Commit 423c733d authored by Skirmantas Lauzikas's avatar Skirmantas Lauzikas Committed by Arnim Läuger
Browse files

ESP32: Add SJSON module (#2562)

* Esp32: Add SJSON module

This adds SJSON module taken directly from master

* ESP32: Fixes for sjson lib
Fixed compilation not including config header, thus braking some of libs functionality

* ESP32: Upgraded SJSON to master
parent c9e7dde5
...@@ -173,6 +173,12 @@ config LUA_MODULE_SIGMA_DELTA ...@@ -173,6 +173,12 @@ config LUA_MODULE_SIGMA_DELTA
Includes the sigma_delta module. This module provides access to the Includes the sigma_delta module. This module provides access to the
sigma-delta hardware. sigma-delta hardware.
config LUA_MODULE_SJSON
bool "SJSON module"
default "n"
help
Includes the sjson module.
config LUA_MODULE_SODIUM config LUA_MODULE_SODIUM
bool "Sodium module" bool "Sodium module"
default "n" default "n"
......
#define LUA_LIB
#include "lua.h"
#include "lauxlib.h"
#include "lstring.h"
#ifndef LOCAL_LUA
#include "module.h"
#include <string.h>
#include <math.h>
#include <limits.h>
#endif
#include "json_config.h"
#include "jsonsl.h"
#define LUA_SJSONLIBNAME "sjson"
#define DEFAULT_DEPTH 20
#define DBG_PRINTF(...)
typedef struct {
jsonsl_t jsn;
int result_ref;
int hkey_ref;
int null_ref;
int metatable;
int pos_ref;
uint8_t complete;
const char *error;
lua_State *L;
size_t min_needed;
size_t min_available;
size_t buffer_len;
const char *buffer; // Points into buffer_ref
int buffer_ref;
} JSN_DATA;
#define get_parent_object_ref() ((state->level == 1) ? data->result_ref : state[-1].lua_object_ref)
#define get_parent_object_used_count_pre_inc() ((state->level == 1) ? 1 : ++state[-1].used_count)
static const char* get_state_buffer(JSN_DATA *ctx, struct jsonsl_state_st *state)
{
size_t offset = state->pos_begin - ctx->min_available;
return ctx->buffer + offset;
}
// The elem data is a ref
static int error_callback(jsonsl_t jsn,
jsonsl_error_t err,
struct jsonsl_state_st *state,
char *at)
{
JSN_DATA *data = (JSN_DATA *) jsn->data;
if (!data->complete) {
data->error = jsonsl_strerror(err);
}
//fprintf(stderr, "Got error at pos %lu: %s\n", jsn->pos, jsonsl_strerror(err));
return 0;
}
static void
create_table(JSN_DATA *data) {
lua_newtable(data->L);
if (data->metatable != LUA_NOREF) {
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->metatable);
lua_setmetatable(data->L, -2);
}
}
static void
create_new_element(jsonsl_t jsn,
jsonsl_action_t action,
struct jsonsl_state_st *state,
const char *buf)
{
JSN_DATA *data = jsn->data;
DBG_PRINTF("L%d: new action %d @ %d state->type %s\n", state->level, action, state->pos_begin, jsonsl_strtype(state->type));
DBG_PRINTF("buf: '%s' ('%.10s')\n", buf, get_state_buffer(data, state));
state->lua_object_ref = LUA_NOREF;
switch(state->type) {
case JSONSL_T_SPECIAL:
case JSONSL_T_STRING:
case JSONSL_T_HKEY:
break;
case JSONSL_T_LIST:
case JSONSL_T_OBJECT:
create_table(data);
state->lua_object_ref = lua_ref(data->L, 1);
state->used_count = 0;
lua_rawgeti(data->L, LUA_REGISTRYINDEX, get_parent_object_ref());
if (data->hkey_ref == LUA_NOREF) {
// list, so append
lua_pushnumber(data->L, get_parent_object_used_count_pre_inc());
DBG_PRINTF("Adding array element\n");
} else {
// object, so
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->hkey_ref);
lua_unref(data->L, data->hkey_ref);
data->hkey_ref = LUA_NOREF;
DBG_PRINTF("Adding hash element\n");
}
if (data->pos_ref != LUA_NOREF && state->level > 1) {
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->pos_ref);
lua_pushnumber(data->L, state->level - 1);
lua_pushvalue(data->L, -3); // get the key
lua_settable(data->L, -3);
lua_pop(data->L, 1);
}
// At this point, the stack:
// top: index/hash key
// : table
int want_value = 1;
// Invoke the checkpath method if possible
if (data->pos_ref != LUA_NOREF) {
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->metatable);
lua_getfield(data->L, -1, "checkpath");
if (lua_type(data->L, -1) != LUA_TNIL) {
// Call with the new table and the path as arguments
lua_rawgeti(data->L, LUA_REGISTRYINDEX, state->lua_object_ref);
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->pos_ref);
lua_call(data->L, 2, 1);
want_value = lua_toboolean(data->L, -1);
}
lua_pop(data->L, 2); // Discard the metatable and either the getfield result or retval
}
if (want_value) {
lua_rawgeti(data->L, LUA_REGISTRYINDEX, state->lua_object_ref);
lua_settable(data->L, -3);
lua_pop(data->L, 1); // the table
} else {
lua_pop(data->L, 2); // the index and table
}
break;
default:
DBG_PRINTF("Unhandled type %c\n", state->type);
luaL_error(data->L, "Unhandled type");
break;
}
data->min_needed = state->pos_begin;
}
static void push_number(JSN_DATA *data, struct jsonsl_state_st *state) {
lua_pushlstring(data->L, get_state_buffer(data, state), state->pos_cur - state->pos_begin);
LUA_NUMBER r = lua_tonumber(data->L, -1);
lua_pop(data->L, 1);
lua_pushnumber(data->L, r);
}
static int fromhex(char c) {
if (c <= '9') {
return c & 0xf;
}
return ((c - 'A' + 10) & 0xf);
}
static void output_utf8(luaL_Buffer *buf, int c) {
char space[4];
char *b = space;
if (c<0x80) *b++=c;
else if (c<0x800) *b++=192+c/64, *b++=128+c%64;
else if (c-0xd800u<0x800) *b++ = '?';
else if (c<0x10000) *b++=224+c/4096, *b++=128+c/64%64, *b++=128+c%64;
else if (c<0x110000) *b++=240+c/262144, *b++=128+c/4096%64, *b++=128+c/64%64, *b++=128+c%64;
else *b++ = '?';
luaL_addlstring(buf, space, b - space);
}
static void push_string(JSN_DATA *data, struct jsonsl_state_st *state) {
luaL_Buffer b;
luaL_buffinit(data->L, &b);
int i;
const char *c = get_state_buffer(data, state) + 1;
for (i = 0; i < state->pos_cur - state->pos_begin - 1; i++) {
int nc = c[i];
if (nc == '\\') {
i++;
nc = c[i] & 255;
switch (c[i]) {
case 'b':
nc = '\b';
break;
case 'f':
nc = '\f';
break;
case 'n':
nc = '\n';
break;
case 'r':
nc = '\r';
break;
case 't':
nc = '\t';
break;
case 'u':
nc = fromhex(c[++i]) << 12;
nc += fromhex(c[++i]) << 8;
nc += fromhex(c[++i]) << 4;
nc += fromhex(c[++i]) ;
output_utf8(&b, nc);
continue;
}
}
luaL_putchar(&b, nc);
}
luaL_pushresult(&b);
}
static void
cleanup_closing_element(jsonsl_t jsn,
jsonsl_action_t action,
struct jsonsl_state_st *state,
const char *at)
{
JSN_DATA *data = (JSN_DATA *) jsn->data;
DBG_PRINTF( "L%d: cc action %d state->type %s\n", state->level, action, jsonsl_strtype(state->type));
DBG_PRINTF( "buf (%d - %d): '%.*s'\n", state->pos_begin, state->pos_cur, state->pos_cur - state->pos_begin, get_state_buffer(data, state));
DBG_PRINTF( "at: '%s'\n", at);
switch (state->type) {
case JSONSL_T_HKEY:
push_string(data, state);
data->hkey_ref = lua_ref(data->L, 1);
break;
case JSONSL_T_STRING:
lua_rawgeti(data->L, LUA_REGISTRYINDEX, get_parent_object_ref());
if (data->hkey_ref == LUA_NOREF) {
// list, so append
lua_pushnumber(data->L, get_parent_object_used_count_pre_inc());
} else {
// object, so
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->hkey_ref);
lua_unref(data->L, data->hkey_ref);
data->hkey_ref = LUA_NOREF;
}
push_string(data, state);
lua_settable(data->L, -3);
lua_pop(data->L, 1);
break;
case JSONSL_T_SPECIAL:
DBG_PRINTF("Special flags = 0x%x\n", state->special_flags);
// need to deal with true/false/null
if (state->special_flags & (JSONSL_SPECIALf_TRUE|JSONSL_SPECIALf_FALSE|JSONSL_SPECIALf_NUMERIC|JSONSL_SPECIALf_NULL)) {
if (state->special_flags & JSONSL_SPECIALf_TRUE) {
lua_pushboolean(data->L, 1);
} else if (state->special_flags & JSONSL_SPECIALf_FALSE) {
lua_pushboolean(data->L, 0);
} else if (state->special_flags & JSONSL_SPECIALf_NULL) {
DBG_PRINTF("Outputting null\n");
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->null_ref);
} else if (state->special_flags & JSONSL_SPECIALf_NUMERIC) {
push_number(data, state);
}
lua_rawgeti(data->L, LUA_REGISTRYINDEX, get_parent_object_ref());
if (data->hkey_ref == LUA_NOREF) {
// list, so append
lua_pushnumber(data->L, get_parent_object_used_count_pre_inc());
} else {
// object, so
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->hkey_ref);
lua_unref(data->L, data->hkey_ref);
data->hkey_ref = LUA_NOREF;
}
lua_pushvalue(data->L, -3);
lua_remove(data->L, -4);
lua_settable(data->L, -3);
lua_pop(data->L, 1);
}
break;
case JSONSL_T_OBJECT:
case JSONSL_T_LIST:
lua_unref(data->L, state->lua_object_ref);
state->lua_object_ref = LUA_NOREF;
if (data->pos_ref != LUA_NOREF) {
lua_rawgeti(data->L, LUA_REGISTRYINDEX, data->pos_ref);
lua_pushnumber(data->L, state->level);
lua_pushnil(data->L);
lua_settable(data->L, -3);
lua_pop(data->L, 1);
}
if (state->level == 1) {
data->complete = 1;
}
break;
}
}
static int sjson_decoder_int(lua_State *L, int argno) {
int nlevels = DEFAULT_DEPTH;
if (lua_type(L, argno) == LUA_TTABLE) {
lua_getfield(L, argno, "depth");
nlevels = lua_tointeger(L, argno);
if (nlevels == 0) {
nlevels = DEFAULT_DEPTH;
}
if (nlevels < 4) {
nlevels = 4;
}
if (nlevels > 1000) {
nlevels = 1000;
}
lua_pop(L, 1);
}
JSN_DATA *data = (JSN_DATA *) lua_newuserdata(L, sizeof(JSN_DATA) + jsonsl_get_size(nlevels));
//
// Associate its metatable
luaL_getmetatable(L, "sjson.decoder");
lua_setmetatable(L, -2);
jsonsl_t jsn = jsonsl_init((jsonsl_t) (data + 1), nlevels);
int i;
for (i = 0; i < jsn->levels_max; i++) {
jsn->stack[i].lua_object_ref = LUA_NOREF;
}
data->jsn = jsn;
data->result_ref = LUA_NOREF;
data->null_ref = LUA_REFNIL;
data->metatable = LUA_NOREF;
data->hkey_ref = LUA_NOREF;
data->pos_ref = LUA_NOREF;
data->buffer_ref = LUA_NOREF;
data->complete = 0;
data->error = NULL;
data->L = L;
data->buffer_len = 0;
data->min_needed = data->min_available = jsn->pos;
lua_pushlightuserdata(L, 0);
data->null_ref = lua_ref(L, 1);
// This may throw...
lua_newtable(L);
data->result_ref = luaL_ref(L, LUA_REGISTRYINDEX);
if (lua_type(L, argno) == LUA_TTABLE) {
luaL_unref(L, LUA_REGISTRYINDEX, data->null_ref);
data->null_ref = LUA_NOREF;
lua_getfield(L, argno, "null");
data->null_ref = lua_ref(L, 1);
lua_getfield(L, argno, "metatable");
lua_pushvalue(L, -1);
data->metatable = lua_ref(L, 1);
if (lua_type(L, -1) != LUA_TNIL) {
lua_getfield(L, -1, "checkpath");
if (lua_type(L, -1) != LUA_TNIL) {
lua_newtable(L);
data->pos_ref = lua_ref(L, 1);
}
lua_pop(L, 1); // Throw away the checkpath value
}
lua_pop(L, 1); // Throw away the metatable
}
jsonsl_enable_all_callbacks(data->jsn);
jsn->action_callback = NULL;
jsn->action_callback_PUSH = create_new_element;
jsn->action_callback_POP = cleanup_closing_element;
jsn->error_callback = error_callback;
jsn->data = data;
jsn->max_callback_level = nlevels;
return 1;
}
static int sjson_decoder(lua_State *L) {
return sjson_decoder_int(L, 1);
}
static int sjson_decoder_result_int(lua_State *L, JSN_DATA *data) {
if (!data->complete) {
luaL_error(L, "decode not complete");
}
lua_rawgeti(L, LUA_REGISTRYINDEX, data->result_ref);
lua_rawgeti(L, -1, 1);
lua_remove(L, -2);
return 1;
}
static int sjson_decoder_result(lua_State *L) {
JSN_DATA *data = (JSN_DATA *)luaL_checkudata(L, 1, "sjson.decoder");
return sjson_decoder_result_int(L, data);
}
static void sjson_free_working_data(lua_State *L, JSN_DATA *data) {
jsonsl_t jsn = data->jsn;
int i;
for (i = 0; i < jsn->levels_max; i++) {
luaL_unref(L, LUA_REGISTRYINDEX, jsn->stack[i].lua_object_ref);
jsn->stack[i].lua_object_ref = LUA_NOREF;
}
luaL_unref(L, LUA_REGISTRYINDEX, data->metatable);
data->metatable = LUA_NOREF;
luaL_unref(L, LUA_REGISTRYINDEX, data->hkey_ref);
data->hkey_ref = LUA_NOREF;
luaL_unref(L, LUA_REGISTRYINDEX, data->null_ref);
data->null_ref = LUA_NOREF;
luaL_unref(L, LUA_REGISTRYINDEX, data->pos_ref);
data->pos_ref = LUA_NOREF;
luaL_unref(L, LUA_REGISTRYINDEX, data->buffer_ref);
data->buffer_ref = LUA_NOREF;
}
static int sjson_decoder_write_int(lua_State *L, int udata_pos, int string_pos) {
JSN_DATA *data = (JSN_DATA *)luaL_checkudata(L, udata_pos, "sjson.decoder");
size_t len;
const char *str = luaL_checklstring(L, string_pos, &len);
if (data->error) {
luaL_error(L, "JSON parse error: previous call");
}
if (!data->complete) {
data->L = L;
// Merge into any existing buffer and deal with discard
if (data->buffer_ref != LUA_NOREF) {
luaL_Buffer b;
luaL_buffinit(L, &b);
lua_rawgeti(L, LUA_REGISTRYINDEX, data->buffer_ref);
size_t prev_len;
const char *prev_buffer = luaL_checklstring(L, -1, &prev_len);
lua_pop(L, 1); // But string still referenced so it cannot move
int discard = data->min_needed - data->min_available;
prev_buffer += discard;
prev_len -= discard;
if (prev_len > 0) {
luaL_addlstring(&b, prev_buffer, prev_len);
}
data->min_available += discard;
luaL_unref(L, LUA_REGISTRYINDEX, data->buffer_ref);
data->buffer_ref = LUA_NOREF;
lua_pushvalue(L, string_pos);
luaL_addvalue(&b);
luaL_pushresult(&b);
} else {
lua_pushvalue(L, string_pos);
}
size_t blen;
data->buffer = luaL_checklstring(L, -1, &blen);
data->buffer_len = blen;
data->buffer_ref = lua_ref(L, 1);
jsonsl_feed(data->jsn, str, len);
if (data->error) {
luaL_error(L, "JSON parse error: %s", data->error);
}
}
if (data->complete) {
// We no longer need the buffer
sjson_free_working_data(L, data);
return sjson_decoder_result_int(L, data);
}
return 0;
}
static int sjson_decoder_write(lua_State *L) {
return sjson_decoder_write_int(L, 1, 2);
}
static int sjson_decode(lua_State *L) {
int push_count = sjson_decoder_int(L, 2);
if (push_count != 1) {
luaL_error(L, "Internal error in sjson.deocder");
}
luaL_checkudata(L, -1, "sjson.decoder");
push_count = sjson_decoder_write_int(L, -1, 1);
if (push_count != 1) {
luaL_error(L, "Incomplete JSON object passed to sjson.decode");
}
// Now we have two items on the stack -- the udata and the result
lua_remove(L, -2);
return 1;
}
static int sjson_decoder_destructor(lua_State *L) {
JSN_DATA *data = (JSN_DATA *)luaL_checkudata(L, 1, "sjson.decoder");
sjson_free_working_data(L, data);
data->jsn = NULL;
luaL_unref(L, LUA_REGISTRYINDEX, data->result_ref);
data->result_ref = LUA_NOREF;
DBG_PRINTF("Destructor called\n");
return 0;
}
//
//--------------------------------- ENCODER BELOW
//
//
//
//#undef DBG_PRINTF
//#define DBG_PRINTF printf
typedef struct {
int lua_object_ref;
// for arrays
// 0 -> [
// 1 -> first element
// 2 -> ,
// 3 -> second element
// 4 -> ]
// for objects
// 0 -> { firstkey :
// 1 -> first value
// 2 -> , secondkey :
// 3 -> second value
// 4 -> }
short offset;
// -1 for objects
// 0 -> n maximum integer key = n
short size;
int lua_key_ref;
} ENC_DATA_STATE;
typedef struct {
ENC_DATA_STATE *stack;
int nlevels;
int level;
int current_str_ref;
int null_ref;
int offset;
} ENC_DATA;
static int sjson_encoder_get_table_size(lua_State *L, int argno) {
// Returns -1 for object, otherwise the maximum integer key value found.
lua_pushvalue(L, argno);
// stack now contains: -1 => table
lua_pushnil(L);
// stack now contains: -1 => nil; -2 => table
//
int maxkey = 0;
while (lua_next(L, -2)) {
lua_pop(L, 1);
// stack now contains: -1 => key; -2 => table
if (lua_type(L, -1) == LUA_TNUMBER) {
int val = lua_tointeger(L, -1);
if (val > maxkey) {
maxkey = val;
} else if (val <= 0) {
maxkey = -1;
lua_pop(L, 1);
break;
}
} else {
maxkey = -1;
lua_pop(L, 1);
break;
}
}
lua_pop(L, 1);
return maxkey;
}
static void enc_pop_stack(lua_State *L, ENC_DATA *data) {
if (data->level < 0) {
luaL_error(L, "encoder stack underflow");
}
ENC_DATA_STATE *state = &data->stack[data->level];
lua_unref(L, state->lua_object_ref);
state->lua_object_ref = LUA_NOREF;
lua_unref(L, state->lua_key_ref);
state->lua_key_ref = LUA_REFNIL;
data->level--;
}
static void enc_push_stack(lua_State *L, ENC_DATA *data, int argno) {
if (++data->level >= data->nlevels) {
luaL_error(L, "encoder stack overflow");
}
lua_pushvalue(L, argno);
ENC_DATA_STATE *state = &data->stack[data->level];
state->lua_object_ref = lua_ref(L, 1);
state->size = sjson_encoder_get_table_size(L, argno);
state->offset = 0; // We haven't started on this one yet
}
static int sjson_encoder(lua_State *L) {
int nlevels = DEFAULT_DEPTH;
int argno = 1;
// Validate first arg is a table
luaL_checktype(L, argno++, LUA_TTABLE);
if (lua_type(L, argno) == LUA_TTABLE) {
lua_getfield(L, argno, "depth");
nlevels = lua_tointeger(L, argno);
if (nlevels == 0) {
nlevels = DEFAULT_DEPTH;
}
if (nlevels < 4) {
nlevels = 4;
}
if (nlevels > 1000) {
nlevels = 1000;
}
lua_pop(L, 1);
}
ENC_DATA *data = (ENC_DATA *) lua_newuserdata(L, sizeof(ENC_DATA) + nlevels * sizeof(ENC_DATA_STATE));
// Associate its metatable
luaL_getmetatable(L, "sjson.encoder");
lua_setmetatable(L, -2);
data->nlevels = nlevels;
data->level = -1;
data->stack = (ENC_DATA_STATE *) (data + 1);
data->current_str_ref = LUA_NOREF;
int i;
for (i = 0; i < nlevels; i++) {
data->stack[i].lua_object_ref = LUA_NOREF;
data->stack[i].lua_key_ref = LUA_REFNIL;
}
enc_push_stack(L, data, 1);
data->null_ref = LUA_REFNIL;
if (lua_type(L, argno) == LUA_TTABLE) {
luaL_unref(L, LUA_REGISTRYINDEX, data->null_ref);
data->null_ref = LUA_NOREF;
lua_getfield(L, argno, "null");
data->null_ref = lua_ref(L, 1);
}
return 1;
}
static void encode_lua_object(lua_State *L, ENC_DATA *data, int argno, const char *prefix, const char *suffix) {
luaL_Buffer b;
luaL_buffinit(L, &b);
luaL_addstring(&b, prefix);
int type = lua_type(L, argno);
if (type == LUA_TSTRING) {
// Check to see if it is the NULL value
if (data->null_ref != LUA_REFNIL) {
lua_rawgeti(L, LUA_REGISTRYINDEX, data->null_ref);
if (lua_equal(L, -1, -2)) {
type = LUA_TNIL;
}
lua_pop(L, 1);
}
}
switch (type) {
default:
luaL_error(L, "Cannot encode type %d", type);
break;
case LUA_TLIGHTUSERDATA:
case LUA_TNIL:
luaL_addstring(&b, "null");
break;
case LUA_TBOOLEAN:
luaL_addstring(&b, lua_toboolean(L, argno) ? "true" : "false");
break;
case LUA_TNUMBER:
{
lua_pushvalue(L, argno);
size_t len;
const char *str = lua_tolstring(L, -1, &len);
char value[len + 1];
strcpy(value, str);
lua_pop(L, 1);
luaL_addstring(&b, value);
break;
}
case LUA_TSTRING:
{
luaL_addchar(&b, '"');
size_t len;
const char *str = lua_tolstring(L, argno, &len);
while (len > 0) {
if ((*str & 0xff) < 0x20) {
char value[8];
value[0] = '\\';
char *d = value + 1;
switch(*str) {
case '\f':
*d++ = 'f';
break;
case '\n':
*d++ = 'n';
break;
case '\t':
*d++ = 't';
break;
case '\r':
*d++ = 'r';
break;
case '\b':
*d++ = 'b';
break;
default:
*d++ = 'u';
*d++ = '0';
*d++ = '0';
*d++ = "0123456789abcdef"[(*str >> 4) & 0xf];
*d++ = "0123456789abcdef"[(*str ) & 0xf];
break;
}
*d = '\0';
luaL_addstring(&b, value);
} else if (*str == '"') {
luaL_addstring(&b, "\\\"");
} else {
luaL_addchar(&b, *str);
}
str++;
len--;
}
luaL_addchar(&b, '"');
break;
}
}
luaL_addstring(&b, suffix);
luaL_pushresult(&b);
}
static int sjson_encoder_next_value_is_table(lua_State *L) {
int count = 10;
while ((lua_type(L, -1) == LUA_TFUNCTION
#ifdef LUA_TLIGHTFUNCTION
|| lua_type(L, -1) == LUA_TLIGHTFUNCTION
#endif
) && count-- > 0) {
// call it and use the return value
lua_call(L, 0, 1); // Expecting replacement value
}
return (lua_type(L, -1) == LUA_TTABLE);
}
static void sjson_encoder_make_next_chunk(lua_State *L, ENC_DATA *data) {
if (data->level < 0) {
return;
}
luaL_Buffer b;
luaL_buffinit(L, &b);
// Ending condition
while (data->level >= 0 && !b.lvl) {
ENC_DATA_STATE *state = &data->stack[data->level];
int finished = 0;
if (state->size >= 0) {
if (state->offset == 0) {
// start of object or whatever
luaL_addchar(&b, '[');
}
if (state->offset == state->size << 1) {
luaL_addchar(&b, ']');
finished = 1;
} else if ((state->offset & 1) == 0) {
if (state->offset > 0) {
luaL_addchar(&b, ',');
}
} else {
// output the value
lua_rawgeti(L, LUA_REGISTRYINDEX, state->lua_object_ref);
lua_rawgeti(L, -1, (state->offset >> 1) + 1);
if (sjson_encoder_next_value_is_table(L)) {
enc_push_stack(L, data, -1);
lua_pop(L, 2);
state->offset++;
continue;
}
encode_lua_object(L, data, -1, "", "");
lua_remove(L, -2);
lua_remove(L, -2);
luaL_addvalue(&b);
}
state->offset++;
} else {
lua_rawgeti(L, LUA_REGISTRYINDEX, state->lua_object_ref);
// stack now contains: -1 => table
lua_rawgeti(L, LUA_REGISTRYINDEX, state->lua_key_ref);
// stack now contains: -1 => nil or key; -2 => table
if (lua_next(L, -2)) {
// save the key
if (state->offset & 1) {
lua_unref(L, state->lua_key_ref);
state->lua_key_ref = LUA_NOREF;
// Duplicate the key
lua_pushvalue(L, -2);
state->lua_key_ref = lua_ref(L, 1);
}
if ((state->offset & 1) == 0) {
// copy the key so that lua_tostring does not modify the original
lua_pushvalue(L, -2);
// stack now contains: -1 => key; -2 => value; -3 => key; -4 => table
// key
lua_tostring(L, -1);
encode_lua_object(L, data, -1, state->offset ? "," : "{", ":");
lua_remove(L, -2);
lua_remove(L, -2);
lua_remove(L, -2);
lua_remove(L, -2);
} else {
if (sjson_encoder_next_value_is_table(L)) {
enc_push_stack(L, data, -1);
lua_pop(L, 3);
state->offset++;
continue;
}
encode_lua_object(L, data, -1, "", "");
lua_remove(L, -2);
lua_remove(L, -2);
lua_remove(L, -2);
}
luaL_addvalue(&b);
} else {
lua_pop(L, 1);
// We have got to the end
luaL_addchar(&b, '}');
finished = 1;
}
state->offset++;
}
if (finished) {
enc_pop_stack(L, data);
}
}
luaL_pushresult(&b);
data->current_str_ref = lua_ref(L, 1);
data->offset = 0;
}
static int sjson_encoder_read_int(lua_State *L, ENC_DATA *data, int readsize) {
luaL_Buffer b;
luaL_buffinit(L, &b);
size_t len;
do {
// Fill the buffer with (up to) readsize characters
if (data->current_str_ref != LUA_NOREF) {
// this is not allowed
lua_rawgeti(L, LUA_REGISTRYINDEX, data->current_str_ref);
const char *str = lua_tolstring(L, -1, &len);
lua_pop(L, 1); // Note that we still have the string referenced so it can't go away
int amnt = len - data->offset;;
if (amnt > readsize) {
amnt = readsize;
}
luaL_addlstring(&b, str + data->offset, amnt);
data->offset += amnt;
readsize -= amnt;
if (data->offset == len) {
lua_unref(L, data->current_str_ref);
data->current_str_ref = LUA_NOREF;
}
}
if (readsize > 0) {
// Make the next chunk
sjson_encoder_make_next_chunk(L, data);
}
} while (readsize > 0 && data->current_str_ref != LUA_NOREF);
luaL_pushresult(&b);
lua_tolstring(L, -1, &len);
if (len == 0) {
// we have got to the end
lua_pop(L, 1);
return 0;
}
return 1;
}
static int sjson_encoder_read(lua_State *L) {
ENC_DATA *data = (ENC_DATA *)luaL_checkudata(L, 1, "sjson.encoder");
int readsize = 1024;
if (lua_type(L, 2) == LUA_TNUMBER) {
readsize = lua_tointeger(L, 2);
if (readsize < 1) {
readsize = 1;
}
}
return sjson_encoder_read_int(L, data, readsize);
}
static int sjson_encode(lua_State *L) {
sjson_encoder(L);
ENC_DATA *data = (ENC_DATA *)luaL_checkudata(L, -1, "sjson.encoder");
int rc = sjson_encoder_read_int(L, data, 1000000);
lua_remove(L, -(rc + 1));
return rc;
}
static int sjson_encoder_destructor(lua_State *L) {
ENC_DATA *data = (ENC_DATA *)luaL_checkudata(L, 1, "sjson.encoder");
int i;
for (i = 0; i < data->nlevels; i++) {
luaL_unref(L, LUA_REGISTRYINDEX, data->stack[i].lua_object_ref);
luaL_unref(L, LUA_REGISTRYINDEX, data->stack[i].lua_key_ref);
}
luaL_unref(L, LUA_REGISTRYINDEX, data->null_ref);
luaL_unref(L, LUA_REGISTRYINDEX, data->current_str_ref);
DBG_PRINTF("Destructor called\n");
return 0;
}
#ifdef LOCAL_LUA
static const luaL_Reg sjson_encoder_map[] = {
{ "read", sjson_encoder_read },
{ "__gc", sjson_encoder_destructor },
{ NULL, NULL }
};
static const luaL_Reg sjson_decoder_map[] = {
{ "write", sjson_decoder_write },
{ "result", sjson_decoder_result },
{ "__gc", sjson_decoder_destructor },
{ NULL, NULL }
};
static const luaL_Reg sjsonlib[] = {
{ "decode", sjson_decode },
{ "decoder", sjson_decoder },
{ "encode", sjson_encode },
{ "encoder", sjson_encoder },
{NULL, NULL}
};
#else
static const LUA_REG_TYPE sjson_encoder_map[] = {
{ LSTRKEY( "read" ), LFUNCVAL( sjson_encoder_read ) },
{ LSTRKEY( "__gc" ), LFUNCVAL( sjson_encoder_destructor ) },
{ LSTRKEY( "__index" ), LROVAL( sjson_encoder_map ) },
{ LNILKEY, LNILVAL }
};
static const LUA_REG_TYPE sjson_decoder_map[] = {
{ LSTRKEY( "write" ), LFUNCVAL( sjson_decoder_write ) },
{ LSTRKEY( "result" ), LFUNCVAL( sjson_decoder_result ) },
{ LSTRKEY( "__gc" ), LFUNCVAL( sjson_decoder_destructor ) },
{ LSTRKEY( "__index" ), LROVAL( sjson_decoder_map ) },
{ LNILKEY, LNILVAL }
};
static const LUA_REG_TYPE sjson_map[] = {
{ LSTRKEY( "encode" ), LFUNCVAL( sjson_encode ) },
{ LSTRKEY( "decode" ), LFUNCVAL( sjson_decode ) },
{ LSTRKEY( "encoder" ), LFUNCVAL( sjson_encoder ) },
{ LSTRKEY( "decoder" ), LFUNCVAL( sjson_decoder ) },
{ LSTRKEY( "NULL" ), LUDATA( 0 ) },
{ LNILKEY, LNILVAL }
};
#endif
LUALIB_API int luaopen_sjson (lua_State *L) {
#ifdef LOCAL_LUA
luaL_register(L, LUA_SJSONLIBNAME, sjsonlib);
lua_getglobal(L, LUA_SJSONLIBNAME);
lua_pushstring(L, "NULL");
lua_pushlightuserdata(L, 0);
lua_settable(L, -3);
lua_pop(L, 1);
luaL_newmetatable(L, "sjson.encoder");
luaL_register(L, NULL, sjson_encoder_map);
lua_setfield(L, -1, "__index");
luaL_newmetatable(L, "sjson.decoder");
luaL_register(L, NULL, sjson_decoder_map);
lua_setfield(L, -1, "__index");
#else
luaL_rometatable(L, "sjson.decoder", (void *)sjson_decoder_map);
luaL_rometatable(L, "sjson.encoder", (void *)sjson_encoder_map);
#endif
return 1;
}
#ifndef LOCAL_LUA
NODEMCU_MODULE(SJSON, "sjson", sjson_map, luaopen_sjson);
#endif
\ No newline at end of file
COMPONENT_ADD_INCLUDEDIRS:=.
CFLAGS+=-imacros json_config.h
\ No newline at end of file
#ifndef __JSON_CONFIG_H__
#define __JSON_CONFIG_H__
#define JSONSL_STATE_USER_FIELDS int lua_object_ref; int used_count;
#define JSONSL_NO_JPR
#endif
/* Copyright (C) 2012-2015 Mark Nunberg.
*
* See included LICENSE file for license details.
*/
#include "jsonsl.h"
#include <assert.h>
#include <limits.h>
#include <stdint.h>
#include <stddef.h>
#include "ctype.h"
#include "platform.h"
#ifdef JSONSL_USE_METRICS
#define XMETRICS \
X(STRINGY_INSIGNIFICANT) \
X(STRINGY_SLOWPATH) \
X(ALLOWED_WHITESPACE) \
X(QUOTE_FASTPATH) \
X(SPECIAL_FASTPATH) \
X(SPECIAL_WSPOP) \
X(SPECIAL_SLOWPATH) \
X(GENERIC) \
X(STRUCTURAL_TOKEN) \
X(SPECIAL_SWITCHFIRST) \
X(STRINGY_CATCH) \
X(NUMBER_FASTPATH) \
X(ESCAPES) \
X(TOTAL) \
struct jsonsl_metrics_st {
#define X(m) \
unsigned long metric_##m;
XMETRICS
#undef X
};
static struct jsonsl_metrics_st GlobalMetrics = { 0 };
static unsigned long GenericCounter[0x100] = { 0 };
static unsigned long StringyCatchCounter[0x100] = { 0 };
#define INCR_METRIC(m) \
GlobalMetrics.metric_##m++;
#define INCR_GENERIC(c) \
INCR_METRIC(GENERIC); \
GenericCounter[c]++; \
#define INCR_STRINGY_CATCH(c) \
INCR_METRIC(STRINGY_CATCH); \
StringyCatchCounter[c]++;
JSONSL_API
void jsonsl_dump_global_metrics(void)
{
int ii;
printf("JSONSL Metrics:\n");
#define X(m) \
printf("\t%-30s %20lu (%0.2f%%)\n", #m, GlobalMetrics.metric_##m, \
(float)((float)(GlobalMetrics.metric_##m/(float)GlobalMetrics.metric_TOTAL)) * 100);
XMETRICS
#undef X
printf("Generic Characters:\n");
for (ii = 0; ii < 0xff; ii++) {
if (GenericCounter[ii]) {
printf("\t[ %c ] %lu\n", ii, GenericCounter[ii]);
}
}
printf("Weird string loop\n");
for (ii = 0; ii < 0xff; ii++) {
if (StringyCatchCounter[ii]) {
printf("\t[ %c ] %lu\n", ii, StringyCatchCounter[ii]);
}
}
}
#else
#define INCR_METRIC(m)
#define INCR_GENERIC(c)
#define INCR_STRINGY_CATCH(c)
JSONSL_API
void jsonsl_dump_global_metrics(void) { }
#endif /* JSONSL_USE_METRICS */
#define CASE_DIGITS \
case '1': \
case '2': \
case '3': \
case '4': \
case '5': \
case '6': \
case '7': \
case '8': \
case '9': \
case '0':
static unsigned extract_special(unsigned);
static int is_special_end(unsigned);
static int is_allowed_whitespace(unsigned);
static int is_allowed_escape(unsigned);
static int is_simple_char(unsigned);
static char get_escape_equiv(unsigned);
JSONSL_API
size_t jsonsl_get_size(int nlevels)
{
return sizeof (struct jsonsl_st) + ( (nlevels-1) * sizeof (struct jsonsl_state_st)) ;
}
JSONSL_API
jsonsl_t jsonsl_init(jsonsl_t jsn, int nlevels)
{
unsigned int ii;
memset(jsn, 0, jsonsl_get_size(nlevels));
jsn->levels_max = nlevels;
jsn->max_callback_level = -1;
jsonsl_reset(jsn);
for (ii = 0; ii < jsn->levels_max; ii++) {
jsn->stack[ii].level = ii;
}
return jsn;
}
JSONSL_API
jsonsl_t jsonsl_new(int nlevels)
{
struct jsonsl_st *jsn = (struct jsonsl_st *)
calloc(1, jsonsl_get_size(nlevels));
if (jsn) {
jsonsl_init(jsn, nlevels);
}
return jsn;
}
JSONSL_API
void jsonsl_reset(jsonsl_t jsn)
{
jsn->tok_last = 0;
jsn->can_insert = 1;
jsn->pos = 0;
jsn->level = 0;
jsn->stopfl = 0;
jsn->in_escape = 0;
jsn->expecting = 0;
}
JSONSL_API
void jsonsl_destroy(jsonsl_t jsn)
{
if (jsn) {
free(jsn);
}
}
#define FASTPARSE_EXHAUSTED 1
#define FASTPARSE_BREAK 0
/*
* This function is meant to accelerate string parsing, reducing the main loop's
* check if we are indeed a string.
*
* @param jsn the parser
* @param[in,out] bytes_p A pointer to the current buffer (i.e. current position)
* @param[in,out] nbytes_p A pointer to the current size of the buffer
* @return true if all bytes have been exhausted (and thus the main loop can
* return), false if a special character was examined which requires greater
* examination.
*/
static int
jsonsl__str_fastparse(jsonsl_t jsn,
const jsonsl_uchar_t **bytes_p, size_t *nbytes_p)
{
const jsonsl_uchar_t *bytes = *bytes_p;
const jsonsl_uchar_t *end;
for (end = bytes + *nbytes_p; bytes != end; bytes++) {
if (
#ifdef JSONSL_USE_WCHAR
*bytes >= 0x100 ||
#endif /* JSONSL_USE_WCHAR */
(is_simple_char(*bytes))) {
INCR_METRIC(TOTAL);
INCR_METRIC(STRINGY_INSIGNIFICANT);
} else {
/* Once we're done here, re-calculate the position variables */
jsn->pos += (bytes - *bytes_p);
*nbytes_p -= (bytes - *bytes_p);
*bytes_p = bytes;
return FASTPARSE_BREAK;
}
}
/* Once we're done here, re-calculate the position variables */
jsn->pos += (bytes - *bytes_p);
return FASTPARSE_EXHAUSTED;
}
/* Functions exactly like str_fastparse, except it also accepts a 'state'
* argument, since the number's value is updated in the state. */
static int
jsonsl__num_fastparse(jsonsl_t jsn,
const jsonsl_uchar_t **bytes_p, size_t *nbytes_p,
struct jsonsl_state_st *state)
{
int exhausted = 1;
size_t nbytes = *nbytes_p;
const jsonsl_uchar_t *bytes = *bytes_p;
for (; nbytes; nbytes--, bytes++) {
jsonsl_uchar_t c = *bytes;
if (isdigit(c)) {
INCR_METRIC(TOTAL);
INCR_METRIC(NUMBER_FASTPATH);
state->nelem = (state->nelem * 10) + (c - 0x30);
} else {
exhausted = 0;
break;
}
}
jsn->pos += (*nbytes_p - nbytes);
if (exhausted) {
return FASTPARSE_EXHAUSTED;
}
*nbytes_p = nbytes;
*bytes_p = bytes;
return FASTPARSE_BREAK;
}
JSONSL_API
void
jsonsl_feed(jsonsl_t jsn, const jsonsl_char_t *bytes, size_t nbytes)
{
#define INVOKE_ERROR(eb) \
if (jsn->error_callback(jsn, JSONSL_ERROR_##eb, state, (char*)c)) { \
goto GT_AGAIN; \
} \
return;
#define STACK_PUSH \
if (jsn->level >= (levels_max-1)) { \
jsn->error_callback(jsn, JSONSL_ERROR_LEVELS_EXCEEDED, state, (char*)c); \
return; \
} \
state = jsn->stack + (++jsn->level); \
state->ignore_callback = jsn->stack[jsn->level-1].ignore_callback; \
state->pos_begin = jsn->pos;
#define STACK_POP_NOPOS \
state->pos_cur = jsn->pos; \
state = jsn->stack + (--jsn->level);
#define STACK_POP \
STACK_POP_NOPOS; \
state->pos_cur = jsn->pos;
#define CALLBACK_AND_POP_NOPOS(T) \
state->pos_cur = jsn->pos; \
DO_CALLBACK(T, POP); \
state->nescapes = 0; \
state = jsn->stack + (--jsn->level);
#define CALLBACK_AND_POP(T) \
CALLBACK_AND_POP_NOPOS(T); \
state->pos_cur = jsn->pos;
#define SPECIAL_POP \
CALLBACK_AND_POP(SPECIAL); \
jsn->expecting = 0; \
jsn->tok_last = 0; \
#define CUR_CHAR (*(jsonsl_uchar_t*)c)
#define DO_CALLBACK(T, action) \
if (jsn->call_##T && \
jsn->max_callback_level > state->level && \
state->ignore_callback == 0) { \
\
if (jsn->action_callback_##action) { \
jsn->action_callback_##action(jsn, JSONSL_ACTION_##action, state, (jsonsl_char_t*)c); \
} else if (jsn->action_callback) { \
jsn->action_callback(jsn, JSONSL_ACTION_##action, state, (jsonsl_char_t*)c); \
} \
if (jsn->stopfl) { return; } \
}
/**
* Verifies that we are able to insert the (non-string) item into a hash.
*/
#define ENSURE_HVAL \
if (state->nelem % 2 == 0 && state->type == JSONSL_T_OBJECT) { \
INVOKE_ERROR(HKEY_EXPECTED); \
}
#define VERIFY_SPECIAL(lit) \
if (CUR_CHAR != (lit)[jsn->pos - state->pos_begin]) { \
INVOKE_ERROR(SPECIAL_EXPECTED); \
}
#define STATE_SPECIAL_LENGTH \
(state)->nescapes
#define IS_NORMAL_NUMBER \
((state)->special_flags == JSONSL_SPECIALf_UNSIGNED || \
(state)->special_flags == JSONSL_SPECIALf_SIGNED)
#define STATE_NUM_LAST jsn->tok_last
#define CONTINUE_NEXT_CHAR() continue
const jsonsl_uchar_t *c = (jsonsl_uchar_t*)bytes;
size_t levels_max = jsn->levels_max;
struct jsonsl_state_st *state = jsn->stack + jsn->level;
jsn->base = bytes;
for (; nbytes; nbytes--, jsn->pos++, c++) {
unsigned state_type;
INCR_METRIC(TOTAL);
GT_AGAIN:
state_type = state->type;
/* Most common type is typically a string: */
if (state_type & JSONSL_Tf_STRINGY) {
/* Special escape handling for some stuff */
if (jsn->in_escape) {
jsn->in_escape = 0;
if (!is_allowed_escape(CUR_CHAR)) {
INVOKE_ERROR(ESCAPE_INVALID);
} else if (CUR_CHAR == 'u') {
DO_CALLBACK(UESCAPE, UESCAPE);
if (jsn->return_UESCAPE) {
return;
}
}
CONTINUE_NEXT_CHAR();
}
if (jsonsl__str_fastparse(jsn, &c, &nbytes) ==
FASTPARSE_EXHAUSTED) {
/* No need to readjust variables as we've exhausted the iterator */
return;
} else {
if (CUR_CHAR == '"') {
goto GT_QUOTE;
} else if (CUR_CHAR == '\\') {
goto GT_ESCAPE;
} else {
INVOKE_ERROR(WEIRD_WHITESPACE);
}
}
INCR_METRIC(STRINGY_SLOWPATH);
} else if (state_type == JSONSL_T_SPECIAL) {
/* Fast track for signed/unsigned */
if (IS_NORMAL_NUMBER) {
if (jsonsl__num_fastparse(jsn, &c, &nbytes, state) ==
FASTPARSE_EXHAUSTED) {
return;
} else {
goto GT_SPECIAL_NUMERIC;
}
} else if (state->special_flags == JSONSL_SPECIALf_DASH) {
if (!isdigit(CUR_CHAR)) {
INVOKE_ERROR(INVALID_NUMBER);
}
if (CUR_CHAR == '0') {
state->special_flags = JSONSL_SPECIALf_ZERO|JSONSL_SPECIALf_SIGNED;
} else if (isdigit(CUR_CHAR)) {
state->special_flags = JSONSL_SPECIALf_SIGNED;
state->nelem = CUR_CHAR - 0x30;
} else {
INVOKE_ERROR(INVALID_NUMBER);
}
CONTINUE_NEXT_CHAR();
} else if (state->special_flags == JSONSL_SPECIALf_ZERO) {
if (isdigit(CUR_CHAR)) {
/* Following a zero! */
INVOKE_ERROR(INVALID_NUMBER);
}
/* Unset the 'zero' flag: */
if (state->special_flags & JSONSL_SPECIALf_SIGNED) {
state->special_flags = JSONSL_SPECIALf_SIGNED;
} else {
state->special_flags = JSONSL_SPECIALf_UNSIGNED;
}
goto GT_SPECIAL_NUMERIC;
}
if (state->special_flags & JSONSL_SPECIALf_NUMERIC) {
GT_SPECIAL_NUMERIC:
switch (CUR_CHAR) {
CASE_DIGITS
STATE_NUM_LAST = '1';
CONTINUE_NEXT_CHAR();
case '.':
if (state->special_flags & JSONSL_SPECIALf_FLOAT) {
INVOKE_ERROR(INVALID_NUMBER);
}
state->special_flags |= JSONSL_SPECIALf_FLOAT;
STATE_NUM_LAST = '.';
CONTINUE_NEXT_CHAR();
case 'e':
case 'E':
if (state->special_flags & JSONSL_SPECIALf_EXPONENT) {
INVOKE_ERROR(INVALID_NUMBER);
}
state->special_flags |= JSONSL_SPECIALf_EXPONENT;
STATE_NUM_LAST = 'e';
CONTINUE_NEXT_CHAR();
case '-':
case '+':
if (STATE_NUM_LAST != 'e') {
INVOKE_ERROR(INVALID_NUMBER);
}
STATE_NUM_LAST = '-';
CONTINUE_NEXT_CHAR();
default:
if (is_special_end(CUR_CHAR)) {
goto GT_SPECIAL_POP;
}
INVOKE_ERROR(INVALID_NUMBER);
break;
}
}
/* else if (!NUMERIC) */
if (!is_special_end(CUR_CHAR)) {
STATE_SPECIAL_LENGTH++;
/* Verify TRUE, FALSE, NULL */
if (state->special_flags == JSONSL_SPECIALf_TRUE) {
VERIFY_SPECIAL("true");
} else if (state->special_flags == JSONSL_SPECIALf_FALSE) {
VERIFY_SPECIAL("false");
} else if (state->special_flags == JSONSL_SPECIALf_NULL) {
VERIFY_SPECIAL("null");
}
INCR_METRIC(SPECIAL_FASTPATH);
CONTINUE_NEXT_CHAR();
}
GT_SPECIAL_POP:
jsn->can_insert = 0;
if (IS_NORMAL_NUMBER) {
/* Nothing */
} else if (state->special_flags == JSONSL_SPECIALf_ZERO ||
state->special_flags == (JSONSL_SPECIALf_ZERO|JSONSL_SPECIALf_SIGNED)) {
/* 0 is unsigned! */
state->special_flags = JSONSL_SPECIALf_UNSIGNED;
} else if (state->special_flags == JSONSL_SPECIALf_DASH) {
/* Still in dash! */
INVOKE_ERROR(INVALID_NUMBER);
} else if (state->special_flags & JSONSL_SPECIALf_NUMERIC) {
/* Check that we're not at the end of a token */
if (STATE_NUM_LAST != '1') {
INVOKE_ERROR(INVALID_NUMBER);
}
} else if (state->special_flags == JSONSL_SPECIALf_TRUE) {
if (STATE_SPECIAL_LENGTH != 4) {
INVOKE_ERROR(SPECIAL_INCOMPLETE);
}
state->nelem = 1;
} else if (state->special_flags == JSONSL_SPECIALf_FALSE) {
if (STATE_SPECIAL_LENGTH != 5) {
INVOKE_ERROR(SPECIAL_INCOMPLETE);
}
} else if (state->special_flags == JSONSL_SPECIALf_NULL) {
if (STATE_SPECIAL_LENGTH != 4) {
INVOKE_ERROR(SPECIAL_INCOMPLETE);
}
}
SPECIAL_POP;
jsn->expecting = ',';
if (is_allowed_whitespace(CUR_CHAR)) {
CONTINUE_NEXT_CHAR();
}
/**
* This works because we have a non-whitespace token
* which is not a special token. If this is a structural
* character then it will be gracefully handled by the
* switch statement. Otherwise it will default to the 'special'
* state again,
*/
goto GT_STRUCTURAL_TOKEN;
} else if (is_allowed_whitespace(CUR_CHAR)) {
INCR_METRIC(ALLOWED_WHITESPACE);
/* So we're not special. Harmless insignificant whitespace
* passthrough
*/
CONTINUE_NEXT_CHAR();
} else if (extract_special(CUR_CHAR)) {
/* not a string, whitespace, or structural token. must be special */
goto GT_SPECIAL_BEGIN;
}
INCR_GENERIC(CUR_CHAR);
if (CUR_CHAR == '"') {
GT_QUOTE:
jsn->can_insert = 0;
switch (state_type) {
/* the end of a string or hash key */
case JSONSL_T_STRING:
CALLBACK_AND_POP(STRING);
CONTINUE_NEXT_CHAR();
case JSONSL_T_HKEY:
CALLBACK_AND_POP(HKEY);
CONTINUE_NEXT_CHAR();
case JSONSL_T_OBJECT:
state->nelem++;
if ( (state->nelem-1) % 2 ) {
/* Odd, this must be a hash value */
if (jsn->tok_last != ':') {
INVOKE_ERROR(MISSING_TOKEN);
}
jsn->expecting = ','; /* Can't figure out what to expect next */
jsn->tok_last = 0;
STACK_PUSH;
state->type = JSONSL_T_STRING;
DO_CALLBACK(STRING, PUSH);
} else {
/* hash key */
if (jsn->expecting != '"') {
INVOKE_ERROR(STRAY_TOKEN);
}
jsn->tok_last = 0;
jsn->expecting = ':';
STACK_PUSH;
state->type = JSONSL_T_HKEY;
DO_CALLBACK(HKEY, PUSH);
}
CONTINUE_NEXT_CHAR();
case JSONSL_T_LIST:
state->nelem++;
STACK_PUSH;
state->type = JSONSL_T_STRING;
jsn->expecting = ',';
jsn->tok_last = 0;
DO_CALLBACK(STRING, PUSH);
CONTINUE_NEXT_CHAR();
case JSONSL_T_SPECIAL:
INVOKE_ERROR(STRAY_TOKEN);
break;
default:
INVOKE_ERROR(STRING_OUTSIDE_CONTAINER);
break;
} /* switch(state->type) */
} else if (CUR_CHAR == '\\') {
GT_ESCAPE:
INCR_METRIC(ESCAPES);
/* Escape */
if ( (state->type & JSONSL_Tf_STRINGY) == 0 ) {
INVOKE_ERROR(ESCAPE_OUTSIDE_STRING);
}
state->nescapes++;
jsn->in_escape = 1;
CONTINUE_NEXT_CHAR();
} /* " or \ */
GT_STRUCTURAL_TOKEN:
switch (CUR_CHAR) {
case ':':
INCR_METRIC(STRUCTURAL_TOKEN);
if (jsn->expecting != CUR_CHAR) {
INVOKE_ERROR(STRAY_TOKEN);
}
jsn->tok_last = ':';
jsn->can_insert = 1;
jsn->expecting = '"';
CONTINUE_NEXT_CHAR();
case ',':
INCR_METRIC(STRUCTURAL_TOKEN);
/**
* The comma is one of the more generic tokens.
* In the context of an OBJECT, the can_insert flag
* should never be set, and no other action is
* necessary.
*/
if (jsn->expecting != CUR_CHAR) {
/* make this branch execute only when we haven't manually
* just placed the ',' in the expecting register.
*/
INVOKE_ERROR(STRAY_TOKEN);
}
if (state->type == JSONSL_T_OBJECT) {
/* end of hash value, expect a string as a hash key */
jsn->expecting = '"';
} else {
jsn->can_insert = 1;
}
jsn->tok_last = ',';
jsn->expecting = '"';
CONTINUE_NEXT_CHAR();
/* new list or object */
/* hashes are more common */
case '{':
case '[':
INCR_METRIC(STRUCTURAL_TOKEN);
if (!jsn->can_insert) {
INVOKE_ERROR(CANT_INSERT);
}
ENSURE_HVAL;
state->nelem++;
STACK_PUSH;
/* because the constants match the opening delimiters, we can do this: */
state->type = CUR_CHAR;
state->nelem = 0;
jsn->can_insert = 1;
if (CUR_CHAR == '{') {
/* If we're a hash, we expect a key first, which is quouted */
jsn->expecting = '"';
}
if (CUR_CHAR == JSONSL_T_OBJECT) {
DO_CALLBACK(OBJECT, PUSH);
} else {
DO_CALLBACK(LIST, PUSH);
}
jsn->tok_last = 0;
CONTINUE_NEXT_CHAR();
/* closing of list or object */
case '}':
case ']':
INCR_METRIC(STRUCTURAL_TOKEN);
if (jsn->tok_last == ',' && jsn->options.allow_trailing_comma == 0) {
INVOKE_ERROR(TRAILING_COMMA);
}
jsn->can_insert = 0;
jsn->level--;
jsn->expecting = ',';
jsn->tok_last = 0;
if (CUR_CHAR == ']') {
if (state->type != '[') {
INVOKE_ERROR(BRACKET_MISMATCH);
}
DO_CALLBACK(LIST, POP);
} else {
if (state->type != '{') {
INVOKE_ERROR(BRACKET_MISMATCH);
} else if (state->nelem && state->nelem % 2 != 0) {
INVOKE_ERROR(VALUE_EXPECTED);
}
DO_CALLBACK(OBJECT, POP);
}
state = jsn->stack + jsn->level;
state->pos_cur = jsn->pos;
CONTINUE_NEXT_CHAR();
default:
GT_SPECIAL_BEGIN:
/**
* Not a string, not a structural token, and not benign whitespace.
* Technically we should iterate over the character always, but since
* we are not doing full numerical/value decoding anyway (but only hinting),
* we only check upon entry.
*/
if (state->type != JSONSL_T_SPECIAL) {
int special_flags = extract_special(CUR_CHAR);
if (!special_flags) {
/**
* Try to do some heuristics here anyway to figure out what kind of
* error this is. The 'special' case is a fallback scenario anyway.
*/
if (CUR_CHAR == '\0') {
INVOKE_ERROR(FOUND_NULL_BYTE);
} else if (CUR_CHAR < 0x20) {
INVOKE_ERROR(WEIRD_WHITESPACE);
} else {
INVOKE_ERROR(SPECIAL_EXPECTED);
}
}
ENSURE_HVAL;
state->nelem++;
if (!jsn->can_insert) {
INVOKE_ERROR(CANT_INSERT);
}
STACK_PUSH;
state->type = JSONSL_T_SPECIAL;
state->special_flags = special_flags;
STATE_SPECIAL_LENGTH = 1;
if (special_flags == JSONSL_SPECIALf_UNSIGNED) {
state->nelem = CUR_CHAR - 0x30;
STATE_NUM_LAST = '1';
} else {
STATE_NUM_LAST = '-';
state->nelem = 0;
}
DO_CALLBACK(SPECIAL, PUSH);
}
CONTINUE_NEXT_CHAR();
}
}
}
JSONSL_API
const char* jsonsl_strerror(jsonsl_error_t err)
{
if (err == JSONSL_ERROR_SUCCESS) {
return "SUCCESS";
}
#define X(t) \
if (err == JSONSL_ERROR_##t) \
return #t;
JSONSL_XERR;
#undef X
return "<UNKNOWN_ERROR>";
}
JSONSL_API
const char *jsonsl_strtype(jsonsl_type_t type)
{
#define X(o,c) \
if (type == JSONSL_T_##o) \
return #o;
JSONSL_XTYPE
#undef X
return "UNKNOWN TYPE";
}
/*
*
* JPR/JSONPointer functions
*
*
*/
#ifndef JSONSL_NO_JPR
static
jsonsl_jpr_type_t
populate_component(char *in,
struct jsonsl_jpr_component_st *component,
char **next,
jsonsl_error_t *errp)
{
unsigned long pctval;
char *c = NULL, *outp = NULL, *end = NULL;
size_t input_len;
jsonsl_jpr_type_t ret = JSONSL_PATH_NONE;
if (*next == NULL || *(*next) == '\0') {
return JSONSL_PATH_NONE;
}
/* Replace the next / with a NULL */
*next = strstr(in, "/");
if (*next != NULL) {
*(*next) = '\0'; /* drop the forward slash */
input_len = *next - in;
end = *next;
*next += 1; /* next character after the '/' */
} else {
input_len = strlen(in);
end = in + input_len + 1;
}
component->pstr = in;
/* Check for special components of interest */
if (*in == JSONSL_PATH_WILDCARD_CHAR && input_len == 1) {
/* Lone wildcard */
ret = JSONSL_PATH_WILDCARD;
goto GT_RET;
} else if (isdigit(*in)) {
/* ASCII Numeric */
char *endptr;
component->idx = strtoul(in, &endptr, 10);
if (endptr && *endptr == '\0') {
ret = JSONSL_PATH_NUMERIC;
goto GT_RET;
}
}
/* Default, it's a string */
ret = JSONSL_PATH_STRING;
for (c = outp = in; c < end; c++, outp++) {
char origc;
if (*c != '%') {
goto GT_ASSIGN;
}
/*
* c = { [+0] = '%', [+1] = 'b', [+2] = 'e', [+3] = '\0' }
*/
/* Need %XX */
if (c+2 >= end) {
*errp = JSONSL_ERROR_PERCENT_BADHEX;
return JSONSL_PATH_INVALID;
}
if (! (isxdigit(*(c+1)) && isxdigit(*(c+2))) ) {
*errp = JSONSL_ERROR_PERCENT_BADHEX;
return JSONSL_PATH_INVALID;
}
/* Temporarily null-terminate the characters */
origc = *(c+3);
*(c+3) = '\0';
pctval = strtoul(c+1, NULL, 16);
*(c+3) = origc;
*outp = (char) pctval;
c += 2;
continue;
GT_ASSIGN:
*outp = *c;
}
/* Null-terminate the string */
for (; outp < c; outp++) {
*outp = '\0';
}
GT_RET:
component->ptype = ret;
if (ret != JSONSL_PATH_WILDCARD) {
component->len = strlen(component->pstr);
}
return ret;
}
JSONSL_API
jsonsl_jpr_t
jsonsl_jpr_new(const char *path, jsonsl_error_t *errp)
{
char *my_copy = NULL;
int count, curidx;
struct jsonsl_jpr_st *ret = NULL;
struct jsonsl_jpr_component_st *components = NULL;
size_t origlen;
jsonsl_error_t errstacked;
#define JPR_BAIL(err) *errp = err; goto GT_ERROR;
if (errp == NULL) {
errp = &errstacked;
}
if (path == NULL || *path != '/') {
JPR_BAIL(JSONSL_ERROR_JPR_NOROOT);
return NULL;
}
count = 1;
path++;
{
const char *c = path;
for (; *c; c++) {
if (*c == '/') {
count++;
if (*(c+1) == '/') {
JPR_BAIL(JSONSL_ERROR_JPR_DUPSLASH);
}
}
}
}
if(*path) {
count++;
}
components = (struct jsonsl_jpr_component_st *)
malloc(sizeof(*components) * count);
if (!components) {
JPR_BAIL(JSONSL_ERROR_ENOMEM);
}
my_copy = (char *)malloc(strlen(path) + 1);
if (!my_copy) {
JPR_BAIL(JSONSL_ERROR_ENOMEM);
}
strcpy(my_copy, path);
components[0].ptype = JSONSL_PATH_ROOT;
if (*my_copy) {
char *cur = my_copy;
int pathret = JSONSL_PATH_STRING;
curidx = 1;
while (pathret > 0 && curidx < count) {
pathret = populate_component(cur, components + curidx, &cur, errp);
if (pathret > 0) {
curidx++;
} else {
break;
}
}
if (pathret == JSONSL_PATH_INVALID) {
JPR_BAIL(JSONSL_ERROR_JPR_BADPATH);
}
} else {
curidx = 1;
}
path--; /*revert path to leading '/' */
origlen = strlen(path) + 1;
ret = (struct jsonsl_jpr_st *)malloc(sizeof(*ret));
if (!ret) {
JPR_BAIL(JSONSL_ERROR_ENOMEM);
}
ret->orig = (char *)malloc(origlen);
if (!ret->orig) {
JPR_BAIL(JSONSL_ERROR_ENOMEM);
}
ret->components = components;
ret->ncomponents = curidx;
ret->basestr = my_copy;
ret->norig = origlen-1;
strcpy(ret->orig, path);
return ret;
GT_ERROR:
free(my_copy);
free(components);
if (ret) {
free(ret->orig);
}
free(ret);
return NULL;
#undef JPR_BAIL
}
void jsonsl_jpr_destroy(jsonsl_jpr_t jpr)
{
free(jpr->components);
free(jpr->basestr);
free(jpr->orig);
free(jpr);
}
/**
* Call when there is a possibility of a match, either as a final match or
* as a path within a match
* @param jpr The JPR path
* @param component Component corresponding to the current element
* @param prlevel The level of the *parent*
* @param chtype The type of the child
* @return Match status
*/
static jsonsl_jpr_match_t
jsonsl__match_continue(jsonsl_jpr_t jpr,
const struct jsonsl_jpr_component_st *component,
unsigned prlevel, unsigned chtype)
{
const struct jsonsl_jpr_component_st *next_comp = component + 1;
if (prlevel == jpr->ncomponents - 1) {
/* This is the match. Check the expected type of the match against
* the child */
if (jpr->match_type == 0 || jpr->match_type == chtype) {
return JSONSL_MATCH_COMPLETE;
} else {
return JSONSL_MATCH_TYPE_MISMATCH;
}
}
if (chtype == JSONSL_T_LIST) {
if (next_comp->ptype == JSONSL_PATH_NUMERIC) {
return JSONSL_MATCH_POSSIBLE;
} else {
return JSONSL_MATCH_TYPE_MISMATCH;
}
} else if (chtype == JSONSL_T_OBJECT) {
if (next_comp->ptype == JSONSL_PATH_NUMERIC) {
return JSONSL_MATCH_TYPE_MISMATCH;
} else {
return JSONSL_MATCH_POSSIBLE;
}
} else {
return JSONSL_MATCH_TYPE_MISMATCH;
}
}
JSONSL_API
jsonsl_jpr_match_t
jsonsl_path_match(jsonsl_jpr_t jpr,
const struct jsonsl_state_st *parent,
const struct jsonsl_state_st *child,
const char *key, size_t nkey)
{
const struct jsonsl_jpr_component_st *comp;
if (!parent) {
/* No parent. Return immediately since it's always a match */
return jsonsl__match_continue(jpr, jpr->components, 0, child->type);
}
comp = jpr->components + parent->level;
/* note that we don't need to verify the type of the match, this is
* always done through the previous call to jsonsl__match_continue.
* If we are in a POSSIBLE tree then we can be certain the types (at
* least at this level) are correct */
if (parent->type == JSONSL_T_OBJECT) {
if (comp->len != nkey || strncmp(key, comp->pstr, nkey) != 0) {
return JSONSL_MATCH_NOMATCH;
}
} else {
if (comp->idx != parent->nelem - 1) {
return JSONSL_MATCH_NOMATCH;
}
}
return jsonsl__match_continue(jpr, comp, parent->level, child->type);
}
JSONSL_API
jsonsl_jpr_match_t
jsonsl_jpr_match(jsonsl_jpr_t jpr,
unsigned int parent_type,
unsigned int parent_level,
const char *key,
size_t nkey)
{
/* find our current component. This is the child level */
int cmpret;
struct jsonsl_jpr_component_st *p_component;
p_component = jpr->components + parent_level;
if (parent_level >= jpr->ncomponents) {
return JSONSL_MATCH_NOMATCH;
}
/* Lone query for 'root' element. Always matches */
if (parent_level == 0) {
if (jpr->ncomponents == 1) {
return JSONSL_MATCH_COMPLETE;
} else {
return JSONSL_MATCH_POSSIBLE;
}
}
/* Wildcard, always matches */
if (p_component->ptype == JSONSL_PATH_WILDCARD) {
if (parent_level == jpr->ncomponents-1) {
return JSONSL_MATCH_COMPLETE;
} else {
return JSONSL_MATCH_POSSIBLE;
}
}
/* Check numeric array index. This gets its special block so we can avoid
* string comparisons */
if (p_component->ptype == JSONSL_PATH_NUMERIC) {
if (parent_type == JSONSL_T_LIST) {
if (p_component->idx != nkey) {
/* Wrong index */
return JSONSL_MATCH_NOMATCH;
} else {
if (parent_level == jpr->ncomponents-1) {
/* This is the last element of the path */
return JSONSL_MATCH_COMPLETE;
} else {
/* Intermediate element */
return JSONSL_MATCH_POSSIBLE;
}
}
} else if (p_component->is_arridx) {
/* Numeric and an array index (set explicitly by user). But not
* a list for a parent */
return JSONSL_MATCH_TYPE_MISMATCH;
}
} else if (parent_type == JSONSL_T_LIST) {
return JSONSL_MATCH_TYPE_MISMATCH;
}
/* Check lengths */
if (p_component->len != nkey) {
return JSONSL_MATCH_NOMATCH;
}
/* Check string comparison */
cmpret = strncmp(p_component->pstr, key, nkey);
if (cmpret == 0) {
if (parent_level == jpr->ncomponents-1) {
return JSONSL_MATCH_COMPLETE;
} else {
return JSONSL_MATCH_POSSIBLE;
}
}
return JSONSL_MATCH_NOMATCH;
}
JSONSL_API
void jsonsl_jpr_match_state_init(jsonsl_t jsn,
jsonsl_jpr_t *jprs,
size_t njprs)
{
size_t ii, *firstjmp;
if (njprs == 0) {
return;
}
jsn->jprs = (jsonsl_jpr_t *)malloc(sizeof(jsonsl_jpr_t) * njprs);
jsn->jpr_count = njprs;
jsn->jpr_root = (size_t*)calloc(1, sizeof(size_t) * njprs * jsn->levels_max);
memcpy(jsn->jprs, jprs, sizeof(jsonsl_jpr_t) * njprs);
/* Set the initial jump table values */
firstjmp = jsn->jpr_root;
for (ii = 0; ii < njprs; ii++) {
firstjmp[ii] = ii+1;
}
}
JSONSL_API
void jsonsl_jpr_match_state_cleanup(jsonsl_t jsn)
{
if (jsn->jpr_count == 0) {
return;
}
free(jsn->jpr_root);
free(jsn->jprs);
jsn->jprs = NULL;
jsn->jpr_root = NULL;
jsn->jpr_count = 0;
}
/**
* This function should be called exactly once on each element...
* This should also be called in recursive order, since we rely
* on the parent having been initalized for a match.
*
* Since the parent is checked for a match as well, we maintain a 'serial' counter.
* Whenever we traverse an element, we expect the serial to be the same as a global
* integer. If they do not match, we re-initialize the context, and set the serial.
*
* This ensures a type of consistency without having a proactive reset by the
* main lexer itself.
*
*/
JSONSL_API
jsonsl_jpr_t jsonsl_jpr_match_state(jsonsl_t jsn,
struct jsonsl_state_st *state,
const char *key,
size_t nkey,
jsonsl_jpr_match_t *out)
{
struct jsonsl_state_st *parent_state;
jsonsl_jpr_t ret = NULL;
/* Jump and JPR tables for our own state and the parent state */
size_t *jmptable, *pjmptable;
size_t jmp_cur, ii, ourjmpidx;
if (!jsn->jpr_root) {
*out = JSONSL_MATCH_NOMATCH;
return NULL;
}
pjmptable = jsn->jpr_root + (jsn->jpr_count * (state->level-1));
jmptable = pjmptable + jsn->jpr_count;
/* If the parent cannot match, then invalidate it */
if (*pjmptable == 0) {
*jmptable = 0;
*out = JSONSL_MATCH_NOMATCH;
return NULL;
}
parent_state = jsn->stack + state->level - 1;
if (parent_state->type == JSONSL_T_LIST) {
nkey = (size_t) parent_state->nelem;
}
*jmptable = 0;
ourjmpidx = 0;
memset(jmptable, 0, sizeof(int) * jsn->jpr_count);
for (ii = 0; ii < jsn->jpr_count; ii++) {
jmp_cur = pjmptable[ii];
if (jmp_cur) {
jsonsl_jpr_t jpr = jsn->jprs[jmp_cur-1];
*out = jsonsl_jpr_match(jpr,
parent_state->type,
parent_state->level,
key, nkey);
if (*out == JSONSL_MATCH_COMPLETE) {
ret = jpr;
*jmptable = 0;
return ret;
} else if (*out == JSONSL_MATCH_POSSIBLE) {
jmptable[ourjmpidx] = ii+1;
ourjmpidx++;
}
} else {
break;
}
}
if (!*jmptable) {
*out = JSONSL_MATCH_NOMATCH;
}
return NULL;
}
JSONSL_API
const char *jsonsl_strmatchtype(jsonsl_jpr_match_t match)
{
#define X(T,v) \
if ( match == JSONSL_MATCH_##T ) \
return #T;
JSONSL_XMATCH
#undef X
return "<UNKNOWN>";
}
#endif /* JSONSL_WITH_JPR */
static char *
jsonsl__writeutf8(uint32_t pt, char *out)
{
#define ADD_OUTPUT(c) *out = (char)(c); out++;
if (pt < 0x80) {
ADD_OUTPUT(pt);
} else if (pt < 0x800) {
ADD_OUTPUT((pt >> 6) | 0xC0);
ADD_OUTPUT((pt & 0x3F) | 0x80);
} else if (pt < 0x10000) {
ADD_OUTPUT((pt >> 12) | 0xE0);
ADD_OUTPUT(((pt >> 6) & 0x3F) | 0x80);
ADD_OUTPUT((pt & 0x3F) | 0x80);
} else {
ADD_OUTPUT((pt >> 18) | 0xF0);
ADD_OUTPUT(((pt >> 12) & 0x3F) | 0x80);
ADD_OUTPUT(((pt >> 6) & 0x3F) | 0x80);
ADD_OUTPUT((pt & 0x3F) | 0x80);
}
return out;
#undef ADD_OUTPUT
}
/* Thanks snej (https://github.com/mnunberg/jsonsl/issues/9) */
static int
jsonsl__digit2int(char ch) {
int d = ch - '0';
if ((unsigned) d < 10) {
return d;
}
d = ch - 'a';
if ((unsigned) d < 6) {
return d + 10;
}
d = ch - 'A';
if ((unsigned) d < 6) {
return d + 10;
}
return -1;
}
/* Assume 's' is at least 4 bytes long */
static int
jsonsl__get_uescape_16(const char *s)
{
int ret = 0;
int cur;
#define GET_DIGIT(off) \
cur = jsonsl__digit2int(s[off]); \
if (cur == -1) { return -1; } \
ret |= (cur << (12 - (off * 4)));
GET_DIGIT(0);
GET_DIGIT(1);
GET_DIGIT(2);
GET_DIGIT(3);
#undef GET_DIGIT
return ret;
}
/**
* Utility function to convert escape sequences
*/
JSONSL_API
size_t jsonsl_util_unescape_ex(const char *in,
char *out,
size_t len,
const int toEscape[128],
unsigned *oflags,
jsonsl_error_t *err,
const char **errat)
{
const unsigned char *c = (const unsigned char*)in;
char *begin_p = out;
unsigned oflags_s;
uint16_t last_codepoint = 0;
if (!oflags) {
oflags = &oflags_s;
}
*oflags = 0;
#define UNESCAPE_BAIL(e,offset) \
*err = JSONSL_ERROR_##e; \
if (errat) { \
*errat = (const char*)(c+ (ptrdiff_t)(offset)); \
} \
return 0;
for (; len; len--, c++, out++) {
int uescval;
if (*c != '\\') {
/* Not an escape, so we don't care about this */
goto GT_ASSIGN;
}
if (len < 2) {
UNESCAPE_BAIL(ESCAPE_INVALID, 0);
}
if (!is_allowed_escape(c[1])) {
UNESCAPE_BAIL(ESCAPE_INVALID, 1)
}
if ((toEscape && toEscape[(unsigned char)c[1] & 0x7f] == 0 &&
c[1] != '\\' && c[1] != '"')) {
/* if we don't want to unescape this string, write the escape sequence to the output */
*out++ = *c++;
if (--len == 0)
break;
goto GT_ASSIGN;
}
if (c[1] != 'u') {
/* simple skip-and-replace using pre-defined maps.
* TODO: should the maps actually reflect the desired
* replacement character in toEscape?
*/
char esctmp = get_escape_equiv(c[1]);
if (esctmp) {
/* Check if there is a corresponding replacement */
*out = esctmp;
} else {
/* Just gobble up the 'reverse-solidus' */
*out = c[1];
}
len--;
c++;
/* do not assign, just continue */
continue;
}
/* next == 'u' */
if (len < 6) {
/* Need at least six characters.. */
UNESCAPE_BAIL(UESCAPE_TOOSHORT, 2);
}
uescval = jsonsl__get_uescape_16((const char *)c + 2);
if (uescval == -1) {
UNESCAPE_BAIL(PERCENT_BADHEX, -1);
} else if (uescval == 0) {
UNESCAPE_BAIL(INVALID_CODEPOINT, 2);
}
if (last_codepoint) {
uint16_t w1 = last_codepoint, w2 = (uint16_t)uescval;
uint32_t cp;
if (uescval < 0xDC00 || uescval > 0xDFFF) {
UNESCAPE_BAIL(INVALID_CODEPOINT, -1);
}
cp = (w1 & 0x3FF) << 10;
cp |= (w2 & 0x3FF);
cp += 0x10000;
out = jsonsl__writeutf8(cp, out) - 1;
last_codepoint = 0;
} else if (uescval < 0xD800 || uescval > 0xDFFF) {
*oflags |= JSONSL_SPECIALf_NONASCII;
out = jsonsl__writeutf8(uescval, out) - 1;
} else if (uescval > 0xD7FF && uescval < 0xDC00) {
*oflags |= JSONSL_SPECIALf_NONASCII;
last_codepoint = (uint16_t)uescval;
out--;
} else {
UNESCAPE_BAIL(INVALID_CODEPOINT, 2);
}
/* Post uescape cleanup */
len -= 5; /* Gobble up 5 chars after 'u' */
c += 5;
continue;
/* Only reached by previous branches */
GT_ASSIGN:
*out = *c;
}
if (last_codepoint) {
*err = JSONSL_ERROR_INVALID_CODEPOINT;
return 0;
}
*err = JSONSL_ERROR_SUCCESS;
return out - begin_p;
}
/**
* Character Table definitions.
* These were all generated via srcutil/genchartables.pl
*/
/**
* This table contains the beginnings of non-string
* allowable (bareword) values.
*/
static const unsigned short Special_Table[0x80] = {
/* 0x00 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
/* 0x20 */ 0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x2c */
/* 0x2d */ JSONSL_SPECIALf_DASH /* <-> */, /* 0x2d */
/* 0x2e */ 0,0, /* 0x2f */
/* 0x30 */ JSONSL_SPECIALf_ZERO /* <0> */, /* 0x30 */
/* 0x31 */ JSONSL_SPECIALf_UNSIGNED /* <1> */, /* 0x31 */
/* 0x32 */ JSONSL_SPECIALf_UNSIGNED /* <2> */, /* 0x32 */
/* 0x33 */ JSONSL_SPECIALf_UNSIGNED /* <3> */, /* 0x33 */
/* 0x34 */ JSONSL_SPECIALf_UNSIGNED /* <4> */, /* 0x34 */
/* 0x35 */ JSONSL_SPECIALf_UNSIGNED /* <5> */, /* 0x35 */
/* 0x36 */ JSONSL_SPECIALf_UNSIGNED /* <6> */, /* 0x36 */
/* 0x37 */ JSONSL_SPECIALf_UNSIGNED /* <7> */, /* 0x37 */
/* 0x38 */ JSONSL_SPECIALf_UNSIGNED /* <8> */, /* 0x38 */
/* 0x39 */ JSONSL_SPECIALf_UNSIGNED /* <9> */, /* 0x39 */
/* 0x3a */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x59 */
/* 0x5a */ 0,0,0,0,0,0,0,0,0,0,0,0, /* 0x65 */
/* 0x66 */ JSONSL_SPECIALf_FALSE /* <f> */, /* 0x66 */
/* 0x67 */ 0,0,0,0,0,0,0, /* 0x6d */
/* 0x6e */ JSONSL_SPECIALf_NULL /* <n> */, /* 0x6e */
/* 0x6f */ 0,0,0,0,0, /* 0x73 */
/* 0x74 */ JSONSL_SPECIALf_TRUE /* <t> */ /* 0x74 */
};
// Bit tables are order such that the MSB is bit 0.
//
/**
* Contains characters which signal the termination of any of the 'special' bareword
* values.
*/
static const char Special_Endings[0x100] = {
/* 0x00 */ 0,0,0,0,0,0,0,0,0, /* 0x08 */
/* 0x09 */ 1 /* <TAB> */, /* 0x09 */
/* 0x0a */ 1 /* <LF> */, /* 0x0a */
/* 0x0b */ 0,0, /* 0x0c */
/* 0x0d */ 1 /* <CR> */, /* 0x0d */
/* 0x0e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
/* 0x20 */ 1 /* <SP> */, /* 0x20 */
/* 0x21 */ 0, /* 0x21 */
/* 0x22 */ 1 /* " */, /* 0x22 */
/* 0x23 */ 0,0,0,0,0,0,0,0,0, /* 0x2b */
/* 0x2c */ 1 /* , */, /* 0x2c */
/* 0x2d */ 0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x39 */
/* 0x3a */ 1 /* : */, /* 0x3a */
/* 0x3b */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x5a */
/* 0x5b */ 1 /* [ */, /* 0x5b */
/* 0x5c */ 1 /* \ */, /* 0x5c */
/* 0x5d */ 1 /* ] */, /* 0x5d */
/* 0x5e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x7a */
/* 0x7b */ 1 /* { */, /* 0x7b */
/* 0x7c */ 0, /* 0x7c */
/* 0x7d */ 1 /* } */, /* 0x7d */
/* 0x7e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x9d */
/* 0x9e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xbd */
/* 0xbe */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xdd */
/* 0xde */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xfd */
/* 0xfe */ 0 /* 0xfe */
};
static const uint32_t Special_Endings_bits[0x80 / 32] = {
0b00000000011001000000000000000000,
0b10100000000010000000000000100000,
0b00000000000000000000000000011100,
0b00000000000000000000000000010100
};
/**
* This table contains entries for the allowed whitespace as per RFC 4627
*/
static const char Allowed_Whitespace[0x100] = {
/* 0x00 */ 0,0,0,0,0,0,0,0,0, /* 0x08 */
/* 0x09 */ 1 /* <TAB> */, /* 0x09 */
/* 0x0a */ 1 /* <LF> */, /* 0x0a */
/* 0x0b */ 0,0, /* 0x0c */
/* 0x0d */ 1 /* <CR> */, /* 0x0d */
/* 0x0e */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
/* 0x20 */ 1 /* <SP> */, /* 0x20 */
/* 0x21 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x40 */
/* 0x41 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x60 */
/* 0x61 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x80 */
/* 0x81 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xa0 */
/* 0xa1 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xc0 */
/* 0xc1 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xe0 */
/* 0xe1 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 /* 0xfe */
};
static const uint32_t Allowed_Whitespace_bits = 0b00000000011001000000000000000000;
static const char String_No_Passthrough[0x100] = {
/* 0x00 */ 1 /* <NUL> */, /* 0x00 */
/* 0x01 */ 1 /* <SOH> */, /* 0x01 */
/* 0x02 */ 1 /* <STX> */, /* 0x02 */
/* 0x03 */ 1 /* <ETX> */, /* 0x03 */
/* 0x04 */ 1 /* <EOT> */, /* 0x04 */
/* 0x05 */ 1 /* <ENQ> */, /* 0x05 */
/* 0x06 */ 1 /* <ACK> */, /* 0x06 */
/* 0x07 */ 1 /* <BEL> */, /* 0x07 */
/* 0x08 */ 1 /* <BS> */, /* 0x08 */
/* 0x09 */ 1 /* <HT> */, /* 0x09 */
/* 0x0a */ 1 /* <LF> */, /* 0x0a */
/* 0x0b */ 1 /* <VT> */, /* 0x0b */
/* 0x0c */ 1 /* <FF> */, /* 0x0c */
/* 0x0d */ 1 /* <CR> */, /* 0x0d */
/* 0x0e */ 1 /* <SO> */, /* 0x0e */
/* 0x0f */ 1 /* <SI> */, /* 0x0f */
/* 0x10 */ 1 /* <DLE> */, /* 0x10 */
/* 0x11 */ 1 /* <DC1> */, /* 0x11 */
/* 0x12 */ 1 /* <DC2> */, /* 0x12 */
/* 0x13 */ 1 /* <DC3> */, /* 0x13 */
/* 0x14 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x21 */
/* 0x22 */ 1 /* <"> */, /* 0x22 */
/* 0x23 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x42 */
/* 0x43 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x5b */
/* 0x5c */ 1 /* <\> */, /* 0x5c */
/* 0x5d */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x7c */
/* 0x7d */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x9c */
/* 0x9d */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xbc */
/* 0xbd */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xdc */
/* 0xdd */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xfc */
/* 0xfd */ 0,0, /* 0xfe */
};
/**
* Allowable two-character 'common' escapes:
*/
static const char Allowed_Escapes[0x100] = {
/* 0x00 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
/* 0x20 */ 0,0, /* 0x21 */
/* 0x22 */ 1 /* <"> */, /* 0x22 */
/* 0x23 */ 0,0,0,0,0,0,0,0,0,0,0,0, /* 0x2e */
/* 0x2f */ 1 /* </> */, /* 0x2f */
/* 0x30 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x4f */
/* 0x50 */ 0,0,0,0,0,0,0,0,0,0,0,0, /* 0x5b */
/* 0x5c */ 1 /* <\> */, /* 0x5c */
/* 0x5d */ 0,0,0,0,0, /* 0x61 */
/* 0x62 */ 1 /* <b> */, /* 0x62 */
/* 0x63 */ 0,0,0, /* 0x65 */
/* 0x66 */ 1 /* <f> */, /* 0x66 */
/* 0x67 */ 0,0,0,0,0,0,0, /* 0x6d */
/* 0x6e */ 1 /* <n> */, /* 0x6e */
/* 0x6f */ 0,0,0, /* 0x71 */
/* 0x72 */ 1 /* <r> */, /* 0x72 */
/* 0x73 */ 0, /* 0x73 */
/* 0x74 */ 1 /* <t> */, /* 0x74 */
/* 0x75 */ 1 /* <u> */, /* 0x75 */
/* 0x76 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x95 */
/* 0x96 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xb5 */
/* 0xb6 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xd5 */
/* 0xd6 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xf5 */
/* 0xf6 */ 0,0,0,0,0,0,0,0,0, /* 0xfe */
};
static const uint32_t Allowed_Escapes_bits[0x80 / 32] = {
0b00000000000000000000000000000000,
0b00100000000000010000000000000000,
0b00000000000000000000000000001000,
0b00100010000000100010110000000000
};
/**
* This table contains the _values_ for a given (single) escaped character.
*/
static unsigned char Escape_Equivs[0x100] = {
/* 0x00 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x1f */
/* 0x20 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x3f */
/* 0x40 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x5f */
/* 0x60 */ 0,0, /* 0x61 */
/* 0x62 */ 8 /* <b> */, /* 0x62 */
/* 0x63 */ 0,0,0, /* 0x65 */
/* 0x66 */ 12 /* <f> */, /* 0x66 */
/* 0x67 */ 0,0,0,0,0,0,0, /* 0x6d */
/* 0x6e */ 10 /* <n> */, /* 0x6e */
/* 0x6f */ 0,0,0, /* 0x71 */
/* 0x72 */ 13 /* <r> */, /* 0x72 */
/* 0x73 */ 0, /* 0x73 */
/* 0x74 */ 9 /* <t> */, /* 0x74 */
/* 0x75 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0x94 */
/* 0x95 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xb4 */
/* 0xb5 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xd4 */
/* 0xd5 */ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xf4 */
/* 0xf5 */ 0,0,0,0,0,0,0,0,0,0 /* 0xfe */
};
/* Definitions of above-declared static functions */
static char get_escape_equiv(unsigned c) {
switch(c) {
case 'b':
return '\b';
case 'n':
return '\n';
case 'r':
return '\r';
case 't':
return '\t';
case 'f':
return '\f';
}
return 0;
}
static unsigned extract_special(unsigned c) {
return (c < 0x80) ? Special_Table[c & 0xff] : 0;
}
static int is_special_end(unsigned c) {
return (c < 0x80) && (Special_Endings_bits[c >> 5] & (1 << (31 - (c & 31))));
}
static int is_allowed_whitespace(unsigned c) {
return c == ' ' || (c < 0x20 && (Allowed_Whitespace_bits & (1 << (31 - c))));
}
static int is_allowed_escape(unsigned c) {
return (c < 0x80) && (Allowed_Escapes_bits[c >> 5] & (1 << (31 - (c & 31))));
}
static int is_simple_char(unsigned c) {
return !(c < 0x14 || c == '"' || c == '\\');
}
/* Clean up all our macros! */
#undef INCR_METRIC
#undef INCR_GENERIC
#undef INCR_STRINGY_CATCH
#undef CASE_DIGITS
#undef INVOKE_ERROR
#undef STACK_PUSH
#undef STACK_POP_NOPOS
#undef STACK_POP
#undef CALLBACK_AND_POP_NOPOS
#undef CALLBACK_AND_POP
#undef SPECIAL_POP
#undef CUR_CHAR
#undef DO_CALLBACK
#undef ENSURE_HVAL
#undef VERIFY_SPECIAL
#undef STATE_SPECIAL_LENGTH
#undef IS_NORMAL_NUMBER
#undef STATE_NUM_LAST
#undef FASTPARSE_EXHAUSTED
#undef FASTPARSE_BREAK
/**
* JSON Simple/Stacked/Stateful Lexer.
* - Does not buffer data
* - Maintains state
* - Callback oriented
* - Lightweight and fast. One source file and one header file
*
* Copyright (C) 2012-2015 Mark Nunberg
* See included LICENSE file for license details.
*/
#ifndef JSONSL_H_
#define JSONSL_H_
#include <stdio.h>
#include <stdlib.h>
#include <stddef.h>
#include <string.h>
#include <sys/types.h>
#include <wchar.h>
#ifdef __cplusplus
extern "C" {
#endif /* __cplusplus */
#ifdef JSONSL_USE_WCHAR
typedef jsonsl_char_t wchar_t;
typedef jsonsl_uchar_t unsigned wchar_t;
#else
typedef char jsonsl_char_t;
typedef unsigned char jsonsl_uchar_t;
#endif /* JSONSL_USE_WCHAR */
/* Stolen from http-parser.h, and possibly others */
#if defined(_WIN32) && !defined(__MINGW32__) && (!defined(_MSC_VER) || _MSC_VER<1600)
typedef __int8 int8_t;
typedef unsigned __int8 uint8_t;
typedef __int16 int16_t;
typedef unsigned __int16 uint16_t;
typedef __int32 int32_t;
typedef unsigned __int32 uint32_t;
typedef __int64 int64_t;
typedef unsigned __int64 uint64_t;
#if !defined(_MSC_VER) || _MSC_VER<1400
typedef unsigned int size_t;
typedef int ssize_t;
#endif
#else
#include <stdint.h>
#endif
#if (!defined(JSONSL_STATE_GENERIC)) && (!defined(JSONSL_STATE_USER_FIELDS))
#define JSONSL_STATE_GENERIC
#endif /* !defined JSONSL_STATE_GENERIC */
#ifdef JSONSL_STATE_GENERIC
#define JSONSL_STATE_USER_FIELDS
#endif /* JSONSL_STATE_GENERIC */
/* Additional fields for component object */
#ifndef JSONSL_JPR_COMPONENT_USER_FIELDS
#define JSONSL_JPR_COMPONENT_USER_FIELDS
#endif
#ifndef JSONSL_API
/**
* We require a /DJSONSL_DLL so that users already using this as a static
* or embedded library don't get confused
*/
#if defined(_WIN32) && defined(JSONSL_DLL)
#define JSONSL_API __declspec(dllexport)
#else
#define JSONSL_API
#endif /* _WIN32 */
#endif /* !JSONSL_API */
#ifndef JSONSL_INLINE
#if defined(_MSC_VER)
#define JSONSL_INLINE __inline
#elif defined(__GNUC__)
#define JSONSL_INLINE __inline__
#else
#define JSONSL_INLINE inline
#endif /* _MSC_VER or __GNUC__ */
#endif /* JSONSL_INLINE */
#define JSONSL_MAX_LEVELS 512
struct jsonsl_st;
typedef struct jsonsl_st *jsonsl_t;
typedef struct jsonsl_jpr_st* jsonsl_jpr_t;
/**
* This flag is true when AND'd against a type whose value
* must be in "quoutes" i.e. T_HKEY and T_STRING
*/
#define JSONSL_Tf_STRINGY 0xffff00
/**
* Constant representing the special JSON types.
* The values are special and aid in speed (the OBJECT and LIST
* values are the char literals of their openings).
*
* Their actual value is a character which attempts to resemble
* some mnemonic reference to the actual type.
*
* If new types are added, they must fit into the ASCII printable
* range (so they should be AND'd with 0x7f and yield something
* meaningful)
*/
#define JSONSL_XTYPE \
X(STRING, '"'|JSONSL_Tf_STRINGY) \
X(HKEY, '#'|JSONSL_Tf_STRINGY) \
X(OBJECT, '{') \
X(LIST, '[') \
X(SPECIAL, '^') \
X(UESCAPE, 'u')
typedef enum {
#define X(o, c) \
JSONSL_T_##o = c,
JSONSL_XTYPE
JSONSL_T_UNKNOWN = '?',
/* Abstract 'root' object */
JSONSL_T_ROOT = 0
#undef X
} jsonsl_type_t;
/**
* Subtypes for T_SPECIAL. We define them as flags
* because more than one type can be applied to a
* given object.
*/
#define JSONSL_XSPECIAL \
X(NONE, 0) \
X(SIGNED, 1<<0) \
X(UNSIGNED, 1<<1) \
X(TRUE, 1<<2) \
X(FALSE, 1<<3) \
X(NULL, 1<<4) \
X(FLOAT, 1<<5) \
X(EXPONENT, 1<<6) \
X(NONASCII, 1<<7)
typedef enum {
#define X(o,b) \
JSONSL_SPECIALf_##o = b,
JSONSL_XSPECIAL
#undef X
/* Handy flags for checking */
JSONSL_SPECIALf_UNKNOWN = 1 << 8,
/** @private Private */
JSONSL_SPECIALf_ZERO = 1 << 9 | JSONSL_SPECIALf_UNSIGNED,
/** @private */
JSONSL_SPECIALf_DASH = 1 << 10,
/** Type is numeric */
JSONSL_SPECIALf_NUMERIC = (JSONSL_SPECIALf_SIGNED| JSONSL_SPECIALf_UNSIGNED),
/** Type is a boolean */
JSONSL_SPECIALf_BOOLEAN = (JSONSL_SPECIALf_TRUE|JSONSL_SPECIALf_FALSE),
/** Type is an "extended", not integral type (but numeric) */
JSONSL_SPECIALf_NUMNOINT = (JSONSL_SPECIALf_FLOAT|JSONSL_SPECIALf_EXPONENT)
} jsonsl_special_t;
/**
* These are the various types of stack (or other) events
* which will trigger a callback.
* Like the type constants, this are also mnemonic
*/
#define JSONSL_XACTION \
X(PUSH, '+') \
X(POP, '-') \
X(UESCAPE, 'U') \
X(ERROR, '!')
typedef enum {
#define X(a,c) \
JSONSL_ACTION_##a = c,
JSONSL_XACTION
JSONSL_ACTION_UNKNOWN = '?'
#undef X
} jsonsl_action_t;
/**
* Various errors which may be thrown while parsing JSON
*/
#define JSONSL_XERR \
/* Trailing garbage characters */ \
X(GARBAGE_TRAILING) \
/* We were expecting a 'special' (numeric, true, false, null) */ \
X(SPECIAL_EXPECTED) \
/* The 'special' value was incomplete */ \
X(SPECIAL_INCOMPLETE) \
/* Found a stray token */ \
X(STRAY_TOKEN) \
/* We were expecting a token before this one */ \
X(MISSING_TOKEN) \
/* Cannot insert because the container is not ready */ \
X(CANT_INSERT) \
/* Found a '\' outside a string */ \
X(ESCAPE_OUTSIDE_STRING) \
/* Found a ':' outside of a hash */ \
X(KEY_OUTSIDE_OBJECT) \
/* found a string outside of a container */ \
X(STRING_OUTSIDE_CONTAINER) \
/* Found a null byte in middle of string */ \
X(FOUND_NULL_BYTE) \
/* Current level exceeds limit specified in constructor */ \
X(LEVELS_EXCEEDED) \
/* Got a } as a result of an opening [ or vice versa */ \
X(BRACKET_MISMATCH) \
/* We expected a key, but got something else instead */ \
X(HKEY_EXPECTED) \
/* We got an illegal control character (bad whitespace or something) */ \
X(WEIRD_WHITESPACE) \
/* Found a \u-escape, but there were less than 4 following hex digits */ \
X(UESCAPE_TOOSHORT) \
/* Invalid two-character escape */ \
X(ESCAPE_INVALID) \
/* Trailing comma */ \
X(TRAILING_COMMA) \
/* An invalid number was passed in a numeric field */ \
X(INVALID_NUMBER) \
/* Value is missing for object */ \
X(VALUE_EXPECTED) \
/* The following are for JPR Stuff */ \
\
/* Found a literal '%' but it was only followed by a single valid hex digit */ \
X(PERCENT_BADHEX) \
/* jsonpointer URI is malformed '/' */ \
X(JPR_BADPATH) \
/* Duplicate slash */ \
X(JPR_DUPSLASH) \
/* No leading root */ \
X(JPR_NOROOT) \
/* Allocation failure */ \
X(ENOMEM) \
/* Invalid unicode codepoint detected (in case of escapes) */ \
X(INVALID_CODEPOINT)
typedef enum {
JSONSL_ERROR_SUCCESS = 0,
#define X(e) \
JSONSL_ERROR_##e,
JSONSL_XERR
#undef X
JSONSL_ERROR_GENERIC
} jsonsl_error_t;
/**
* A state is a single level of the stack.
* Non-private data (i.e. the 'data' field, see the STATE_GENERIC section)
* will remain in tact until the item is popped.
*
* As a result, it means a parent state object may be accessed from a child
* object, (the parents fields will all be valid). This allows a user to create
* an ad-hoc hierarchy on top of the JSON one.
*
*/
struct jsonsl_state_st {
/**
* The JSON object type
*/
unsigned int type;
/**
* The position (in terms of number of bytes since the first call to
* jsonsl_feed()) at which the state was first pushed. This includes
* opening tokens, if applicable.
*
* @note For strings (i.e. type & JSONSL_Tf_STRINGY is nonzero) this will
* be the position of the first quote.
*
* @see jsonsl_st::pos which contains the _current_ position and can be
* used during a POP callback to get the length of the element.
*/
size_t pos_begin;
/**FIXME: This is redundant as the same information can be derived from
* jsonsl_st::pos at pop-time */
size_t pos_cur;
/** If this element is special, then its extended type is here */
unsigned short special_flags;
/**
* Level of recursion into nesting. This is mainly a convenience
* variable, as this can technically be deduced from the lexer's
* level parameter (though the logic is not that simple)
*/
unsigned short level;
/**
* how many elements in the object/list.
* For objects (hashes), an element is either
* a key or a value. Thus for one complete pair,
* nelem will be 2.
*
* For special types, this will hold the sum of the digits.
* This only holds true for values which are simple signed/unsigned
* numbers. Otherwise a special flag is set, and extra handling is not
* performed.
*/
uint32_t nelem;
/*TODO: merge this and special_flags into a union */
/**
* Useful for an opening nest, this will prevent a callback from being
* invoked on this item or any of its children
*/
int ignore_callback : 1;
/**
* Counter which is incremented each time an escape ('\') is encountered.
* This is used internally for non-string types and should only be
* inspected by the user if the state actually represents a string
* type.
*/
unsigned int nescapes : 31;
/**
* Put anything you want here. if JSONSL_STATE_USER_FIELDS is here, then
* the macro expansion happens here.
*
* You can use these fields to store hierarchical or 'tagging' information
* for specific objects.
*
* See the documentation above for the lifetime of the state object (i.e.
* if the private data points to allocated memory, it should be freed
* when the object is popped, as the state object will be re-used)
*/
#ifndef JSONSL_STATE_GENERIC
JSONSL_STATE_USER_FIELDS
#else
/**
* Otherwise, this is a simple void * pointer for anything you want
*/
void *data;
#endif /* JSONSL_STATE_USER_FIELDS */
};
/**Gets the number of elements in the list.
* @param st The state. Must be of type JSONSL_T_LIST
* @return number of elements in the list
*/
#define JSONSL_LIST_SIZE(st) ((st)->nelem)
/**Gets the number of key-value pairs in an object
* @param st The state. Must be of type JSONSL_T_OBJECT
* @return the number of key-value pairs in the object
*/
#define JSONSL_OBJECT_SIZE(st) ((st)->nelem / 2)
/**Gets the numeric value.
* @param st The state. Must be of type JSONSL_T_SPECIAL and
* special_flags must have the JSONSL_SPECIALf_NUMERIC flag
* set.
* @return the numeric value of the state.
*/
#define JSONSL_NUMERIC_VALUE(st) ((st)->nelem)
/*
* So now we need some special structure for keeping the
* JPR info in sync. Preferrably all in a single block
* of memory (there's no need for separate allocations.
* So we will define a 'table' with the following layout
*
* Level nPosbl JPR1_last JPR2_last JPR3_last
*
* 0 1 NOMATCH POSSIBLE POSSIBLE
* 1 0 NOMATCH NOMATCH COMPLETE
* [ table ends here because no further path is possible]
*
* Where the JPR..n corresponds to the number of JPRs
* requested, and nPosble is a quick flag to determine
*
* the number of possibilities. In the future this might
* be made into a proper 'jump' table,
*
* Since we always mark JPRs from the higher levels descending
* into the lower ones, a prospective child match would first
* look at the parent table to check the possibilities, and then
* see which ones were possible..
*
* Thus, the size of this blob would be (and these are all ints here)
* nLevels * nJPR * 2.
*
* the 'Width' of the table would be nJPR*2, and the 'height' would be
* nlevels
*/
/**
* This is called when a stack change ocurs.
*
* @param jsn The lexer
* @param action The type of action, this can be PUSH or POP
* @param state A pointer to the stack currently affected by the action
* @param at A pointer to the position of the input buffer which triggered
* this action.
*/
typedef void (*jsonsl_stack_callback)(
jsonsl_t jsn,
jsonsl_action_t action,
struct jsonsl_state_st* state,
const jsonsl_char_t *at);
/**
* This is called when an error is encountered.
* Sometimes it's possible to 'erase' characters (by replacing them
* with whitespace). If you think you have corrected the error, you
* can return a true value, in which case the parser will backtrack
* and try again.
*
* @param jsn The lexer
* @param error The error which was thrown
* @param state the current state
* @param a pointer to the position of the input buffer which triggered
* the error. Note that this is not const, this is because you have the
* possibility of modifying the character in an attempt to correct the
* error
*
* @return zero to bail, nonzero to try again (this only makes sense if
* the input buffer has been modified by this callback)
*/
typedef int (*jsonsl_error_callback)(
jsonsl_t jsn,
jsonsl_error_t error,
struct jsonsl_state_st* state,
jsonsl_char_t *at);
struct jsonsl_st {
/** Public, read-only */
/** This is the current level of the stack */
unsigned int level;
/** Flag set to indicate we should stop processing */
unsigned int stopfl;
/**
* This is the current position, relative to the beginning
* of the stream.
*/
size_t pos;
/** This is the 'bytes' variable passed to feed() */
const jsonsl_char_t *base;
/** Callback invoked for PUSH actions */
jsonsl_stack_callback action_callback_PUSH;
/** Callback invoked for POP actions */
jsonsl_stack_callback action_callback_POP;
/** Default callback for any action, if neither PUSH or POP callbacks are defined */
jsonsl_stack_callback action_callback;
/**
* Do not invoke callbacks for objects deeper than this level.
* NOTE: This field establishes the lower bound for ignored callbacks,
* and is thus misnamed. `min_ignore_level` would actually make more
* sense, but we don't want to break API.
*/
unsigned int max_callback_level;
/** The error callback. Invoked when an error happens. Should not be NULL */
jsonsl_error_callback error_callback;
/* these are boolean flags you can modify. You will be called
* about notification for each of these types if the corresponding
* variable is true.
*/
/**
* @name Callback Booleans.
* These determine whether a callback is to be invoked for certain types of objects
* @{*/
/** Boolean flag to enable or disable the invokcation for events on this type*/
int call_SPECIAL;
int call_OBJECT;
int call_LIST;
int call_STRING;
int call_HKEY;
/*@}*/
/**
* @name u-Escape handling
* Special handling for the \\u-f00d type sequences. These are meant
* to be translated back into the corresponding octet(s).
* A special callback (if set) is invoked with *at=='u'. An application
* may wish to temporarily suspend parsing and handle the 'u-' sequence
* internally (or not).
*/
/*@{*/
/** Callback to be invoked for a u-escape */
jsonsl_stack_callback action_callback_UESCAPE;
/** Boolean flag, whether to invoke the callback */
int call_UESCAPE;
/** Boolean flag, whether we should return after encountering a u-escape:
* the callback is invoked and then we return if this is true
*/
int return_UESCAPE;
/*@}*/
struct {
int allow_trailing_comma;
} options;
/** Put anything here */
void *data;
/*@{*/
/** Private */
int in_escape;
char expecting;
char tok_last;
int can_insert;
unsigned int levels_max;
#ifndef JSONSL_NO_JPR
size_t jpr_count;
jsonsl_jpr_t *jprs;
/* Root pointer for JPR matching information */
size_t *jpr_root;
#endif /* JSONSL_NO_JPR */
/*@}*/
/**
* This is the stack. Its upper bound is levels_max, or the
* nlevels argument passed to jsonsl_new. If you modify this structure,
* make sure that this member is last.
*/
struct jsonsl_state_st stack[1];
};
/**
* Creates a new lexer object, with capacity for recursion up to nlevels
*
* @param nlevels maximum recursion depth
*/
JSONSL_API
jsonsl_t jsonsl_new(int nlevels);
JSONSL_API
jsonsl_t jsonsl_init(jsonsl_t jsn, int nlevels);
JSONSL_API
size_t jsonsl_get_size(int nlevels);
/**
* Feeds data into the lexer.
*
* @param jsn the lexer object
* @param bytes new data to be fed
* @param nbytes size of new data
*/
JSONSL_API
void jsonsl_feed(jsonsl_t jsn, const jsonsl_char_t *bytes, size_t nbytes);
/**
* Resets the internal parser state. This does not free the parser
* but does clean it internally, so that the next time feed() is called,
* it will be treated as a new stream
*
* @param jsn the lexer
*/
JSONSL_API
void jsonsl_reset(jsonsl_t jsn);
/**
* Frees the lexer, cleaning any allocated memory taken
*
* @param jsn the lexer
*/
JSONSL_API
void jsonsl_destroy(jsonsl_t jsn);
/**
* Gets the 'parent' element, given the current one
*
* @param jsn the lexer
* @param cur the current nest, which should be a struct jsonsl_nest_st
*/
static JSONSL_INLINE
struct jsonsl_state_st *jsonsl_last_state(const jsonsl_t jsn,
const struct jsonsl_state_st *state)
{
/* Don't complain about overriding array bounds */
if (state->level > 1) {
return jsn->stack + state->level - 1;
} else {
return NULL;
}
}
/**
* Gets the state of the last fully consumed child of this parent. This is
* only valid in the parent's POP callback.
*
* @param the lexer
* @return A pointer to the child.
*/
static JSONSL_INLINE
struct jsonsl_state_st *jsonsl_last_child(const jsonsl_t jsn,
const struct jsonsl_state_st *parent)
{
return jsn->stack + (parent->level + 1);
}
/**Call to instruct the parser to stop parsing and return. This is valid
* only from within a callback */
static JSONSL_INLINE
void jsonsl_stop(jsonsl_t jsn)
{
jsn->stopfl = 1;
}
/**
* This enables receiving callbacks on all events. Doesn't do
* anything special but helps avoid some boilerplate.
* This does not touch the UESCAPE callbacks or flags.
*/
static JSONSL_INLINE
void jsonsl_enable_all_callbacks(jsonsl_t jsn)
{
jsn->call_HKEY = 1;
jsn->call_STRING = 1;
jsn->call_OBJECT = 1;
jsn->call_SPECIAL = 1;
jsn->call_LIST = 1;
}
/**
* A macro which returns true if the current state object can
* have children. This means a list type or an object type.
*/
#define JSONSL_STATE_IS_CONTAINER(state) \
(state->type == JSONSL_T_OBJECT || state->type == JSONSL_T_LIST)
/**
* These two functions, dump a string representation
* of the error or type, respectively. They will never
* return NULL
*/
JSONSL_API
const char* jsonsl_strerror(jsonsl_error_t err);
JSONSL_API
const char* jsonsl_strtype(jsonsl_type_t jt);
/**
* Dumps global metrics to the screen. This is a noop unless
* jsonsl was compiled with JSONSL_USE_METRICS
*/
JSONSL_API
void jsonsl_dump_global_metrics(void);
/* This macro just here for editors to do code folding */
#ifndef JSONSL_NO_JPR
/**
* @name JSON Pointer API
*
* JSONPointer API. This isn't really related to the lexer (at least not yet)
* JSONPointer provides an extremely simple specification for providing
* locations within JSON objects. We will extend it a bit and allow for
* providing 'wildcard' characters by which to be able to 'query' the stream.
*
* See http://tools.ietf.org/html/draft-pbryan-zyp-json-pointer-00
*
* Currently I'm implementing the 'single query' API which can only use a single
* query component. In the future I will integrate my yet-to-be-published
* Boyer-Moore-esque prefix searching implementation, in order to allow
* multiple paths to be merged into one for quick and efficient searching.
*
*
* JPR (as we'll refer to it within the source) can be used by splitting
* the components into mutliple sections, and incrementally 'track' each
* component. When JSONSL delivers a 'pop' callback for a string, or a 'push'
* callback for an object, we will check to see whether the index matching
* the component corresponding to the current level contains a match
* for our path.
*
* In order to do this properly, a structure must be maintained within the
* parent indicating whether its children are possible matches. This flag
* will be 'inherited' by call children which may conform to the match
* specification, and discarded by all which do not (thereby eliminating
* their children from inheriting it).
*
* A successful match is a complete one. One can provide multiple paths with
* multiple levels of matches e.g.
* /foo/bar/baz/^/blah
*
* @{
*/
/** The wildcard character */
#ifndef JSONSL_PATH_WILDCARD_CHAR
#define JSONSL_PATH_WILDCARD_CHAR '^'
#endif /* WILDCARD_CHAR */
#define JSONSL_XMATCH \
X(COMPLETE,1) \
X(POSSIBLE,0) \
X(NOMATCH,-1) \
X(TYPE_MISMATCH, -2)
typedef enum {
#define X(T,v) \
JSONSL_MATCH_##T = v,
JSONSL_XMATCH
#undef X
JSONSL_MATCH_UNKNOWN
} jsonsl_jpr_match_t;
typedef enum {
JSONSL_PATH_STRING = 1,
JSONSL_PATH_WILDCARD,
JSONSL_PATH_NUMERIC,
JSONSL_PATH_ROOT,
/* Special */
JSONSL_PATH_INVALID = -1,
JSONSL_PATH_NONE = 0
} jsonsl_jpr_type_t;
struct jsonsl_jpr_component_st {
/** The string the component points to */
char *pstr;
/** if this is a numeric type, the number is 'cached' here */
unsigned long idx;
/** The length of the string */
size_t len;
/** The type of component (NUMERIC or STRING) */
jsonsl_jpr_type_t ptype;
/** Set this to true to enforce type checking between dict keys and array
* indices. jsonsl_jpr_match() will return TYPE_MISMATCH if it detects
* that an array index is actually a child of a dictionary. */
short is_arridx;
/* Extra fields (for more advanced searches. Default is empty) */
JSONSL_JPR_COMPONENT_USER_FIELDS
};
struct jsonsl_jpr_st {
/** Path components */
struct jsonsl_jpr_component_st *components;
size_t ncomponents;
/**Type of the match to be expected. If nonzero, will be compared against
* the actual type */
unsigned match_type;
/** Base of allocated string for components */
char *basestr;
/** The original match string. Useful for returning to the user */
char *orig;
size_t norig;
};
/**
* Create a new JPR object.
*
* @param path the JSONPointer path specification.
* @param errp a pointer to a jsonsl_error_t. If this function returns NULL,
* then more details will be in this variable.
*
* @return a new jsonsl_jpr_t object, or NULL on error.
*/
JSONSL_API
jsonsl_jpr_t jsonsl_jpr_new(const char *path, jsonsl_error_t *errp);
/**
* Destroy a JPR object
*/
JSONSL_API
void jsonsl_jpr_destroy(jsonsl_jpr_t jpr);
/**
* Match a JSON object against a type and specific level
*
* @param jpr the JPR object
* @param parent_type the type of the parent (should be T_LIST or T_OBJECT)
* @param parent_level the level of the parent
* @param key the 'key' of the child. If the parent is an array, this should be
* empty.
* @param nkey - the length of the key. If the parent is an array (T_LIST), then
* this should be the current index.
*
* NOTE: The key of the child means any kind of associative data related to the
* element. Thus: <<< { "foo" : [ >>,
* the opening array's key is "foo".
*
* @return a status constant. This indicates whether a match was excluded, possible,
* or successful.
*/
JSONSL_API
jsonsl_jpr_match_t jsonsl_jpr_match(jsonsl_jpr_t jpr,
unsigned int parent_type,
unsigned int parent_level,
const char *key, size_t nkey);
/**
* Alternate matching algorithm. This matching algorithm does not use
* JSONPointer but relies on a more structured searching mechanism. It
* assumes that there is a clear distinction between array indices and
* object keys. In this case, the jsonsl_path_component_st::ptype should
* be set to @ref JSONSL_PATH_NUMERIC for an array index (the
* jsonsl_path_comonent_st::is_arridx field will be removed in a future
* version).
*
* @param jpr The path
* @param parent The parent structure. Can be NULL if this is the root object
* @param child The child structure. Should not be NULL
* @param key Object key, if an object
* @param nkey Length of object key
* @return Status constant if successful
*
* @note
* For successful matching, both the key and the path itself should be normalized
* to contain 'proper' utf8 sequences rather than utf16 '\uXXXX' escapes. This
* should currently be done in the application. Another version of this function
* may use a temporary buffer in such circumstances (allocated by the application).
*
* Since this function also checks the state of the child, it should only
* be called on PUSH callbacks, and not POP callbacks
*/
JSONSL_API
jsonsl_jpr_match_t
jsonsl_path_match(jsonsl_jpr_t jpr,
const struct jsonsl_state_st *parent,
const struct jsonsl_state_st *child,
const char *key, size_t nkey);
/**
* Associate a set of JPR objects with a lexer instance.
* This should be called before the lexer has been fed any data (and
* behavior is undefined if you don't adhere to this).
*
* After using this function, you may subsequently call match_state() on
* given states (presumably from within the callbacks).
*
* Note that currently the first JPR is the quickest and comes
* pre-allocated with the state structure. Further JPR objects
* are chained.
*
* @param jsn The lexer
* @param jprs An array of jsonsl_jpr_t objects
* @param njprs How many elements in the jprs array.
*/
JSONSL_API
void jsonsl_jpr_match_state_init(jsonsl_t jsn,
jsonsl_jpr_t *jprs,
size_t njprs);
/**
* This follows the same semantics as the normal match,
* except we infer parent and type information from the relevant state objects.
* The match status (for all possible JPR objects) is set in the *out parameter.
*
* If a match has succeeded, then its JPR object will be returned. In all other
* instances, NULL is returned;
*
* @param jpr The jsonsl_jpr_t handle
* @param state The jsonsl_state_st which is a candidate
* @param key The hash key (if applicable, can be NULL if parent is list)
* @param nkey Length of hash key (if applicable, can be zero if parent is list)
* @param out A pointer to a jsonsl_jpr_match_t. This will be populated with
* the match result
*
* @return If a match was completed in full, then the JPR object containing
* the matching path will be returned. Otherwise, the return is NULL (note, this
* does not mean matching has failed, it can still be part of the match: check
* the out parameter).
*/
JSONSL_API
jsonsl_jpr_t jsonsl_jpr_match_state(jsonsl_t jsn,
struct jsonsl_state_st *state,
const char *key,
size_t nkey,
jsonsl_jpr_match_t *out);
/**
* Cleanup any memory allocated and any states set by
* match_state_init() and match_state()
* @param jsn The lexer
*/
JSONSL_API
void jsonsl_jpr_match_state_cleanup(jsonsl_t jsn);
/**
* Return a string representation of the match result returned by match()
*/
JSONSL_API
const char *jsonsl_strmatchtype(jsonsl_jpr_match_t match);
/* @}*/
/**
* Utility function to convert escape sequences into their original form.
*
* The decoders I've sampled do not seem to specify a standard behavior of what
* to escape/unescape.
*
* RFC 4627 Mandates only that the quoute, backslash, and ASCII control
* characters (0x00-0x1f) be escaped. It is often common for applications
* to escape a '/' - however this may also be desired behavior. the JSON
* spec is not clear on this, and therefore jsonsl leaves it up to you.
*
* Additionally, sometimes you may wish to _normalize_ JSON. This is specifically
* true when dealing with 'u-escapes' which can be expressed perfectly fine
* as utf8. One use case for normalization is JPR string comparison, in which
* case two effectively equivalent strings may not match because one is using
* u-escapes and the other proper utf8. To normalize u-escapes only, pass in
* an empty `toEscape` table, enabling only the `u` index.
*
* @param in The input string.
* @param out An allocated output (should be the same size as in)
* @param len the size of the buffer
* @param toEscape - A sparse array of characters to unescape. Characters
* which are not present in this array, e.g. toEscape['c'] == 0 will be
* ignored and passed to the output in their original form.
* @param oflags If not null, and a \uXXXX escape expands to a non-ascii byte,
* then this variable will have the SPECIALf_NONASCII flag on.
*
* @param err A pointer to an error variable. If an error ocurrs, it will be
* set in this variable
* @param errat If not null and an error occurs, this will be set to point
* to the position within the string at which the offending character was
* encountered.
*
* @return The effective size of the output buffer.
*
* @note
* This function now encodes the UTF8 equivalents of utf16 escapes (i.e.
* 'u-escapes'). Previously this would encode the escapes as utf16 literals,
* which while still correct in some sense was confusing for many (especially
* considering that the inputs were variations of char).
*
* @note
* The output buffer will never be larger than the input buffer, since
* standard escape sequences (i.e. '\t') occupy two bytes in the source
* but only one byte (when unescaped) in the output. Likewise u-escapes
* (i.e. \uXXXX) will occupy six bytes in the source, but at the most
* two bytes when escaped.
*/
JSONSL_API
size_t jsonsl_util_unescape_ex(const char *in,
char *out,
size_t len,
const int toEscape[128],
unsigned *oflags,
jsonsl_error_t *err,
const char **errat);
/**
* Convenience macro to avoid passing too many parameters
*/
#define jsonsl_util_unescape(in, out, len, toEscape, err) \
jsonsl_util_unescape_ex(in, out, len, toEscape, NULL, err, NULL)
#endif /* JSONSL_NO_JPR */
#ifdef __cplusplus
}
#endif /* __cplusplus */
#endif /* JSONSL_H_ */
# SJSON Module
| Since | Origin / Contributor | Maintainer | Source |
| :----- | :-------------------- | :---------- | :------ |
| 2017-02-01 | [Philip Gladstone](https://github.com/pjsg) | [Philip Gladstone](https://github.com/pjsg) | [sjson](../../../app/modules/sjson.c) |
The JSON support module. Allows encoding and decoding to/from JSON.
Please note that nested tables can require a lot of memory to encode. To catch out-of-memory errors, use `pcall()`.
This code using the streaming json library [jsonsl](https://github.com/mnunberg/jsonsl) to do the parsing of the string.
This module can be used in two ways. The simpler way is to use it as a direct drop-in for cjson (you can just do `_G.cjson = sjson`).
The more advanced approach is to use the streaming interface. This allows encoding and decoding of significantly larger objects.
The handling of json null is as follows:
- By default, the decoder represents null as sjson.NULL (which is a userdata object). This is the behavior of cjson.
- The encoder always converts any userdata object into null.
- Optionally, a single string can be specified in both the encoder and decoder. This string will be used in encoding/decoding to represent json null values. This string should not be used
anywhere else in your data structures. A suitable value might be `"\0"`.
When encoding a Lua object, if a function is found, then it is invoked (with no arguments) and the (single) returned value is encoded in the place of the function.
!!! note
All examples below use in-memory JSON or content read from the SPIFFS file system. However, where a streaming implementation really shines is in fetching large JSON structures from the remote resources and extracting values on-the-fly. An elaborate streaming example can be found in the [`/lua_examples`](../../../lua_examples/sjson-streaming.lua) folder.
## sjson.encoder()
This creates an encoder object that can convert a Lua object into a JSON encoded string.
####Syntax
`sjson.encoder(table [, opts])`
####Parameters
- `table` data to encode
- `opts` an optional table of options. The possible entries are:
- `depth` the maximum encoding depth needed to encode the table. The default is 20 which should be enough for nearly all situations.
- `null` the string value to treat as null.
####Returns
A `sjson.encoder` object.
## sjson.encoder:read
This gets a chunk of JSON encoded data.
####Syntax
`encoder:read([size])`
####Parameters
- `size` an optional value for the number of bytes to return. The default is 1024.
####Returns
A string of up to `size` bytes, or `nil` if the encoding is complete and all data has been returned.
#### Example
The following example prints out (in 64 byte chunks) a JSON encoded string containing the first 4k of every file in the file system. The total string
can be bigger than the total amount of memory on the NodeMCU.
```lua
function files()
result = {}
for k,v in pairs(file.list()) do
result[k] = function() return file.open(k):read(4096) end
end
return result
end
local encoder = sjson.encoder(files())
while true do
data = encoder:read(64)
if not data then
break
end
print(data)
end
```
## sjson.encode()
Encode a Lua table to a JSON string. This is a convenience method provided for backwards compatibility with `cjson`.
####Syntax
`sjson.encode(table [, opts])`
####Parameters
- `table` data to encode
- `opts` an optional table of options. The possible entries are:
- `depth` the maximum encoding depth needed to encode the table. The default is 20 which should be enough for nearly all situations.
- `null` the string value to treat as null.
####Returns
JSON string
####Example
```lua
ok, json = pcall(sjson.encode, {key="value"})
if ok then
print(json)
else
print("failed to encode!")
end
```
## sjson.decoder()
This makes a decoder object that can parse a JSON encoded string into a Lua object. A metatable can be specified for all the newly created Lua tables. This allows
you to handle each value as it is inserted into each table (by implementing the `__newindex` method).
####Syntax
`sjson.decoder([opts])`
#### Parameters
- `opts` an optional table of options. The possible entries are:
- `depth` the maximum encoding depth needed to encode the table. The default is 20 which should be enough for nearly all situations.
- `null` the string value to treat as null.
- `metatable` a table to use as the metatable for all the new tables in the returned object.
#### Returns
A `sjson.decoder` object
####Metatable
There are two principal methods that are invoked in the metatable (if it is present).
- `__newindex` this is the standard method invoked whenever a new table element is created.
- `checkpath` this is invoked (if defined) whenever a new table is created. It is invoked with two arguments:
- `table` this is the newly created table
- `path` this is a list of the keys from the root.
It must return `true` if this object is wanted in the result, or `false` otherwise.
For example, when decoding `{ "foo": [1, 2, []] }` the checkpath will be invoked as follows:
- `checkpath({}, {})` the `table` argument is the object that will correspond with the value of the JSON object.
- `checkpath({}, {"foo"})` the `table` argument is the object that will correspond with the value of the outer JSON array.
- `checkpath({}, {"foo", 3})` the `table` argument is the object that will correspond to the empty inner JSON array.
When the `checkpath` method is called, the metatable has already be associated with the new table. Thus the `checkpath` method can replace it
if desired. For example, if you are decoding `{ "foo": { "bar": [1,2,3,4], "cat": [5] } }` and, for some reason, you did not want to capture the
value of the `"bar"` key, then there are various ways to do this:
* In the `__newindex` metamethod, just check for the value of the key and skip the `rawset` if the key is `"bar"`. This only works if you want to skip all the
`"bar"` keys.
* In the `checkpath` method, if the path is `["foo"]`, then return `false`.
* Use the following `checkpath`: `checkpath=function(tab, path) tab['__json_path'] = path return true end` This will save the path in each constructed object. Now the `__newindex` method can perform more sophisticated filtering.
The reason for being able to filter is that it enables processing of very large JSON responses on a memory constrained platform. Many APIs return lots of information
which would exceed the memory budget of the platform. For example, `https://api.github.com/repos/nodemcu/nodemcu-firmware/contents` is over 13kB, and yet, if
you only need the `download_url` keys, then the total size is around 600B. This can be handled with a simple `__newindex` method.
## sjson.decoder:write
This provides more data to be parsed into the Lua object.
####Syntax
`decoder:write(string)`
####Parameters
- `string` the next piece of JSON encoded data
####Returns
The constructed Lua object or `nil` if the decode is not yet complete.
####Errors
If a parse error occurrs during this decode, then an error is thrown and the parse is aborted. The object cannot be used again.
## sjson.decoder:result
This gets the decoded Lua object, or raises an error if the decode is not yet complete. This can be called multiple times and will return the
same object each time.
####Syntax
`decoder:result()`
####Errors
If the decode is not complete, then an error is thrown.
####Example
```
local decoder = sjson.decoder()
decoder:write("[10, 1")
decoder:write("1")
decoder:write(", \"foo\"]")
for k,v in pairs(decoder:result()) do
print (k, v)
end
```
The next example demonstrates the use of the metatable argument. In this case it just prints out the operations, but it could suppress the assignment
altogether if desired.
```
local decoder = sjson.decoder({metatable=
{__newindex=function(t,k,v) print("Setting '" .. k .. "' = '" .. tostring(v) .."'")
rawset(t,k,v) end}})
decoder:write('[1, 2, {"foo":"bar"}]')
```
## sjson.decode()
Decode a JSON string to a Lua table. This is a convenience method provided for backwards compatibility with `cjson`.
####Syntax
`sjson.decode(str[, opts])`
####Parameters
- `str` JSON string to decode
- `opts` an optional table of options. The possible entries are:
- `depth` the maximum encoding depth needed to encode the table. The default is 20 which should be enough for nearly all situations.
- `null` the string value to treat as null.
- `metatable` a table to use as the metatable for all the new tables in the returned object. See the metatable section in the description of `sjson.decoder()` above.
####Returns
Lua table representation of the JSON data
####Errors
If the string is not valid JSON, then an error is thrown.
####Example
```lua
t = sjson.decode('{"key":"value"}')
for k,v in pairs(t) do print(k,v) end
```
##Constants
There is one constant, `sjson.NULL`, which is used in Lua structures to represent the presence of a JSON null.
...@@ -51,6 +51,7 @@ pages: ...@@ -51,6 +51,7 @@ pages:
- 'qrcodegen': 'en/modules/qrcodegen.md' - 'qrcodegen': 'en/modules/qrcodegen.md'
- 'sdmmc': 'en/modules/sdmmc.md' - 'sdmmc': 'en/modules/sdmmc.md'
- 'sigma delta': 'en/modules/sigma-delta.md' - 'sigma delta': 'en/modules/sigma-delta.md'
- 'sjson': 'en/modules/sjson.md'
- 'sodium': 'en/modules/sodium.md' - 'sodium': 'en/modules/sodium.md'
- 'spi': 'en/modules/spi.md' - 'spi': 'en/modules/spi.md'
- 'struct': 'en/modules/struct.md' - 'struct': 'en/modules/struct.md'
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment