crater/src/assembler/preprocessor.c

/* Copyright (C) 2014-2015 Ben Kurtovic <ben.kurtovic@gmail.com>
   Released under the terms of the MIT License. See LICENSE for details. */

#include <libgen.h>
#include <limits.h>
#include <stdbool.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#include "preprocessor.h"
#include "directives.h"
#include "errors.h"
#include "io.h"
#include "parse_util.h"
#include "../logging.h"
#include "../util.h"

/* Helper macros for preprocess() */

#define FAIL_ON_COND_(cond, err_desc)                                         \
    if ((cond)) {                                                             \
        ei = error_info_create(line, ET_PREPROC, err_desc);                   \
        goto cleanup;                                                         \
    }

#define CALL_GENERIC_PARSER_(arg_type)                                        \
    parse_##arg_type((arg_type*) &arg, line, directive)

#define CALL_SPECIFIC_PARSER_(arg_type, parser)                               \
    parse_##parser((arg_type*) &arg, line)

#define DISPATCH_(first, second, target, ...) target

#define CALL_PARSER_(...)                                                     \
    DISPATCH_(__VA_ARGS__, CALL_SPECIFIC_PARSER_, CALL_GENERIC_PARSER_,       \
              __VA_ARGS__)(__VA_ARGS__)

#define VALIDATE(func)                                                        \
    FAIL_ON_COND_(!(func(arg)), ED_PP_BAD_ARG)

#define CHECK_RANGE(bound)                                                    \
    FAIL_ON_COND_(arg > bound, ED_PP_ARG_RANGE)

#define USE_PARSER(...)                                                       \
    FAIL_ON_COND_(!CALL_PARSER_(__VA_ARGS__), ED_PP_BAD_ARG)

#define PARSER_BRANCH(arg_type, true_part, false_part)                        \
    if (CALL_PARSER_(arg_type)) {true_part} else {false_part}

#define BEGIN_DIRECTIVE_BLOCK                                                 \
    ssize_t first_ctr = -1;                                                   \
    if (0) {}

#define BEGIN_DIRECTIVE(d, arg_type, dest_loc, auto_val)                      \
    else if (first_ctr++, IS_DIRECTIVE(line, d)) {                            \
        directive = d;                                                        \
        FAIL_ON_COND_(!DIRECTIVE_HAS_ARG(line, directive), ED_PP_NO_ARG)      \
        arg_type arg;                                                         \
        arg_type* dest = &(dest_loc);                                         \
        if (DIRECTIVE_IS_AUTO(line, directive)) {                             \
            arg = auto_val;                                                   \
        } else {

#define END_DIRECTIVE                                                         \
        }                                                                     \
        if (firsts[first_ctr] && *dest != arg) {                              \
            ei = error_info_create(line, ET_PREPROC, ED_PP_DUPLICATE);        \
            error_info_append(ei, firsts[first_ctr]);                         \
            goto cleanup;                                                     \
        }                                                                     \
        *dest = arg;                                                          \
        firsts[first_ctr] = line;                                             \
    }

#define END_DIRECTIVE_BLOCK                                                   \
    else FAIL_ON_COND_(true, ED_PP_UNKNOWN)

/*
    Preprocess a single source line (source, length) into a normalized ASMLine.

    *Only* the data and length fields in the ASMLine object are populated. The
    normalization process converts tabs to spaces, lowercases all alphabetical
    characters, and removes runs of multiple spaces (outside of string
    literals), strips comments, and other things.

    Return NULL if an ASM line was not generated from the source, i.e. if it is
    blank after being stripped.
*/
static ASMLine* normalize_line(const char *source, size_t length)
{
    char *data = malloc(sizeof(char) * length);
    if (!data)
        OUT_OF_MEMORY()

    size_t si, di, slashes = 0;
    bool has_content = false, space_pending = false, in_string = false;
    for (si = di = 0; si < length; si++) {
        char c = source[si];

        if (c == '\\')
            slashes++;
        else
            slashes = 0;

        if (in_string) {
            if (c == '"' && (slashes % 2) == 0)
                in_string = false;

            data[di++] = c;
        } else {
            if (c == ';')
                break;
            if (c == '"' && (slashes % 2) == 0)
                in_string = true;
            if (c >= 'A' && c <= 'Z')
                c += 'a' - 'A';

            if (c == ' ' || c == '\t')
                space_pending = true;
            else {
                if (space_pending) {
                    if (has_content)
                        data[di++] = ' ';
                    space_pending = false;
                }
                has_content = true;
                data[di++] = c;
            }
        }
    }

    if (!has_content) {
        free(data);
        return NULL;
    }

    ASMLine *line = malloc(sizeof(ASMLine));
    if (!line)
        OUT_OF_MEMORY()

    data = realloc(data, sizeof(char) * di);
    if (!data)
        OUT_OF_MEMORY()

    line->data = data;
    line->length = di;
    return line;
}

/*
    Read and return the target path from an include directive.

    This function allocates a buffer to store the filename; it must be free()'d
    after calling read_source_file(). If a syntax error occurs while trying to
    read the path, it returns NULL.
*/
static char* read_include_path(const ASMLine *line)
{
    size_t maxlen = strlen(line->filename) + line->length, i, start, slashes;
    if (maxlen >= INT_MAX)  // Allows us to safely downcast to int later
        return NULL;

    char *path = malloc(sizeof(char) * maxlen);
    if (!path)
        OUT_OF_MEMORY()

    if (!(i = DIRECTIVE_OFFSET(line, DIR_INCLUDE)))
        goto error;
    if (line->length - i <= 4)  // Not long enough to hold a non-zero argument
        goto error;
    if (line->data[i++] != ' ' || line->data[i++] != '"')
        goto error;

    // TODO: parse escaped characters properly
    for (start = i, slashes = 0; i < line->length; i++) {
        if (line->data[i] == '"' && (slashes % 2) == 0)
            break;
        if (line->data[i] == '\\')
            slashes++;
        else
            slashes = 0;
    }

    if (i != line->length - 1)  // Junk present after closing quote
        goto error;

    char *dup = strdup(line->filename);
    if (!dup)
        OUT_OF_MEMORY()

    // TODO: should normalize filenames in some way to prevent accidental dupes
    snprintf(path, maxlen, "%s/%.*s", dirname(dup), (int) (i - start),
             line->data + start);
    free(dup);
    return path;

    error:
    free(path);
    return NULL;
}

/*
    Return whether the given path has already been loaded.
*/
static bool path_has_been_loaded(
    const char *path, const LineBuffer *root, const ASMInclude *include)
{
    if (!strcmp(path, root->filename))
        return true;

    while (include) {
        if (!strcmp(path, include->lines->filename))
            return true;
        include = include->next;
    }
    return false;
}

/*
    Build a LineBuffer into a ASMLines, normalizing them along the way.

    This function operates recursively to handle includes, but handles no other
    preprocessor directives.

    On success, NULL is returned; *head points to the head of the new ASMLine
    list, and *tail to its tail (assuming it is non-NULL). On error, an
    ErrorInfo object is returned, and *head and *tail are not modified.
    *includes may be updated in either case.
*/
static ErrorInfo* build_asm_lines(
    const LineBuffer *root, const LineBuffer *source, ASMLine **head,
    ASMLine **tail, ASMInclude **includes)
{
    ASMLine dummy = {.next = NULL};
    ASMLine *line, *prev = &dummy;
    const Line *orig, *next_orig = source->lines;

    while ((orig = next_orig)) {
        line = normalize_line(orig->data, orig->length);
        next_orig = orig->next;
        if (!line)
            continue;

        // Populate ASMLine fields not set by normalize_line():
        line->original = orig;
        line->filename = source->filename;
        line->next = NULL;

        if (IS_DIRECTIVE(line, DIR_INCLUDE)) {
            ErrorInfo *ei;
            char *path = read_include_path(line);
            if (!path) {
                ei = error_info_create(line, ET_INCLUDE, ED_INC_BAD_ARG);
                asm_lines_free(line);
                asm_lines_free(dummy.next);
                return ei;
            }

            if (path_has_been_loaded(path, root, *includes)) {
                ei = error_info_create(line, ET_INCLUDE, ED_INC_RECURSION);
                asm_lines_free(line);
                asm_lines_free(dummy.next);
                free(path);
                return ei;
            }

            DEBUG("- reading included file: %s", path)
            LineBuffer *incbuffer = read_source_file(path, false);
            free(path);
            if (!incbuffer) {
                ei = error_info_create(line, ET_INCLUDE, ED_INC_FILE_READ);
                asm_lines_free(line);
                asm_lines_free(dummy.next);
                return ei;
            }

            ASMInclude *include = malloc(sizeof(ASMInclude));
            if (!include)
                OUT_OF_MEMORY()

            include->lines = incbuffer;
            include->next = *includes;
            *includes = include;

            ASMLine *inchead, *inctail;
            if ((ei = build_asm_lines(root, incbuffer, &inchead, &inctail,
                                      includes))) {
                error_info_append(ei, line);
                asm_lines_free(line);
                asm_lines_free(dummy.next);
                return ei;
            }

            prev->next = inchead;
            prev = inctail;
            asm_lines_free(line);  // Destroy only the .include line
        }
        else {
            prev->next = line;
            prev = line;
        }
    }

    *head = dummy.next;
    if (tail)
        *tail = prev;
    return NULL;
}

/*
    Return whether the given header offset is a valid location.
*/
static inline bool is_header_offset_valid(uint16_t offset)
{
    return offset == 0x7FF0 || offset == 0x3FF0 || offset == 0x1FF0;
}

/*
    Preprocess the LineBuffer into ASMLines. Change some state along the way.

    This function processes include directives, so read_source_file() may be
    called multiple times (along with the implications that has), and
    state->includes may be modified.

    On success, NULL is returned. On error, an ErrorInfo object is returned.
    state->lines and state->includes may still be modified.
*/
ErrorInfo* preprocess(AssemblerState *state, const LineBuffer *source)
{
    ErrorInfo* ei = NULL;
    DEBUG("Running preprocessor:")

    if ((ei = build_asm_lines(source, source, &state->lines, NULL,
                              &state->includes)))
        return ei;

    const ASMLine *firsts[NUM_DIRECTIVES];
    for (size_t i = 0; i < NUM_DIRECTIVES; i++)
        firsts[i] = NULL;

    ASMLine dummy = {.next = state->lines};
    ASMLine *prev, *line = &dummy, *next = state->lines, *condemned = NULL;
    const char *directive;

    while ((prev = line, line = next)) {
        next = line->next;
        if (line->data[0] != DIRECTIVE_MARKER)
            continue;
        if (IS_DIRECTIVE(line, DIR_ORIGIN))
            continue;  // Origins are handled by tokenizer

        DEBUG("- handling directive: %.*s", (int) line->length, line->data)

        BEGIN_DIRECTIVE_BLOCK

        BEGIN_DIRECTIVE(DIR_OPTIMIZER, bool, state->optimizer, false)
            USE_PARSER(bool)
        END_DIRECTIVE

        BEGIN_DIRECTIVE(DIR_ROM_SIZE, size_t, state->rom_size, 0)
            USE_PARSER(uint32_t, rom_size)
        END_DIRECTIVE

        BEGIN_DIRECTIVE(DIR_ROM_HEADER, size_t, state->header.offset, DEFAULT_HEADER_OFFSET)
            USE_PARSER(uint16_t)
            VALIDATE(is_header_offset_valid)
        END_DIRECTIVE

        BEGIN_DIRECTIVE(DIR_ROM_CHECKSUM, bool, state->header.checksum, true)
            USE_PARSER(bool)
        END_DIRECTIVE

        BEGIN_DIRECTIVE(DIR_ROM_PRODUCT, uint32_t, state->header.product_code, 0)
            USE_PARSER(uint32_t)
            CHECK_RANGE(160000)
        END_DIRECTIVE

        BEGIN_DIRECTIVE(DIR_ROM_VERSION, uint8_t, state->header.version, 0)
            USE_PARSER(uint8_t)
            CHECK_RANGE(0x10)
        END_DIRECTIVE

        BEGIN_DIRECTIVE(DIR_ROM_REGION, uint8_t, state->header.region, DEFAULT_REGION)
            PARSER_BRANCH(uint8_t, {
                CHECK_RANGE(0x10)
                VALIDATE(region_code_to_string)
            }, {
                USE_PARSER(uint8_t, region_string)
            })
        END_DIRECTIVE

        BEGIN_DIRECTIVE(DIR_ROM_DECLSIZE, uint8_t, state->header.rom_size, 0)
            PARSER_BRANCH(uint8_t, {
                CHECK_RANGE(0x10)
                VALIDATE(size_code_to_bytes)
            }, {
                USE_PARSER(uint8_t, size_code)
            })
        END_DIRECTIVE

        END_DIRECTIVE_BLOCK

        // Remove directive from lines, and schedule it for deletion:
        line->next = condemned;
        condemned = line;
        prev->next = next;
        line = prev;
    }

    // TODO: if giving rom size, check header offset is in rom size range
    // TODO: if giving reported and actual rom size, check reported is <= actual

    state->rom_size = 8;  // TODO

    cleanup:
    asm_lines_free(condemned);
    state->lines = dummy.next;  // Fix list head if first line was a directive

#ifdef DEBUG_MODE
    DEBUG("Dumping ASMLines:")
    const ASMLine *temp = state->lines;
    while (temp) {
        DEBUG("- %-40.*s [%s:%02zu]", (int) temp->length, temp->data,
              temp->filename, temp->original->lineno)
        temp = temp->next;
    }
#endif

    return ei;
}