parser: almost working

src/parser.c

@@ -37,34 +37,109 @@ char** tokenize(char* str) {
   return tokens;
 }
 
-sc_ast* sc_parse(char* input) {
-  char** tokens = tokenize(input);
-  sc_ast* parent = malloc(sizeof(sc_ast));
-  parent->tag = 0;
-  parent->value = NULL;
-  parent->children = NULL;
-  parent->n_children = 0;
+sc_ast* make_node(int tag, char* value) {
+  sc_ast* node = malloc(sizeof(sc_ast));
+  node->tag = tag;
+  node->value = value;
+  node->n_children = 0;
+  node->children = NULL;
 
-  while (*tokens) {
-    parent->n_children++;
+  return node;
+}
 
-    sc_ast* child = malloc(sizeof(sc_ast));
-    parent->children = realloc(parent->children, parent->n_children*sizeof(sc_ast*));
-    child->tag = 1;
-    child->value = *tokens;
-    child->children = NULL;
-    child->n_children = 0;
-    parent->children[parent->n_children-1] = child;
-    tokens++;
+sc_ast* make_pseudo() {
+  return make_node(PSEUDO, NULL);
+}
+
+sc_ast* make_list() {
+  return make_node(LIST, NULL);
+}
+
+sc_ast* make_int(char* value) {
+  return make_node(INT, value);
+}
+
+sc_ast* make_float(char* value) {
+  return make_node(FLOAT, value);
+}
+
+sc_ast* make_atom(char* value) {
+  return make_node(ATOM, value);
+}
+
+typedef struct parse_state {
+  char** tokens;
+  sc_ast* node;
+} parse_state;
+
+short is_numerical(char* inp) {
+  int i;
+  int len = strlen(inp);
+
+  for (i = 0; i < len; i++) if (!isdigit(inp[i])) return 0;
+
+  return 1;
+}
+
+sc_ast* read_token(char* inp) {
+  if (is_numerical(inp)) return make_int(inp);
+  return make_atom(inp);
+}
+
+parse_state* read_tokens(parse_state* state) {
+  char* token = state->tokens[0];
+  state->tokens++;
+  int n = state->node->n_children++;
+
+  state->node->children = realloc(state->node->children, n+1);
+
+  if (!strncmp(token, "(", 2)) {
+    sc_ast* list = make_list();
+    parse_state* nstate = malloc(sizeof(parse_state));
+    nstate->node = list;
+    nstate->tokens = state->tokens;
+
+    while (strncmp(nstate->tokens[0], ")", 2)) {
+      nstate = read_tokens(nstate);
     }
 
-  return parent;
+    state->node->children[n] = nstate->node;
+    state->tokens = nstate->tokens;
+    free(nstate);
+  } else {
+    state->node->children[n] = read_token(token);
+  }
+
+  return state;
+}
+
+sc_ast* sc_parse(char* input) {
+  char** tokens = tokenize(input);
+  parse_state* state = malloc(sizeof(parse_state));
+  state->tokens = tokens;
+  state->node = make_pseudo();
+
+  state =  read_tokens(state);
+
+  return state->node->children[0];
+}
+
+const char* tag_to_string(int x) {
+  switch(x) {
+    case PSEUDO: return "pseudo";
+    case ATOM: return "atom";
+    case STRING: return "string";
+    case INT: return "int";
+    case FLOAT: return "float";
+    case LIST: return "list";
+    default: {char* t=malloc(3); snprintf(t, 3, "%d", x); return t;}
+  }
 }
 
 void sc_ast_print(sc_ast* ast) {
   int i;
 
-  printf("(%d) %s (%lu)\n", ast->tag, ast->value, ast->value ? strlen(ast->value) : 0);
+  printf("%s: %s\n", tag_to_string(ast->tag), ast->value);
 
   for (i = 0; i < ast->n_children; ++i) sc_ast_print(ast->children[i]);
 }

src/parser.h

@@ -1,7 +1,18 @@
+#include <ctype.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 
+enum tag {
+  PSEUDO,
+  ATOM,
+  LIST,
+  INT,
+  FLOAT,
+  STRING,
+};
+
+
 typedef struct sc_ast {
   short tag;
   char* value;