* font/devlj4/generate/text.map: Add more MSL numbers.

* src/utils/hpftodit/hpftodit.cpp (is_decomposed): New macro.
(text_symbol_sets): Add more symbol sets.
(debug_flag): New static variable (moved from `main').
(get_printcode, show_symset): New functions.
(main): Remove `debug_flag'.
(output_charset): Use `show_symset'.
(dump_symbols):  Use `show_symset'.
Print symsets for all glyphs.
(read_map): Actually call `unicode_to_ucode_make' but this time
correctly.

4 files changed, 144 insertions, 40 deletions

diff --git a/ChangeLog b/ChangeLog
index 94f9c511..138c8454 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,20 @@
+2004-01-02  Werner LEMBERG  <wl@gnu.org>
+
+	* font/devlj4/generate/text.map: Add more MSL numbers.
+
+2004-01-02  Jeff Conrad  <jeff_conrad@msn.com>
+
+	* src/utils/hpftodit/hpftodit.cpp (is_decomposed): New macro.
+	(text_symbol_sets): Add more symbol sets.
+	(debug_flag): New static variable (moved from `main').
+	(get_printcode, show_symset): New functions.
+	(main): Remove `debug_flag'.
+	(output_charset): Use `show_symset'.
+	(dump_symbols):  Use `show_symset'.
+	Print symsets for all glyphs.
+	(read_map): Actually call `unicode_to_ucode_make' but this time
+	correctly.
+
 2004-01-01  Werner LEMBERG  <wl@gnu.org>
 
 	* font/devlj4/generate/text.map, font/devlj4/generate/special.map:
diff --git a/font/devlj4/generate/special.map b/font/devlj4/generate/special.map
index 65f4b850..fe947814 100644
--- a/font/devlj4/generate/special.map
+++ b/font/devlj4/generate/special.map
@@ -48,7 +48,6 @@
 225	2193	da
 226	2192	->
 227	2190	<-
-# 228 ?
 228	2319	u2319
 229	2194	<>
 230	25B2	u25B2
diff --git a/font/devlj4/generate/text.map b/font/devlj4/generate/text.map
index 06625744..c596a217 100644
--- a/font/devlj4/generate/text.map
+++ b/font/devlj4/generate/text.map
@@ -129,7 +129,7 @@
 # 96 *
 96	007E	ti
 # 97 medium shading *
-# 98 ?
+# 98 uppercase greek delta -- unused
 99	00C0	`A
 
 100	00C2	^A
@@ -277,7 +277,7 @@
 # 225 down arrow *
 # 226 right arrow *
 # 227 left arrow *
-# 228 tick up at left *
+# 228 tick up at left (`logicalnotflopped') *
 # 229 left/right arrow *
 
 # 230 up solid arrowhead *
@@ -337,13 +337,13 @@
 
 # 280 top 1/2 solid rectangle *
 # 281 lowercase greek alpha -- unused
-# 282 ?
+# 282 lowercase greek beta -- unused
 # 283 uppercase greek gamma -- unused
 # 284 lowercase greek pi -- unused
 # 285 uppercase greek sigma -- unused
 # 286 lowercase greek sigma -- unused
 # 287 lowercase greek tau -- unused
-# 288 ?
+# 288 uppercase greek phi -- unused
 # 289 uppercase greek theta -- unused
 
 # 290 uppercase omega, ohms -- unused
@@ -362,7 +362,7 @@
 # 302 *
 302	00B7	pc
 # 303 radical symbol *
-# 304 ?
+# 304 lowercase greek eta -- unused
 # 305 small solid box *
 306	013F	u013F
 307	0140	u0140
@@ -405,7 +405,13 @@
 340	014A	u014A
 341	014B	u014B
 342	0111	u0111
-# 343-399 ?
+# 343-369 ?
+
+# 370 `lira' ?
+# 371-374 ?
+# 375 `asciigrave' ?
+
+# 376-399 ?
 
 400	0102	u0041_0306
 401	0103	u0061_0306
@@ -789,12 +795,50 @@
 # 1110 medium open bullet *
 # 1111 large solid bullet ?
 # 1112 visible carriage return *
-# 1113 visible tab ?
+# 1113 visible tab (`arrowtab') ?
 # 1114 visible space *
 # 1115 visible soft hyphen -- unused
-# 1116 visible end-of-file ?
-
-# 1117-3811 ?
+# 1116 visible end-of-file (`embox')?
+# 1117 `enspacereverse' ?
+# 1118 `emspacereverse' ?
+# 1119 ?
+
+# 1120-1399 ?
+
+1400 u0301
+1401 u0300
+1402 u0302
+1403 u0308
+1404 u0303
+1405 u030C
+1406 u0306
+1407 u030B
+1408 u030A
+1409 u0307
+
+1410 u0304
+1411 u0327
+1412 u0328
+# 1413 combining uppercase acute accent -- unused
+# 1414 combining uppercase grave accent -- unused
+# 1415 combining uppercase circumflex accent -- unused
+# 1416 combining uppercase dieresis accent -- unused
+# 1417 combining uppercase tilde accent -- unused
+# 1418 combining uppercase caron accent -- unused
+# 1419 combining uppercase breve accent -- unused
+
+# 1420 combining uppercase double acute accent -- unused
+# 1421 combining uppercase ring accent -- unused
+# 1422 combining uppercase dot accent -- unused
+# 1423 combining uppercase macron accent -- unused
+# 1424 combining uppercase cedilla -- unused
+# 1425 combining uppercase ogonek -- unused
+# 1426 ?
+# 1427 uppercase middle dot -- unused (no Unicode mapping)
+# 1428 ?
+# 1429 ?
+
+# 1430-3811 ?
 
 # 3812 ornament, apple ?
 
diff --git a/src/utils/hpftodit/hpftodit.cpp b/src/utils/hpftodit/hpftodit.cpp
index 779d3c1b..a8dccde0 100644
--- a/src/utils/hpftodit/hpftodit.cpp
+++ b/src/utils/hpftodit/hpftodit.cpp
@@ -45,7 +45,9 @@ extern "C" const char *Version_string;
 extern const char *hp_msl_to_unicode_code(const char *);
 
 #define SIZEOF(v) (sizeof(v)/sizeof(v[0]))
-#define equal(a, b)      (strcmp(a, b) == 0)
+#define equal(a, b) (strcmp(a, b) == 0)
+// only valid if is_uname(c) has returned true
+#define is_decomposed(c) strchr(c, '_')
 
 #define NO       0
 #define YES      1
@@ -206,7 +208,7 @@ uint16 text_symbol_sets[] = {
   SYMBOL_SET(7, 'J'),		// Desktop
   SYMBOL_SET(6, 'J'),		// Microsoft Publishing
   SYMBOL_SET(9, 'E'),		// Windows Latin 2, Code Page 1250
-  SYMBOL_SET(2, 'N'),		// Latin 2 (subset of 9E,
+  SYMBOL_SET(2, 'N'),		// Latin 2 (subset of 9M,
 				// so we should never get here)
   SYMBOL_SET(0, 'N'),		// Latin 1 (subset of 19U,
 				// so we should never get here)
@@ -214,10 +216,9 @@ uint16 text_symbol_sets[] = {
   SYMBOL_SET(10, 'J'),		// PS Standard
   SYMBOL_SET(9, 'U'),		// Windows 3.0 "ANSI"
 
-#if 0
-  SYMBOL_SET(13, 'J'),		// Ventura International
-  SYMBOL_SET(6, 'M'),		// Ventura Math
-  SYMBOL_SET(14, 'J'),		// Ventura US
+//SYMBOL_SET(13, 'J'),		// Ventura International (deprecated)
+//SYMBOL_SET(6, 'M'),		// Ventura Math (deprecated)
+//SYMBOL_SET(14, 'J'),		// Ventura US (deprecated)
   SYMBOL_SET(5, 'T'),		// Code Page 1254
   SYMBOL_SET(0, 'D'),		// ISO 60, 7-bit Norwegian version 1
   SYMBOL_SET(5, 'N'),		// ISO 8859-9, Latin 5
@@ -235,7 +236,6 @@ uint16 text_symbol_sets[] = {
   SYMBOL_SET(0, 'S'),		// ISO 11, 7-bit Swedish
   SYMBOL_SET(1, 'E'),		// ISO 4, 7-bit UK English
   SYMBOL_SET(0, 'U'),		// ISO 6, 7-bit ASCII English
-#endif
   0
 };
 
@@ -259,6 +259,7 @@ name_list **charcode_name_table = NULL;
 unsigned int n_symbol_sets;
 symbol_set *symbol_set_table;
 
+static int debug_flag = NO;
 static int special_flag = NO;	// not a special font
 static int italic_flag = NO;	// don't add italic correction
 static int italic_sep;
@@ -268,6 +269,8 @@ static int quiet_flag = NO;	// don't suppress warnings about symbols not found
 static char *hp_msl_to_ucode_name(int);
 static char *unicode_to_ucode_name(int);
 static int is_uname(char *);
+static int get_printcode(uint32, uint16 *, uint16 *);
+static char *show_symset(unsigned int);
 static void usage(FILE *);
 static void usage();
 static const char *xbasename(const char *);
@@ -304,7 +307,6 @@ int main(int argc, char **argv)
   program_name = argv[0];
 
   int opt;
-  int debug_flag = NO;
   int res = 1200;		// PCL unit of measure for cursor moves
   int scalesize = 4;		// LaserJet 4 only allows 1/4 point increments
   int unitwidth = 6350;		
@@ -332,13 +334,11 @@ int main(int argc, char **argv)
       italic_sep = atoi(optarg);	// design units
       break;
     case 'q':
-	quiet_flag = YES;	// suppress warnings about symbols not found
-	break;
+      quiet_flag = YES;		// suppress warnings about symbols not found
+      break;
     case 'v':
-      {
-	printf("GNU hpftodit (groff) version %s\n", Version_string);
-	exit(0);
-      }
+      printf("GNU hpftodit (groff) version %s\n", Version_string);
+      exit(0);
       break;
     case CHAR_MAX + 1: // --help
       usage(stdout);
@@ -619,9 +619,8 @@ void read_symbol_sets(File &f)
   for (i = 0; i < nchars; i++)
     char_table[i].symbol_set = NO_SYMBOL_SET;
 
-  uint16 *symbol_set_selectors = (special_flag
-				  ? special_symbol_sets
-				  : text_symbol_sets);
+  uint16 *symbol_set_selectors = special_flag ? special_symbol_sets
+					      : text_symbol_sets;
   for (i = 0; symbol_set_selectors[i] != 0; i++) {
     unsigned int j;
     for (j = 0; j < n_symbol_sets; j++)
@@ -836,7 +835,7 @@ void output_charset(const int tfm_type)
   for (i = 0; i < nchars; i++) {
     uint16 charcode = char_table[i].charcode;
 
-    // the TFM supports the character
+    // the glyph is bound to one of the searched symbol sets
     if (char_table[i].symbol_set != NO_SYMBOL_SET) {
       // the character was in the map file
       if (charcode < charcode_name_table_size && charcode_name_table[charcode])
@@ -899,9 +898,8 @@ void output_charset(const int tfm_type)
       }
       else
 	printf("\t\t-- MSL %4d", charcode);
-      int symset = char_table[i].symbol_set;
-      printf(" (%2d%c %3d)\n",
-	     symset / 32, (symset & 31) + 64, char_table[i].code);
+      printf(" (%3s %3d)\n",
+	     show_symset(char_table[i].symbol_set), char_table[i].code);
 
       if (charcode < charcode_name_table_size
 	  && charcode_name_table[charcode])
@@ -925,7 +923,7 @@ void output_charset(const int tfm_type)
 	fprintf(stderr, ") not in any searched symbol set\n");
       }
       else if (!quiet_flag && !equal(name, UNNAMED) && !is_uname(name)) {
-	fprintf(stderr, "%s:warning: symbol MSL %d (%s",
+	fprintf(stderr, "%s: warning: symbol MSL %d (%s",
 		program_name, charcode, name);
 	for (name_list *p = charcode_name_table[charcode]->next;
 	     p; p = p->next)
@@ -1204,11 +1202,10 @@ void dump_symbols(int tfm_type)
     if (charcode < charcode_name_table_size
 	&& charcode_name_table[charcode]) {
       if (char_table[i].symbol_set != NO_SYMBOL_SET) {
-	printf(tfm_type == UNICODE ? "%4d (U+%04X)   (%2d%c %3d)  %s"
-				   : "%4d (MSL %4d) (%2d%c %3d)  %s",
+	printf(tfm_type == UNICODE ? "%4d (U+%04X)   (%3s %3d)  %s"
+				   : "%4d (MSL %4d) (%3s %3d)  %s",
 	       i, charcode,
-	       char_table[i].symbol_set / 32,
-	       (char_table[i].symbol_set % 32) + 64,
+	       show_symset(char_table[i].symbol_set),
 	       char_table[i].code,
 	       charcode_name_table[charcode]->name);
 	for (name_list *p = charcode_name_table[charcode]->next;
@@ -1217,14 +1214,58 @@ void dump_symbols(int tfm_type)
 	putchar('\n');
       }
     }
-    else
-      printf(tfm_type == UNICODE ? "%4d (U+%04X)\n"
-				 : "%4d (MSL %4d)\n",
+    else {
+      printf(tfm_type == UNICODE ? "%4d (U+%04X)   "
+				 : "%4d (MSL %4d) ",
 	     i, charcode);
+      uint16 symset, code;
+      if (char_table[i].symbol_set != NO_SYMBOL_SET)
+	printf("(%3s %3d)",
+	       show_symset(char_table[i].symbol_set), char_table[i].code);
+      else if (get_printcode(i, &symset, &code))
+	printf("[%3s %3d]  (set not searched)", show_symset(symset), code);
+      putchar('\n');
+    }
   }
 }
 
 static char *
+show_symset(unsigned int symset)
+{
+   static char symset_str[8];
+
+   sprintf(symset_str, "%d%c", symset / 32, (symset & 31) + 64);
+   return symset_str;
+}
+
+// search symbol sets opposite the state of the 's' option
+static int
+get_printcode(uint32 index, uint16 *symset, uint16 *code)
+{
+  uint16 *symbol_set_selectors = special_flag ? text_symbol_sets
+					      : special_symbol_sets;
+  for (unsigned int i = 0; symbol_set_selectors[i] != 0; i++) {
+    unsigned int j;
+    for (j = 0; j < n_symbol_sets; j++)
+      if (symbol_set_table[j].select == symbol_set_selectors[i])
+	break;
+    if (j < n_symbol_sets) {
+      for (unsigned int k = 0; k < 256; k++) {
+	uint16 ndx = symbol_set_table[j].index[k];
+	if (ndx == index) {
+	  *symset = symbol_set_table[j].select;
+	  *code = k;
+	  return 1;
+	}
+      }
+    }
+  }
+
+  // not found in text or special symbol sets
+  return 0;
+}
+
+static char *
 hp_msl_to_ucode_name(int msl)
 {
   char codestr[8];
@@ -1367,6 +1408,9 @@ int read_map(const char *file, const int tfm_type)
       fclose(fp);
       return 0;
     }
+    // leave decomposed Unicode values alone
+    else if (is_uname(ptr) && !is_decomposed(ptr))
+      ptr = unicode_to_ucode_name(strtol(ptr + 1, &nonum, 16));
 
     if (size_t(n) >= charcode_name_table_size) {
       size_t old_size = charcode_name_table_size;