (strchr(charName, '_')) { // possibly a ligature?
+	  // names of the form A_a (e.g. f_i, T_h, l_quotesingle)
+	  char *lig_copy = charName = copyString(charName);
+	  n = 0;
+	  do {
+	    if ((buf = strchr(charName, '_')))
+	      *buf = 0;
+	    if (!strcmp(charName, ".notdef"))
+	      ;
+	    else if ((uBuf[n] = globalParams->mapNameToUnicode(charName)))
+	      n++;
+	    else if (globalParams->getMapNumericCharNames() &&
+		     (i = parseNumericCharCode(charName, hex, uBuf + n, m - n)))
+	      n += i;
+	    else
+	      fprintf(stderr, "Could not parse ligature component in charref "
+		      "for nameToUnicode: %s\n", charName);
+	    charName = buf + 1;
+	  } while (buf && n < m);
+	  if (n >= 1) {
+	    ctu->setMapping((CharCode)code, uBuf, n);
+	  }
+	  gfree(lig_copy);
+	} else {
+	  fprintf(stderr, "Could not parse charref for nameToUnicode: %s\n",
+		  charName);
 	}
       }
     }
   }
 
-  // construct the char code -> Unicode mapping object
-  ctu = CharCodeToUnicode::make8BitToUnicode(toUnicode);
-
   // merge in a ToUnicode CMap, if there is one -- this overwrites
   // existing entries in ctu, i.e., the ToUnicode CMap takes
   // precedence, but the other encoding info is allowed to fill in any
@@ -938,6 +948,44 @@ Gfx8BitFont::~Gfx8BitFont() {
   }
 }
 
+int Gfx8BitFont::parseNumericCharCode(char *charName, GBool hex,
+				      Unicode *uBuf, int uLen) {
+  int n, code2;
+  if (uLen < 1)
+    return 0;
+  // look for names of the form 'Axx', 'xx', 'Ann', 'ABnn', or 'nn', where 
+  // 'A' and 'B' are any letters, 'xx' is two hex digits, and 'nn' is 2-4
+  // decimal digits
+  // look for names of the form 'uniXXXX' where 'XXXX' is 4 hex digits
+  n = strlen(charName);
+  code2 = -1;
+  if (hex && n == 3 && isalpha(charName[0]) &&
+      isxdigit(charName[1]) && isxdigit(charName[2])) {
+    sscanf(charName+1, "%x", &code2);
+  } else if (hex && n == 2 &&
+	     isxdigit(charName[0]) && isxdigit(charName[1])) {
+    sscanf(charName, "%x", &code2);
+  } else if (!hex && n >= 2 && n <= 4 &&
+	     isdigit(charName[0]) && isdigit(charName[1])) {
+    code2 = atoi(charName);
+  } else if (n >= 3 && n <= 5 &&
+	     isdigit(charName[1]) && isdigit(charName[2])) {
+    code2 = atoi(charName+1);
+  } else if (n >= 4 && n <= 6 &&
+	     isdigit(charName[2]) && isdigit(charName[3])) {
+    code2 = atoi(charName+2);
+  } else if (n == 7 && !strncmp(charName, "uni", 3) && 
+	     isxdigit(charName[3]) && isxdigit(charName[4]) && 
+	     isxdigit(charName[5]) && isxdigit(charName[6])) {
+    sscanf(charName + 3, "%x", &code2);
+  }
+  if (code2 != -1) {
+    uBuf[0] = (Unicode)code2;
+    return 1;
+  }
+  return 0;
+}
+
 int Gfx8BitFont::getNextChar(char *s, int len, CharCode *code,
 			     Unicode *u, int uSize, int *uLen,
 			     double *dx, double *dy, double *ox, double *oy) {
--- poppler/GfxFont.h	2006/11/14 23:02:19	1.1
+++ poppler/GfxFont.h	2006/11/15 07:54:00
@@ -271,6 +271,9 @@ public:
 
 private:
 
+  // Parse a character name as a numeric reference.
+  int parseNumericCharCode(char *charName, GBool hex, Unicode *uBuf, int uLen);
+
   char *enc[256];		// char code --> char name
   char encFree[256];		// boolean for each char name: if set,
 				//   the string is malloc'ed