diff options
Diffstat (limited to 'src')
| -rw-r--r-- | src/cjson/fpconv.c | 211 | ||||
| -rw-r--r-- | src/cjson/fpconv.h | 22 | ||||
| -rw-r--r-- | src/cjson/lua_cjson.c | 1609 | ||||
| -rw-r--r-- | src/cjson/lua_cjson.h | 10 | ||||
| -rw-r--r-- | src/cjson/strbuf.c | 251 | ||||
| -rw-r--r-- | src/cjson/strbuf.h | 159 | ||||
| -rw-r--r-- | src/nvim/CMakeLists.txt | 6 | ||||
| -rw-r--r-- | src/nvim/lua/executor.c | 4 |
8 files changed, 2269 insertions, 3 deletions
diff --git a/src/cjson/fpconv.c b/src/cjson/fpconv.c new file mode 100644 index 0000000000..b2f7a214c2 --- /dev/null +++ b/src/cjson/fpconv.c @@ -0,0 +1,211 @@ +/* fpconv - Floating point conversion routines + * + * Copyright (c) 2011-2012 Mark Pulford <mark@kyne.com.au> + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +/* JSON uses a '.' decimal separator. strtod() / sprintf() under C libraries + * with locale support will break when the decimal separator is a comma. + * + * fpconv_* will around these issues with a translation buffer if required. + */ + +#include <stdio.h> +#include <stdlib.h> +#include <assert.h> +#include <string.h> + +#include "fpconv.h" + +/* Workaround for MSVC */ +#ifdef _MSC_VER +#define inline __inline +#define snprintf sprintf_s +#endif + +/* Lua CJSON assumes the locale is the same for all threads within a + * process and doesn't change after initialisation. + * + * This avoids the need for per thread storage or expensive checks + * for call. */ +static char locale_decimal_point = '.'; + +/* In theory multibyte decimal_points are possible, but + * Lua CJSON only supports UTF-8 and known locales only have + * single byte decimal points ([.,]). + * + * localconv() may not be thread safe (=>crash), and nl_langinfo() is + * not supported on some platforms. Use sprintf() instead - if the + * locale does change, at least Lua CJSON won't crash. */ +static void fpconv_update_locale(void) +{ + char buf[8]; + + snprintf(buf, sizeof(buf), "%g", 0.5); + + /* Failing this test might imply the platform has a buggy dtoa + * implementation or wide characters */ + if (buf[0] != '0' || buf[2] != '5' || buf[3] != 0) { + fprintf(stderr, "Error: wide characters found or printf() bug."); + abort(); + } + + locale_decimal_point = buf[1]; +} + +/* Check for a valid number character: [-+0-9a-yA-Y.] + * Eg: -0.6e+5, infinity, 0xF0.F0pF0 + * + * Used to find the probable end of a number. It doesn't matter if + * invalid characters are counted - strtod() will find the valid + * number if it exists. The risk is that slightly more memory might + * be allocated before a parse error occurs. */ +static inline int valid_number_character(char ch) +{ + char lower_ch; + + if ('0' <= ch && ch <= '9') + return 1; + if (ch == '-' || ch == '+' || ch == '.') + return 1; + + /* Hex digits, exponent (e), base (p), "infinity",.. */ + lower_ch = ch | 0x20; + if ('a' <= lower_ch && lower_ch <= 'y') + return 1; + + return 0; +} + +/* Calculate the size of the buffer required for a strtod locale + * conversion. */ +static int strtod_buffer_size(const char *s) +{ + const char *p = s; + + while (valid_number_character(*p)) + p++; + + return p - s; +} + +/* Similar to strtod(), but must be passed the current locale's decimal point + * character. Guaranteed to be called at the start of any valid number in a string */ +double fpconv_strtod(const char *nptr, char **endptr) +{ + char localbuf[FPCONV_G_FMT_BUFSIZE]; + char *buf, *endbuf, *dp; + int buflen; + double value; + + /* System strtod() is fine when decimal point is '.' */ + if (locale_decimal_point == '.') + return strtod(nptr, endptr); + + buflen = strtod_buffer_size(nptr); + if (!buflen) { + /* No valid characters found, standard strtod() return */ + *endptr = (char *)nptr; + return 0; + } + + /* Duplicate number into buffer */ + if (buflen >= FPCONV_G_FMT_BUFSIZE) { + /* Handle unusually large numbers */ + buf = malloc(buflen + 1); + if (!buf) { + fprintf(stderr, "Out of memory"); + abort(); + } + } else { + /* This is the common case.. */ + buf = localbuf; + } + memcpy(buf, nptr, buflen); + buf[buflen] = 0; + + /* Update decimal point character if found */ + dp = strchr(buf, '.'); + if (dp) + *dp = locale_decimal_point; + + value = strtod(buf, &endbuf); + *endptr = (char *)&nptr[endbuf - buf]; + if (buflen >= FPCONV_G_FMT_BUFSIZE) + free(buf); + + return value; +} + +/* "fmt" must point to a buffer of at least 6 characters */ +static void set_number_format(char *fmt, int precision) +{ + int d1, d2, i; + + assert(1 <= precision && precision <= 16); + + /* Create printf format (%.14g) from precision */ + d1 = precision / 10; + d2 = precision % 10; + fmt[0] = '%'; + fmt[1] = '.'; + i = 2; + if (d1) { + fmt[i++] = '0' + d1; + } + fmt[i++] = '0' + d2; + fmt[i++] = 'g'; + fmt[i] = 0; +} + +/* Assumes there is always at least 32 characters available in the target buffer */ +int fpconv_g_fmt(char *str, double num, int precision) +{ + char buf[FPCONV_G_FMT_BUFSIZE]; + char fmt[6]; + int len; + char *b; + + set_number_format(fmt, precision); + + /* Pass through when decimal point character is dot. */ + if (locale_decimal_point == '.') + return snprintf(str, FPCONV_G_FMT_BUFSIZE, fmt, num); + + /* snprintf() to a buffer then translate for other decimal point characters */ + len = snprintf(buf, FPCONV_G_FMT_BUFSIZE, fmt, num); + + /* Copy into target location. Translate decimal point if required */ + b = buf; + do { + *str++ = (*b == locale_decimal_point ? '.' : *b); + } while(*b++); + + return len; +} + +void fpconv_init() +{ + fpconv_update_locale(); +} + +/* vi:ai et sw=4 ts=4: + */ diff --git a/src/cjson/fpconv.h b/src/cjson/fpconv.h new file mode 100644 index 0000000000..6ac97808b6 --- /dev/null +++ b/src/cjson/fpconv.h @@ -0,0 +1,22 @@ +/* Lua CJSON floating point conversion routines */ + +/* Buffer required to store the largest string representation of a double. + * + * Longest double printed with %.14g is 21 characters long: + * -1.7976931348623e+308 */ +# define FPCONV_G_FMT_BUFSIZE 32 + +#ifdef USE_INTERNAL_FPCONV +static inline void fpconv_init() +{ + /* Do nothing - not required */ +} +#else +extern void fpconv_init(void); +#endif + +extern int fpconv_g_fmt(char*, double, int); +extern double fpconv_strtod(const char*, char**); + +/* vi:ai et sw=4 ts=4: + */ diff --git a/src/cjson/lua_cjson.c b/src/cjson/lua_cjson.c new file mode 100644 index 0000000000..92d07963bd --- /dev/null +++ b/src/cjson/lua_cjson.c @@ -0,0 +1,1609 @@ +/* Lua CJSON - JSON support for Lua + * + * Copyright (c) 2010-2012 Mark Pulford <mark@kyne.com.au> + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +/* Caveats: + * - JSON "null" values are represented as lightuserdata since Lua + * tables cannot contain "nil". Compare with cjson.null. + * - Invalid UTF-8 characters are not detected and will be passed + * untouched. If required, UTF-8 error checking should be done + * outside this library. + * - Javascript comments are not part of the JSON spec, and are not + * currently supported. + * + * Note: Decoding is slower than encoding. Lua spends significant + * time (30%) managing tables when parsing JSON since it is + * difficult to know object/array sizes ahead of time. + */ + +#include <assert.h> +#include <stdint.h> +#include <string.h> +#include <math.h> +#include <limits.h> +#include <lua.h> +#include <lauxlib.h> + +#include "nvim/lua/executor.h" + +#include "lua_cjson.h" +#include "strbuf.h" +#include "fpconv.h" + +#ifndef CJSON_MODNAME +#define CJSON_MODNAME "cjson" +#endif + +#ifndef CJSON_VERSION +#define CJSON_VERSION "2.1.0.9" +#endif + +#ifdef _MSC_VER +#define snprintf sprintf_s + +#ifndef isnan +#include <float.h> +#define isnan(x) _isnan(x) +#endif + +#endif + +/* Workaround for Solaris platforms missing isinf() */ +#if !defined(isinf) && (defined(USE_INTERNAL_ISINF) || defined(MISSING_ISINF)) +#define isinf(x) (!isnan(x) && isnan((x) - (x))) +#endif + +#define DEFAULT_SPARSE_CONVERT 0 +#define DEFAULT_SPARSE_RATIO 2 +#define DEFAULT_SPARSE_SAFE 10 +#define DEFAULT_ENCODE_MAX_DEPTH 1000 +#define DEFAULT_DECODE_MAX_DEPTH 1000 +#define DEFAULT_ENCODE_INVALID_NUMBERS 0 +#define DEFAULT_DECODE_INVALID_NUMBERS 1 +#define DEFAULT_ENCODE_KEEP_BUFFER 1 +#define DEFAULT_ENCODE_NUMBER_PRECISION 14 +#define DEFAULT_ENCODE_EMPTY_TABLE_AS_OBJECT 0 +#define DEFAULT_DECODE_ARRAY_WITH_ARRAY_MT 0 +#define DEFAULT_ENCODE_ESCAPE_FORWARD_SLASH 1 + +#ifdef DISABLE_INVALID_NUMBERS +#undef DEFAULT_DECODE_INVALID_NUMBERS +#define DEFAULT_DECODE_INVALID_NUMBERS 0 +#endif + +#ifdef _MSC_VER +/* Microsoft C compiler lacks strncasecmp and strcasecmp. */ +#define strncasecmp _strnicmp +#define strcasecmp _stricmp +#endif + +#if LONG_MAX > ((1UL << 31) - 1) +#define json_lightudata_mask(ludata) \ + ((void *) ((uintptr_t) (ludata) & ((1UL << 47) - 1))) + +#else +#define json_lightudata_mask(ludata) (ludata) +#endif + +#if LUA_VERSION_NUM > 501 +#define lua_objlen(L,i) lua_rawlen(L, (i)) +#endif + +static const char * const *json_empty_array; +static const char * const *json_array; + +typedef enum { + T_OBJ_BEGIN, + T_OBJ_END, + T_ARR_BEGIN, + T_ARR_END, + T_STRING, + T_NUMBER, + T_BOOLEAN, + T_NULL, + T_COLON, + T_COMMA, + T_END, + T_WHITESPACE, + T_ERROR, + T_UNKNOWN +} json_token_type_t; + +static const char *json_token_type_name[] = { + "T_OBJ_BEGIN", + "T_OBJ_END", + "T_ARR_BEGIN", + "T_ARR_END", + "T_STRING", + "T_NUMBER", + "T_BOOLEAN", + "T_NULL", + "T_COLON", + "T_COMMA", + "T_END", + "T_WHITESPACE", + "T_ERROR", + "T_UNKNOWN", + NULL +}; + +typedef struct { + json_token_type_t ch2token[256]; + char escape2char[256]; /* Decoding */ + + /* encode_buf is only allocated and used when + * encode_keep_buffer is set */ + strbuf_t encode_buf; + + int encode_sparse_convert; + int encode_sparse_ratio; + int encode_sparse_safe; + int encode_max_depth; + int encode_invalid_numbers; /* 2 => Encode as "null" */ + int encode_number_precision; + int encode_keep_buffer; + int encode_empty_table_as_object; + int encode_escape_forward_slash; + + int decode_invalid_numbers; + int decode_max_depth; + int decode_array_with_array_mt; +} json_config_t; + +typedef struct { + const char *data; + const char *ptr; + strbuf_t *tmp; /* Temporary storage for strings */ + json_config_t *cfg; + int current_depth; +} json_parse_t; + +typedef struct { + json_token_type_t type; + int index; + union { + const char *string; + double number; + int boolean; + } value; + int string_len; +} json_token_t; + +static const char *char2escape[256] = { + "\\u0000", "\\u0001", "\\u0002", "\\u0003", + "\\u0004", "\\u0005", "\\u0006", "\\u0007", + "\\b", "\\t", "\\n", "\\u000b", + "\\f", "\\r", "\\u000e", "\\u000f", + "\\u0010", "\\u0011", "\\u0012", "\\u0013", + "\\u0014", "\\u0015", "\\u0016", "\\u0017", + "\\u0018", "\\u0019", "\\u001a", "\\u001b", + "\\u001c", "\\u001d", "\\u001e", "\\u001f", + NULL, NULL, "\\\"", NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, "\\/", + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, "\\\\", NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, "\\u007f", + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, +}; + +/* ===== CONFIGURATION ===== */ + +static json_config_t *json_fetch_config(lua_State *l) +{ + json_config_t *cfg; + + cfg = lua_touserdata(l, lua_upvalueindex(1)); + if (!cfg) + luaL_error(l, "BUG: Unable to fetch CJSON configuration"); + + return cfg; +} + +/* Ensure the correct number of arguments have been provided. + * Pad with nil to allow other functions to simply check arg[i] + * to find whether an argument was provided */ +static json_config_t *json_arg_init(lua_State *l, int args) +{ + luaL_argcheck(l, lua_gettop(l) <= args, args + 1, + "found too many arguments"); + + while (lua_gettop(l) < args) + lua_pushnil(l); + + return json_fetch_config(l); +} + +/* Process integer options for configuration functions */ +static int json_integer_option(lua_State *l, int optindex, int *setting, + int min, int max) +{ + char errmsg[64]; + int value; + + if (!lua_isnil(l, optindex)) { + value = luaL_checkinteger(l, optindex); + snprintf(errmsg, sizeof(errmsg), "expected integer between %d and %d", min, max); + luaL_argcheck(l, min <= value && value <= max, 1, errmsg); + *setting = value; + } + + lua_pushinteger(l, *setting); + + return 1; +} + +/* Process enumerated arguments for a configuration function */ +static int json_enum_option(lua_State *l, int optindex, int *setting, + const char **options, int bool_true) +{ + static const char *bool_options[] = { "off", "on", NULL }; + + if (!options) { + options = bool_options; + bool_true = 1; + } + + if (!lua_isnil(l, optindex)) { + if (bool_true && lua_isboolean(l, optindex)) + *setting = lua_toboolean(l, optindex) * bool_true; + else + *setting = luaL_checkoption(l, optindex, NULL, options); + } + + if (bool_true && (*setting == 0 || *setting == bool_true)) + lua_pushboolean(l, *setting); + else + lua_pushstring(l, options[*setting]); + + return 1; +} + +/* Configures handling of extremely sparse arrays: + * convert: Convert extremely sparse arrays into objects? Otherwise error. + * ratio: 0: always allow sparse; 1: never allow sparse; >1: use ratio + * safe: Always use an array when the max index <= safe */ +static int json_cfg_encode_sparse_array(lua_State *l) +{ + json_config_t *cfg = json_arg_init(l, 3); + + json_enum_option(l, 1, &cfg->encode_sparse_convert, NULL, 1); + json_integer_option(l, 2, &cfg->encode_sparse_ratio, 0, INT_MAX); + json_integer_option(l, 3, &cfg->encode_sparse_safe, 0, INT_MAX); + + return 3; +} + +/* Configures the maximum number of nested arrays/objects allowed when + * encoding */ +static int json_cfg_encode_max_depth(lua_State *l) +{ + json_config_t *cfg = json_arg_init(l, 1); + + return json_integer_option(l, 1, &cfg->encode_max_depth, 1, INT_MAX); +} + +/* Configures the maximum number of nested arrays/objects allowed when + * encoding */ +static int json_cfg_decode_max_depth(lua_State *l) +{ + json_config_t *cfg = json_arg_init(l, 1); + + return json_integer_option(l, 1, &cfg->decode_max_depth, 1, INT_MAX); +} + +/* Configures number precision when converting doubles to text */ +static int json_cfg_encode_number_precision(lua_State *l) +{ + json_config_t *cfg = json_arg_init(l, 1); + + return json_integer_option(l, 1, &cfg->encode_number_precision, 1, 16); +} + +/* Configures how to treat empty table when encode lua table */ +static int json_cfg_encode_empty_table_as_object(lua_State *l) +{ + json_config_t *cfg = json_arg_init(l, 1); + + return json_enum_option(l, 1, &cfg->encode_empty_table_as_object, NULL, 1); +} + +/* Configures how to decode arrays */ +static int json_cfg_decode_array_with_array_mt(lua_State *l) +{ + json_config_t *cfg = json_arg_init(l, 1); + + json_enum_option(l, 1, &cfg->decode_array_with_array_mt, NULL, 1); + + return 1; +} + +/* Configures JSON encoding buffer persistence */ +static int json_cfg_encode_keep_buffer(lua_State *l) +{ + json_config_t *cfg = json_arg_init(l, 1); + int old_value; + + old_value = cfg->encode_keep_buffer; + + json_enum_option(l, 1, &cfg->encode_keep_buffer, NULL, 1); + + /* Init / free the buffer if the setting has changed */ + if (old_value ^ cfg->encode_keep_buffer) { + if (cfg->encode_keep_buffer) + strbuf_init(&cfg->encode_buf, 0); + else + strbuf_free(&cfg->encode_buf); + } + + return 1; +} + +#if defined(DISABLE_INVALID_NUMBERS) && !defined(USE_INTERNAL_FPCONV) +void json_verify_invalid_number_setting(lua_State *l, int *setting) +{ + if (*setting == 1) { + *setting = 0; + luaL_error(l, "Infinity, NaN, and/or hexadecimal numbers are not supported."); + } +} +#else +#define json_verify_invalid_number_setting(l, s) do { } while(0) +#endif + +static int json_cfg_encode_invalid_numbers(lua_State *l) +{ + static const char *options[] = { "off", "on", "null", NULL }; + json_config_t *cfg = json_arg_init(l, 1); + + json_enum_option(l, 1, &cfg->encode_invalid_numbers, options, 1); + + json_verify_invalid_number_setting(l, &cfg->encode_invalid_numbers); + + return 1; +} + +static int json_cfg_decode_invalid_numbers(lua_State *l) +{ + json_config_t *cfg = json_arg_init(l, 1); + + json_enum_option(l, 1, &cfg->decode_invalid_numbers, NULL, 1); + + json_verify_invalid_number_setting(l, &cfg->encode_invalid_numbers); + + return 1; +} + +static int json_cfg_encode_escape_forward_slash(lua_State *l) +{ + int ret; + json_config_t *cfg = json_arg_init(l, 1); + + ret = json_enum_option(l, 1, &cfg->encode_escape_forward_slash, NULL, 1); + if (cfg->encode_escape_forward_slash) { + char2escape['/'] = "\\/"; + } else { + char2escape['/'] = NULL; + } + return ret; +} + +static int json_destroy_config(lua_State *l) +{ + json_config_t *cfg; + + cfg = lua_touserdata(l, 1); + if (cfg) + strbuf_free(&cfg->encode_buf); + cfg = NULL; + + return 0; +} + +static void json_create_config(lua_State *l) +{ + json_config_t *cfg; + int i; + + cfg = lua_newuserdata(l, sizeof(*cfg)); + + /* Create GC method to clean up strbuf */ + lua_newtable(l); + lua_pushcfunction(l, json_destroy_config); + lua_setfield(l, -2, "__gc"); + lua_setmetatable(l, -2); + + cfg->encode_sparse_convert = DEFAULT_SPARSE_CONVERT; + cfg->encode_sparse_ratio = DEFAULT_SPARSE_RATIO; + cfg->encode_sparse_safe = DEFAULT_SPARSE_SAFE; + cfg->encode_max_depth = DEFAULT_ENCODE_MAX_DEPTH; + cfg->decode_max_depth = DEFAULT_DECODE_MAX_DEPTH; + cfg->encode_invalid_numbers = DEFAULT_ENCODE_INVALID_NUMBERS; + cfg->decode_invalid_numbers = DEFAULT_DECODE_INVALID_NUMBERS; + cfg->encode_keep_buffer = DEFAULT_ENCODE_KEEP_BUFFER; + cfg->encode_number_precision = DEFAULT_ENCODE_NUMBER_PRECISION; + cfg->encode_empty_table_as_object = DEFAULT_ENCODE_EMPTY_TABLE_AS_OBJECT; + cfg->decode_array_with_array_mt = DEFAULT_DECODE_ARRAY_WITH_ARRAY_MT; + cfg->encode_escape_forward_slash = DEFAULT_ENCODE_ESCAPE_FORWARD_SLASH; + +#if DEFAULT_ENCODE_KEEP_BUFFER > 0 + strbuf_init(&cfg->encode_buf, 0); +#endif + + /* Decoding init */ + + /* Tag all characters as an error */ + for (i = 0; i < 256; i++) + cfg->ch2token[i] = T_ERROR; + + /* Set tokens that require no further processing */ + cfg->ch2token['{'] = T_OBJ_BEGIN; + cfg->ch2token['}'] = T_OBJ_END; + cfg->ch2token['['] = T_ARR_BEGIN; + cfg->ch2token[']'] = T_ARR_END; + cfg->ch2token[','] = T_COMMA; + cfg->ch2token[':'] = T_COLON; + cfg->ch2token['\0'] = T_END; + cfg->ch2token[' '] = T_WHITESPACE; + cfg->ch2token['\t'] = T_WHITESPACE; + cfg->ch2token['\n'] = T_WHITESPACE; + cfg->ch2token['\r'] = T_WHITESPACE; + + /* Update characters that require further processing */ + cfg->ch2token['f'] = T_UNKNOWN; /* false? */ + cfg->ch2token['i'] = T_UNKNOWN; /* inf, ininity? */ + cfg->ch2token['I'] = T_UNKNOWN; + cfg->ch2token['n'] = T_UNKNOWN; /* null, nan? */ + cfg->ch2token['N'] = T_UNKNOWN; + cfg->ch2token['t'] = T_UNKNOWN; /* true? */ + cfg->ch2token['"'] = T_UNKNOWN; /* string? */ + cfg->ch2token['+'] = T_UNKNOWN; /* number? */ + cfg->ch2token['-'] = T_UNKNOWN; + for (i = 0; i < 10; i++) + cfg->ch2token['0' + i] = T_UNKNOWN; + + /* Lookup table for parsing escape characters */ + for (i = 0; i < 256; i++) + cfg->escape2char[i] = 0; /* String error */ + cfg->escape2char['"'] = '"'; + cfg->escape2char['\\'] = '\\'; + cfg->escape2char['/'] = '/'; + cfg->escape2char['b'] = '\b'; + cfg->escape2char['t'] = '\t'; + cfg->escape2char['n'] = '\n'; + cfg->escape2char['f'] = '\f'; + cfg->escape2char['r'] = '\r'; + cfg->escape2char['u'] = 'u'; /* Unicode parsing required */ +} + +/* ===== ENCODING ===== */ + +static void json_encode_exception(lua_State *l, json_config_t *cfg, strbuf_t *json, int lindex, + const char *reason) +{ + if (!cfg->encode_keep_buffer) + strbuf_free(json); + luaL_error(l, "Cannot serialise %s: %s", + lua_typename(l, lua_type(l, lindex)), reason); +} + +/* json_append_string args: + * - lua_State + * - JSON strbuf + * - String (Lua stack index) + * + * Returns nothing. Doesn't remove string from Lua stack */ +static void json_append_string(lua_State *l, strbuf_t *json, int lindex) +{ + const char *escstr; + unsigned i; + const char *str; + size_t len; + + str = lua_tolstring(l, lindex, &len); + + /* Worst case is len * 6 (all unicode escapes). + * This buffer is reused constantly for small strings + * If there are any excess pages, they won't be hit anyway. + * This gains ~5% speedup. */ + strbuf_ensure_empty_length(json, len * 6 + 2); + + strbuf_append_char_unsafe(json, '\"'); + for (i = 0; i < len; i++) { + escstr = char2escape[(unsigned char)str[i]]; + if (escstr) + strbuf_append_string(json, escstr); + else + strbuf_append_char_unsafe(json, str[i]); + } + strbuf_append_char_unsafe(json, '\"'); +} + +/* Find the size of the array on the top of the Lua stack + * -1 object (not a pure array) + * >=0 elements in array + */ +static int lua_array_length(lua_State *l, json_config_t *cfg, strbuf_t *json) +{ + double k; + int max; + int items; + + max = 0; + items = 0; + + lua_pushnil(l); + /* table, startkey */ + while (lua_next(l, -2) != 0) { + /* table, key, value */ + if (lua_type(l, -2) == LUA_TNUMBER && + (k = lua_tonumber(l, -2))) { + /* Integer >= 1 ? */ + if (floor(k) == k && k >= 1) { + if (k > max) + max = k; + items++; + lua_pop(l, 1); + continue; + } + } + + /* Must not be an array (non integer key) */ + lua_pop(l, 2); + return -1; + } + + /* Encode excessively sparse arrays as objects (if enabled) */ + if (cfg->encode_sparse_ratio > 0 && + max > items * cfg->encode_sparse_ratio && + max > cfg->encode_sparse_safe) { + if (!cfg->encode_sparse_convert) + json_encode_exception(l, cfg, json, -1, "excessively sparse array"); + + return -1; + } + + return max; +} + +static void json_check_encode_depth(lua_State *l, json_config_t *cfg, + int current_depth, strbuf_t *json) +{ + /* Ensure there are enough slots free to traverse a table (key, + * value) and push a string for a potential error message. + * + * Unlike "decode", the key and value are still on the stack when + * lua_checkstack() is called. Hence an extra slot for luaL_error() + * below is required just in case the next check to lua_checkstack() + * fails. + * + * While this won't cause a crash due to the EXTRA_STACK reserve + * slots, it would still be an improper use of the API. */ + if (current_depth <= cfg->encode_max_depth && lua_checkstack(l, 3)) + return; + + if (!cfg->encode_keep_buffer) + strbuf_free(json); + + luaL_error(l, "Cannot serialise, excessive nesting (%d)", + current_depth); +} + +static void json_append_data(lua_State *l, json_config_t *cfg, + int current_depth, strbuf_t *json); + +/* json_append_array args: + * - lua_State + * - JSON strbuf + * - Size of passwd Lua array (top of stack) */ +static void json_append_array(lua_State *l, json_config_t *cfg, int current_depth, + strbuf_t *json, int array_length) +{ + int comma, i; + + strbuf_append_char(json, '['); + + comma = 0; + for (i = 1; i <= array_length; i++) { + if (comma) + strbuf_append_char(json, ','); + else + comma = 1; + + lua_rawgeti(l, -1, i); + json_append_data(l, cfg, current_depth, json); + lua_pop(l, 1); + } + + strbuf_append_char(json, ']'); +} + +static void json_append_number(lua_State *l, json_config_t *cfg, + strbuf_t *json, int lindex) +{ + double num = lua_tonumber(l, lindex); + int len; + + if (cfg->encode_invalid_numbers == 0) { + /* Prevent encoding invalid numbers */ + if (isinf(num) || isnan(num)) + json_encode_exception(l, cfg, json, lindex, + "must not be NaN or Infinity"); + } else if (cfg->encode_invalid_numbers == 1) { + /* Encode NaN/Infinity separately to ensure Javascript compatible + * values are used. */ + if (isnan(num)) { + strbuf_append_mem(json, "NaN", 3); + return; + } + if (isinf(num)) { + if (num < 0) + strbuf_append_mem(json, "-Infinity", 9); + else + strbuf_append_mem(json, "Infinity", 8); + return; + } + } else { + /* Encode invalid numbers as "null" */ + if (isinf(num) || isnan(num)) { + strbuf_append_mem(json, "null", 4); + return; + } + } + + strbuf_ensure_empty_length(json, FPCONV_G_FMT_BUFSIZE); + len = fpconv_g_fmt(strbuf_empty_ptr(json), num, cfg->encode_number_precision); + strbuf_extend_length(json, len); +} + +static void json_append_object(lua_State *l, json_config_t *cfg, + int current_depth, strbuf_t *json) +{ + int comma, keytype; + + /* Object */ + strbuf_append_char(json, '{'); + + lua_pushnil(l); + /* table, startkey */ + comma = 0; + while (lua_next(l, -2) != 0) { + if (comma) + strbuf_append_char(json, ','); + else + comma = 1; + + /* table, key, value */ + keytype = lua_type(l, -2); + if (keytype == LUA_TNUMBER) { + strbuf_append_char(json, '"'); + json_append_number(l, cfg, json, -2); + strbuf_append_mem(json, "\":", 2); + } else if (keytype == LUA_TSTRING) { + json_append_string(l, json, -2); + strbuf_append_char(json, ':'); + } else { + json_encode_exception(l, cfg, json, -2, + "table key must be a number or string"); + /* never returns */ + } + + /* table, key, value */ + json_append_data(l, cfg, current_depth, json); + lua_pop(l, 1); + /* table, key */ + } + + strbuf_append_char(json, '}'); +} + +/* Serialise Lua data into JSON string. */ +static void json_append_data(lua_State *l, json_config_t *cfg, + int current_depth, strbuf_t *json) +{ + int len; + int as_array = 0; + int as_empty_dict = 0; + int has_metatable; + + switch (lua_type(l, -1)) { + case LUA_TSTRING: + json_append_string(l, json, -1); + break; + case LUA_TNUMBER: + json_append_number(l, cfg, json, -1); + break; + case LUA_TBOOLEAN: + if (lua_toboolean(l, -1)) + strbuf_append_mem(json, "true", 4); + else + strbuf_append_mem(json, "false", 5); + break; + case LUA_TTABLE: + current_depth++; + json_check_encode_depth(l, cfg, current_depth, json); + + has_metatable = lua_getmetatable(l, -1); + + if (has_metatable) { + + nlua_pushref(l, nlua_empty_dict_ref); + if (lua_rawequal(l, -2, -1)) { + as_empty_dict = true; + } else { + lua_pop(l, 1); + lua_pushlightuserdata(l, json_lightudata_mask(&json_array)); + lua_rawget(l, LUA_REGISTRYINDEX); + as_array = lua_rawequal(l, -1, -2); + } + lua_pop(l, 2); + } + + if (as_array) { + len = lua_objlen(l, -1); + json_append_array(l, cfg, current_depth, json, len); + } else { + len = lua_array_length(l, cfg, json); + + if (len > 0 || (len == 0 && !cfg->encode_empty_table_as_object && !as_empty_dict)) { + json_append_array(l, cfg, current_depth, json, len); + } else { + if (has_metatable) { + lua_getmetatable(l, -1); + lua_pushlightuserdata(l, json_lightudata_mask( + &json_empty_array)); + lua_rawget(l, LUA_REGISTRYINDEX); + as_array = lua_rawequal(l, -1, -2); + lua_pop(l, 2); /* pop pointer + metatable */ + if (as_array) { + json_append_array(l, cfg, current_depth, json, 0); + break; + } + } + json_append_object(l, cfg, current_depth, json); + } + } + break; + case LUA_TNIL: + strbuf_append_mem(json, "null", 4); + break; + case LUA_TLIGHTUSERDATA: + if (lua_touserdata(l, -1) == &json_array) { + json_append_array(l, cfg, current_depth, json, 0); + } + break; + case LUA_TUSERDATA: + nlua_pushref(l, nlua_nil_ref); + bool is_nil = lua_rawequal(l, -2, -1); + lua_pop(l, 1); + if (is_nil) { + strbuf_append_mem(json, "null", 4); + break; + } else { + FALLTHROUGH; + } + default: + /* Remaining types (LUA_TFUNCTION, LUA_TUSERDATA, LUA_TTHREAD, + * and LUA_TLIGHTUSERDATA) cannot be serialised */ + json_encode_exception(l, cfg, json, -1, "type not supported"); + /* never returns */ + } +} + +static int json_encode(lua_State *l) +{ + json_config_t *cfg = json_fetch_config(l); + strbuf_t local_encode_buf; + strbuf_t *encode_buf; + char *json; + int len; + + luaL_argcheck(l, lua_gettop(l) == 1, 1, "expected 1 argument"); + + if (!cfg->encode_keep_buffer) { + /* Use private buffer */ + encode_buf = &local_encode_buf; + strbuf_init(encode_buf, 0); + } else { + /* Reuse existing buffer */ + encode_buf = &cfg->encode_buf; + strbuf_reset(encode_buf); + } + + json_append_data(l, cfg, 0, encode_buf); + json = strbuf_string(encode_buf, &len); + + lua_pushlstring(l, json, len); + + if (!cfg->encode_keep_buffer) + strbuf_free(encode_buf); + + return 1; +} + +/* ===== DECODING ===== */ + +static void json_process_value(lua_State *l, json_parse_t *json, + json_token_t *token); + +static int hexdigit2int(char hex) +{ + if ('0' <= hex && hex <= '9') + return hex - '0'; + + /* Force lowercase */ + hex |= 0x20; + if ('a' <= hex && hex <= 'f') + return 10 + hex - 'a'; + + return -1; +} + +static int decode_hex4(const char *hex) +{ + int digit[4]; + int i; + + /* Convert ASCII hex digit to numeric digit + * Note: this returns an error for invalid hex digits, including + * NULL */ + for (i = 0; i < 4; i++) { + digit[i] = hexdigit2int(hex[i]); + if (digit[i] < 0) { + return -1; + } + } + + return (digit[0] << 12) + + (digit[1] << 8) + + (digit[2] << 4) + + digit[3]; +} + +/* Converts a Unicode codepoint to UTF-8. + * Returns UTF-8 string length, and up to 4 bytes in *utf8 */ +static int codepoint_to_utf8(char *utf8, int codepoint) +{ + /* 0xxxxxxx */ + if (codepoint <= 0x7F) { + utf8[0] = codepoint; + return 1; + } + + /* 110xxxxx 10xxxxxx */ + if (codepoint <= 0x7FF) { + utf8[0] = (codepoint >> 6) | 0xC0; + utf8[1] = (codepoint & 0x3F) | 0x80; + return 2; + } + + /* 1110xxxx 10xxxxxx 10xxxxxx */ + if (codepoint <= 0xFFFF) { + utf8[0] = (codepoint >> 12) | 0xE0; + utf8[1] = ((codepoint >> 6) & 0x3F) | 0x80; + utf8[2] = (codepoint & 0x3F) | 0x80; + return 3; + } + + /* 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */ + if (codepoint <= 0x1FFFFF) { + utf8[0] = (codepoint >> 18) | 0xF0; + utf8[1] = ((codepoint >> 12) & 0x3F) | 0x80; + utf8[2] = ((codepoint >> 6) & 0x3F) | 0x80; + utf8[3] = (codepoint & 0x3F) | 0x80; + return 4; + } + + return 0; +} + + +/* Called when index pointing to beginning of UTF-16 code escape: \uXXXX + * \u is guaranteed to exist, but the remaining hex characters may be + * missing. + * Translate to UTF-8 and append to temporary token string. + * Must advance index to the next character to be processed. + * Returns: 0 success + * -1 error + */ +static int json_append_unicode_escape(json_parse_t *json) +{ + char utf8[4]; /* Surrogate pairs require 4 UTF-8 bytes */ + int codepoint; + int surrogate_low; + int len; + int escape_len = 6; + + /* Fetch UTF-16 code unit */ + codepoint = decode_hex4(json->ptr + 2); + if (codepoint < 0) + return -1; + + /* UTF-16 surrogate pairs take the following 2 byte form: + * 11011 x yyyyyyyyyy + * When x = 0: y is the high 10 bits of the codepoint + * x = 1: y is the low 10 bits of the codepoint + * + * Check for a surrogate pair (high or low) */ + if ((codepoint & 0xF800) == 0xD800) { + /* Error if the 1st surrogate is not high */ + if (codepoint & 0x400) + return -1; + + /* Ensure the next code is a unicode escape */ + if (*(json->ptr + escape_len) != '\\' || + *(json->ptr + escape_len + 1) != 'u') { + return -1; + } + + /* Fetch the next codepoint */ + surrogate_low = decode_hex4(json->ptr + 2 + escape_len); + if (surrogate_low < 0) + return -1; + + /* Error if the 2nd code is not a low surrogate */ + if ((surrogate_low & 0xFC00) != 0xDC00) + return -1; + + /* Calculate Unicode codepoint */ + codepoint = (codepoint & 0x3FF) << 10; + surrogate_low &= 0x3FF; + codepoint = (codepoint | surrogate_low) + 0x10000; + escape_len = 12; + } + + /* Convert codepoint to UTF-8 */ + len = codepoint_to_utf8(utf8, codepoint); + if (!len) + return -1; + + /* Append bytes and advance parse index */ + strbuf_append_mem_unsafe(json->tmp, utf8, len); + json->ptr += escape_len; + + return 0; +} + +static void json_set_token_error(json_token_t *token, json_parse_t *json, + const char *errtype) +{ + token->type = T_ERROR; + token->index = json->ptr - json->data; + token->value.string = errtype; +} + +static void json_next_string_token(json_parse_t *json, json_token_t *token) +{ + char *escape2char = json->cfg->escape2char; + char ch; + + /* Caller must ensure a string is next */ + assert(*json->ptr == '"'); + + /* Skip " */ + json->ptr++; + + /* json->tmp is the temporary strbuf used to accumulate the + * decoded string value. + * json->tmp is sized to handle JSON containing only a string value. + */ + strbuf_reset(json->tmp); + + while ((ch = *json->ptr) != '"') { + if (!ch) { + /* Premature end of the string */ + json_set_token_error(token, json, "unexpected end of string"); + return; + } + + /* Handle escapes */ + if (ch == '\\') { + /* Fetch escape character */ + ch = *(json->ptr + 1); + + /* Translate escape code and append to tmp string */ + ch = escape2char[(unsigned char)ch]; + if (ch == 'u') { + if (json_append_unicode_escape(json) == 0) + continue; + + json_set_token_error(token, json, + "invalid unicode escape code"); + return; + } + if (!ch) { + json_set_token_error(token, json, "invalid escape code"); + return; + } + + /* Skip '\' */ + json->ptr++; + } + /* Append normal character or translated single character + * Unicode escapes are handled above */ + strbuf_append_char_unsafe(json->tmp, ch); + json->ptr++; + } + json->ptr++; /* Eat final quote (") */ + + strbuf_ensure_null(json->tmp); + + token->type = T_STRING; + token->value.string = strbuf_string(json->tmp, &token->string_len); +} + +/* JSON numbers should take the following form: + * -?(0|[1-9]|[1-9][0-9]+)(.[0-9]+)?([eE][-+]?[0-9]+)? + * + * json_next_number_token() uses strtod() which allows other forms: + * - numbers starting with '+' + * - NaN, -NaN, infinity, -infinity + * - hexadecimal numbers + * - numbers with leading zeros + * + * json_is_invalid_number() detects "numbers" which may pass strtod()'s + * error checking, but should not be allowed with strict JSON. + * + * json_is_invalid_number() may pass numbers which cause strtod() + * to generate an error. + */ +static int json_is_invalid_number(json_parse_t *json) +{ + const char *p = json->ptr; + + /* Reject numbers starting with + */ + if (*p == '+') + return 1; + + /* Skip minus sign if it exists */ + if (*p == '-') + p++; + + /* Reject numbers starting with 0x, or leading zeros */ + if (*p == '0') { + int ch2 = *(p + 1); + + if ((ch2 | 0x20) == 'x' || /* Hex */ + ('0' <= ch2 && ch2 <= '9')) /* Leading zero */ + return 1; + + return 0; + } else if (*p <= '9') { + return 0; /* Ordinary number */ + } + + /* Reject inf/nan */ + if (!strncasecmp(p, "inf", 3)) + return 1; + if (!strncasecmp(p, "nan", 3)) + return 1; + + /* Pass all other numbers which may still be invalid, but + * strtod() will catch them. */ + return 0; +} + +static void json_next_number_token(json_parse_t *json, json_token_t *token) +{ + char *endptr; + + token->type = T_NUMBER; + token->value.number = fpconv_strtod(json->ptr, &endptr); + if (json->ptr == endptr) + json_set_token_error(token, json, "invalid number"); + else + json->ptr = endptr; /* Skip the processed number */ + + return; +} + +/* Fills in the token struct. + * T_STRING will return a pointer to the json_parse_t temporary string + * T_ERROR will leave the json->ptr pointer at the error. + */ +static void json_next_token(json_parse_t *json, json_token_t *token) +{ + const json_token_type_t *ch2token = json->cfg->ch2token; + int ch; + + /* Eat whitespace. */ + while (1) { + ch = (unsigned char)*(json->ptr); + token->type = ch2token[ch]; + if (token->type != T_WHITESPACE) + break; + json->ptr++; + } + + /* Store location of new token. Required when throwing errors + * for unexpected tokens (syntax errors). */ + token->index = json->ptr - json->data; + + /* Don't advance the pointer for an error or the end */ + if (token->type == T_ERROR) { + json_set_token_error(token, json, "invalid token"); + return; + } + + if (token->type == T_END) { + return; + } + + /* Found a known single character token, advance index and return */ + if (token->type != T_UNKNOWN) { + json->ptr++; + return; + } + + /* Process characters which triggered T_UNKNOWN + * + * Must use strncmp() to match the front of the JSON string. + * JSON identifier must be lowercase. + * When strict_numbers if disabled, either case is allowed for + * Infinity/NaN (since we are no longer following the spec..) */ + if (ch == '"') { + json_next_string_token(json, token); + return; + } else if (ch == '-' || ('0' <= ch && ch <= '9')) { + if (!json->cfg->decode_invalid_numbers && json_is_invalid_number(json)) { + json_set_token_error(token, json, "invalid number"); + return; + } + json_next_number_token(json, token); + return; + } else if (!strncmp(json->ptr, "true", 4)) { + token->type = T_BOOLEAN; + token->value.boolean = 1; + json->ptr += 4; + return; + } else if (!strncmp(json->ptr, "false", 5)) { + token->type = T_BOOLEAN; + token->value.boolean = 0; + json->ptr += 5; + return; + } else if (!strncmp(json->ptr, "null", 4)) { + token->type = T_NULL; + json->ptr += 4; + return; + } else if (json->cfg->decode_invalid_numbers && + json_is_invalid_number(json)) { + /* When decode_invalid_numbers is enabled, only attempt to process + * numbers we know are invalid JSON (Inf, NaN, hex) + * This is required to generate an appropriate token error, + * otherwise all bad tokens will register as "invalid number" + */ + json_next_number_token(json, token); + return; + } + + /* Token starts with t/f/n but isn't recognised above. */ + json_set_token_error(token, json, "invalid token"); +} + +/* This function does not return. + * DO NOT CALL WITH DYNAMIC MEMORY ALLOCATED. + * The only supported exception is the temporary parser string + * json->tmp struct. + * json and token should exist on the stack somewhere. + * luaL_error() will long_jmp and release the stack */ +static void json_throw_parse_error(lua_State *l, json_parse_t *json, + const char *exp, json_token_t *token) +{ + const char *found; + + strbuf_free(json->tmp); + + if (token->type == T_ERROR) + found = token->value.string; + else + found = json_token_type_name[token->type]; + + /* Note: token->index is 0 based, display starting from 1 */ + luaL_error(l, "Expected %s but found %s at character %d", + exp, found, token->index + 1); +} + +static inline void json_decode_ascend(json_parse_t *json) +{ + json->current_depth--; +} + +static void json_decode_descend(lua_State *l, json_parse_t *json, int slots) +{ + json->current_depth++; + + if (json->current_depth <= json->cfg->decode_max_depth && + lua_checkstack(l, slots)) { + return; + } + + strbuf_free(json->tmp); + luaL_error(l, "Found too many nested data structures (%d) at character %d", + json->current_depth, json->ptr - json->data); +} + +static void json_parse_object_context(lua_State *l, json_parse_t *json) +{ + json_token_t token; + + /* 3 slots required: + * .., table, key, value */ + json_decode_descend(l, json, 3); + + lua_newtable(l); + + json_next_token(json, &token); + + /* Handle empty objects */ + if (token.type == T_OBJ_END) { + nlua_pushref(l, nlua_empty_dict_ref); \ + lua_setmetatable(l, -2); \ + json_decode_ascend(json); + return; + } + + while (1) { + if (token.type != T_STRING) + json_throw_parse_error(l, json, "object key string", &token); + + /* Push key */ + lua_pushlstring(l, token.value.string, token.string_len); + + json_next_token(json, &token); + if (token.type != T_COLON) + json_throw_parse_error(l, json, "colon", &token); + + /* Fetch value */ + json_next_token(json, &token); + json_process_value(l, json, &token); + + /* Set key = value */ + lua_rawset(l, -3); + + json_next_token(json, &token); + + if (token.type == T_OBJ_END) { + json_decode_ascend(json); + return; + } + + if (token.type != T_COMMA) + json_throw_parse_error(l, json, "comma or object end", &token); + + json_next_token(json, &token); + } +} + +/* Handle the array context */ +static void json_parse_array_context(lua_State *l, json_parse_t *json) +{ + json_token_t token; + int i; + + /* 2 slots required: + * .., table, value */ + json_decode_descend(l, json, 2); + + lua_newtable(l); + + /* set array_mt on the table at the top of the stack */ + if (json->cfg->decode_array_with_array_mt) { + lua_pushlightuserdata(l, json_lightudata_mask(&json_array)); + lua_rawget(l, LUA_REGISTRYINDEX); + lua_setmetatable(l, -2); + } + + json_next_token(json, &token); + + /* Handle empty arrays */ + if (token.type == T_ARR_END) { + json_decode_ascend(json); + return; + } + + for (i = 1; ; i++) { + json_process_value(l, json, &token); + lua_rawseti(l, -2, i); /* arr[i] = value */ + + json_next_token(json, &token); + + if (token.type == T_ARR_END) { + json_decode_ascend(json); + return; + } + + if (token.type != T_COMMA) + json_throw_parse_error(l, json, "comma or array end", &token); + + json_next_token(json, &token); + } +} + +/* Handle the "value" context */ +static void json_process_value(lua_State *l, json_parse_t *json, + json_token_t *token) +{ + switch (token->type) { + case T_STRING: + lua_pushlstring(l, token->value.string, token->string_len); + break;; + case T_NUMBER: + lua_pushnumber(l, token->value.number); + break;; + case T_BOOLEAN: + lua_pushboolean(l, token->value.boolean); + break;; + case T_OBJ_BEGIN: + json_parse_object_context(l, json); + break;; + case T_ARR_BEGIN: + json_parse_array_context(l, json); + break;; + case T_NULL: + nlua_pushref(l, nlua_nil_ref); + break;; + default: + json_throw_parse_error(l, json, "value", token); + } +} + +static int json_decode(lua_State *l) +{ + json_parse_t json; + json_token_t token; + size_t json_len; + + luaL_argcheck(l, lua_gettop(l) == 1, 1, "expected 1 argument"); + + json.cfg = json_fetch_config(l); + json.data = luaL_checklstring(l, 1, &json_len); + json.current_depth = 0; + json.ptr = json.data; + + /* Detect Unicode other than UTF-8 (see RFC 4627, Sec 3) + * + * CJSON can support any simple data type, hence only the first + * character is guaranteed to be ASCII (at worst: '"'). This is + * still enough to detect whether the wrong encoding is in use. */ + if (json_len >= 2 && (!json.data[0] || !json.data[1])) + luaL_error(l, "JSON parser does not support UTF-16 or UTF-32"); + + /* Ensure the temporary buffer can hold the entire string. + * This means we no longer need to do length checks since the decoded + * string must be smaller than the entire json string */ + json.tmp = strbuf_new(json_len); + + json_next_token(&json, &token); + json_process_value(l, &json, &token); + + /* Ensure there is no more input left */ + json_next_token(&json, &token); + + if (token.type != T_END) + json_throw_parse_error(l, &json, "the end", &token); + + strbuf_free(json.tmp); + + return 1; +} + +/* ===== INITIALISATION ===== */ + +#if !defined(LUA_VERSION_NUM) || LUA_VERSION_NUM < 502 +/* Compatibility for Lua 5.1 and older LuaJIT. + * + * compat_luaL_setfuncs() is used to create a module table where the functions + * have json_config_t as their first upvalue. Code borrowed from Lua 5.2 + * source's luaL_setfuncs(). + */ +static void compat_luaL_setfuncs(lua_State *l, const luaL_Reg *reg, int nup) +{ + int i; + + luaL_checkstack(l, nup, "too many upvalues"); + for (; reg->name != NULL; reg++) { /* fill the table with given functions */ + for (i = 0; i < nup; i++) /* copy upvalues to the top */ + lua_pushvalue(l, -nup); + lua_pushcclosure(l, reg->func, nup); /* closure with those upvalues */ + lua_setfield(l, -(nup + 2), reg->name); + } + lua_pop(l, nup); /* remove upvalues */ +} +#else +#define compat_luaL_setfuncs(L, reg, nup) luaL_setfuncs(L, reg, nup) +#endif + +/* Call target function in protected mode with all supplied args. + * Assumes target function only returns a single non-nil value. + * Convert and return thrown errors as: nil, "error message" */ +static int json_protect_conversion(lua_State *l) +{ + int err; + + /* Deliberately throw an error for invalid arguments */ + luaL_argcheck(l, lua_gettop(l) == 1, 1, "expected 1 argument"); + + /* pcall() the function stored as upvalue(1) */ + lua_pushvalue(l, lua_upvalueindex(1)); + lua_insert(l, 1); + err = lua_pcall(l, 1, 1, 0); + if (!err) + return 1; + + if (err == LUA_ERRRUN) { + lua_pushnil(l); + lua_insert(l, -2); + return 2; + } + + /* Since we are not using a custom error handler, the only remaining + * errors are memory related */ + return luaL_error(l, "Memory allocation error in CJSON protected call"); +} + +/* Return cjson module table */ +int lua_cjson_new(lua_State *l) +{ + luaL_Reg reg[] = { + { "encode", json_encode }, + { "decode", json_decode }, + { "encode_empty_table_as_object", json_cfg_encode_empty_table_as_object }, + { "decode_array_with_array_mt", json_cfg_decode_array_with_array_mt }, + { "encode_sparse_array", json_cfg_encode_sparse_array }, + { "encode_max_depth", json_cfg_encode_max_depth }, + { "decode_max_depth", json_cfg_decode_max_depth }, + { "encode_number_precision", json_cfg_encode_number_precision }, + { "encode_keep_buffer", json_cfg_encode_keep_buffer }, + { "encode_invalid_numbers", json_cfg_encode_invalid_numbers }, + { "decode_invalid_numbers", json_cfg_decode_invalid_numbers }, + { "encode_escape_forward_slash", json_cfg_encode_escape_forward_slash }, + { "new", lua_cjson_new }, + { NULL, NULL } + }; + + /* Initialise number conversions */ + fpconv_init(); + + /* Test if array metatables are in registry */ + lua_pushlightuserdata(l, json_lightudata_mask(&json_empty_array)); + lua_rawget(l, LUA_REGISTRYINDEX); + if (lua_isnil(l, -1)) { + /* Create array metatables. + * + * If multiple calls to lua_cjson_new() are made, + * this prevents overriding the tables at the given + * registry's index with a new one. + */ + lua_pop(l, 1); + + /* empty_array_mt */ + lua_pushlightuserdata(l, json_lightudata_mask(&json_empty_array)); + lua_newtable(l); + lua_rawset(l, LUA_REGISTRYINDEX); + + /* array_mt */ + lua_pushlightuserdata(l, json_lightudata_mask(&json_array)); + lua_newtable(l); + lua_rawset(l, LUA_REGISTRYINDEX); + } + + /* cjson module table */ + lua_newtable(l); + + /* Register functions with config data as upvalue */ + json_create_config(l); + compat_luaL_setfuncs(l, reg, 1); + + /* Set cjson.null */ + nlua_pushref(l, nlua_nil_ref); + lua_setfield(l, -2, "null"); + + /* Set cjson.empty_array_mt */ + lua_pushlightuserdata(l, json_lightudata_mask(&json_empty_array)); + lua_rawget(l, LUA_REGISTRYINDEX); + lua_setfield(l, -2, "empty_array_mt"); + + /* Set cjson.array_mt */ + lua_pushlightuserdata(l, json_lightudata_mask(&json_array)); + lua_rawget(l, LUA_REGISTRYINDEX); + lua_setfield(l, -2, "array_mt"); + + /* Set cjson.empty_array */ + lua_pushlightuserdata(l, json_lightudata_mask(&json_array)); + lua_setfield(l, -2, "empty_array"); + + /* Set module name / version fields */ + lua_pushliteral(l, CJSON_MODNAME); + lua_setfield(l, -2, "_NAME"); + lua_pushliteral(l, CJSON_VERSION); + lua_setfield(l, -2, "_VERSION"); + + return 1; +} + +/* Return cjson.safe module table */ +static int lua_cjson_safe_new(lua_State *l) +{ + const char *func[] = { "decode", "encode", NULL }; + int i; + + lua_cjson_new(l); + + /* Fix new() method */ + lua_pushcfunction(l, lua_cjson_safe_new); + lua_setfield(l, -2, "new"); + + for (i = 0; func[i]; i++) { + lua_getfield(l, -1, func[i]); + lua_pushcclosure(l, json_protect_conversion, 1); + lua_setfield(l, -2, func[i]); + } + + return 1; +} + +int luaopen_cjson(lua_State *l) +{ + lua_cjson_new(l); + +#ifdef ENABLE_CJSON_GLOBAL + /* Register a global "cjson" table. */ + lua_pushvalue(l, -1); + lua_setglobal(l, CJSON_MODNAME); +#endif + + /* Return cjson table */ + return 1; +} + +int luaopen_cjson_safe(lua_State *l) +{ + lua_cjson_safe_new(l); + + /* Return cjson.safe table */ + return 1; +} + +/* vi:ai et sw=4 ts=4: + */ diff --git a/src/cjson/lua_cjson.h b/src/cjson/lua_cjson.h new file mode 100644 index 0000000000..3f70b679be --- /dev/null +++ b/src/cjson/lua_cjson.h @@ -0,0 +1,10 @@ +#ifndef CJSON_LUACJSON_H +#define CJSON_LUACJSON_H + +#include "lua.h" + +int lua_cjson_new(lua_State *l); +int luaopen_cjson(lua_State *l); +int luaopen_cjson_safe(lua_State *l); + +#endif // CJSON_LUACJSON_H diff --git a/src/cjson/strbuf.c b/src/cjson/strbuf.c new file mode 100644 index 0000000000..f0f7f4b9a3 --- /dev/null +++ b/src/cjson/strbuf.c @@ -0,0 +1,251 @@ +/* strbuf - String buffer routines + * + * Copyright (c) 2010-2012 Mark Pulford <mark@kyne.com.au> + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +#include <stdio.h> +#include <stdlib.h> +#include <stdarg.h> +#include <string.h> + +#include "strbuf.h" + +static void die(const char *fmt, ...) +{ + va_list arg; + + va_start(arg, fmt); + vfprintf(stderr, fmt, arg); + va_end(arg); + fprintf(stderr, "\n"); + + exit(-1); +} + +void strbuf_init(strbuf_t *s, int len) +{ + int size; + + if (len <= 0) + size = STRBUF_DEFAULT_SIZE; + else + size = len + 1; /* \0 terminator */ + + s->buf = NULL; + s->size = size; + s->length = 0; + s->increment = STRBUF_DEFAULT_INCREMENT; + s->dynamic = 0; + s->reallocs = 0; + s->debug = 0; + + s->buf = malloc(size); + if (!s->buf) + die("Out of memory"); + + strbuf_ensure_null(s); +} + +strbuf_t *strbuf_new(int len) +{ + strbuf_t *s; + + s = malloc(sizeof(strbuf_t)); + if (!s) + die("Out of memory"); + + strbuf_init(s, len); + + /* Dynamic strbuf allocation / deallocation */ + s->dynamic = 1; + + return s; +} + +void strbuf_set_increment(strbuf_t *s, int increment) +{ + /* Increment > 0: Linear buffer growth rate + * Increment < -1: Exponential buffer growth rate */ + if (increment == 0 || increment == -1) + die("BUG: Invalid string increment"); + + s->increment = increment; +} + +static inline void debug_stats(strbuf_t *s) +{ + if (s->debug) { + fprintf(stderr, "strbuf(%lx) reallocs: %d, length: %d, size: %d\n", + (long)s, s->reallocs, s->length, s->size); + } +} + +/* If strbuf_t has not been dynamically allocated, strbuf_free() can + * be called any number of times strbuf_init() */ +void strbuf_free(strbuf_t *s) +{ + debug_stats(s); + + if (s->buf) { + free(s->buf); + s->buf = NULL; + } + if (s->dynamic) + free(s); +} + +char *strbuf_free_to_string(strbuf_t *s, int *len) +{ + char *buf; + + debug_stats(s); + + strbuf_ensure_null(s); + + buf = s->buf; + if (len) + *len = s->length; + + if (s->dynamic) + free(s); + + return buf; +} + +static int calculate_new_size(strbuf_t *s, int len) +{ + int reqsize, newsize; + + if (len <= 0) + die("BUG: Invalid strbuf length requested"); + + /* Ensure there is room for optional NULL termination */ + reqsize = len + 1; + + /* If the user has requested to shrink the buffer, do it exactly */ + if (s->size > reqsize) + return reqsize; + + newsize = s->size; + if (s->increment < 0) { + /* Exponential sizing */ + while (newsize < reqsize) + newsize *= -s->increment; + } else { + /* Linear sizing */ + newsize = ((newsize + s->increment - 1) / s->increment) * s->increment; + } + + return newsize; +} + + +/* Ensure strbuf can handle a string length bytes long (ignoring NULL + * optional termination). */ +void strbuf_resize(strbuf_t *s, int len) +{ + int newsize; + + newsize = calculate_new_size(s, len); + + if (s->debug > 1) { + fprintf(stderr, "strbuf(%lx) resize: %d => %d\n", + (long)s, s->size, newsize); + } + + s->size = newsize; + s->buf = realloc(s->buf, s->size); + if (!s->buf) + die("Out of memory"); + s->reallocs++; +} + +void strbuf_append_string(strbuf_t *s, const char *str) +{ + int space, i; + + space = strbuf_empty_length(s); + + for (i = 0; str[i]; i++) { + if (space < 1) { + strbuf_resize(s, s->length + 1); + space = strbuf_empty_length(s); + } + + s->buf[s->length] = str[i]; + s->length++; + space--; + } +} + +/* strbuf_append_fmt() should only be used when an upper bound + * is known for the output string. */ +void strbuf_append_fmt(strbuf_t *s, int len, const char *fmt, ...) +{ + va_list arg; + int fmt_len; + + strbuf_ensure_empty_length(s, len); + + va_start(arg, fmt); + fmt_len = vsnprintf(s->buf + s->length, len, fmt, arg); + va_end(arg); + + if (fmt_len < 0) + die("BUG: Unable to convert number"); /* This should never happen.. */ + + s->length += fmt_len; +} + +/* strbuf_append_fmt_retry() can be used when the there is no known + * upper bound for the output string. */ +void strbuf_append_fmt_retry(strbuf_t *s, const char *fmt, ...) +{ + va_list arg; + int fmt_len, try; + int empty_len; + + /* If the first attempt to append fails, resize the buffer appropriately + * and try again */ + for (try = 0; ; try++) { + va_start(arg, fmt); + /* Append the new formatted string */ + /* fmt_len is the length of the string required, excluding the + * trailing NULL */ + empty_len = strbuf_empty_length(s); + /* Add 1 since there is also space to store the terminating NULL. */ + fmt_len = vsnprintf(s->buf + s->length, empty_len + 1, fmt, arg); + va_end(arg); + + if (fmt_len <= empty_len) + break; /* SUCCESS */ + if (try > 0) + die("BUG: length of formatted string changed"); + + strbuf_resize(s, s->length + fmt_len); + } + + s->length += fmt_len; +} + +/* vi:ai et sw=4 ts=4: + */ diff --git a/src/cjson/strbuf.h b/src/cjson/strbuf.h new file mode 100644 index 0000000000..5df0b7bea3 --- /dev/null +++ b/src/cjson/strbuf.h @@ -0,0 +1,159 @@ +/* strbuf - String buffer routines + * + * Copyright (c) 2010-2012 Mark Pulford <mark@kyne.com.au> + * + * Permission is hereby granted, free of charge, to any person obtaining + * a copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sublicense, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ + +#include <stdlib.h> +#include <stdarg.h> + +/* Workaround for MSVC */ +#ifdef _MSC_VER +#define inline __inline +#endif + +/* Size: Total bytes allocated to *buf + * Length: String length, excluding optional NULL terminator. + * Increment: Allocation increments when resizing the string buffer. + * Dynamic: True if created via strbuf_new() + */ + +typedef struct { + char *buf; + int size; + int length; + int increment; + int dynamic; + int reallocs; + int debug; +} strbuf_t; + +#ifndef STRBUF_DEFAULT_SIZE +#define STRBUF_DEFAULT_SIZE 1023 +#endif +#ifndef STRBUF_DEFAULT_INCREMENT +#define STRBUF_DEFAULT_INCREMENT -2 +#endif + +/* Initialise */ +extern strbuf_t *strbuf_new(int len); +extern void strbuf_init(strbuf_t *s, int len); +extern void strbuf_set_increment(strbuf_t *s, int increment); + +/* Release */ +extern void strbuf_free(strbuf_t *s); +extern char *strbuf_free_to_string(strbuf_t *s, int *len); + +/* Management */ +extern void strbuf_resize(strbuf_t *s, int len); +static int strbuf_empty_length(strbuf_t *s); +static int strbuf_length(strbuf_t *s); +static char *strbuf_string(strbuf_t *s, int *len); +static void strbuf_ensure_empty_length(strbuf_t *s, int len); +static char *strbuf_empty_ptr(strbuf_t *s); +static void strbuf_extend_length(strbuf_t *s, int len); + +/* Update */ +extern void strbuf_append_fmt(strbuf_t *s, int len, const char *fmt, ...); +extern void strbuf_append_fmt_retry(strbuf_t *s, const char *format, ...); +static void strbuf_append_mem(strbuf_t *s, const char *c, int len); +extern void strbuf_append_string(strbuf_t *s, const char *str); +static void strbuf_append_char(strbuf_t *s, const char c); +static void strbuf_ensure_null(strbuf_t *s); + +/* Reset string for before use */ +static inline void strbuf_reset(strbuf_t *s) +{ + s->length = 0; +} + +static inline int strbuf_allocated(strbuf_t *s) +{ + return s->buf != NULL; +} + +/* Return bytes remaining in the string buffer + * Ensure there is space for a NULL terminator. */ +static inline int strbuf_empty_length(strbuf_t *s) +{ + return s->size - s->length - 1; +} + +static inline void strbuf_ensure_empty_length(strbuf_t *s, int len) +{ + if (len > strbuf_empty_length(s)) + strbuf_resize(s, s->length + len); +} + +static inline char *strbuf_empty_ptr(strbuf_t *s) +{ + return s->buf + s->length; +} + +static inline void strbuf_extend_length(strbuf_t *s, int len) +{ + s->length += len; +} + +static inline int strbuf_length(strbuf_t *s) +{ + return s->length; +} + +static inline void strbuf_append_char(strbuf_t *s, const char c) +{ + strbuf_ensure_empty_length(s, 1); + s->buf[s->length++] = c; +} + +static inline void strbuf_append_char_unsafe(strbuf_t *s, const char c) +{ + s->buf[s->length++] = c; +} + +static inline void strbuf_append_mem(strbuf_t *s, const char *c, int len) +{ + strbuf_ensure_empty_length(s, len); + memcpy(s->buf + s->length, c, len); + s->length += len; +} + +static inline void strbuf_append_mem_unsafe(strbuf_t *s, const char *c, int len) +{ + memcpy(s->buf + s->length, c, len); + s->length += len; +} + +static inline void strbuf_ensure_null(strbuf_t *s) +{ + s->buf[s->length] = 0; +} + +static inline char *strbuf_string(strbuf_t *s, int *len) +{ + if (len) + *len = s->length; + + return s->buf; +} + +/* vi:ai et sw=4 ts=4: + */ diff --git a/src/nvim/CMakeLists.txt b/src/nvim/CMakeLists.txt index 331ab16dd7..e35e7ce2d4 100644 --- a/src/nvim/CMakeLists.txt +++ b/src/nvim/CMakeLists.txt @@ -87,8 +87,8 @@ file(MAKE_DIRECTORY ${LINT_SUPPRESSES_ROOT}/src) file(GLOB NVIM_SOURCES *.c) file(GLOB NVIM_HEADERS *.h) -file(GLOB EXTERNAL_SOURCES ../xdiff/*.c ../mpack/*.c) -file(GLOB EXTERNAL_HEADERS ../xdiff/*.h ../mpack/*.h) +file(GLOB EXTERNAL_SOURCES ../xdiff/*.c ../mpack/*.c ../cjson/*.c) +file(GLOB EXTERNAL_HEADERS ../xdiff/*.h ../mpack/*.h ../cjson/*.h) foreach(subdir os @@ -171,7 +171,7 @@ foreach(sfile ${CONV_SOURCES}) message(FATAL_ERROR "${sfile} doesn't exist (it was added to CONV_SOURCES)") endif() endforeach() -# xdiff, mpack: inlined external project, we don't maintain it. #9306 +# xdiff, mpack, lua-cjson: inlined external project, we don't maintain it. #9306 list(APPEND CONV_SOURCES ${EXTERNAL_SOURCES}) if(NOT MSVC) diff --git a/src/nvim/lua/executor.c b/src/nvim/lua/executor.c index 8c7dc90111..9333d781cd 100644 --- a/src/nvim/lua/executor.c +++ b/src/nvim/lua/executor.c @@ -40,6 +40,7 @@ #include "nvim/undo.h" #include "nvim/version.h" #include "nvim/vim.h" +#include "cjson/lua_cjson.h" static int in_fast_callback = 0; @@ -531,6 +532,9 @@ static int nlua_state_init(lua_State *const lstate) FUNC_ATTR_NONNULL_ALL lua_pushcfunction(lstate, &nlua_xdl_diff); lua_setfield(lstate, -2, "diff"); + lua_cjson_new(lstate); + lua_setfield(lstate, -2, "json"); + lua_setglobal(lstate, "vim"); { |