42sh/src/parser/grammar_basic.c

#include <stdbool.h>
#define _POSIX_C_SOURCE 200809L

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

#include "../utils/lists/lists.h"
#include "grammar.h"
#include "grammar_advanced.h"
#include "grammar_basic.h"

// === Static functions

static enum ast_and_or_type and_or_tok_to_ast(enum token_type tok_type)
{
    switch (tok_type)
    {
    case TOKEN_AND:
        return AST_AND_OR_TYPE_AND;
    case TOKEN_OR:
        return AST_AND_OR_TYPE_OR;
    default:
        fprintf(stderr, "and_or impossible to init, wrong token type");
        return AST_AND_OR_NULL;
    }
}

// === Functions

struct ast *parse_list(struct lexer_context *ctx)
{
    struct list *result_list = NULL;
    struct ast *current_node = NULL;

    struct token *token = PEEK_TOKEN();

    // and_or
    current_node = parse_and_or(ctx);
    if (current_node == NULL)
        return NULL;
    result_list = list_append(result_list, current_node);

    // Following and_or commands
    token = PEEK_TOKEN();
    while (token->type == TOKEN_SEMICOLON)
    {
        token = POP_TOKEN();
        token = PEEK_TOKEN();
        if (is_first(*token, RULE_AND_OR))
        {
            current_node = parse_and_or(ctx);
            if (current_node == NULL)
            {
                struct ast *tmp = ast_create_list(result_list);
                ast_free(&tmp);
                return NULL;
            }
            result_list = list_append(result_list, current_node);
            token = PEEK_TOKEN();
        }
    }

    return ast_create_list(result_list);
}

struct ast *parse_and_or(struct lexer_context *ctx)
{
    struct ast *result = parse_pipeline(ctx);
    if (result == NULL)
        return NULL;
    struct token *token = PEEK_TOKEN();

    while (token->type == TOKEN_AND || token->type == TOKEN_OR)
    {
        // Set left part

        struct ast *left = result;

        // eat and_or token
        token = POP_TOKEN();

        // Set type
        enum ast_and_or_type type = and_or_tok_to_ast(token->type);

        token = PEEK_TOKEN();

        // Skip newlines
        while (token->type == TOKEN_NEWLINE)
        {
            token = POP_TOKEN();
            token = PEEK_TOKEN();
        }

        // Right part
        struct ast *right = parse_pipeline(ctx);

        result = ast_create_and_or(left, right, type);
        if (result == NULL)
        {
            ast_free(&left);
            ast_free(&right);
            return NULL;
        }
    }

    return result;
}

struct ast *parse_pipeline(struct lexer_context *ctx)
{
    bool negation = false;
    struct token *token = PEEK_TOKEN();

    // Eventual '!'
    if (token->type == TOKEN_NEGATION)
    {
        negation = true;
        POP_TOKEN();
        token = PEEK_TOKEN();
    }

    struct ast *left = parse_command(ctx);
    if (negation)
    {
        left = ast_create_neg(negation, left);
    }

    token = PEEK_TOKEN();
    while (token->type == TOKEN_PIPE)
    {
        POP_TOKEN();

        // skip newlines
        token = PEEK_TOKEN();
        while (token->type == TOKEN_NEWLINE)
        {
            POP_TOKEN();
            token = PEEK_TOKEN();
        }

        struct ast *right = parse_command(ctx);

        // Create AST
        left = ast_create_pipe(left, right);
        token = PEEK_TOKEN();
    }

    return left;
}

struct ast *parse_command(struct lexer_context *ctx)
{
    struct token *token = PEEK_TOKEN();

    if (token->type == TOKEN_WORD)
    {
        return parse_simple_command(ctx);
    }
    else if (token->type == TOKEN_IF)
    {
        return parse_shell_command(ctx);
    }
    else
    {
        perror("Syntax error: unexpected token");
        return NULL;
    }
}

struct ast *parse_simple_command(struct lexer_context *ctx)
{
    struct list *command_elements = NULL;

    // WORD
    struct token *token = POP_TOKEN();
    if (token->type != TOKEN_WORD)
    {
        perror("Expected a command but got a different token type");
        return NULL;
    }
    char *command = strdup(token->data);
    command_elements = list_append(command_elements, command);

    token = PEEK_TOKEN();

    // Eventual elements
    while (is_first(*token, RULE_ELEMENT))
    {
        // Get element
        struct ast *element = parse_element(ctx);
        if (element == NULL)
        {
            list_deep_destroy(command_elements);
            return NULL;
        }

        // Get element type
        if (ast_is_word(element))
        {
            struct ast_word *element_word = ast_get_word(element);
            command_elements =
                list_append(command_elements, element_word->word);
        }
        else if (ast_is_redir(element))
        {
            // TODO
            perror("NOT IMPLEMENTED");
            return NULL;
        }
        else
        {
            perror("Internal error: unexpected return value from parse_element "
                 "in parse_simple_command");
            list_deep_destroy(command_elements);
            return NULL;
        }

        // Forward
        token = PEEK_TOKEN();
    }

    // Result
    struct ast *result = ast_create_command(command_elements);
    if (result == NULL)
    {
        list_deep_destroy(command_elements);
        return NULL;
    }
    return result;
}

struct ast *parse_element(struct lexer_context *ctx)
{
    struct token *token = PEEK_TOKEN();
    if (token->type == TOKEN_WORD)
    {
        token = POP_TOKEN();
        return ast_create_word(token->data);
    }
    else if (token->type == TOKEN_IONUMBER || is_token_redir(token))
    {
        return parse_redirection(ctx);
    }
    else
    {
        perror("Syntax error: unexpected token at parse_element");
        return NULL;
    }
}

struct ast *parse_shell_command(struct lexer_context *ctx)
{
    return parse_if_rule(ctx);
}

struct ast *parse_if_rule(struct lexer_context *ctx)
{
    // If keyword
    struct token *token = POP_TOKEN();
    if (token->type != TOKEN_IF)
    {
        perror("Internal error: expected a if rule but token has different "
             "type");
        return NULL;
    }

    // Condition content
    struct ast *condition_content = parse_compound_list(ctx);

    // Then keyword
    token = POP_TOKEN();
    if (token->type != TOKEN_THEN)
    {
        ast_free(&condition_content);
        perror("Expected the 'then' keyword but token has different type");
        return NULL;
    }

    // Then content
    struct ast *then_content = parse_compound_list(ctx);
    if (then_content == NULL)
    {
        ast_free(&condition_content);
        ast_free(&then_content);
        return NULL;
    }

    // Eventual else/elif clause(s)
    struct ast *else_content = parse_else_clause(ctx);
    if (else_content == NULL)
    {
        ast_free(&condition_content);
        ast_free(&then_content);
        return NULL;
    }

    // Fi keyword
    token = POP_TOKEN();
    if (token->type != TOKEN_FI)
    {
        ast_free(&condition_content);
        ast_free(&then_content);
        ast_free(&else_content);
        perror("Expected the 'fi' keyword but token has different type");
        return NULL;
    }

    // Result
    struct ast *result =
        ast_create_if(condition_content, then_content, else_content);
    if (result == NULL)
    {
        ast_free(&condition_content);
        ast_free(&then_content);
        ast_free(&else_content);
        perror("Internal error: could not create a new AST (AST_IF)");
        return NULL;
    }

    return result;
}

struct ast *parse_compound_list(struct lexer_context *ctx)
{
    struct list *result_list = NULL; // ast* list
    struct ast *current_cmd = NULL;
    struct token *token = PEEK_TOKEN();

    // Skip newlines
    while (token->type == TOKEN_NEWLINE)
    {
        token = POP_TOKEN();
        token = PEEK_TOKEN();
    }

    // And/or
    current_cmd = parse_and_or(ctx);
    if (current_cmd == NULL)
        return NULL;
    result_list = list_append(result_list, current_cmd);

    // Following commands
    token = PEEK_TOKEN();
    while (token->type == TOKEN_SEMICOLON || token->type == TOKEN_NEWLINE)
    {
        POP_TOKEN();
        token = PEEK_TOKEN();

        // Skip newlines
        while (token->type == TOKEN_NEWLINE)
        {
            token = POP_TOKEN();
            token = PEEK_TOKEN();
        }

        // And/or
        if (is_first(*token, RULE_AND_OR))
        {
            current_cmd = parse_and_or(ctx);
            if (current_cmd == NULL)
                return NULL;
            result_list = list_append(result_list, current_cmd);
        }

        token = PEEK_TOKEN();
    }

    // Eventual semicolons
    if (token->type == TOKEN_SEMICOLON)
    {
        token = POP_TOKEN();
        token = PEEK_TOKEN();
    }

    // Skip newlines
    while (token->type == TOKEN_NEWLINE)
    {
        token = POP_TOKEN();
        token = PEEK_TOKEN();
    }

    struct ast *result = ast_create_list(result_list);
    return result;
}

struct ast *parse_else_clause(struct lexer_context *ctx)
{
    struct token *token = PEEK_TOKEN();

    // Eventual elif content
    while (token->type == TOKEN_ELIF)
    {
        // Condition
        token = POP_TOKEN();
        struct ast *condition = parse_compound_list(ctx);

        // Then keyword
        token = POP_TOKEN();
        if (token->type != TOKEN_THEN)
        {
            perror("Expected the 'then' keyword but got a different token type");
            return NULL;
        }

        // Then clause
        struct ast *then_content = parse_compound_list(ctx);

        // Eventual else clause (recursive)
        struct ast *else_content = NULL;
        token = PEEK_TOKEN();
        if (token->type == TOKEN_ELSE || token->type == TOKEN_ELIF)
        {
            else_content = parse_else_clause(ctx);
        }

        struct ast *result =
            ast_create_if(condition, then_content, else_content);
        return result;
    }

    // Eventual else content

    struct ast *result = NULL;

    if (token->type == TOKEN_ELSE)
    {
        token = POP_TOKEN(); // eat else
        result = parse_compound_list(ctx);
    }

    if (result == NULL)
        result = ast_create_void();

    return result;
}