path: root/shell.c

#include "queue.h"
#include "dynamic_array.h"

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <string.h>
#include <sys/wait.h>
#include <fcntl.h>

enum modes { word_separation, whole_word };
enum {
    new_line        = 10,
    whitespace      = ' ',
    tab             = 9,
    backslash       = '\\',
    double_quotes   = '"'
};

/* two-letter tokens */
enum {
    append          = '>' + 1,
    and             = '&' + 1,
    or              = '|' + 1
};

/* storing file names to redirect standard input/output streams */
struct io_type {
    char *input_stream;
    char *output_stream;
    char *output_stream_to_append;
};

struct param_type {
    int is_word;
    int escape_sequences;
    unsigned int double_quotes_counter;
    char stored_symbol;
    int empty_word_flag;
    int tokens;
    int wrong_command;
    struct io_type streams;
    int last_execution_status;
};

static void show_invitation()
{
    printf("> ");
}

static void print_error()
{
    fprintf(stderr, "Error: unmatched quotes\n");
}

static void init_params(struct param_type *params, enum modes *current_mode)
{
    params->is_word = 0;
    params->escape_sequences = 0;
    params->double_quotes_counter = 0;
    params->stored_symbol = ' ';
    params->empty_word_flag = 0;
    params->tokens = 0;
    params->wrong_command = 0;
    params->streams.input_stream = NULL;
    params->streams.output_stream = NULL;
    params->streams.output_stream_to_append = NULL;
    /* 0 - success, 1 - error */
    params->last_execution_status = 0;
    *current_mode = word_separation;
}

static void clear_filename(struct param_type *params)
{
    if(params->streams.input_stream)
        free(params->streams.input_stream);
    if(params->streams.output_stream)
        free(params->streams.output_stream);
    if(params->streams.output_stream_to_append)
        free(params->streams.output_stream_to_append);
    params->streams.input_stream = NULL;
    params->streams.output_stream = NULL;
    params->streams.output_stream_to_append = NULL;
}

static void reset_params(struct param_type *params,
                         enum modes *current_mode,
                         struct queue *word_chain,
                         struct dynamic_array *tmp_word)
{
    queue_clear(word_chain);
    dynarr_drop_word(tmp_word);
    clear_filename(params);
    init_params(params, current_mode);
}

static int is_double_quotes_pair(struct param_type params)
{
    return !(params.double_quotes_counter % 2);
}

static int check_separation(int ch, struct param_type params)
{
    return (ch == whitespace || ch == tab) && params.is_word &&
            !params.escape_sequences;
}

static int is_stream_redirection_set(const struct param_type *params)
{
    return params->streams.input_stream || params->streams.output_stream ||
            params->streams.output_stream_to_append;
}

static int filename_waiting(const struct param_type *params)
{
    return params->tokens == '<' || params->tokens == '>' ||
        params->tokens == append;
}

static void add_filename(struct dynamic_array *tmp_word,
                         struct param_type *params)
{
    dynarr_push_back(tmp_word, '\0');
    switch(params->tokens) {
    case '<':
        params->streams.input_stream = malloc(tmp_word->last_element_index+1);
        dynarr_copy_array(tmp_word, params->streams.input_stream);
        break;
    case '>':
        params->streams.output_stream = malloc(tmp_word->last_element_index+1);
        dynarr_copy_array(tmp_word, params->streams.output_stream);
        break;
    case append:
        params->streams.output_stream_to_append =
            malloc(tmp_word->last_element_index+1);
        dynarr_copy_array(tmp_word, params->streams.output_stream_to_append);
    }
    dynarr_drop_word(tmp_word);
    params->is_word = 0;
}

static void add_word(struct queue *word_chain,
                     struct dynamic_array *tmp_word,
                     struct param_type *params)
{
    dynarr_push_back(tmp_word, 0);
    char *word = malloc(tmp_word->last_element_index+1);
    dynarr_copy_array(tmp_word, word);
    queue_push(word_chain, word);
    dynarr_drop_word(tmp_word);
    params->is_word = 0;
}

static int is_double_token(struct param_type *params)
{
    return params->tokens == and || params->tokens == or ||
            params->tokens == append;
}

static int is_redirect_token(int ch, int next_ch)
{
    return ch == '<' || ch == '>' || (ch == '>' && next_ch == '>');
}

static int validate_redirections(int ch, int next_ch, struct param_type *params)
{
    return (ch == '<' && params->streams.input_stream == NULL) ||
            ((ch == '>' || (ch == '>' && next_ch == '>')) &&
             params->streams.output_stream == NULL &&
             params->streams.output_stream_to_append == NULL);
}

static int stream_redirect_tokens(struct queue *word_chain,
                                  struct dynamic_array *tmp_word, int ch,
                                  struct param_type *params)
{
    int next_ch;
    next_ch = getchar();
    ungetc(next_ch, stdin);

    if(is_redirect_token(ch, next_ch)) {
        /* filenames */
        if(filename_waiting(params))
            if(params->is_word)
                add_filename(tmp_word, params);
            else {
                fprintf(stderr, "syntax error\n");
                params->wrong_command = 1;
                return 0;
            }
            /* execute command */
        else if(params->is_word) {
            add_word(word_chain, tmp_word, params);
            params->is_word = 0;
        }

        if(validate_redirections(ch, next_ch, params)) {
            params->tokens = (ch == '>' && next_ch == '>') ? append : ch;
            if(is_double_token(params))
                getchar();
            return 1;
        } else {
            fprintf(stderr, "syntax error\n");
            params->wrong_command = 1;
        }
    }
    return 0;
}

static void clean_input_buffer()
{
    int ch;
    while((ch = getchar()) != new_line)
    {}
}

static int is_special_token(int ch)
{
    return ch == and || ch == or || ch == '&' || ch == ';' || ch == '|';
}

static int special_tokens_allowed(int ch, struct param_type *params)
{
    return (!filename_waiting(params) ||
            (filename_waiting(params) && params->is_word));
}

static int special_tokens(struct dynamic_array *tmp_word, int ch,
                               struct param_type *params)
{
    int next_ch;
    next_ch = getchar();
    ungetc(next_ch, stdin);

    if(is_special_token(ch)) {
        if(special_tokens_allowed(ch, params)) {
            if(filename_waiting(params) && params->is_word) {
                add_filename(tmp_word, params);
                params->is_word = 0;
            }
            if(ch == '|')
                params->tokens = '|';
            else if(ch == '|' && next_ch == '|')
                params->tokens = or;
            else if(ch == '&' && next_ch == '&')
                params->tokens = and;
            else if(ch == '&') {
                while((ch = getchar()) != new_line) {
                    if(ch != whitespace && ch != tab) {
                        fprintf(stderr, "incorrect command\n");
                        params->wrong_command = 1;
                        return 0;
                    }
                }
                params->tokens = '&';
            }
            if(is_double_token(params))
                    getchar();
            return 1;
        } else {
            params->wrong_command = 1;
            fprintf(stderr, "filename expected\n");
        }
    }
    return 0;
}

static int special_token_handling(struct queue *word_chain,
                        struct dynamic_array *tmp_word, int ch,
                        struct param_type *params)
{
    return
        stream_redirect_tokens(word_chain, tmp_word, ch, params) ?
            1 : special_tokens(tmp_word, ch, params);
}

static int ignore_spaces(int ch, struct param_type params)
{
    return (ch == whitespace || ch == tab) && !params.escape_sequences;
}

static int change_mode(int ch, struct param_type params)
{
    return ch == '"' && !params.escape_sequences;
}

static int start_escape_sequence(int ch, struct param_type params)
{
    return ch == backslash && !params.escape_sequences;
}

/* name of file for opening stream already exists */

static int excessive_words(int ch, const struct param_type *params)
{
    int next_ch;
#if 0
    if(filename_waiting(params))
        return
            (params->tokens == '<' && params->streams.input_stream != NULL) ||
            (params->tokens == '>' && params->streams.output_stream !=NULL) ||
            (params->tokens == append &&
                params->streams.output_stream_to_append != NULL);
    else
        return 0;
#endif
    if(filename_waiting(params)) {
        if(ch == new_line)
            return 0;
        while((next_ch = getchar()) != new_line) {
            if(next_ch == ' ')
                continue;
            if(!is_special_token(next_ch) && next_ch != '<' && next_ch != '>')
                return 1;
            else
                break;
        }
        ungetc(next_ch, stdin);
    }
    return 0;
}

#if 0
static void print_word(char *word)
{
    putchar('[');
    while(*word) {
        putchar(*word);
        ++word;
    }
    printf("]\n");
}

static void print_line(const struct queue *word_chain)
{
    queue_processing(word_chain, print_word);
}
#endif

static char** create_cmdline(const struct queue *word_chain,
                                   int word_counter)
{
    char **cmdline = malloc((word_counter + 1) * sizeof(char*));
    queue_copy_words_to_args(word_chain, cmdline);
    return cmdline;
}

static int is_cd_command(const char *arg)
{
    return !strcmp(arg, "cd");
}

static void change_directory(char **cmdline)
{
    int result;
    char *first_arg = cmdline[1];
    char *path = NULL;
    /* change to user home directory */
    if(first_arg == NULL) {
        path = getenv("HOME");
        if(!path) {
            perror("I don't know where's your home...");
            return;
        }
    } else {
        if(cmdline[2]) {
            perror("cd: too many arguments");
            return;
        }
        path = first_arg;
    }
    result = chdir(path);
    if(result == -1)
        perror(path);
}

static void clean_up_memory(struct queue *word_chain, char **cmdline,
                            struct param_type *params)
{
    queue_clear(word_chain);
    free(cmdline);
    clear_filename(params);
}

static void clean_up_zombie_process()
{
    int pid;
    do {
        pid = wait4(-1, NULL, WNOHANG, NULL);
    } while(pid > 0);
}

static void open_files(const struct param_type *params, int *input_fd,
                      int *output_fd)
{
    if(params->streams.input_stream)
        *input_fd = open(params->streams.input_stream, O_RDONLY);
    if(params->streams.output_stream)
        *output_fd = open(params->streams.output_stream, O_WRONLY | O_CREAT |
                          O_TRUNC, 0666);
    if(params->streams.output_stream_to_append)
        *output_fd = open(params->streams.output_stream_to_append, O_WRONLY |
                          O_APPEND, 0666);
}

static void change_streams(int input_fd, int output_fd)
{
    if(input_fd) {
        dup2(input_fd, 0);
        close(input_fd);
    }
    if(output_fd) {
        dup2(output_fd, 1);
        close(output_fd);
    }
}

static void close_files(int input_fd, int output_fd)
{
    if(input_fd)
        close(input_fd);
    if(output_fd)
        close(output_fd);
}

static void run_external_program(struct queue *word_chain,
                                 struct param_type *params)
{
    int pid, wait_pid, result, input_fd, output_fd;
    if(word_chain->first == NULL) {
        fprintf(stderr, "empty command\n");
        return;
    }

    if(is_stream_redirection_set(params))
        open_files(params, &input_fd, &output_fd);

    if(input_fd == -1 || output_fd == -1) {
        perror("can't open file");
        return;
    }
    char **cmdline = create_cmdline(word_chain,
                                    queue_get_word_count(word_chain));
    if(is_cd_command(cmdline[0])) {
        change_directory(cmdline);
        clean_up_memory(word_chain, cmdline, params);
    }
    else {
        pid = fork();
        if(pid == -1) {
            perror("fork error");
            exit(1);
        }
        /* child process  */
        if(pid == 0) {
            if(is_stream_redirection_set(params))
                change_streams(input_fd, output_fd);
            execvp(cmdline[0], cmdline);
            params->last_execution_status = 1;
            perror(cmdline[0]);
            exit(1);
        }
        /* parent process */

        close_files(input_fd, output_fd);
        clean_up_memory(word_chain, cmdline, params);

        /* waiting for forground process */
        if(params->tokens != '&') {
            do {
                wait_pid = wait(&result);
            } while(wait_pid != pid);
        } else
            clean_up_zombie_process();
    }
}

static void command_processing(struct param_type *params,
                                enum modes *current_mode,
                                struct queue *word_chain,
                                struct dynamic_array *tmp_word, int ch)
{
    /* and odd number of double quotes  */
    if(!is_double_quotes_pair(*params)) {
        print_error();
        reset_params(params, current_mode, word_chain, tmp_word);
        show_invitation();
        return;
    }

    if(params->empty_word_flag || params->is_word) {
        if(filename_waiting(params))
            if(excessive_words(ch, params)) {
                fprintf(stderr, "too many args\n");
                goto clean;
            } else
                add_filename(tmp_word, params);
        else
            add_word(word_chain, tmp_word, params);
    } else if(filename_waiting(params)) {
        fprintf(stderr, "filename expected\n");
        goto clean;
    }

    if((params->tokens == and || params->tokens == or) &&
            params->last_execution_status == 1)
        fprintf(stderr, "cannot be performed\n");
    else
        run_external_program(word_chain, params);

clean:
    if(params->tokens == '&' || params->tokens == 0 || ch == new_line)
        show_invitation();
    reset_params(params, current_mode, word_chain, tmp_word);
}

static int is_empty_word(int ch, struct param_type params)
{
    return (ch == whitespace || ch == tab) && !params.is_word &&
            params.empty_word_flag;
}

static void word_separation_processing(int ch, struct param_type *params,
                                  enum modes *current_mode,
                                  struct queue *word_chain,
                                  struct dynamic_array *tmp_word)
{
    /* could be a marker for the beginning of a blank word */
    params->stored_symbol = ch;
    if(is_empty_word(ch, *params)) {
        dynarr_push_back(tmp_word, '\0');
        add_word(word_chain, tmp_word, params);
        params->empty_word_flag = 0;
        return;
    }
    params->empty_word_flag = 0;
    if(change_mode(ch, *params)) {
        ++params->double_quotes_counter;
        *current_mode = whole_word;
        return;
    }
    if(start_escape_sequence(ch, *params)) {
        params->escape_sequences = 1;
        return;
    }
    if(special_token_handling(word_chain, tmp_word, ch, params)) {
        if(!filename_waiting(params))
            command_processing(params, current_mode, word_chain, tmp_word, ch);
        return;
    }
    if(params->wrong_command) {
        clean_input_buffer();
        reset_params(params, current_mode, word_chain, tmp_word);
        show_invitation();
        return;
    }
    if(check_separation(ch, *params)) {
        if(excessive_words(ch, params)) {
            clean_input_buffer();
            reset_params(params, current_mode, word_chain, tmp_word);
            fprintf(stderr, "too many args\n");
            show_invitation();
            return;
        }
        if(!filename_waiting(params)) {
            add_word(word_chain, tmp_word, params);
            params->is_word = 0;
        }
        return;
    }
    if(ignore_spaces(ch, *params))
        return;

    dynarr_push_back(tmp_word, ch);
    params->is_word = 1;
    params->escape_sequences = 0;
}

static int escape_double_quotes_or_backslash(int ch, struct param_type params)
{
    return params.escape_sequences &&
        (ch == double_quotes || ch == backslash);
}

static int double_quotes_again(int ch, struct param_type params)
{
    return ch == double_quotes && !params.is_word &&
        params.stored_symbol == '"';
}

static void whole_word_processing(int ch, struct param_type *params,
                                  enum modes *current_mode,
                                  struct queue *word_chain,
                                  struct dynamic_array *tmp_word)
{
    if(double_quotes_again(ch, *params)) {
        params->stored_symbol = '\0';
        params->empty_word_flag = 1;
        ++params->double_quotes_counter;
        *current_mode = word_separation;
        return;
    }
    params->stored_symbol = '\0';
    if(change_mode(ch, *params)) {
        ++params->double_quotes_counter;
        *current_mode = word_separation;
        return;
    }
    if(escape_double_quotes_or_backslash(ch, *params)) {
        dynarr_push_back(tmp_word, ch);
        params->escape_sequences = 0;
        params->is_word = 1;
        return;
    }
    if(start_escape_sequence(ch, *params)) {
        params->escape_sequences = 1;
        return;
    }
    /* backslash recovery */
    if(params->escape_sequences)
        dynarr_push_back(tmp_word, backslash);
    dynarr_push_back(tmp_word, ch);
    params->is_word = 1;
    params->escape_sequences = 0;
}

int main()
{
    int ch;
    struct param_type params;
    struct queue word_chain;
    struct dynamic_array tmp_word;
    enum modes current_mode = word_separation;
    queue_init(&word_chain);
    dynarr_create_array(&tmp_word);
    init_params(&params, &current_mode);
    show_invitation();
    while((ch = getchar()) != EOF) {
        clean_up_zombie_process();
        if(ch == new_line)
            command_processing(&params, &current_mode, &word_chain,
                               &tmp_word, ch);
        else if(current_mode == word_separation)
            word_separation_processing(ch, &params, &current_mode, &word_chain,
                                       &tmp_word);
        else if(current_mode == whole_word)
            whole_word_processing(ch, &params, &current_mode, &word_chain,
                                  &tmp_word);
    }
    putchar(new_line);
    dynarr_clear(&tmp_word);
    return 0;
}