start of grammar parsing

1 files changed, 149 insertions, 0 deletions

diff --git a/demos/sample-files/gram-skeleton.c b/demos/sample-files/gram-skeleton.c
new file mode 100644
index 0000000..89ef6b4
--- /dev/null
+++ b/demos/sample-files/gram-skeleton.c
@@ -0,0 +1,149 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <ctype.h>
+
+#define ARENA_IMPLEMENTATION
+#include "util/arena.h"
+
+static char buf[1024];
+static struct arena_ctx global_arena = ARENA_CTX_INIT(buf, sizeof(buf));
+static void *xalloc(size_t sz) {
+    void *addr = arena_allocate(&global_arena, sz);
+    if(!addr) {
+        fprintf(stderr, "ERROR: Arena empty\n"); exit(1);
+    }
+
+    return addr;
+}
+
+// generated
+#include "bin/gram.h"
+#include "bin/gram.c"
+
+#include "parts/toklist.h"
+struct token {
+    symbol s;
+    intptr_t v;
+} tok;
+
+static char *next_token(char *str);
+
+symbol token_sym(struct token *t) { return t->s; }
+intptr_t token_val(struct token *t) { return t->v; }
+
+static char *input = (char []){
+    "-left B;"
+    "-right C;"
+    "-left D;"
+    ""
+    "A: B {a}"
+    " | C N {d}."
+};
+
+struct token *toklist_eat()
+{
+    static struct token t;
+    t = tok;
+    input = next_token(input);
+    return &t;
+}
+
+struct token *toklist_peek() { return &tok; }
+
+#include "lr-parser.c"
+
+int main(void)
+{
+    input = next_token(input);
+
+    intptr_t value;
+    if(lr_parser(&value)) {
+        return 1;
+    }
+
+    printf("OUTPUT: %jd\n", value);
+    return 0;
+}
+
+// STR UTIL
+
+#define strdup(...) _strdup(__VA_ARGS__)
+static inline char *_strdup(char *str)
+{
+    return memcpy(xalloc(strlen(str) + 1), str, strlen(str)+1);
+}
+
+static inline char *substring(char *str, size_t sub_end)
+{
+    static char sub[128];
+    if(sub_end+1 > sizeof(sub)) return NULL;
+
+    sub[sub_end] = '\0';
+    return memcpy(sub, str, sub_end);
+}
+
+// LEXER
+
+static inline int issep(char c)
+{
+    return isspace(c) || c == '\0' || c == ':' || c == '|' || c == ';' || c == '.' || c == '-' || c == '{';
+}
+
+static inline int tillsep(char *str)
+{
+    size_t i = 0;
+    while(!issep(str[i++]));
+    return i-1;
+}
+
+static char *next_token(char *str)
+{
+    if(!str) return str;
+
+    size_t off = 0;
+    char c0 = str[0];
+
+    if(c0 == '\0') tok.s = END_INPUT;
+    if(isspace(c0)) return next_token(str+1);
+    else {
+        off = tillsep(str);
+        if(off == 0) { // sep
+            switch(str[off++]) {
+            case ':': tok.s = COLON; break;
+            case '|': tok.s = PIPE; break;
+            case ';': tok.s = SEMICOL; break;
+            case '.': tok.s = DOT; break;
+            case '-':
+                off = tillsep(++str);
+                char *s = substring(str, off);
+                if(strcmp(s, "left") == 0) tok.s = D_LEFT;
+                else if(strcmp(s, "right") == 0) tok.s = D_RIGHT;
+                else if(strcmp(s, "terminal") == 0) tok.s = D_TERMINAL;
+                else if(strcmp(s, "nonterminal") == 0) tok.s = D_NONTERM;
+                else { fprintf(stderr, "ERROR: Unknown directive '-%s'\n", s); goto fail; }
+                break;
+            case '{':
+                for(int c = 1; c != 0; off++)
+                    if(str[off] == '\0') { fprintf(stderr, "ERROR: No closing '{'\n"); goto fail; }
+                    else if(str[off] == '{') c++;
+                    else if(str[off] == '}') c--;
+                tok.s = ACTION;
+                tok.v = (intptr_t)strdup(substring(str, off));
+                break;
+            }
+        } else if(isalpha(c0)) { // iden or named symbol
+            tok.s = IDEN;
+            tok.v = (intptr_t)strdup(substring(str, off));
+        } else if(c0 >= '0' && c0 <= '9') { // num
+            tok.s = NUM;
+            tok.v = (intptr_t)atoi(substring(str, off));
+        }
+    }
+
+    return str+off;
+
+fail:
+    tok.s = END_INPUT;
+    return NULL;
+}