Allow chains in any form like <name1><name2> or \xf1\xf2, not binary

representation only.

3 files changed, 37 insertions, 37 deletions

diff --git a/usr.bin/colldef/colldef.1 b/usr.bin/colldef/colldef.1
index 47a7584..8136b4f 100644
--- a/usr.bin/colldef/colldef.1
+++ b/usr.bin/colldef/colldef.1
@@ -175,15 +175,12 @@ The symbol itself (for example,
 for the lower-case letter
 .Ar a )
 .It
-The symbol chain (for example,
-.Ar abc )
-.It
-In octal representation (for example,
+The symbol in octal representation (for example,
 .Ar \e141
 for the letter
 .Ar a )
 .It
-In hexadecimal representation (for example,
+The symbol in hexadecimal representation (for example,
 .Ar \ex61
 for the letter
 .Ar a )
@@ -191,9 +188,9 @@ for the letter
 The symbol name as defined in the
 .Ar charmap
 file (for example,
-.Ar <abc>
+.Ar <letterA>
 for
-.Ar abc \e023
+.Ar letterA \e023
 record in
 .Ar charmapfile ) .
 If character map name have
@@ -214,6 +211,11 @@ Symbols
 .Ar \ev
 are permitted in its usual C-language meaning
 .It
+The symbol chain (for example:
+.Ar abc ,
+.Ar <letterA><letterB> ,
+.Ar \exf1\exf2 )
+.It
 The symbol range (for example,
 .Ar a;...;z )
 .It
diff --git a/usr.bin/colldef/parse.y b/usr.bin/colldef/parse.y
index d3abb18..3d5f7bc 100644
--- a/usr.bin/colldef/parse.y
+++ b/usr.bin/colldef/parse.y
@@ -47,6 +47,7 @@ static void usage(void);
 static void collate_print_tables(void);
 
 char map_name[FILENAME_MAX] = ".";
+char curr_chain[STR_LEN];
 
 char __collate_version[STR_LEN];
 u_char charmap_table[UCHAR_MAX + 1][CHARMAP_SYMBOL_LEN];
@@ -72,7 +73,6 @@ const char *out_file = "LC_COLLATE";
 }
 %token SUBSTITUTE WITH ORDER RANGE
 %token <str> STRING
-%token <str> CHAIN
 %token <str> DEFN
 %token <ch> CHAR
 %%
@@ -139,17 +139,33 @@ order : ORDER order_list {
 order_list : item
 	| order_list ';' item
 ;
+chain : CHAR CHAR {
+	curr_chain[0] = $1;
+	curr_chain[1] = $2;
+	if (curr_chain[0] == '\0' || curr_chain[1] == '\0')
+		yyerror("\\0 can't be chained");
+	curr_chain[2] = '\0';
+}
+	| chain CHAR {
+	static char tb[2];
+
+	tb[0] = $2;
+	if (tb[0] == '\0')
+		yyerror("\\0 can't be chained");
+	if (strlen(curr_chain) + 2 > STR_LEN)
+		yyerror("Chain '%s' grows too long", curr_chain);
+	(void)strcat(curr_chain, tb);
+}
+;
 item :  CHAR {
 	if (__collate_char_pri_table[$1].prim)
 		yyerror("Char 0x%02x duplicated", $1);
 	__collate_char_pri_table[$1].prim = prim_pri++;
 }
-	| CHAIN {
+	| chain {
 	if (chain_index >= TABLE_SIZE - 1)
 		yyerror("__collate_chain_pri_table overflow");
-	if (strlen($1) + 1 > STR_LEN)
-		yyerror("Chain %d is too long", chain_index);
-	strcpy(__collate_chain_pri_table[chain_index].str, $1);
+	(void)strcpy(__collate_chain_pri_table[chain_index].str, curr_chain);
 	__collate_chain_pri_table[chain_index++].prim = prim_pri++;
 }
 	| CHAR RANGE CHAR {
@@ -196,12 +212,10 @@ prim_sub_item : CHAR {
 		__collate_char_pri_table[(u_char)i].prim = prim_pri;
 	}
 }
-	| CHAIN {
+	| chain {
 	if (chain_index >= TABLE_SIZE - 1)
 		yyerror("__collate_chain_pri_table overflow");
-	if (strlen($1) + 1 > STR_LEN)
-		yyerror("Chain %d is too long", chain_index);
-	strcpy(__collate_chain_pri_table[chain_index].str, $1);
+	(void)strcpy(__collate_chain_pri_table[chain_index].str, curr_chain);
 	__collate_chain_pri_table[chain_index++].prim = prim_pri;
 }
 ;
@@ -225,12 +239,10 @@ sec_sub_item : CHAR {
 		__collate_char_pri_table[(u_char)i].sec = sec_pri++;
 	}
 }
-	| CHAIN {
+	| chain {
 	if (chain_index >= TABLE_SIZE - 1)
 		yyerror("__collate_chain_pri_table overflow");
-	if (strlen($1) + 1 > STR_LEN)
-		yyerror("Chain %d is too long", chain_index);
-	strcpy(__collate_chain_pri_table[chain_index].str, $1);
+	(void)strcpy(__collate_chain_pri_table[chain_index].str, curr_chain);
 	__collate_chain_pri_table[chain_index].prim = prim_pri;
 	__collate_chain_pri_table[chain_index++].sec = sec_pri++;
 }
diff --git a/usr.bin/colldef/scan.l b/usr.bin/colldef/scan.l
index 10b17c8..aeeff14 100644
--- a/usr.bin/colldef/scan.l
+++ b/usr.bin/colldef/scan.l
@@ -63,7 +63,6 @@ int yylex(void);
 <INITIAL,nchar,subs>\\v      { yylval.ch = '\v'; return CHAR; }
 <INITIAL,nchar,subs>\\r      { yylval.ch = '\r'; return CHAR; }
 <INITIAL,nchar,subs>\\a      { yylval.ch = '\a'; return CHAR; }
-<INITIAL,nchar,subs>\\.      { yylval.ch = yytext[1]; return CHAR; }
 <subs2>\n               {
 	line_no++;
 	BEGIN(INITIAL);
@@ -90,28 +89,15 @@ int yylex(void);
 	yylval.ch = (u_char)v;
 	return CHAR;
 }
-<INITIAL,nchar,subs>\\x[0-9a-z]{2}   {
+<INITIAL,nchar,subs>\\x[0-9a-fA-F]{2}   {
 	u_int v;
 
 	sscanf(&yytext[2], "%x", &v);
 	yylval.ch = (u_char)v;
 	return CHAR;
 }
-<INITIAL>[^;,{}() \t\n"<]+      {
-	if(yyleng == 1) {
-		yylval.ch = *yytext;
-		return CHAR;
-	}
-	if(yyleng > BUFSIZE - 1)
-		errx(EX_UNAVAILABLE, "chain buffer overflow near line %u",
-		     line_no);
-	strcpy(yylval.str, yytext);
-	return CHAIN;
-}
-<nchar,subs>.                {
-	yylval.ch = *yytext;
-	return CHAR;
-}
+<INITIAL,nchar,subs>\\. { yylval.ch = yytext[1]; return CHAR; }
+<INITIAL,nchar,subs>.   { yylval.ch = *yytext; return CHAR; }
 <defn>^#.*\n            line_no++;
 <defn>[ \t]+            {
 	if (ptr == buf)