WIP - implementing b or u prefix for a string or a character literal

lang.txt

@@ -23,6 +23,9 @@
 		b"..." or B"..." for an byte string constant notation
 		u"..." or U"..." for an explicit unicode string constant notation?
 
+		#b[ ] byte array??
+		#[ ] normal array?
+
 	make basic branded types to an object if possible.
 	for example (#[10 20]:at 1)
 

lib/hcl-prv.h

@@ -632,6 +632,12 @@ struct hcl_flx_st_t
 };
 
 
+typedef struct hcl_flx_bu_t hcl_flx_bu_t;
+struct hcl_flx_bu_t
+{
+	hcl_ooch_t start_c;
+};
+
 enum hcl_flx_state_t
 {
 	HCL_FLX_START,
@@ -645,7 +651,8 @@ enum hcl_flx_state_t
 	HCL_FLX_PLAIN_IDENT,    /* plain identifier */
 	HCL_FLX_PLAIN_NUMBER,   /* plain number */
 	HCL_FLX_QUOTED_TOKEN,   /* string, character */
-	HCL_FLX_SIGNED_TOKEN    /* prefixed with + or - */
+	HCL_FLX_SIGNED_TOKEN,   /* prefixed with + or - */
+	HCL_FLX_BU              /* beginning with B or U */
 };
 typedef enum hcl_flx_state_t hcl_flx_state_t;
 
@@ -735,6 +742,7 @@ struct hcl_compiler_t
 				hcl_flx_pn_t pn; /* plain number */
 				hcl_flx_qt_t qt; /* quoted token */
 				hcl_flx_st_t st; /* signed token */
+				hcl_flx_st_t bu; /* b or u prefix */
 			} u;
 		} lx;
 

lib/read.c

@@ -1719,6 +1719,7 @@ static int feed_continue_with_char (hcl_t* hcl, hcl_ooci_t c, hcl_flx_state_t st
 #define FLX_PN(hcl) (&((hcl)->c->feed.lx.u.pn))
 #define FLX_QT(hcl) (&((hcl)->c->feed.lx.u.qt))
 #define FLX_ST(hcl) (&((hcl)->c->feed.lx.u.st))
+#define FLX_BU(hcl) (&((hcl)->c->feed.lx.u.bu))
 
 static HCL_INLINE void init_flx_hc (hcl_flx_hc_t* hc)
 {
@@ -1765,6 +1766,12 @@ static HCL_INLINE void init_flx_st (hcl_flx_st_t* st, hcl_ooch_t sign_c)
 	st->sign_c = sign_c;
 }
 
+static HCL_INLINE void init_flx_bu (hcl_flx_bu_t* bu, hcl_ooch_t start_c)
+{
+	HCL_MEMSET (bu, 0, HCL_SIZEOF(*bu));
+	bu->start_c = start_c;
+}
+
 static void reset_flx_token (hcl_t* hcl)
 {
 	/* clear the token name, reset its location */
@@ -1852,6 +1859,14 @@ static int flx_start (hcl_t* hcl, hcl_ooci_t c)
 			FEED_CONTINUE (hcl, HCL_FLX_PLAIN_NUMBER);
 			goto not_consumed;
 
+		case 'B':
+		case 'b':
+		case 'U':
+		case 'u':
+			init_flx_bu(FLX_BU(hcl), c);
+			FEED_CONTINUE_WITH_CHAR(hcl, c, HCL_FLX_BU);
+			goto consumed;
+
 		default:
 		/* TODO: limit the identifier characters and cause syntax error for other characters.. */
 			init_flx_pi (FLX_PI(hcl));
@@ -2564,6 +2579,47 @@ not_consumed:
 	return 0;
 }
 
+static int flx_bu (hcl_t* hcl, hcl_ooci_t c)
+{
+	hcl_flx_bu_t* bu = FLX_BU(hcl);
+
+	if (c == '\"')
+	{
+/* TODO: determine type based on the start_c */
+		reset_flx_token (hcl);
+		init_flx_qt (FLX_QT(hcl), HCL_TOK_STRLIT, HCL_SYNERR_STRLIT, c, '\\', 0, HCL_TYPE_MAX(hcl_oow_t));
+		FEED_CONTINUE (hcl, HCL_FLX_QUOTED_TOKEN); /* discard prefix, quote and move on */
+		goto consumed;
+	}
+	else if (c == '\'')
+	{
+/* TODO: determine type based on the start_c */
+		reset_flx_token (hcl);
+		init_flx_qt (FLX_QT(hcl), HCL_TOK_CHARLIT, HCL_SYNERR_CHARLIT, c, '\\', 1, 1);
+		FEED_CONTINUE (hcl, HCL_FLX_QUOTED_TOKEN); /* dicard prefix, quote, and move on */
+		goto consumed;
+	}
+	else
+	{
+		/* not followed by a quote. switch to the plain identifier */
+		init_flx_pi (FLX_PI(hcl));
+
+		/* the prefix is already in the token buffer. just adjust state data */
+		FLX_PI(hcl)->char_count++;
+		FLX_PI(hcl)->seg_len++;
+
+		/* refeed c */
+		FEED_CONTINUE (hcl, HCL_FLX_PLAIN_IDENT);
+		goto not_consumed;
+	}
+
+consumed:
+	return 1;
+
+not_consumed:
+	return 0;
+}
+
 /* ------------------------------------------------------------------------ */
 
 static int feed_char (hcl_t* hcl, hcl_ooci_t c)
@@ -2583,6 +2639,7 @@ static int feed_char (hcl_t* hcl, hcl_ooci_t c)
 		case HCL_FLX_PLAIN_NUMBER:     return flx_plain_number(hcl, c);
 		case HCL_FLX_QUOTED_TOKEN:     return flx_quoted_token(hcl, c);
 		case HCL_FLX_SIGNED_TOKEN:     return flx_signed_token(hcl, c);
+		case HCL_FLX_BU:               return flx_bu(hcl, c);
 
 		default:
 			/* unknown state */

t/feed-01.hcl

@@ -1,4 +1,4 @@
-## this file is to test the reader/feeder againsst weirdly formatted input text.
+## this file is to test the reader/feeder against weirdly formatted input text.
 
 {  ## START
 
@@ -7,7 +7,7 @@
 defun xxx (x y z
 	::: r ) {
 
-	| k 
+	| k
 	b
 	s |
 
@@ -25,9 +25,9 @@ defun xxx (x y z
 [
 j
 ] \
-	:= (xxx 
+	:= (xxx
-	10 
+	10
-	20 
+	20
 	30)
 
 if (eqv? j 12960000) \