tiffread/src/tiff/tiffread.c

582 lines
16 KiB
C

#include "tiffread.h"
#include "aliases.h"
#include "endianness.h"
#include "image.h"
#include "mem_arena.h"
#include <assert.h>
#include <math.h>
#include <netinet/in.h>
#include <stdbool.h>
#include <stddef.h>
#include <stdio.h>
#include <string.h>
#define TIFF_MAGIC 0x002a
#define TIFF_FILENAME_MIN_LENGTH 6
#define TIFF_EXT_MAX_LENGTH 5
#define IS_TIFF_EXTENSION(EXT) \
(strncmp(EXT, ".tif", 4) == 0 || strncmp(EXT, ".tiff", 5) == 0)
#define NULL_TIFF_HEADER ((TiffHdr){0})
#define IS_NULL_HEADER(HDR) (HDR.order == 0)
#define NULL_TIFF_IFD ((TiffIFD){0})
#define IS_NULL_IFD(IFD) (IFD.count == 0)
#define NULL_TIFF_IMAGE ((TiffImage){0})
#define IS_NULL_IMAGE(IMG) (IMG.type == TIFF_IMAGE_TYPE_INVALID)
#define INVALID_SAMPLE_COUNT 0
#define INVALID_ROWS_PER_STRIP 0
#define INVALID_STRIP_COUNT 0
#define INVALID_ALPHA_OFFSET -1
#define TEMP_ARENA_CAPACITY (20 * 1024 * 1024)
typedef struct tiff_reader TiffReader;
struct tiff_reader {
FILE *fp;
TiffHdr header;
TiffIFD ifd;
TiffImage img;
Pixel *pixels;
};
typedef struct strip_data_field StripDataField;
struct strip_data_field {
u32 *strip_value;
const ShortLongValue *value_from_file;
u32 strip_index;
u32 length_in_bytes;
u16 type_byte_count;
};
internal TiffHdr read_tiff_header(const TiffReader *reader);
internal TiffIFD read_ifd(const TiffReader *reader, Arena *arena);
internal TiffImage read_ifd_fields(const TiffReader *reader);
internal Pixel *load_image_pixels(TiffReader *reader, Arena *arena);
internal bool read_strip_data(TiffReader *reader, Arena *arena);
internal void read_strips(const TiffReader *reader, Pixel *buf);
internal void read_strip_data_field(const TiffReader *reader,
StripDataField *field);
internal bool read_field(const TiffField *field, TiffImage *img);
internal bool validate_image_type(const TiffImage *img);
internal void read_from_file_with_offset(FILE *fp, void *dst, u64 count,
u64 offset);
Image *read_baseline_tiff(const char *file, Arena *arena) {
Image *img_out = NULL;
if (!file || !arena) {
goto READ_BASELINE_RETURN_IMG;
}
u64 name_length = strlen(file);
if (name_length < TIFF_FILENAME_MIN_LENGTH) {
goto READ_BASELINE_RETURN_IMG;
}
const char *ext = NULL;
for (u32 i = name_length - 1; i >= name_length - TIFF_FILENAME_MIN_LENGTH;
--i) {
if (file[i] == '.') {
ext = &(file[i]);
break;
}
}
if (!ext || !IS_TIFF_EXTENSION(ext)) {
goto READ_BASELINE_RETURN_IMG;
}
TiffReader reader = {0};
reader.fp = fopen(file, "rb");
if (!reader.fp) {
goto READ_BASELINE_RETURN_IMG;
}
Arena *temp_arena = NULL;
if (!wapp_mem_arena_init(&temp_arena, TEMP_ARENA_CAPACITY)) {
goto READ_BASELINE_FILE_CLEANUP;
}
reader.header = read_tiff_header(&reader);
if (IS_NULL_HEADER(reader.header)) {
goto READ_BASELINE_DESTROY_ARENA;
}
reader.ifd = read_ifd(&reader, temp_arena);
if (IS_NULL_IFD(reader.ifd)) {
goto READ_BASELINE_DESTROY_ARENA;
}
reader.img = read_ifd_fields(&reader);
if (IS_NULL_IMAGE(reader.img)) {
goto READ_BASELINE_DESTROY_ARENA;
}
assert((reader.img.type == TIFF_IMAGE_TYPE_RGB) &&
"Currently, only RGB images are supported");
reader.pixels = load_image_pixels(&reader, temp_arena);
if (!reader.pixels) {
goto READ_BASELINE_DESTROY_ARENA;
}
img_out = create_image(reader.img.image_width, reader.img.image_length,
reader.pixels, arena);
READ_BASELINE_DESTROY_ARENA:
wapp_mem_arena_destroy(&temp_arena);
READ_BASELINE_FILE_CLEANUP:
fclose(reader.fp);
READ_BASELINE_RETURN_IMG:
return img_out;
}
internal TiffHdr read_tiff_header(const TiffReader *reader) {
TiffHdr header = NULL_TIFF_HEADER;
read_from_file_with_offset(reader->fp, &header, sizeof(TiffHdr), 0);
switch (header.order) {
case TIFF_ORDER_LITTLE_ENDIAN:
if (IS_BIG_ENDIAN) {
header.magic = htons(header.magic);
header.first_ifd_offset = htonl(header.first_ifd_offset);
}
break;
case TIFF_ORDER_BIG_ENDIAN:
if (IS_LITTLE_ENDIAN) {
header.magic = ntohs(header.magic);
header.first_ifd_offset = ntohl(header.first_ifd_offset);
}
break;
default:
return NULL_TIFF_HEADER;
}
if (header.magic != TIFF_MAGIC) {
return NULL_TIFF_HEADER;
}
return header;
}
internal TiffIFD read_ifd(const TiffReader *reader, Arena *arena) {
TiffIFD ifd = NULL_TIFF_IFD;
read_from_file_with_offset(reader->fp, &(ifd.count), sizeof(ifd.count),
reader->header.first_ifd_offset);
switch (reader->header.order) {
case TIFF_ORDER_LITTLE_ENDIAN:
if (IS_BIG_ENDIAN) {
ifd.count = htons(ifd.count);
}
break;
case TIFF_ORDER_BIG_ENDIAN:
if (IS_LITTLE_ENDIAN) {
ifd.count = ntohs(ifd.count);
}
break;
}
u64 field_byte_count = sizeof(TiffField) * ifd.count;
ifd.fields = (TiffField *)wapp_mem_arena_alloc(arena, field_byte_count);
if (!(ifd.fields)) {
return NULL_TIFF_IFD;
}
fread(ifd.fields, field_byte_count, 1, reader->fp);
fread(&(ifd.next_ifd), sizeof(ifd.next_ifd), 1, reader->fp);
return ifd;
}
internal TiffImage read_ifd_fields(const TiffReader *reader) {
TiffImage img_out = NULL_TIFF_IMAGE;
img_out.type = TIFF_IMAGE_TYPE_BILEVEL;
for (u64 i = 0; i < reader->ifd.count; ++i) {
TiffField *field = &(reader->ifd.fields[i]);
switch (reader->header.order) {
case TIFF_ORDER_BIG_ENDIAN:
if (IS_LITTLE_ENDIAN) {
field->tag = ntohs(field->tag);
field->type = ntohs(field->type);
field->count = ntohl(field->count);
if (field_types[field->type].byte_count > 2 ||
field_types[field->type].byte_count * field->count > 4) {
field->value_offset = ntohl(field->value_offset);
} else {
field->value_offset = ntohs(field->value_offset);
}
}
break;
case TIFF_ORDER_LITTLE_ENDIAN:
if (IS_BIG_ENDIAN) {
field->tag = htons(field->tag);
field->type = htons(field->type);
field->count = htonl(field->count);
if (field_types[field->type].byte_count > 2 ||
field_types[field->type].byte_count * field->count > 4) {
field->value_offset = htonl(field->value_offset);
} else {
field->value_offset = htons(field->value_offset);
}
}
break;
}
if (!read_field(field, &img_out)) {
img_out = NULL_TIFF_IMAGE;
goto READ_FIELDS_RETURN_IMAGE;
}
}
if (!validate_image_type(&img_out)) {
img_out = NULL_TIFF_IMAGE;
goto READ_FIELDS_RETURN_IMAGE;
}
#ifdef DEBUG
// clang-format off
printf("SIZE (width, height) : %u, %u\n", img_out.image_width, img_out.image_length);
printf("SAMPLES (bits per sample, count, is offset) : %u, %u, %u\n", img_out.bits_per_sample, img_out.sample_count, img_out.bits_per_sample_offset);
printf("EXTRA SAMPLES (samples, count, is offset) : %u, %u, %u\n", img_out.extra_samples, img_out.extra_samples_count, img_out.extra_samples_offset);
printf("SAMPLES PER PIXEL : %u\n", img_out.sample_count);
printf("PHOTOMETRIC INTERPRETATION : %u\n", img_out.photometric_interpretation);
printf("ROWS PER STRIP (rows, strip count) : %u, %u\n", img_out.rows_per_strip, img_out.strip_count);
printf("STRIP OFFSETS (offsets, is offset) : %u, %u\n", img_out.strip_offsets.long_val, img_out.strip_offsets_offset);
printf("STRIP BYTES (byte count, is offset) : %u, %u\n", img_out.strip_byte_counts.long_val, img_out.strip_byte_counts_offset);
// clang-format on
#endif
READ_FIELDS_RETURN_IMAGE:
return img_out;
}
internal Pixel *load_image_pixels(TiffReader *reader, Arena *arena) {
Pixel *buf = NULL;
u64 img_byte_count =
sizeof(Pixel) * reader->img.image_width * reader->img.image_length;
if (!read_strip_data(reader, arena)) {
goto LOAD_IMAGE_PIXELS_RETURN;
}
#ifdef DEBUG
for (u32 i = 0; i < reader->img.strip_count; ++i) {
printf("%u, %u\n", reader->img.strips[i].offset,
reader->img.strips[i].byte_count);
}
#endif
buf = wapp_mem_arena_alloc(arena, img_byte_count);
if (!buf) {
goto LOAD_IMAGE_PIXELS_RETURN;
}
read_strips(reader, buf);
LOAD_IMAGE_PIXELS_RETURN:
return buf;
}
internal bool read_strip_data(TiffReader *reader, Arena *arena) {
reader->img.strips =
wapp_mem_arena_alloc(arena, sizeof(TiffStrip) * reader->img.strip_count);
if (!(reader->img.strips)) {
return false;
}
u32 offsets_total_bytes =
reader->img.strip_count * reader->img.strip_offsets_type_byte_count;
u32 byte_count_total_bytes =
reader->img.strip_count * reader->img.strip_byte_count_type_byte_count;
if ((!(reader->img.strip_offsets_offset) &&
offsets_total_bytes > sizeof(u32)) ||
(!(reader->img.strip_byte_counts_offset) &&
byte_count_total_bytes > sizeof(u32))) {
return false;
}
u16 offset_size = reader->img.strip_offsets_type_byte_count;
u16 counts_size = reader->img.strip_byte_count_type_byte_count;
for (u32 i = 0; i < reader->img.strip_count; ++i) {
TiffStrip *strip = &(reader->img.strips[i]);
StripDataField offsets = {
.strip_value = &(strip->offset),
.value_from_file = &(reader->img.strip_offsets),
.strip_index = i,
.length_in_bytes = offsets_total_bytes,
.type_byte_count = offset_size,
};
StripDataField counts = {
.strip_value = &(strip->byte_count),
.value_from_file = &(reader->img.strip_byte_counts),
.strip_index = i,
.length_in_bytes = byte_count_total_bytes,
.type_byte_count = counts_size,
};
read_strip_data_field(reader, &offsets);
read_strip_data_field(reader, &counts);
}
return true;
}
internal void read_strips(const TiffReader *reader, Pixel *buf) {
u64 position = 0;
for (u64 i = 0; i < reader->img.strip_count; ++i) {
const TiffStrip *strip = &(reader->img.strips[i]);
for (u64 j = 0; j < strip->byte_count / reader->img.sample_count; ++j) {
Pixel *p = &(buf[position]);
read_from_file_with_offset(reader->fp, p, reader->img.sample_count,
strip->offset + j * reader->img.sample_count);
if (reader->img.sample_count == 3) {
p->a = 255;
}
++position;
}
}
}
internal void read_strip_data_field(const TiffReader *reader,
StripDataField *field) {
u16 tiff_long_byte_count = field_types[TIFF_FIELD_TYPE_LONG].byte_count;
bool value_is_file_offset = field->length_in_bytes > tiff_long_byte_count;
u32 offset = field->type_byte_count * field->strip_index;
if (!value_is_file_offset) {
u8 *value = (u8 *)&(field->value_from_file->long_val);
memcpy(field->strip_value, value + offset, field->type_byte_count);
return;
}
read_from_file_with_offset(reader->fp, field->strip_value,
field->type_byte_count,
field->value_from_file->long_val + offset);
u16 tiff_short_byte_count = field_types[TIFF_FIELD_TYPE_SHORT].byte_count;
switch (reader->header.order) {
case TIFF_ORDER_BIG_ENDIAN:
if (IS_LITTLE_ENDIAN) {
*(field->strip_value) = field->type_byte_count > tiff_short_byte_count
? ntohl(*(field->strip_value))
: ntohs(*(field->strip_value));
}
break;
case TIFF_ORDER_LITTLE_ENDIAN:
if (IS_BIG_ENDIAN) {
*(field->strip_value) = field->type_byte_count > tiff_short_byte_count
? htonl(*(field->strip_value))
: htons(*(field->strip_value));
}
break;
}
}
internal bool read_field(const TiffField *field, TiffImage *img) {
switch (field->tag) {
case TIFF_PUBLIC_TAG_IMAGE_WIDTH:
if (field->count > 1) {
return false;
}
img->image_width = field->value_offset;
break;
case TIFF_PUBLIC_TAG_IMAGE_LENGTH:
if (field->count > 1) {
return false;
}
img->image_length = field->value_offset;
break;
case TIFF_PUBLIC_TAG_BITS_PER_SAMPLE:
img->type |= field->tag;
if (img->sample_count == INVALID_SAMPLE_COUNT) {
img->sample_count = field->count;
} else if (img->sample_count != field->count) {
return false;
}
img->bits_per_sample = field->value_offset;
img->bits_per_sample_offset =
img->sample_count * field_types[field->type].byte_count > 4;
break;
case TIFF_PUBLIC_TAG_COMPRESSION:
if (field->count > 1 ||
field->value_offset != TIFF_COMPRESSION_UNCOMPRESSED) {
return false;
}
img->compression = field->value_offset;
break;
case TIFF_PUBLIC_TAG_PHOTOMETRIC_INTERPRETATION:
if (field->count > 1) {
return false;
}
img->photometric_interpretation = field->value_offset;
break;
case TIFF_PUBLIC_TAG_STRIP_OFFSETS:
img->strip_offsets_type_byte_count = field_types[field->type].byte_count;
if (img->strip_offsets_type_byte_count == sizeof(u16)) {
img->strip_offsets.short_val = field->value_offset;
} else if (img->strip_offsets_type_byte_count == sizeof(u32)) {
img->strip_offsets.long_val = field->value_offset;
}
if (img->strip_count == INVALID_STRIP_COUNT) {
img->strip_count = field->count;
} else if (img->strip_count != field->count) {
return false;
}
img->strip_offsets_offset =
img->strip_count * img->strip_offsets_type_byte_count > 4;
break;
case TIFF_PUBLIC_TAG_SAMPLES_PER_PIXEL:
img->type |= field->tag;
if (field->count > 1) {
return false;
}
img->sample_count = field->value_offset;
break;
case TIFF_PUBLIC_TAG_ROWS_PER_STRIP:
if (field->count > 1) {
return false;
}
img->rows_per_strip = field->value_offset;
if (img->image_length > 0) {
img->strip_count =
floor((f64)(img->image_length + img->rows_per_strip - 1) /
img->rows_per_strip);
} else {
img->strip_count = INVALID_STRIP_COUNT;
}
break;
case TIFF_PUBLIC_TAG_STRIP_BYTE_COUNTS:
img->strip_byte_count_type_byte_count = field_types[field->type].byte_count;
if (img->strip_byte_count_type_byte_count == sizeof(u16)) {
img->strip_byte_counts.short_val = field->value_offset;
} else if (img->strip_byte_count_type_byte_count == sizeof(u32)) {
img->strip_byte_counts.long_val = field->value_offset;
}
if (img->strip_count == INVALID_STRIP_COUNT) {
img->strip_count = field->count;
} else if (img->strip_count != field->count) {
return false;
}
img->strip_byte_counts_offset =
img->strip_count * img->strip_byte_count_type_byte_count > 4;
break;
case TIFF_PUBLIC_TAG_COLOR_MAP:
img->type |= field->tag;
break;
case TIFF_PUBLIC_TAG_EXTRA_SAMPLES:
img->extra_samples_count = field->count;
img->extra_samples = field->value_offset;
img->extra_samples_offset =
img->extra_samples_count * field_types[field->type].byte_count > 4;
break;
case TIFF_PUBLIC_TAG_PLANAR_CONFIGURATION:
if (field->count > 1 || field->value_offset != TIFF_PLANAR_CONFIG_CHUNKY) {
return false;
}
break;
}
return true;
}
internal bool validate_image_type(const TiffImage *img) {
if (img->type < TIFF_IMAGE_TYPE_BILEVEL || img->type > TIFF_IMAGE_TYPE_RGB) {
return false;
}
switch (img->type) {
case TIFF_IMAGE_TYPE_BILEVEL:
case TIFF_IMAGE_TYPE_GRAYSCALE:
if (img->photometric_interpretation >
TIFF_PHOTOMETRIC_INTERPRETATION_BLACK_IS_ZERO) {
return false;
}
break;
case TIFF_IMAGE_TYPE_PALETTE:
if (img->photometric_interpretation !=
TIFF_PHOTOMETRIC_INTERPRETATION_RGB_PALETTE) {
return false;
}
break;
case TIFF_IMAGE_TYPE_RGB:
if (img->photometric_interpretation !=
TIFF_PHOTOMETRIC_INTERPRETATION_RGB) {
return false;
}
break;
default:
return false;
}
return true;
}
internal void read_from_file_with_offset(FILE *fp, void *dst, u64 count,
u64 offset) {
if (!fp || !dst) {
return;
}
fseek(fp, offset, SEEK_SET);
fread(dst, count, 1, fp);
}