added QSE_XLI_LEADDIGIT.

enhanced keyword and identifier scanning for xli

qse/cmd/xli/xli.c

@@ -135,6 +135,7 @@ static void print_usage (QSE_FILE* out, int argc, qse_char_t* argv[])
 	qse_fprintf (out, QSE_T(" -f                        keep file inclusion info\n"));
 	qse_fprintf (out, QSE_T(" -t                        keep comment text\n"));
 	qse_fprintf (out, QSE_T(" -s                        allow multi-segmented strings\n"));
+	qse_fprintf (out, QSE_T(" -d                        allow a leading digit in identifiers\n"));
 	qse_fprintf (out, QSE_T(" -v                        perform validation\n"));
 	qse_fprintf (out, QSE_T(" -m                 number specify the maximum amount of memory to use in bytes\n"));
 #if defined(QSE_BUILD_DEBUG)
@@ -162,9 +163,9 @@ static int handle_args (int argc, qse_char_t* argv[])
 	static qse_opt_t opt = 
 	{
 #if defined(QSE_BUILD_DEBUG)
-		QSE_T("hi:o:uaftsvm:X:"),
+		QSE_T("hi:o:uaftsdvm:X:"),
 #else
-		QSE_T("hi:o:uaftsvm:"),
+		QSE_T("hi:o:uaftsdvm:"),
 #endif
 		lng
 	};
@@ -226,6 +227,10 @@ static int handle_args (int argc, qse_char_t* argv[])
 				g_trait |= QSE_XLI_MULSEGSTR;
 				break;
 
+			case QSE_T('d'):
+				g_trait |= QSE_XLI_LEADDIGIT;
+				break;
+
 			case QSE_T('v'):
 				g_trait |= QSE_XLI_VALIDATE;
 				break;

qse/include/qse/xli/xli.h

@@ -50,6 +50,7 @@ enum qse_xli_errnum_t
 	QSE_XLI_ELXCHR,  /**< invalid character '${0} */
 	QSE_XLI_EXKWNR,  /**< @word '${0}' not recognized */
 	QSE_XLI_EXKWEM,  /**< @ not followed by a valid word  */
+	QSE_XLI_EIDENT,  /**< invalid identifier '${0}' */
 	QSE_XLI_EUDKEY,  /**< undefined key '${0}' */
 	QSE_XLI_ENOALI,  /**< no alias for '${0}' */
 	QSE_XLI_EILVAL,  /**< illegal value for '${0}' */
@@ -86,7 +87,8 @@ enum qse_xli_trait_t
 	QSE_XLI_KEEPFILE  = (1 << 4), /**< keep inclusion file info */
 
 	QSE_XLI_MULSEGSTR = (1 << 5), /**< support multi-segmented string */
-	QSE_XLI_VALIDATE  = (1 << 6)
+	QSE_XLI_LEADDIGIT = (1 << 6), /**< allow a leading digit in an identifier */
+	QSE_XLI_VALIDATE  = (1 << 7)
 };
 typedef enum qse_xli_trait_t qse_xli_trait_t;
 

qse/lib/xli/err.c

@@ -49,6 +49,7 @@ const qse_char_t* qse_xli_dflerrstr (
 		QSE_T("invalid character '${0}'"),
 		QSE_T("'${0}' not recognized"),
 		QSE_T("@ not followed by a valid word"),
+		QSE_T("invalid identifier '${0}'"),
 		QSE_T("undefined key '${0}'"),
 		QSE_T("no alias for '${0}'"),
 		QSE_T("illegal value for '${0}'"),

qse/lib/xli/read.c

@@ -104,6 +104,7 @@ struct kwent_t
 	int type; 
 };
 
+/* note that the keyword must start with @. */
 static kwent_t kwtab[] = 
 {
 	/* keep it sorted by the first field for binary search */
@@ -420,50 +421,84 @@ retry:
 	}	
 	else if (c == QSE_T('@'))
 	{
+		/* keyword/directive - start with @ */
+
 		int type;
 
 		ADD_TOKEN_CHAR (xli, tok, c);
 		GET_CHAR_TO (xli, c);
 
-		if (c != QSE_T('_') && !QSE_ISALPHA (c))
+		if (!QSE_ISALPHA (c))
 		{
-			/* this directive is empty, 
+			/* this directive is empty, not followed by a valid word */
-			 * not followed by a valid word */
 			qse_xli_seterror (xli, QSE_XLI_EXKWEM, QSE_NULL, &xli->tok.loc);
 			return -1;
 		}
 
-		/* expect normal identifier starting with an alphabet */
+		/* expect an identifier starting with an alphabet. the identifier 
+		 * forming a keyword/directory is composed of alphabets. */
 		do 
 		{
 			ADD_TOKEN_CHAR (xli, tok, c);
 			GET_CHAR_TO (xli, c);
 		} 
-		while (c == QSE_T('_') || c == QSE_T('-') || QSE_ISALNUM (c));
+		while (QSE_ISALPHA (c));
 
 		type = classify_ident (xli, QSE_STR_CSTR(tok->name));
 		if (type == TOK_IDENT)
 		{
-			/* this directive is not recognized */
+			/* this keyword/directive is not recognized */
 			qse_xli_seterror (xli, QSE_XLI_EXKWNR, QSE_STR_CSTR(xli->tok.name), &xli->tok.loc);
 			return -1;
 		}
 		SET_TOKEN_TYPE (xli, tok, type);
 	}
-	else if (c == QSE_T('_') || QSE_ISALPHA (c))
+	else if (c == QSE_T('_') || QSE_ISALPHA (c) || 
+	         (!(xli->tok_status & TOK_STATUS_ENABLE_NSTR) &&
+	          (xli->opt.trait & QSE_XLI_LEADDIGIT) && 
+	          QSE_ISDIGIT(c)))
 	{
-		int type;
+		int lead_digit = QSE_ISDIGIT(c);
+		int all_digits = 1;
 
-		/* identifier */
+		/* a normal identifier can be composed of wider varieties of characters
+		 * than a keyword/directive */
+		while (1)
+		{
+			ADD_TOKEN_CHAR (xli, tok, c);
+			GET_CHAR_TO (xli, c);
+
+			if (c == QSE_T('_') || c == QSE_T('-') || 
+			    c == QSE_T(':') || c == QSE_T('*') ||
+			    c == QSE_T('/') || QSE_ISALPHA (c)) 
+			{
+				all_digits = 0;
+			}
+			else if (lead_digit && QSE_ISDIGIT(c)) 
+			{
+				/* nothing to do */
+			}
+			else break;
+		} 
+
+		if (lead_digit && all_digits)
+		{
+			/* if an identifier begins with a digit, it must contain a non-digits character */
+			qse_xli_seterror (xli, QSE_XLI_EIDENT, QSE_STR_CSTR(xli->tok.name), &xli->tok.loc);
+			return -1;
+		}
+
+		SET_TOKEN_TYPE (xli, tok, TOK_IDENT);
+	}
+	else if ((xli->tok_status & TOK_STATUS_ENABLE_NSTR) && QSE_ISDIGIT(c))
+	{
+		SET_TOKEN_TYPE (xli, tok, TOK_NSTR);
 		do
 		{
 			ADD_TOKEN_CHAR (xli, tok, c);
 			GET_CHAR_TO (xli, c);
 		}
-		while (c == QSE_T('_') || c == QSE_T('-') || QSE_ISALNUM (c));
+		while (QSE_ISDIGIT(c));
-
-		type = classify_ident (xli, QSE_STR_CSTR(tok->name));
-		SET_TOKEN_TYPE (xli, tok, type);
 	}
 	else if (c == QSE_T('\''))
 	{
@@ -533,16 +568,6 @@ retry:
 			}
 		}
 	}
-	else if ((xli->tok_status & TOK_STATUS_ENABLE_NSTR) && QSE_ISDIGIT(c))
-	{
-		SET_TOKEN_TYPE (xli, tok, TOK_NSTR);
-		do
-		{
-			ADD_TOKEN_CHAR (xli, tok, c);
-			GET_CHAR_TO (xli, c);
-		}
-		while (QSE_ISDIGIT(c));
-	}
 	else
 	{
 		n = get_symbols (xli, c, tok);
@@ -667,6 +692,7 @@ static int read_pair (qse_xli_t* xli)
 		}
 	}
 
+	/* once the key name is read, enable the numeric string for a value */
 	xli->tok_status |= TOK_STATUS_ENABLE_NSTR;
 
 	if (get_token (xli) <= -1) goto oops;
@@ -705,7 +731,7 @@ static int read_pair (qse_xli_t* xli)
 		{
 			/* SCM_KEYALIAS is specified for this particular item. Let the alias be required. 
 			 * If KEYALIAS is globally specified with the specific one, it's optional. */
-			qse_xli_seterrnum (xli, QSE_XLI_ENOALI, &key); 
+			qse_xli_seterrnum (xli, QSE_XLI_ENOALI, (const qse_cstr_t*)&key); 
 			goto oops;
 		}
 	}
@@ -771,6 +797,8 @@ static int read_pair (qse_xli_t* xli)
 				goto oops;	
 			}
 
+			/* semicolon read. turn off NSTR */
+			xli->tok_status &= ~TOK_STATUS_ENABLE_NSTR;
 			if (get_token (xli) <= -1) goto oops;
 		}
 		else