Implement parsing

This commit is contained in:
Aaron Cook 2025-09-15 01:17:36 -05:00
commit cab6950b6e

548
src/empack.c Normal file
View file

@ -0,0 +1,548 @@
#include <stdio.h>
#include <string.h>
#include "empack.h"
#define EMPACK_POS_FIXINT_MAX (0x7f)
#define EMPACK_FIXMAP_MAX (0x8f)
#define EMPACK_FIXARRAY_MAX (0x9f)
#define EMPACK_FIXSTR_MAX (0xbf)
#define EMPACK_NEG_FIXINT_MAX (0xff)
/** Attempt to allocate size bytes from the memory arena, NULL if out of memory */
static void *empack_allocate(empack_t *empack, uint32_t size);
/** Copy src to dst but with reversed byte order */
static void memswap(uint8_t *dst, const uint8_t *src, uint32_t len);
/** Convert a big-endian UINT16 buffer to native uint16_t */
static uint16_t buf_to_u16(const uint8_t *buf);
/** Convert a big-endian UINT32 buffer to native uint32_t */
static uint32_t buf_to_u32(const uint8_t *buf);
/** Convert a big-endian UINT64 buffer to native uint64_t */
static uint64_t buf_to_u64(const uint8_t *buf);
/** Convert a big-endian INT16 buffer to native int16_t */
static int16_t buf_to_i16(const uint8_t *buf);
/** Convert a big-endian INT32 buffer to native int32_t */
static int32_t buf_to_i32(const uint8_t *buf);
/** Convert a big-endian INT64 buffer to native int64_t */
static int64_t buf_to_i64(const uint8_t *buf);
/** Convert a big-endian FLOAT32 buffer to native float */
static float buf_to_float(const uint8_t *buf);
/** Convert a big-endian FLOAT64 buffer to native double */
static double buf_to_double(const uint8_t *buf);
void
empack_init(empack_t *empack, void *arena, uint32_t arena_sz)
{
empack->arena = arena;
empack->arena_sz = arena_sz;
empack->arena_used = 0;
memset(arena, 0, arena_sz);
}
empack_status_e
empack_parse(empack_t *empack, const void *msgpack, uint32_t len)
{
const uint8_t *buf = (const uint8_t *)msgpack;
uint32_t p = 0;
empack_status_e ret;
empack_obj_t *cur = &empack->root;
empack_obj_t *parent = NULL;
uint32_t tokens_needed = 1;
while (tokens_needed > 0) {
uint32_t len_used;
if (parent) {
switch (parent->type) {
case EMPACK_TYPE_FIXARRAY:
case EMPACK_TYPE_ARRAY16:
case EMPACK_TYPE_ARRAY32:
{
empack_array_t *parr = &parent->val.array;
if (parr->object_count == parr->object_count_real) {
ret = EMPACK_STATUS_ERROR;
goto empack_parse_return;
}
cur = &parr->objects[parr->object_count_real];
cur->parent = parent;
parr->object_count_real++;
if (parr->object_count == parr->object_count_real) {
parent = parent->parent;
}
break;
}
case EMPACK_TYPE_FIXMAP:
case EMPACK_TYPE_MAP16:
case EMPACK_TYPE_MAP32:
{
empack_map_t *pmap = &parent->val.map;
if (pmap->pair_count == pmap->pair_count_real) {
ret = EMPACK_STATUS_ERROR;
goto empack_parse_return;
}
if (pmap->pairs[pmap->pair_count_real].key.type == EMPACK_TYPE_ERROR) {
/* We haven't parsed the key yet */
cur = &pmap->pairs[pmap->pair_count_real].key;
} else {
cur = &pmap->pairs[pmap->pair_count_real].value;
pmap->pair_count_real++;
}
cur->parent = parent;
if (pmap->pair_count == pmap->pair_count_real) {
parent = parent->parent;
}
break;
}
default:
ret = EMPACK_STATUS_ERROR;
goto empack_parse_return;
break;
}
}
len_used = len - p;
ret = empack_parse_one(cur, &buf[p], &len_used);
if ((ret != EMPACK_STATUS_COMPLETE) && (ret != EMPACK_STATUS_EXTRA)) {
goto empack_parse_return;
}
p += len_used;
tokens_needed--;
if ((tokens_needed == 0) && (ret == EMPACK_STATUS_EXTRA)) {
goto empack_parse_return;
}
switch (cur->type) {
case EMPACK_TYPE_FIXARRAY:
case EMPACK_TYPE_ARRAY16:
case EMPACK_TYPE_ARRAY32:
{
empack_array_t *parr = &cur->val.array;
parr->objects = empack_allocate(empack, sizeof(empack_obj_t) * parr->object_count);
if (parr->objects == NULL) {
ret = EMPACK_STATUS_NOMEM;
goto empack_parse_return;
}
tokens_needed += parr->object_count;
parent = cur;
break;
}
case EMPACK_TYPE_FIXMAP:
case EMPACK_TYPE_MAP16:
case EMPACK_TYPE_MAP32:
{
empack_map_t *pmap = &cur->val.map;
pmap->pairs = empack_allocate(empack, sizeof(empack_map_t) * pmap->pair_count);
if (pmap->pairs == NULL) {
ret = EMPACK_STATUS_NOMEM;
goto empack_parse_return;
}
/* Initialize the key objects to a type of error so we can tell if
* we've already parsed them when we loop back around to parse the
* value */
for (uint32_t i = 0; i < pmap->pair_count; i++) {
pmap->pairs[i].key.type = EMPACK_TYPE_ERROR;
}
break;
}
default:
break;
}
}
/* prioritize errors over incomplete status */
if ((ret == EMPACK_STATUS_COMPLETE) && (p < len)) {
ret = EMPACK_STATUS_WANT_MORE;
}
empack_parse_return:
return (ret);
}
empack_status_e
empack_parse_one(empack_obj_t *object, const void *msgpack, uint32_t *len)
{
const uint8_t *buf = (const uint8_t *)msgpack;
uint32_t p = 0;
empack_status_e ret = EMPACK_STATUS_EXTRA;
memset(object, 0, sizeof(*object));
#define EMPACK_PARSEONE_ASSERT(need) do { \
if ((p + (need)) > (*len)) { \
ret = EMPACK_STATUS_ERROR; \
goto empack_parse_one_return; \
} \
} while (0)
EMPACK_PARSEONE_ASSERT(1);
/* this casting will be validated for object types with a range */
object->type = (empack_type_e)buf[p];
p += 1;
switch (object->type) {
case EMPACK_TYPE_NIL:
case EMPACK_TYPE_TRUE:
case EMPACK_TYPE_FALSE:
/* no extra value associated for these types */
break;
case EMPACK_TYPE_BIN8:
EMPACK_PARSEONE_ASSERT(1);
object->val.bin.size = buf[p];
p += 1;
EMPACK_PARSEONE_ASSERT(object->val.bin.size);
object->val.bin.data = &buf[p];
p += object->val.bin.size;
break;
case EMPACK_TYPE_BIN16:
EMPACK_PARSEONE_ASSERT(2);
object->val.bin.size = buf_to_u16(&buf[p]);
p += 2;
EMPACK_PARSEONE_ASSERT(object->val.bin.size);
object->val.bin.data = &buf[p];
p += object->val.bin.size;
break;
case EMPACK_TYPE_BIN32:
EMPACK_PARSEONE_ASSERT(4);
object->val.bin.size = buf_to_u32(&buf[p]);
p += 4;
EMPACK_PARSEONE_ASSERT(object->val.bin.size);
object->val.bin.data = &buf[p];
p += object->val.bin.size;
break;
case EMPACK_TYPE_EXT8:
EMPACK_PARSEONE_ASSERT(1);
object->val.ext.type = buf[p];
p += 1;
EMPACK_PARSEONE_ASSERT(1);
object->val.ext.size = buf[p];
p += 1;
EMPACK_PARSEONE_ASSERT(object->val.ext.size);
object->val.ext.data = &buf[p];
p += object->val.ext.size;
break;
case EMPACK_TYPE_EXT16:
EMPACK_PARSEONE_ASSERT(1);
object->val.ext.type = buf[p];
p += 1;
EMPACK_PARSEONE_ASSERT(2);
object->val.ext.size = buf_to_u16(&buf[p]);
p += 2;
EMPACK_PARSEONE_ASSERT(object->val.ext.size);
object->val.ext.data = &buf[p];
p += object->val.ext.size;
break;
case EMPACK_TYPE_EXT32:
EMPACK_PARSEONE_ASSERT(1);
object->val.ext.type = buf[p];
p += 1;
EMPACK_PARSEONE_ASSERT(4);
object->val.ext.size = buf_to_u32(&buf[p]);
p += 4;
EMPACK_PARSEONE_ASSERT(object->val.ext.size);
object->val.ext.data = &buf[p];
p += object->val.ext.size;
break;
case EMPACK_TYPE_FLOAT32:
EMPACK_PARSEONE_ASSERT(4);
object->val.float32 = buf_to_float(&buf[p]);
p += 4;
break;
case EMPACK_TYPE_FLOAT64:
EMPACK_PARSEONE_ASSERT(8);
object->val.float64 = buf_to_double(&buf[p]);
p += 8;
break;
case EMPACK_TYPE_UINT8:
EMPACK_PARSEONE_ASSERT(1);
object->val.uint8 = buf[p];
p += 1;
break;
case EMPACK_TYPE_UINT16:
EMPACK_PARSEONE_ASSERT(2);
object->val.uint16 = buf_to_u16(&buf[p]);
p += 2;
break;
case EMPACK_TYPE_UINT32:
EMPACK_PARSEONE_ASSERT(4);
object->val.uint16 = buf_to_u32(&buf[p]);
p += 4;
break;
case EMPACK_TYPE_UINT64:
EMPACK_PARSEONE_ASSERT(8);
object->val.uint64 = buf_to_u64(&buf[p]);
p += 8;
break;
case EMPACK_TYPE_INT8:
EMPACK_PARSEONE_ASSERT(1);
object->val.int8 = (int8_t)buf[p];
p += 1;
break;
case EMPACK_TYPE_INT16:
EMPACK_PARSEONE_ASSERT(2);
object->val.int16 = buf_to_i16(&buf[p]);
p += 2;
break;
case EMPACK_TYPE_INT32:
EMPACK_PARSEONE_ASSERT(4);
object->val.int32 = buf_to_i32(&buf[p]);
p += 4;
break;
case EMPACK_TYPE_INT64:
EMPACK_PARSEONE_ASSERT(8);
object->val.int64 = buf_to_i64(&buf[p]);
p += 8;
break;
case EMPACK_TYPE_FIXEXT1:
EMPACK_PARSEONE_ASSERT(1);
object->val.ext.type = buf[p];
p += 1;
EMPACK_PARSEONE_ASSERT(1);
object->val.ext.data = &buf[p];
object->val.ext.size = 1;
p += 1;
break;
case EMPACK_TYPE_FIXEXT2:
EMPACK_PARSEONE_ASSERT(1);
object->val.ext.type = buf[p];
p += 1;
EMPACK_PARSEONE_ASSERT(2);
object->val.ext.data = &buf[p];
object->val.ext.size = 2;
p += 2;
break;
case EMPACK_TYPE_FIXEXT4:
EMPACK_PARSEONE_ASSERT(1);
object->val.ext.type = buf[p];
p += 1;
EMPACK_PARSEONE_ASSERT(4);
object->val.ext.data = &buf[p];
object->val.ext.size = 4;
p += 4;
break;
case EMPACK_TYPE_FIXEXT8:
EMPACK_PARSEONE_ASSERT(1);
object->val.ext.type = buf[p];
p += 1;
EMPACK_PARSEONE_ASSERT(8);
object->val.ext.data = &buf[p];
object->val.ext.size = 8;
p += 8;
break;
case EMPACK_TYPE_FIXEXT16:
EMPACK_PARSEONE_ASSERT(1);
object->val.ext.type = buf[p];
p += 1;
EMPACK_PARSEONE_ASSERT(16);
object->val.ext.data = &buf[p];
object->val.ext.size = 16;
p += 16;
break;
case EMPACK_TYPE_STR8:
EMPACK_PARSEONE_ASSERT(1);
object->val.str.size = buf[p];
p += 1;
EMPACK_PARSEONE_ASSERT(object->val.str.size);
object->val.str.data = (const char *)&buf[p];
p += object->val.str.size;
break;
case EMPACK_TYPE_STR16:
EMPACK_PARSEONE_ASSERT(2);
object->val.str.size = buf_to_u16(&buf[p]);
p += 2;
EMPACK_PARSEONE_ASSERT(object->val.str.size);
object->val.str.data = (const char *)&buf[p];
p += object->val.str.size;
break;
case EMPACK_TYPE_STR32:
EMPACK_PARSEONE_ASSERT(4);
object->val.str.size = buf_to_u32(&buf[p]);
p += 4;
EMPACK_PARSEONE_ASSERT(object->val.str.size);
object->val.str.data = (const char *)&buf[p];
p += object->val.str.size;
break;
case EMPACK_TYPE_ARRAY16:
EMPACK_PARSEONE_ASSERT(2);
object->val.array.object_count = buf_to_u16(&buf[p]);
p += 2;
break;
case EMPACK_TYPE_ARRAY32:
EMPACK_PARSEONE_ASSERT(4);
object->val.array.object_count = buf_to_u32(&buf[p]);
p += 4;
break;
case EMPACK_TYPE_MAP16:
EMPACK_PARSEONE_ASSERT(2);
object->val.map.pair_count = buf_to_u16(&buf[p]);
p += 2;
break;
case EMPACK_TYPE_MAP32:
EMPACK_PARSEONE_ASSERT(4);
object->val.map.pair_count = buf_to_u32(&buf[p]);
p += 4;
break;
default:
if ((object->type >= EMPACK_TYPE_POS_FIXINT) && (object->type <= EMPACK_POS_FIXINT_MAX)) {
object->val.uint8 = (uint8_t)object->type;
object->type = EMPACK_TYPE_POS_FIXINT;
}
else if ((object->type >= EMPACK_TYPE_FIXMAP) && (object->type <= EMPACK_FIXMAP_MAX)) {
object->val.map.pair_count = ((uint8_t)object->type) & 0x0f;
object->type = EMPACK_TYPE_FIXMAP;
}
else if ((object->type >= EMPACK_TYPE_FIXSTR) && (object->type <= EMPACK_FIXSTR_MAX)) {
object->val.str.size = ((uint8_t)object->type) & 0x1f;
object->type = EMPACK_TYPE_FIXSTR;
}
else if ((object->type >= EMPACK_TYPE_NEG_FIXINT) && (object->type <= EMPACK_NEG_FIXINT_MAX)) {
object->val.int8 = (int8_t)object->type;
object->type = EMPACK_TYPE_NEG_FIXINT;
}
else {
object->type = EMPACK_TYPE_ERROR;
ret = EMPACK_STATUS_ERROR;
goto empack_parse_one_return;
}
break;
}
empack_parse_one_return:
if ((ret == EMPACK_STATUS_EXTRA) && (p == (*len))) {
ret = EMPACK_STATUS_COMPLETE;
}
(*len) = p;
return ret;
}
static void *
empack_allocate(empack_t *empack, uint32_t size)
{
void *ret;
if (size > (empack->arena_sz - empack->arena_used)) {
return NULL;
}
ret = &((uint8_t *)empack->arena)[empack->arena_used];
empack->arena_used += size;
return ret;
}
static void
memswap(uint8_t *dst, const uint8_t *src, uint32_t len)
{
for (uint32_t i = 0; i < len; i++) {
dst[i] = src[len - i - 1];
}
}
static uint16_t
buf_to_u16(const uint8_t *buf)
{
uint16_t ret;
memswap((uint8_t *)&ret, buf, 2);
return ret;
}
static uint32_t
buf_to_u32(const uint8_t *buf)
{
uint32_t ret;
memswap((uint8_t *)&ret, buf, 4);
return ret;
}
static uint64_t
buf_to_u64(const uint8_t *buf)
{
uint64_t ret;
memswap((uint8_t *)&ret, buf, 8);
return ret;
}
static int16_t
buf_to_i16(const uint8_t *buf)
{
int16_t ret;
memswap((uint8_t *)&ret, buf, 2);
return ret;
}
static int32_t
buf_to_i32(const uint8_t *buf)
{
int32_t ret;
memswap((uint8_t *)&ret, buf, 4);
return ret;
}
static int64_t
buf_to_i64(const uint8_t *buf)
{
int64_t ret;
memswap((uint8_t *)&ret, buf, 8);
return ret;
}
static float
buf_to_float(const uint8_t *buf)
{
float ret;
memswap((uint8_t *)&ret, buf, 4);
return ret;
}
static double
buf_to_double(const uint8_t *buf)
{
double ret;
memswap((uint8_t *)&ret, buf, 8);
return ret;
}