rc/main2.c

/*
 * Mini C Interpreter
 * A concise recursive-descent interpreter in C.
 * Supports: int, char*, pointers, if/else, while, functions, printf.
 * Updates: Pointer declarations, improved printf, escape sequences.
 */

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <ctype.h>
#include <math.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <unistd.h>
#include <arpa/inet.h>

#define MAX_SRC 100000
#define MAX_TOK 10000
#define VAR_MAX 500
#define MEM_SIZE 10000
#define STR_POOL_SIZE 100000

// --- Token Types ---
enum {
    Num = 128, Str, Id, Int, Char, Else, If, While, Return, Printf,
    Assign, Eq, Ne, Lt, Gt, Le, Ge, Or, And
};

// --- Structures ---
typedef struct {
    int type;
    long val;    // Changed to long to hold pointers on 64-bit
    char *text;  
} Token;

typedef struct {
    char name[32];
    int type;
    int addr;
    int is_array;
} Symbol;

typedef struct {
    char name[32];
    int entry_point;
    int param_count;
} Func;

typedef long (*NativeFunc)(long*, int);

typedef struct {
    char name[32];
    NativeFunc func;
} NativeFuncDef;

// --- Globals ---
Token tokens[MAX_TOK];
int tk_idx = 0;
int pc = 0;
long memory[MEM_SIZE]; // Changed to long
int sp = 0;
int bp = 0;

Symbol locals[VAR_MAX];
int loc_cnt = 0;

Func funcs[100];
int func_cnt = 0;

NativeFuncDef native_funcs[100];
int native_func_cnt = 0;

char *src_code;

char str_pool[STR_POOL_SIZE];
int str_pool_idx = 0;

// --- Tokenizer ---
void tokenize(char *src) {
    char *s = src;
    while (*s) {
        if (isspace(*s)) { s++; continue; }
        
        Token *t = &tokens[tk_idx++];
        t->text = s;

        // Comments
        if (*s == '/' && *(s+1) == '/') {
            while (*s && *s != '\n') s++;
            tk_idx--; 
            continue;
        }

        // Keywords & Identifiers
        if (isalpha(*s)) {
            int len = 0;
            while (isalnum(s[len]) || s[len] == '_') len++;
            
            char buf[32];
            strncpy(buf, s, len); buf[len] = 0;
            
            if (!strcmp(buf, "int")) t->type = Int;
            else if (!strcmp(buf, "char")) t->type = Char;
            else if (!strcmp(buf, "if")) t->type = If;
            else if (!strcmp(buf, "else")) t->type = Else;
            else if (!strcmp(buf, "while")) t->type = While;
            else if (!strcmp(buf, "return")) t->type = Return;
            else if (!strcmp(buf, "printf")) t->type = Printf;
            else t->type = Id;
            
            t->val = len;
            s += len;
            continue;
        }

        // Numbers
        if (isdigit(*s)) {
            t->type = Num;
            t->val = strtol(s, &s, 10);
            continue;
        }

        // Strings
        if (*s == '"') {
            s++;
            t->type = Str;
            t->text = s;

            char *d = s;
            while (*s && *s != '"') {
                if (*s == '\\' && *(s+1) == 'n') {
                    *d++ = '\n'; s+=2;
                } else {
                    *d++ = *s++;
                }
            }
            if (*s == '"') s++;
            *d = 0;
            t->val = (long)(d - t->text);
            continue;
        }

 
       // Operators
        if (!strncmp(s, "==", 2)) { t->type = Eq; s += 2; continue; }
        if (!strncmp(s, "!=", 2)) { t->type = Ne; s += 2; continue; }
        if (!strncmp(s, "<=", 2)) { t->type = Le; s += 2; continue; }
        if (!strncmp(s, ">=", 2)) { t->type = Ge; s += 2; continue; }
        if (!strncmp(s, "&&", 2)) { t->type = And; s += 2; continue; }
        if (!strncmp(s, "||", 2)) { t->type = Or; s += 2; continue; }

        if (*s == '<') { t->type = Lt; s++; continue; }
        if (*s == '>') { t->type = Gt; s++; continue; }

        t->type = *s++; 
    }
    tokens[tk_idx].type = 0;
}

// --- Helpers ---
void error(char *msg) {
    printf("Error at token %d ('%c'): %s\n", pc, tokens[pc].type, msg);
    exit(1);
}

void match(int type) {
    if (tokens[pc].type == type) pc++;
    else error("Unexpected token");
}

int find_local(char *name, int len) {
    for (int i = loc_cnt - 1; i >= 0; i--) {
        if (!strncmp(locals[i].name, name, len) && locals[i].name[len] == 0)
            return i;
    }
    return -1;
}

int find_func(char *name, int len) {
    for (int i = 0; i < func_cnt; i++) {
        if (!strncmp(funcs[i].name, name, len) && funcs[i].name[len] == 0)
            return i;
    }
    return -1;
}

int find_native_func(char *name, int len) {
    for (int i = 0; i < native_func_cnt; i++) {
        if (!strncmp(native_funcs[i].name, name, len) && native_funcs[i].name[len] == 0)
            return i;
    }
    return -1;
}

void register_native_func(char *name, NativeFunc func) {
    NativeFuncDef *nf = &native_funcs[native_func_cnt++];
    strncpy(nf->name, name, 31);
    nf->name[31] = 0;
    nf->func = func;
}

// --- Forward Declarations ---
int is_string_ptr(long val);
long slice_string(long str_ptr, int start, int end);

// --- Parser ---
long expression();
void statement();

long factor() {
    Token *t = &tokens[pc];
    long val = 0;

    if (t->type == Num) {
        pc++;
        return t->val;
    } 
    else if (t->type == Str) {
        pc++;
        return (long)t->text; 
    }
    else if (t->type == '(') {
        pc++;
        val = expression();
        match(')');
        return val;
    }
    else if (t->type == Id) {
        if (tokens[pc + 1].type == '(') {
            int nf_idx = find_native_func(t->text, t->val);

            if (nf_idx != -1) {
                pc += 2;
                long args[10];
                int argc = 0;

                if (tokens[pc].type != ')') {
                    do {
                        args[argc++] = expression();
                    } while (tokens[pc].type == ',' && pc++);
                }
                match(')');

                return native_funcs[nf_idx].func(args, argc);
            }

            int f_idx = find_func(t->text, t->val);
            if (f_idx == -1) error("Unknown function");
            pc += 2;

            int old_bp = bp;
            long args[10];
            int argc = 0;

            if (tokens[pc].type != ')') {
                do {
                    args[argc++] = expression();
                } while (tokens[pc].type == ',' && pc++);
            }
            match(')');

            int ret_pc = pc;
            memory[sp] = bp; bp = sp++;
            memory[sp++] = ret_pc;
            for(int i=0; i<argc; i++) memory[sp++] = args[i];

            pc = funcs[f_idx].entry_point;
            statement();

            extern long ax;
            val = ax;

            sp = bp;
            bp = memory[sp];
            pc = ret_pc;
            return val;
        }
        else {
            int idx = find_local(t->text, t->val);
            if (idx == -1) error("Undefined variable");
            pc++;

            Symbol *sym = &locals[idx];

            if (tokens[pc].type == '[') {
                pc++;
                long start_or_index = expression();

                if (tokens[pc].type == ':') {
                    pc++;
                    long end = expression();
                    match(']');

                    long val = memory[sym->addr];
                    if (is_string_ptr(val)) {
                        return slice_string(val, start_or_index, end);
                    } else {
                        error("Slicing only works on strings");
                    }
                } else {
                    match(']');
                    return memory[sym->addr + start_or_index];
                }
            }

            if (sym->is_array) {
                return sym->addr;
            }

            return memory[sym->addr];
        }
    }
    return 0;
}

long ax = 0;

int is_string_ptr(long val) {
    return val > MEM_SIZE * 8;
}

long concat_strings(long ptr1, long ptr2) {
    char *s1 = (char*)ptr1;
    char *s2 = (char*)ptr2;
    char *result = &str_pool[str_pool_idx];

    int len1 = strlen(s1);
    int len2 = strlen(s2);

    if (str_pool_idx + len1 + len2 + 1 >= STR_POOL_SIZE) {
        error("String pool overflow");
    }

    strcpy(result, s1);
    strcat(result, s2);

    str_pool_idx += len1 + len2 + 1;

    return (long)result;
}

long slice_string(long str_ptr, int start, int end) {
    char *str = (char*)str_ptr;
    char *result = &str_pool[str_pool_idx];
    int str_len = strlen(str);

    if (start < 0) start = 0;
    if (end < 0) end = str_len;
    if (end > str_len) end = str_len;
    if (start > end) start = end;

    int length = end - start;

    if (str_pool_idx + length + 1 >= STR_POOL_SIZE) {
        error("String pool overflow");
    }

    strncpy(result, str + start, length);
    result[length] = 0;

    str_pool_idx += length + 1;
    return (long)result;
}

long unary() {
    if (tokens[pc].type == '*') { 
        pc++;
        int addr = unary(); 
        // CAUTION: Simplified access. 
        // In real C, dereferencing char* reads a byte, int* reads int.
        // Here, everything is 'long' cell in memory.
        // This works for the specific request but is not byte-perfect.
        // Since 'b' holds an address to 'src_code' (char*), reading memory[addr] is invalid
        // if addr points to 'src_code'. 'memory' array is only for stack variables.
        // To support reading chars from string literals:
        
        // Hack for reading string characters:
        // If addr is outside virtual memory range, assume it's a pointer to raw C memory (string literal)
        // This is a bit unsafe but needed for "char *b" where b points to tokens.
        if (addr > MEM_SIZE * 8 || addr < 0) { // Rough heuristic for external pointer
             return *(char*)addr; 
        }
        return memory[addr]; 
    }
    else if (tokens[pc].type == '&') {
        pc++;
        Token *t = &tokens[pc];
        if (t->type != Id) error("Expected identifier after &");
        int idx = find_local(t->text, t->val);
        if (idx == -1) error("Undefined variable");
        pc++;
        return locals[idx].addr;
    }
    else if (tokens[pc].type == '-') {
        pc++;
        return -unary();
    }
    return factor();
}

long term() {
    long val = unary();
    while (tokens[pc].type == '*' || tokens[pc].type == '/') {
        int op = tokens[pc++].type;
        long val2 = unary();
        if (op == '*') val = val * val2;
        else val = val / val2;
    }
    return val;
}

long add() {
    long val = term();
    while (tokens[pc].type == '+' || tokens[pc].type == '-') {
        int op = tokens[pc++].type;
        long val2 = term();
        if (op == '+') {
            if (is_string_ptr(val) && is_string_ptr(val2)) {
                val = concat_strings(val, val2);
            } else {
                val = val + val2;
            }
        } else {
            val = val - val2;
        }
    }
    return val;
}

long relational() {
    long val = add();
    while (tokens[pc].type >= Eq && tokens[pc].type <= Ge) {
        int op = tokens[pc++].type;
        long val2 = add();
        if (op == Eq) val = val == val2;
        if (op == Ne) val = val != val2;
        if (op == Lt) val = val < val2;
        if (op == Gt) val = val > val2;
    }
    return val;
}

long expression() {
    // Check for pointer assignment: *ptr = val (Not fully robust for *b = 'x', but ok for int*)
    if (tokens[pc].type == '*') {
        int save_pc = pc;
        unary(); 
        if (tokens[pc].type == '=') {
            pc = save_pc; 
            pc++; 
            long addr = unary(); 
            match('=');
            long val = expression();
            if (addr >= 0 && addr < MEM_SIZE) memory[addr] = val;
            return val;
        }
        pc = save_pc; 
    }

    if (tokens[pc].type == Id) {
        if (tokens[pc+1].type == '[') {
            int idx = find_local(tokens[pc].text, tokens[pc].val);
            if (idx == -1) error("Assign to unknown var");
            pc += 2;
            long start_or_index = expression();

            if (tokens[pc].type == ':') {
                pc++;
                long end = expression();
                match(']');

                long val = memory[locals[idx].addr];
                if (is_string_ptr(val)) {
                    return slice_string(val, start_or_index, end);
                } else {
                    error("Slicing only works on strings");
                }
            }

            match(']');
            int addr = locals[idx].addr;
            if (tokens[pc].type == '=') {
                pc++;
                long val = expression();
                memory[addr + start_or_index] = val;
                return val;
            }
            return memory[addr + start_or_index];
        } else if (tokens[pc+1].type == '=') {
            int idx = find_local(tokens[pc].text, tokens[pc].val);
            if (idx == -1) error("Assign to unknown var");
            pc += 2;
            long val = expression();
            memory[locals[idx].addr] = val;
            return val;
        }
    }

    return relational();
}

void skip_block() {
    int brace = 0;
    do {
        if (tokens[pc].type == '{') brace++;
        if (tokens[pc].type == '}') brace--;
        pc++;
    } while (brace > 0 && tokens[pc].type != 0);
}

void statement() {
    if (tokens[pc].type == '{') {
        pc++;
        while (tokens[pc].type != '}' && tokens[pc].type != 0) {
            statement();
            if (ax == -999) break; 
        }
        match('}');
    }
    else if (tokens[pc].type == Int || tokens[pc].type == Char) {
        pc++;
        while (tokens[pc].type != ';') {
            while (tokens[pc].type == '*') pc++;
            Token *t = &tokens[pc];
            match(Id);

            int addr = sp;
            Symbol *s = &locals[loc_cnt++];
            strncpy(s->name, t->text, t->val); s->name[t->val] = 0;
            s->addr = addr;
            s->is_array = 0;

            if (tokens[pc].type == '[') {
                pc++;
                int size = (int)expression();
                match(']');
                s->is_array = 1;
                sp += size;
            } else {
                sp++;
            }

            if (tokens[pc].type == '=') {
                pc++;
                memory[addr] = expression();
            }
            if (tokens[pc].type == ',') pc++;
        }
        match(';');
    }
    else if (tokens[pc].type == If) {
        pc++;
        match('(');
        long cond = expression();
        match(')');
        if (cond) {
            statement();
            if (ax == -999) return;
            if (tokens[pc].type == Else) { pc++; skip_block(); }
        } else {
            skip_block();
            if (tokens[pc].type == Else) {
                pc++;
                statement();
                if (ax == -999) return;
            }
        }
    }
    else if (tokens[pc].type == While) {
        pc++;
        int loop_start = pc;
        match('(');
        long cond = expression();
        match(')');
        if (!cond) {
            skip_block();
        } else {
            while (1) {
                statement();
                if (ax == -999) return;
                int save_pc = pc;
                pc = loop_start;
                match('(');
                cond = expression();
                match(')');
                if (!cond) { pc = save_pc; break; }
            }
        }
    }
    else if (tokens[pc].type == Return) {
        pc++;
        if (tokens[pc].type != ';') ax = expression();
        else ax = 0;
        match(';');
        ax = -999; 
    }
    else if (tokens[pc].type == Printf) {
        pc++;
        match('(');
        char *fmt = tokens[pc].text; // Get raw format string
        match(Str);
        
        char *p = fmt;
        while (*p) {
            if (*p == '%' && (p[1] == 'd' || p[1] == 's')) {
                p++; // skip %
                match(',');
                long val = expression();
                if (*p == 'd') printf("%ld", val);
                else if (*p == 's') printf("%s", (char*)val);
                p++; // skip d or s
            } else {
                putchar(*p++);
            }
        }
        match(')');
        match(';');
    }
    else {
        expression();
        match(';');
    }
}

void scan_functions() {
    int i = 0;
    while (tokens[i].type != 0) {
        // Simple scan: Type Id ( ...
        if ((tokens[i].type == Int || tokens[i].type == Char) && 
            tokens[i+1].type == Id && tokens[i+2].type == '(') {
            
            Func *f = &funcs[func_cnt++];
            Token *name = &tokens[i+1];
            strncpy(f->name, name->text, name->val); f->name[name->val] = 0;
            
            i += 3; // Type Id (
            int params = 0;
            while(tokens[i].type != ')') {
                if (tokens[i].type == Int || tokens[i].type == Char) {
                    params++;
                    // Skip 'char * name' or 'int name'
                    i++; // Type
                    while (tokens[i].type == '*') i++; 
                    if (tokens[i].type == Id) i++;
                } else i++;
            }
            f->param_count = params;
            i++; 
            f->entry_point = i; 
            
            int brace = 0;
            do {
                if (tokens[i].type == '{') brace++;
                if (tokens[i].type == '}') brace--;
                i++;
            } while (brace > 0 && tokens[i].type != 0);
        } else {
            i++;
        }
    }
}

long native_socket(long *args, int argc) {
    int domain = (int)args[0];
    int type = (int)args[1];
    int protocol = (int)args[2];
    return socket(domain, type, protocol);
}

long native_bind(long *args, int argc) {
    int sockfd = (int)args[0];
    int port = (int)args[1];

    struct sockaddr_in addr;
    memset(&addr, 0, sizeof(addr));
    addr.sin_family = AF_INET;
    addr.sin_addr.s_addr = INADDR_ANY;
    addr.sin_port = htons(port);

    return bind(sockfd, (struct sockaddr*)&addr, sizeof(addr));
}

long native_listen(long *args, int argc) {
    int sockfd = (int)args[0];
    int backlog = (int)args[1];
    return listen(sockfd, backlog);
}

long native_accept(long *args, int argc) {
    int sockfd = (int)args[0];
    return accept(sockfd, NULL, NULL);
}

long native_recv(long *args, int argc) {
    int sockfd = (int)args[0];
    int addr = (int)args[1];
    int len = (int)args[2];
    int flags = (int)args[3];

    char temp_buf[8192];
    if (len > 8192) len = 8192;

    int result = recv(sockfd, temp_buf, len, flags);
    if (result > 0) {
        for (int i = 0; i < result; i++) {
            memory[addr + i] = temp_buf[i];
        }
    }
    return result;
}

long native_send(long *args, int argc) {
    int sockfd = (int)args[0];
    long buf_arg = args[1];
    int len = (int)args[2];
    int flags = (int)args[3];

    if (buf_arg > MEM_SIZE * 8 || buf_arg < 0) {
        return send(sockfd, (char*)buf_arg, len, flags);
    }

    char temp_buf[8192];
    if (len > 8192) len = 8192;

    for (int i = 0; i < len; i++) {
        temp_buf[i] = (char)memory[buf_arg + i];
    }

    return send(sockfd, temp_buf, len, flags);
}

long native_close(long *args, int argc) {
    int fd = (int)args[0];
    return close(fd);
}

long native_strlen(long *args, int argc) {
    char *str = (char*)args[0];
    return strlen(str);
}

long native_AF_INET(long *args, int argc) {
    return AF_INET;
}

long native_SOCK_STREAM(long *args, int argc) {
    return SOCK_STREAM;
}

long native_sqrt(long *args, int argc) {
    return (long)sqrt((double)args[0]);
}

long native_pow(long *args, int argc) {
    return (long)pow((double)args[0], (double)args[1]);
}

long native_sin(long *args, int argc) {
    return (long)(sin((double)args[0]) * 1000000);
}

long native_cos(long *args, int argc) {
    return (long)(cos((double)args[0]) * 1000000);
}

long native_tan(long *args, int argc) {
    return (long)(tan((double)args[0]) * 1000000);
}

long native_abs(long *args, int argc) {
    return (long)abs((int)args[0]);
}

long native_floor(long *args, int argc) {
    return (long)floor((double)args[0]);
}

long native_ceil(long *args, int argc) {
    return (long)ceil((double)args[0]);
}

long native_strpos(long *args, int argc) {
    char *haystack = (char*)args[0];
    char *needle = (char*)args[1];
    char *pos = strstr(haystack, needle);
    if (pos) {
        return pos - haystack;
    }
    return -1;
}

long native_substr(long *args, int argc) {
    char *str = (char*)args[0];
    int start = (int)args[1];
    int length = (int)args[2];

    char *result = &str_pool[str_pool_idx];
    int str_len = strlen(str);

    if (start < 0) start = 0;
    if (start >= str_len) return (long)"";
    if (start + length > str_len) length = str_len - start;
    if (length < 0) return (long)"";

    if (str_pool_idx + length + 1 >= STR_POOL_SIZE) {
        error("String pool overflow");
    }

    strncpy(result, str + start, length);
    result[length] = 0;

    str_pool_idx += length + 1;
    return (long)result;
}

long native_upper(long *args, int argc) {
    char *str = (char*)args[0];
    char *result = &str_pool[str_pool_idx];
    int len = strlen(str);

    if (str_pool_idx + len + 1 >= STR_POOL_SIZE) {
        error("String pool overflow");
    }

    for (int i = 0; i <= len; i++) {
        result[i] = toupper(str[i]);
    }

    str_pool_idx += len + 1;
    return (long)result;
}

long native_lower(long *args, int argc) {
    char *str = (char*)args[0];
    char *result = &str_pool[str_pool_idx];
    int len = strlen(str);

    if (str_pool_idx + len + 1 >= STR_POOL_SIZE) {
        error("String pool overflow");
    }

    for (int i = 0; i <= len; i++) {
        result[i] = tolower(str[i]);
    }

    str_pool_idx += len + 1;
    return (long)result;
}

long native_strip(long *args, int argc) {
    char *str = (char*)args[0];
    char *result = &str_pool[str_pool_idx];

    while (*str && isspace(*str)) str++;

    int len = strlen(str);
    while (len > 0 && isspace(str[len - 1])) len--;

    if (str_pool_idx + len + 1 >= STR_POOL_SIZE) {
        error("String pool overflow");
    }

    strncpy(result, str, len);
    result[len] = 0;

    str_pool_idx += len + 1;
    return (long)result;
}

long native_replace(long *args, int argc) {
    char *str = (char*)args[0];
    char *old_str = (char*)args[1];
    char *new_str = (char*)args[2];

    char *result = &str_pool[str_pool_idx];
    char *src = str;
    char *dst = result;
    int old_len = strlen(old_str);
    int new_len = strlen(new_str);

    while (*src) {
        if (strncmp(src, old_str, old_len) == 0) {
            if (str_pool_idx + (dst - result) + new_len + strlen(src + old_len) + 1 >= STR_POOL_SIZE) {
                error("String pool overflow");
            }
            strcpy(dst, new_str);
            dst += new_len;
            src += old_len;
        } else {
            *dst++ = *src++;
        }
    }
    *dst = 0;

    int total_len = dst - result;
    str_pool_idx += total_len + 1;
    return (long)result;
}

long native_startswith(long *args, int argc) {
    char *str = (char*)args[0];
    char *prefix = (char*)args[1];
    int prefix_len = strlen(prefix);
    return strncmp(str, prefix, prefix_len) == 0;
}

long native_endswith(long *args, int argc) {
    char *str = (char*)args[0];
    char *suffix = (char*)args[1];
    int str_len = strlen(str);
    int suffix_len = strlen(suffix);

    if (suffix_len > str_len) return 0;
    return strcmp(str + str_len - suffix_len, suffix) == 0;
}

void register_native_functions() {
    register_native_func("socket", native_socket);
    register_native_func("bind", native_bind);
    register_native_func("listen", native_listen);
    register_native_func("accept", native_accept);
    register_native_func("recv", native_recv);
    register_native_func("send", native_send);
    register_native_func("close", native_close);
    register_native_func("strlen", native_strlen);
    register_native_func("AF_INET", native_AF_INET);
    register_native_func("SOCK_STREAM", native_SOCK_STREAM);
    register_native_func("sqrt", native_sqrt);
    register_native_func("pow", native_pow);
    register_native_func("sin", native_sin);
    register_native_func("cos", native_cos);
    register_native_func("tan", native_tan);
    register_native_func("abs", native_abs);
    register_native_func("floor", native_floor);
    register_native_func("ceil", native_ceil);
    register_native_func("strpos", native_strpos);
    register_native_func("substr", native_substr);
    register_native_func("upper", native_upper);
    register_native_func("lower", native_lower);
    register_native_func("strip", native_strip);
    register_native_func("replace", native_replace);
    register_native_func("startswith", native_startswith);
    register_native_func("endswith", native_endswith);
}

int main(int argc, char **argv) {
    if (argc < 2) { printf("Usage: ./mini_c file.c\n"); return 1; }
    
    FILE *f = fopen(argv[1], "rb");
    if (!f) { printf("Could not open file.\n"); return 1; }
    
    src_code = malloc(MAX_SRC);
    size_t n = fread(src_code, 1, MAX_SRC, f);
    src_code[n] = 0;
    fclose(f);
    
    register_native_functions();

    tokenize(src_code);
    scan_functions();

    int main_idx = find_func("main", 4);
    if (main_idx == -1) { printf("No main function found.\n"); return 1; }
    
    pc = funcs[main_idx].entry_point;
    memory[sp++] = 0; 
    memory[sp++] = 0; 
    
    ax = 0;
    statement(); 
    
    return 0;
}