From 16a3288af9d214d11feb0c1c3bdb7ecaca2bfdd0 Mon Sep 17 00:00:00 2001 From: _ov4 Date: Sat, 5 Sep 2026 14:45:12 +0000 Subject: refactor very simple tokenizer --- ast.cc | 255 ++--------------------------------------------------------------- 1 file changed, 6 insertions(+), 249 deletions(-) (limited to 'ast.cc') diff --git a/ast.cc b/ast.cc index e8a1c8c..a0af720 100644 --- a/ast.cc +++ b/ast.cc @@ -6,260 +6,17 @@ #include #include +#include "parser/shared.h" +#include "parser/ast.h" +#include "parser/enum_type.h" +#include "parser/token.h" + using namespace std; +using namespace ov4; constexpr int MAXLINE = 1024; -size_t ast_index = 0; - -enum enum_token_type -{ - TEXT, // foo - SINGLE_QUOTATION, // ' - DOUBLE_QUOTATION, // " - LOGIC_AND, // && - LOGIC_OR, // || - ASYNC, // & - PIPE, // | - SEMICOLON, // ; - EXPAND, // $ - LEFT_BRACKET, // ( - RIGHT_BRACKET, // ) -}; - -enum guard_type -{ - GUARD_OFF, - GUARD_WEAK, - GUARD_STRONG, -}; - -class T_token -{ -public: - T_token() = default; - - enum_token_type token_type; - string text; - guard_type guard = GUARD_OFF; - int bracket_depth = 0; -}; - -class T_ast -{ -public: - T_ast() = default; - - enum_token_type token_type; - string command_text; - size_t left; - size_t right; -}; - - - -vector ast; -vector token; - -bool is_symbol(char c) -{ - if (c == '&' || c == '|' || c == '(' || c == ')') - return true; - return false; -} - -bool is_symbol_same(string x, string y) -{ - string tmp = ""; - for (size_t i = 0; i < x.size(); i++) - { - if (x[i] != ' ') - tmp += x[i]; - } - return (tmp == y); -} - -// handle [prev, current) -void token_push(string s, size_t current, size_t prev, int bracket_depth, bool is_symbol_prev) -{ - T_token tmp; - string sub = s.substr(prev, current - prev); - cout << "substr " << sub << endl; - - tmp.bracket_depth = bracket_depth; - if (is_symbol_prev) - { - if (is_symbol_same(sub, "&&")) - { - cout << "&& symbol" << endl; - tmp.token_type = LOGIC_AND; - } - else if (is_symbol_same(sub, "||")) - { - cout << "|| symbol" << endl; - tmp.token_type = LOGIC_OR; - } - else if (is_symbol_same(sub, "&")) - { - cout << "& symbol" << endl; - tmp.token_type = ASYNC; - } - else if (is_symbol_same(sub, "|")) - { - cout << "| symbol" << endl; - tmp.token_type = PIPE; - } - else if (is_symbol_same(sub, "(")) - { - cout << "( symbol" << endl; - tmp.token_type = LEFT_BRACKET; - } - else if (is_symbol_same(sub, ")")) - { - cout << ") symbol" << endl; - tmp.token_type = RIGHT_BRACKET; - } - else - { - cerr << "ERROR: is symbol but not symbol" << endl; - return; - } - token.push_back(tmp); - return; - } - - tmp.token_type = TEXT; - tmp.text = sub; - token.push_back(tmp); -} - -void bracket_depth_changer(char c, int &bracket_depth) -{ - if (c == '(') - { - bracket_depth++; - println("changed depth: {}", bracket_depth); - return; - } - if (c == ')') - { - bracket_depth--; - println("changed depth: {}", bracket_depth); - return; - } -} -int tokenizer(string s) -{ - if (s.size() == 0) return -1; - - size_t len = s.size(); - size_t prev = 0; - size_t token_vector_index = 0; - bool is_in_symbol; - int bracket_depth = 0; - - if (is_symbol(s[0])) - is_in_symbol = true; - else - is_in_symbol = false; - for (size_t i = 0; i < len; i++) - { - println("---"); - println("prev {} current char {}", prev, s[i]); - - - // turn from symbol to text - if (is_in_symbol && !is_symbol(s[i])) - { - cout << s[i] <<" symbol -> text" << endl; - is_in_symbol = false; - token_push(s, i, prev, bracket_depth, true); - prev = i; - bracket_depth_changer(s[i], bracket_depth); - continue; - } - - // turn from text to symbol - if (!is_in_symbol && is_symbol(s[i])) - { - cout << s[i] << " text -> symbol" << endl; - is_in_symbol = true; - token_push(s, i, prev, bracket_depth, false); - prev = i; - bracket_depth_changer(s[i], bracket_depth); - continue; - } - - bracket_depth_changer(s[i], bracket_depth); - - // text continue - if (!is_in_symbol && !is_symbol(s[i])) - { - cout << s[i] <<" still text" << endl; - continue; - } - - // symbol continue - if (is_in_symbol && is_symbol(s[i])) - { - cout << s[i] <<" still symbol" << endl; - continue; - } - } - - if (bracket_depth != 0) - { - cerr << "Brackets are not paired" << endl; - return -1; - } - - // final round text - if (!is_symbol(s[len - 1])) - { - cout << s[len - 1] << " final text" << endl; - token_push(s, len, prev, 0, false); - } - - // final round symbol - if (is_symbol(s[len - 1])) - { - cout << s[len - 1] << " final symbol" << endl; - token_push(s, len, prev, 0, true); - } - return 0; -} - -size_t alloc_ast() -{ - ast.emplace_back(); - return ast_index++; -} - -void parse(size_t cmd_begin, size_t cmd_end, size_t ast_vec) -{ - ; -} -int get_pivot_order(enum_token_type x) -{ - switch(x) - { - case ASYNC: - case SEMICOLON: - return 1000; - - case LOGIC_AND: - case LOGIC_OR: - return 500; - - case PIPE: - return 100; - - default: - return 0; - } - return 0; -} int main() -- cgit v1.3.1-4-g156e