#include <stdio.h>
#include <stdlib.h>
#include <string.h>

enum asm_symbol_type{
    DATA_BYTE,
    DATA_WORD,
    DATA_HALFWORD,
    DATA_DOUBLEWORD,
    DATA_STRING,
    DATA_SYMBLE,
    TEXT_LABEL,
    UNKNOWN,
};

struct asm_symbol_table {
    const char* name;
    enum asm_symbol_type type;
    union {
        long long  data_doubleword;
        long       data_word;
        short      data_halfword;
        char       data_byte;
        char*      data_string;
    } data;
};

static const char comments[]  =  {
    '#',
};

static inline int is_blank(char ch) {
    return ch == ' ' || ch == '\t' || ch == '\n' || ch == '\v' || ch == '\f' || ch == '\r';
}

void parse_line_without_blank(const char* in, const char** out_start, const char** out_end) {
    while (*in == '\0' || is_blank(*in)) {
        in ++;
    }
    *out_start = in;
    while (*in != '\0') {
        if (*in == '\r' || *in == '\n') {
            break;
        }
        for (int i = 0; i < sizeof(comments); i ++) {
            if (*in == comments[i]) {
                break;
            }
        }
        in ++;
    }

    if (*out_end == *out_start) {
        goto END;
    }
    
    while (is_blank(*in)) {
        in --;
    }
    
END:
    *out_end = in;
    return;
}

void parse_data_symbol(const char* start, const char* end, struct asm_symbol_table* table) {
    table->name = start;
    while(start < end) {
        if (*start == ':') {

        }
    }
    table->type = UNKNOWN;
}

#define TMP_BUFF_SIZE 1024

enum parse_state {
    IN_DATA,
    IN_TEXT,
    IN_UNKNOWN,
};
void get_symbol_table(FILE* in) {
    enum parse_state state = IN_UNKNOWN;
    fseek(in, 0, SEEK_SET);
    char buf[TMP_BUFF_SIZE];

    int current_line = 0;
    while (1) {
        current_line ++;
        char *start = fgets(buf, sizeof(buf), in);
        if (start == NULL) {
            return;
        }
        char *end;
        parse_line_without_blank(buf, &start, &end);

        if (start == end) {
            continue;
        }

        if (start[0] == '.') {
            // .data .text and so on
            if (strcmp(start, ".data") == 0) {
                state = IN_DATA;
            } else if (strcmp(start, ".text") == 0) {
                state = IN_TEXT;
            } else {
                printf("unknown directive at line %d\n", current_line);
                state = IN_UNKNOWN;
            }
            continue;
        }

        switch (state) {
            case IN_DATA:
                parse_data_symbol(start, end);
                break;
            case IN_TEXT:
                break;
            case IN_UNKNOWN:
                break;
        }
    }
    
}

void assembler(FILE* in, FILE* out, char *start_symble) {
    char buf[TMP_BUFF_SIZE];

    char *res = fgets(buf, sizeof(buf), in);
    if (res == NULL) {
        return;
    }

    if (res[0] == '.') {
        // maybe .data .text and so on
    } else if (res[0] == ' ') {

    }


}