Add initial tokenizer and tests

2022-10-14 19:37:12 -04:00
parent 9cde08b910
commit 2f8bff9c3b
7 changed files with 516 additions and 3 deletions
--- a/test/test_parse.c
+++ b/test/test_parse.c
@@ -0,0 +1,263 @@
+#include <stdlib.h>
+#include <unity.h>
+
+#include "nihilispm.h"
+#include "nihilispm_internal.h"
+
+/* static struct nl_parse_result *result; */
+static struct nl_object *response;
+
+void setUp(void) {
+    response = NULL;
+}
+
+void tearDown(void) {
+    nl_object_delete(response);
+}
+
+/* void test_parse_null() { */
+/*     result = nl_parse("()"); */
+/*     TEST_ASSERT_EQUAL(result->result, 0); */
+/*     TEST_ASSERT_NOT_NULL(result->statement); */
+/*     TEST_ASSERT_NULL(result->statement->car); */
+/*     TEST_ASSERT_NULL(result->statement->cdr); */
+/* } */
+
+/* void test_parse_error() { */
+/*     result = nl_parse("("); */
+/*     TEST_ASSERT_EQUAL(result->result, 0); */
+/*     TEST_ASSERT_NULL(result->statement); */
+/* } */
+
+void test_token_next_empty_str() {
+    const char *input = "";
+    const char *curr = input;
+
+    response = nl_token_next(&curr);
+
+    TEST_ASSERT_NULL(response);
+    TEST_ASSERT_EQUAL('\0', *curr);
+}
+
+void test_token_next_only_whitespace() {
+    const char *input = " \n";
+    const char *curr = input;
+
+    response = nl_token_next(&curr);
+
+    TEST_ASSERT_NULL(response);
+    TEST_ASSERT_EQUAL('\0', *curr);
+}
+
+void test_token_next_lparen() {
+    const char *input = "(";
+    const char *curr = input;
+
+    response = nl_token_next(&curr);
+
+    TEST_ASSERT_NOT_NULL(response);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, response->type);
+    TEST_ASSERT_EQUAL(NL_TYPE_SYMBOL, response->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING("(", response->cell.car->string);
+    TEST_ASSERT_EQUAL('\0', *curr);
+}
+
+void test_token_next_rparen() {
+    const char *input = ")";
+    const char *curr = input;
+
+    response = nl_token_next(&curr);
+
+    TEST_ASSERT_NOT_NULL(response);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, response->type);
+    TEST_ASSERT_EQUAL(NL_TYPE_SYMBOL, response->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING(")", response->cell.car->string);
+    TEST_ASSERT_EQUAL('\0', *curr);
+}
+
+void test_token_next_lrparen() {
+    const char *input = "()";
+    const char *curr = input;
+
+    response = nl_token_next(&curr);
+
+    TEST_ASSERT_NOT_NULL(response);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, response->type);
+    TEST_ASSERT_EQUAL(NL_TYPE_SYMBOL, response->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING("(", response->cell.car->string);
+    TEST_ASSERT_EQUAL(')', *curr);
+
+    nl_object_delete(response);
+    response = nl_token_next(&curr);
+
+    TEST_ASSERT_NOT_NULL(response);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, response->type);
+    TEST_ASSERT_EQUAL(NL_TYPE_SYMBOL, response->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING(")", response->cell.car->string);
+    TEST_ASSERT_EQUAL('\0', *curr);
+}
+
+void test_token_next_string() {
+    const char *input = "\"foo\"";
+    const char *curr = input;
+
+    response = nl_token_next(&curr);
+
+    TEST_ASSERT_NOT_NULL(response);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, response->type);
+    TEST_ASSERT_EQUAL(NL_TYPE_STRING, response->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING("foo", response->cell.car->string);
+    TEST_ASSERT_EQUAL('\0', *curr);
+}
+
+void test_token_next_string_w_whitespace() {
+    const char *input = " \"foo\"  ";
+    const char *curr = input;
+
+    response = nl_token_next(&curr);
+
+    TEST_ASSERT_NOT_NULL(response);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, response->type);
+    TEST_ASSERT_EQUAL(NL_TYPE_STRING, response->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING("foo", response->cell.car->string);
+    TEST_ASSERT_EQUAL_STRING("  ", curr);
+}
+
+void test_token_next_symbol() {
+    const char *input = "foo";
+    const char *curr = input;
+
+    response = nl_token_next(&curr);
+
+    TEST_ASSERT_NOT_NULL(response);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, response->type);
+    TEST_ASSERT_EQUAL(NL_TYPE_SYMBOL, response->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING("foo", response->cell.car->string);
+    TEST_ASSERT_EQUAL_STRING("", curr);
+}
+
+void test_token_next_symbol_w_whitespace() {
+    const char *input = " foo  ";
+    const char *curr = input;
+
+    response = nl_token_next(&curr);
+
+    TEST_ASSERT_NOT_NULL(response);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, response->type);
+    TEST_ASSERT_EQUAL(NL_TYPE_SYMBOL, response->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING("foo", response->cell.car->string);
+    TEST_ASSERT_EQUAL_STRING("  ", curr);
+}
+
+void test_tokenize_empty_str() {
+    response = nl_tokenize("");
+
+    TEST_ASSERT_NULL(response);
+}
+
+void test_tokenize_nil() {
+    response = nl_tokenize("()");
+
+    TEST_ASSERT_NOT_NULL(response);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, response->type);
+
+    const struct nl_object *token1 = response;
+    const struct nl_object *token2 = response->cell.cdr;
+
+    TEST_ASSERT_EQUAL(NL_TYPE_SYMBOL, token1->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING("(", token1->cell.car->string);
+
+    TEST_ASSERT_NOT_NULL(token2);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, token2->type);
+    TEST_ASSERT_EQUAL(NL_TYPE_SYMBOL, token2->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING(")", token2->cell.car->string);
+
+    TEST_ASSERT_NULL(token2->cell.cdr);
+}
+
+void test_tokenize_statement() {
+    response = nl_tokenize("(foo)");
+
+    TEST_ASSERT_NOT_NULL(response);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, response->type);
+
+    const struct nl_object *token1 = response;
+    const struct nl_object *token2 = response->cell.cdr;
+
+    TEST_ASSERT_EQUAL(NL_TYPE_SYMBOL, token1->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING("(", token1->cell.car->string);
+
+    TEST_ASSERT_NOT_NULL(token2);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, token2->type);
+    TEST_ASSERT_EQUAL(NL_TYPE_SYMBOL, token2->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING("foo", token2->cell.car->string);
+
+    const struct nl_object *token3 = token2->cell.cdr;
+
+    TEST_ASSERT_NOT_NULL(token3);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, token3->type);
+    TEST_ASSERT_EQUAL(NL_TYPE_SYMBOL, token3->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING(")", token3->cell.car->string);
+}
+
+void test_parse_empty_str() {
+    response = nl_parse("");
+    TEST_ASSERT_NULL(response);
+}
+
+void test_parse_empty_str() {
+    response = nl_parse("foo");
+
+    TEST_ASSERT_NOT_NULL(response);
+    TEST_ASSERT_EQUAL(NL_TYPE_SYMBOL, response->type);
+    TEST_ASSERT_EQUAL_STRING(NL_TYPE_SYMBOL, "foo");
+}
+
+void test_parse_nil() {
+    response = nl_parse("()");
+
+    TEST_ASSERT_NOT_NULL(response);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, response->type);
+    TEST_ASSERT_NULL(response->cell.car);
+    TEST_ASSERT_NULL(response->cell.cdr);
+}
+
+void test_parse_list_1elem() {
+    response = nl_parse("(foo)");
+
+    TEST_ASSERT_NOT_NULL(response);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, response->type);
+
+    TEST_ASSERT_NOT_NULL(response->cell.car);
+    TEST_ASSERT_EQUAL(NL_TYPE_SYMBOL, response->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING("foo", response->cell.car->symbol);
+}
+
+void test_parse_list_2elem() {
+    response = nl_parse("(foo bar)");
+
+    TEST_ASSERT_NOT_NULL(response);
+    TEST_ASSERT_EQUAL(NL_TYPE_CELL, response->type);
+
+    TEST_ASSERT_NOT_NULL(response->cell.car);
+    TEST_ASSERT_EQUAL(NL_TYPE_SYMBOL, response->cell.car->type);
+    TEST_ASSERT_EQUAL_STRING("foo", response->cell.car->symbol);
+}
+
+
+int main(void) {
+    UNITY_BEGIN();
+    RUN_TEST(test_token_next_empty_str);
+    RUN_TEST(test_token_next_only_whitespace);
+    RUN_TEST(test_token_next_lparen);
+    RUN_TEST(test_token_next_rparen);
+    RUN_TEST(test_token_next_lrparen);
+    RUN_TEST(test_token_next_string);
+    RUN_TEST(test_token_next_string_w_whitespace);
+    RUN_TEST(test_token_next_symbol);
+    RUN_TEST(test_token_next_symbol_w_whitespace);
+    RUN_TEST(test_tokenize_empty_str);
+    RUN_TEST(test_tokenize_nil);
+    RUN_TEST(test_tokenize_statement);
+    return UNITY_END();
+}