lexer: Add token for numbers

Signed-off-by: Ian Moffett <ian@mirocom.org>

core/lexer.c

@@ -283,6 +283,82 @@ lexer_assert_preproc(struct rifle_state *state, struct token *tok)
     return -1;
 }
 
+/*
+ * Scan for a series of numbers
+ *
+ * @state: Compiler state
+ * @lc:    Last character
+ * @tok:   Token result
+ */
+static int
+lexer_scan_num(struct rifle_state *state, int lc, struct token *tok)
+{
+    char c, buf[22];
+    char prefix = '\0';
+    size_t buf_ind;
+    uint8_t radix;
+
+    if (state == NULL || tok == NULL) {
+        return -1;
+    }
+
+    if (!isdigit(lc)) {
+        return -1;
+    }
+
+    /* Obtain the prefix if any */
+    if (lc == '0') {
+        if ((prefix = lexer_nom(state, false)) == '\0')
+            return -1;
+        if ((lc = lexer_nom(state, false)) == '\0')
+            return -1;
+        if (!isxdigit(lc))
+            return -1;
+    }
+
+    /*
+     * Determine the radix based on the prefix
+     *
+     * 'x'): Base-16
+     * 'o'): Base-8
+     */
+    switch (prefix) {
+    case 'x':
+        radix = 16;
+        break;
+    case 'o':
+        radix = 8;
+        break;
+    default:
+        radix = 10;
+    }
+
+    buf_ind = 0;
+    buf[buf_ind++] = lc;
+
+    for (;;) {
+        c = lexer_nom(state, true);
+        if (c == '\0') {
+            return -1;
+        }
+
+        if (!isxdigit(c) && c != '_') {
+            buf[buf_ind] = '\0';
+            break;
+        }
+
+        buf[buf_ind++] = c;
+        if (buf_ind >= sizeof(buf) - 1) {
+            trace_error(state, "value exceeds width of u64\n");
+            return -1;
+        }
+    }
+
+    tok->type = TT_NUMBER;
+    tok->v = strtoll(buf, NULL, radix);
+    return 0;
+}
+
 int
 lexer_scan(struct rifle_state *state, struct token *res)
 {
@@ -351,6 +427,11 @@ lexer_scan(struct rifle_state *state, struct token *res)
 
         return 0;
     default:
+        if (isdigit(c)) {
+            if (lexer_scan_num(state, c, res) == 0)
+                return 0;
+        }
+
         if (lexer_scan_ident(state, c, res) == 0) {
             lexer_check_kw(state, res);
             return 0;

core/parser.c

@@ -46,6 +46,7 @@
 static const char *toktab[] = {
     [TT_NONE]   = symtok("none"),
     [TT_IDENT]  = symtok("ident"),
+    [TT_NUMBER] = symtok("number"),
     [TT_PLUS]   = qtok("+"),
     [TT_MINUS]  = qtok("-"),
     [TT_STAR]   = qtok("*"),

inc/rifle/token.h

@@ -7,6 +7,7 @@
 typedef enum {
     TT_NONE,    /* <NONE> */
     TT_IDENT,   /* <IDENT> */
+    TT_NUMBER,  /* <NUMBER> */
     TT_PLUS,    /* '+' */
     TT_MINUS,   /* '-' */
     TT_STAR,    /* '*' */
@@ -36,6 +37,7 @@ struct token {
     union {
         char c;
         char *s;
+        ssize_t v;
     };
 };