refactor: separate preprocess.{h,c}

1 files changed, 313 insertions, 0 deletions

diff --git a/src/token.c b/src/token.c
new file mode 100644
index 0000000..3ca34d8
--- /dev/null
+++ b/src/token.c
@@ -0,0 +1,313 @@
+#include "token.h"
+#include "common.h"
+
+const char* token_kind_stringify(TokenKind k) {
+    if (k == TokenKind_eof)
+        return "<eof>";
+    else if (k == TokenKind_hash)
+        return "#";
+    else if (k == TokenKind_hashhash)
+        return "##";
+    else if (k == TokenKind_whitespace)
+        return "<whitespace>";
+    else if (k == TokenKind_newline)
+        return "<new-line>";
+    else if (k == TokenKind_other)
+        return "<other>";
+    else if (k == TokenKind_character_constant)
+        return "<character-constant>";
+    else if (k == TokenKind_header_name)
+        return "<header-name>";
+    else if (k == TokenKind_pp_directive_define)
+        return "#define";
+    else if (k == TokenKind_pp_directive_elif)
+        return "#elif";
+    else if (k == TokenKind_pp_directive_elifdef)
+        return "#elifdef";
+    else if (k == TokenKind_pp_directive_elifndef)
+        return "#elifndef";
+    else if (k == TokenKind_pp_directive_else)
+        return "#else";
+    else if (k == TokenKind_pp_directive_embed)
+        return "#embed";
+    else if (k == TokenKind_pp_directive_endif)
+        return "#endif";
+    else if (k == TokenKind_pp_directive_error)
+        return "#error";
+    else if (k == TokenKind_pp_directive_if)
+        return "#if";
+    else if (k == TokenKind_pp_directive_ifdef)
+        return "#ifdef";
+    else if (k == TokenKind_pp_directive_ifndef)
+        return "#ifndef";
+    else if (k == TokenKind_pp_directive_include)
+        return "#include";
+    else if (k == TokenKind_pp_directive_line)
+        return "#line";
+    else if (k == TokenKind_pp_directive_pragma)
+        return "#pragma";
+    else if (k == TokenKind_pp_directive_undef)
+        return "#undef";
+    else if (k == TokenKind_pp_directive_warning)
+        return "#warning";
+    else if (k == TokenKind_pp_operator_defined)
+        return "defined";
+    else if (k == TokenKind_pp_operator___has_c_attribute)
+        return "__has_c_attribute";
+    else if (k == TokenKind_pp_operator___has_embed)
+        return "__has_embed";
+    else if (k == TokenKind_pp_operator___has_include)
+        return "__has_include";
+    else if (k == TokenKind_keyword_alignas)
+        return "alignas";
+    else if (k == TokenKind_keyword_alignof)
+        return "alignof";
+    else if (k == TokenKind_keyword_auto)
+        return "auto";
+    else if (k == TokenKind_keyword_bool)
+        return "bool";
+    else if (k == TokenKind_keyword_break)
+        return "break";
+    else if (k == TokenKind_keyword_case)
+        return "case";
+    else if (k == TokenKind_keyword_char)
+        return "char";
+    else if (k == TokenKind_keyword_const)
+        return "const";
+    else if (k == TokenKind_keyword_constexpr)
+        return "constexpr";
+    else if (k == TokenKind_keyword_continue)
+        return "continue";
+    else if (k == TokenKind_keyword_default)
+        return "default";
+    else if (k == TokenKind_keyword_do)
+        return "do";
+    else if (k == TokenKind_keyword_double)
+        return "double";
+    else if (k == TokenKind_keyword_else)
+        return "else";
+    else if (k == TokenKind_keyword_enum)
+        return "enum";
+    else if (k == TokenKind_keyword_extern)
+        return "extern";
+    else if (k == TokenKind_keyword_false)
+        return "false";
+    else if (k == TokenKind_keyword_float)
+        return "float";
+    else if (k == TokenKind_keyword_for)
+        return "for";
+    else if (k == TokenKind_keyword_goto)
+        return "goto";
+    else if (k == TokenKind_keyword_if)
+        return "if";
+    else if (k == TokenKind_keyword_inline)
+        return "inline";
+    else if (k == TokenKind_keyword_int)
+        return "int";
+    else if (k == TokenKind_keyword_long)
+        return "long";
+    else if (k == TokenKind_keyword_nullptr)
+        return "nullptr";
+    else if (k == TokenKind_keyword_register)
+        return "register";
+    else if (k == TokenKind_keyword_restrict)
+        return "restrict";
+    else if (k == TokenKind_keyword_return)
+        return "return";
+    else if (k == TokenKind_keyword_short)
+        return "short";
+    else if (k == TokenKind_keyword_signed)
+        return "signed";
+    else if (k == TokenKind_keyword_sizeof)
+        return "sizeof";
+    else if (k == TokenKind_keyword_static)
+        return "static";
+    else if (k == TokenKind_keyword_static_assert)
+        return "static_assert";
+    else if (k == TokenKind_keyword_struct)
+        return "struct";
+    else if (k == TokenKind_keyword_switch)
+        return "switch";
+    else if (k == TokenKind_keyword_thread_local)
+        return "thread_local";
+    else if (k == TokenKind_keyword_true)
+        return "true";
+    else if (k == TokenKind_keyword_typedef)
+        return "typedef";
+    else if (k == TokenKind_keyword_typeof)
+        return "typeof";
+    else if (k == TokenKind_keyword_typeof_unqual)
+        return "typeof_unqual";
+    else if (k == TokenKind_keyword_union)
+        return "union";
+    else if (k == TokenKind_keyword_unsigned)
+        return "unsigned";
+    else if (k == TokenKind_keyword_void)
+        return "void";
+    else if (k == TokenKind_keyword_volatile)
+        return "volatile";
+    else if (k == TokenKind_keyword_while)
+        return "while";
+    else if (k == TokenKind_keyword__Atomic)
+        return "_Atomic";
+    else if (k == TokenKind_keyword__BitInt)
+        return "_BitInt";
+    else if (k == TokenKind_keyword__Complex)
+        return "_Complex";
+    else if (k == TokenKind_keyword__Decimal128)
+        return "_Decimal128";
+    else if (k == TokenKind_keyword__Decimal32)
+        return "_Decimal32";
+    else if (k == TokenKind_keyword__Decimal64)
+        return "_Decimal64";
+    else if (k == TokenKind_keyword__Generic)
+        return "_Generic";
+    else if (k == TokenKind_keyword__Imaginary)
+        return "_Imaginary";
+    else if (k == TokenKind_keyword__Noreturn)
+        return "_Noreturn";
+    else if (k == TokenKind_and)
+        return "&";
+    else if (k == TokenKind_andand)
+        return "&&";
+    else if (k == TokenKind_arrow)
+        return "->";
+    else if (k == TokenKind_assign)
+        return "=";
+    else if (k == TokenKind_assign_add)
+        return "+=";
+    else if (k == TokenKind_assign_and)
+        return "&=";
+    else if (k == TokenKind_assign_div)
+        return "/=";
+    else if (k == TokenKind_assign_lshift)
+        return "<<=";
+    else if (k == TokenKind_assign_mod)
+        return "%=";
+    else if (k == TokenKind_assign_mul)
+        return "*=";
+    else if (k == TokenKind_assign_or)
+        return "|=";
+    else if (k == TokenKind_assign_rshift)
+        return ">>=";
+    else if (k == TokenKind_assign_sub)
+        return "-=";
+    else if (k == TokenKind_assign_xor)
+        return "^=";
+    else if (k == TokenKind_brace_l)
+        return "{";
+    else if (k == TokenKind_brace_r)
+        return "}";
+    else if (k == TokenKind_bracket_l)
+        return "[";
+    else if (k == TokenKind_bracket_r)
+        return "]";
+    else if (k == TokenKind_colon)
+        return ":";
+    else if (k == TokenKind_comma)
+        return ",";
+    else if (k == TokenKind_dot)
+        return ".";
+    else if (k == TokenKind_ellipsis)
+        return "...";
+    else if (k == TokenKind_eq)
+        return "==";
+    else if (k == TokenKind_ge)
+        return ">=";
+    else if (k == TokenKind_gt)
+        return ">";
+    else if (k == TokenKind_ident)
+        return "<identifier>";
+    else if (k == TokenKind_le)
+        return "le";
+    else if (k == TokenKind_literal_int)
+        return "<integer>";
+    else if (k == TokenKind_literal_str)
+        return "<string>";
+    else if (k == TokenKind_lshift)
+        return "<<";
+    else if (k == TokenKind_lt)
+        return "lt";
+    else if (k == TokenKind_minus)
+        return "-";
+    else if (k == TokenKind_minusminus)
+        return "--";
+    else if (k == TokenKind_ne)
+        return "!=";
+    else if (k == TokenKind_not)
+        return "!";
+    else if (k == TokenKind_or)
+        return "|";
+    else if (k == TokenKind_oror)
+        return "||";
+    else if (k == TokenKind_paren_l)
+        return "(";
+    else if (k == TokenKind_paren_r)
+        return ")";
+    else if (k == TokenKind_percent)
+        return "%";
+    else if (k == TokenKind_plus)
+        return "+";
+    else if (k == TokenKind_plusplus)
+        return "++";
+    else if (k == TokenKind_question)
+        return "?";
+    else if (k == TokenKind_rshift)
+        return ">>";
+    else if (k == TokenKind_semicolon)
+        return ";";
+    else if (k == TokenKind_slash)
+        return "/";
+    else if (k == TokenKind_star)
+        return "*";
+    else if (k == TokenKind_tilde)
+        return "~";
+    else if (k == TokenKind_xor)
+        return "^";
+    else
+        unreachable();
+}
+
+const char* token_stringify(Token* t) {
+    TokenKind k = t->kind;
+    if (k == TokenKind_literal_int) {
+        const char* kind_str = token_kind_stringify(k);
+        char* buf = calloc(10 + strlen(kind_str) + 3 + 1, sizeof(char));
+        sprintf(buf, "%d (%s)", t->value.integer, kind_str);
+        return buf;
+    } else if (k == TokenKind_other || k == TokenKind_character_constant || k == TokenKind_ident ||
+               k == TokenKind_literal_int || k == TokenKind_literal_str) {
+        const char* kind_str = token_kind_stringify(k);
+        char* buf = calloc(strlen(t->value.string) + strlen(kind_str) + 3 + 1, sizeof(char));
+        sprintf(buf, "%s (%s)", t->value.string, kind_str);
+        return buf;
+    } else {
+        return token_kind_stringify(k);
+    }
+}
+
+void tokens_init(TokenArray* tokens, size_t capacity) {
+    tokens->len = 0;
+    tokens->capacity = capacity;
+    tokens->data = calloc(tokens->capacity, sizeof(Token));
+}
+
+void tokens_reserve(TokenArray* tokens, size_t size) {
+    if (size <= tokens->capacity)
+        return;
+    while (tokens->capacity < size) {
+        tokens->capacity *= 2;
+    }
+    tokens->data = realloc(tokens->data, tokens->capacity * sizeof(Token));
+    memset(tokens->data + tokens->len, 0, (tokens->capacity - tokens->len) * sizeof(Token));
+}
+
+Token* tokens_push_new(TokenArray* tokens) {
+    tokens_reserve(tokens, tokens->len + 1);
+    return &tokens->data[tokens->len++];
+}
+
+Token* tokens_pop(TokenArray* tokens) {
+    if (tokens->len != 0)
+        tokens->len--;
+}