make ljson more strict

- add new context ARRAY_SINGLE for [singleton] without commas
- add new context ARRAY_END for [] without commas
- add new context OBJECT_END for {} without commas
- call Parse() once before looping for array or object so that the empty
  case is handled easily ([,] should now be wrong)
- narrow use of OBJECT_KEY -- now it is only used to confirm that the
  key is a string
- rename ARRAY_VAL to AFTER_VALUE -- a comma will only be present after
  a value ie either [1,2] or {"a":1,"b":2}, so the context for comma is
  always after value
- reset context when AFTER_VALUE and OBJECT_VAL to avoid extra ,:
- use bitfields instead of equality to check the context value
- TOP_LEVEL is used to define the maximum depth, and now the depth is
  compared to TOP_LEVEL for the base case
This commit is contained in:
ahgamut 2022-07-13 13:06:41 +05:30
parent 30cc2c8dc1
commit f3a76cde5d

View file

@ -31,10 +31,14 @@
#include "third_party/lua/lua.h" #include "third_party/lua/lua.h"
#include "tool/net/ljson.h" #include "tool/net/ljson.h"
#define TOP_LEVEL 0 #define AFTER_VALUE 0x01u
#define ARRAY_VAL 1 #define ARRAY_SINGLE 0x02u
#define OBJECT_KEY 2 #define ARRAY_END 0x04u
#define OBJECT_VAL 3 #define OBJECT_KEY 0x10u
#define OBJECT_VAL 0x20u
#define OBJECT_END 0x40u
#define TOP_LEVEL 64
static struct DecodeJson Parse(struct lua_State *L, const char *p, static struct DecodeJson Parse(struct lua_State *L, const char *p,
const char *e, int context, int depth) { const char *e, int context, int depth) {
@ -58,7 +62,8 @@ static struct DecodeJson Parse(struct lua_State *L, const char *p,
break; break;
case ',': // present in list and object case ',': // present in list and object
if (context == ARRAY_VAL || context == OBJECT_KEY) { if (0 != (context & AFTER_VALUE)) {
context = 0;
a = p; a = p;
break; break;
} else { } else {
@ -66,7 +71,8 @@ static struct DecodeJson Parse(struct lua_State *L, const char *p,
} }
case ':': // present only in object after key case ':': // present only in object after key
if (context == OBJECT_VAL) { if (0 != (context & OBJECT_VAL)) {
context = 0;
a = p; a = p;
break; break;
} else { } else {
@ -74,7 +80,7 @@ static struct DecodeJson Parse(struct lua_State *L, const char *p,
} }
case 'n': // null case 'n': // null
if (UNLIKELY(context == OBJECT_KEY)) goto BadObjectKey; if (UNLIKELY(0 != (context & OBJECT_KEY))) goto BadObjectKey;
if (p + 3 <= e && READ32LE(p - 1) == READ32LE("null")) { if (p + 3 <= e && READ32LE(p - 1) == READ32LE("null")) {
lua_pushnil(L); lua_pushnil(L);
return (struct DecodeJson){1, p + 3}; return (struct DecodeJson){1, p + 3};
@ -83,7 +89,7 @@ static struct DecodeJson Parse(struct lua_State *L, const char *p,
} }
case 'f': // false case 'f': // false
if (UNLIKELY(context == OBJECT_KEY)) goto BadObjectKey; if (UNLIKELY(0 != (context & OBJECT_KEY))) goto BadObjectKey;
if (p + 4 <= e && READ32LE(p) == READ32LE("alse")) { if (p + 4 <= e && READ32LE(p) == READ32LE("alse")) {
lua_pushboolean(L, false); lua_pushboolean(L, false);
return (struct DecodeJson){1, p + 4}; return (struct DecodeJson){1, p + 4};
@ -92,7 +98,7 @@ static struct DecodeJson Parse(struct lua_State *L, const char *p,
} }
case 't': // true case 't': // true
if (UNLIKELY(context == OBJECT_KEY)) goto BadObjectKey; if (UNLIKELY(0 != (context & OBJECT_KEY))) goto BadObjectKey;
if (p + 3 <= e && READ32LE(p - 1) == READ32LE("true")) { if (p + 3 <= e && READ32LE(p - 1) == READ32LE("true")) {
lua_pushboolean(L, true); lua_pushboolean(L, true);
return (struct DecodeJson){1, p + 3}; return (struct DecodeJson){1, p + 3};
@ -104,7 +110,7 @@ static struct DecodeJson Parse(struct lua_State *L, const char *p,
return (struct DecodeJson){-1, "object key must be string"}; return (struct DecodeJson){-1, "object key must be string"};
case '-': // negative case '-': // negative
if (UNLIKELY(context == OBJECT_KEY)) goto BadObjectKey; if (UNLIKELY(0 != (context & OBJECT_KEY))) goto BadObjectKey;
if (p < e && isdigit(*p)) { if (p < e && isdigit(*p)) {
d = -1; d = -1;
break; break;
@ -113,7 +119,7 @@ static struct DecodeJson Parse(struct lua_State *L, const char *p,
} }
case '0': // zero or number case '0': // zero or number
if (UNLIKELY(context == OBJECT_KEY)) goto BadObjectKey; if (UNLIKELY(0 != (context & OBJECT_KEY))) goto BadObjectKey;
if (p < e) { if (p < e) {
if ((*p == '.' || *p == 'e' || *p == 'E')) { if ((*p == '.' || *p == 'e' || *p == 'E')) {
goto UseDubble; goto UseDubble;
@ -125,7 +131,7 @@ static struct DecodeJson Parse(struct lua_State *L, const char *p,
return (struct DecodeJson){1, p}; return (struct DecodeJson){1, p};
case '1' ... '9': // integer case '1' ... '9': // integer
if (UNLIKELY(context == OBJECT_KEY)) goto BadObjectKey; if (UNLIKELY(0 != (context & OBJECT_KEY))) goto BadObjectKey;
for (x = (c - '0') * d; p < e; ++p) { for (x = (c - '0') * d; p < e; ++p) {
c = *p & 255; c = *p & 255;
if (isdigit(c)) { if (isdigit(c)) {
@ -148,11 +154,11 @@ static struct DecodeJson Parse(struct lua_State *L, const char *p,
return (struct DecodeJson){1, a + c}; return (struct DecodeJson){1, a + c};
case '[': // Array case '[': // Array
if (UNLIKELY(context == OBJECT_KEY)) goto BadObjectKey; if (UNLIKELY(0 != (context & OBJECT_KEY))) goto BadObjectKey;
lua_newtable(L); lua_newtable(L);
i = 0; i = 0;
r = Parse(L, p, e, ARRAY_SINGLE | ARRAY_END, depth - 1);
for (;;) { for (;;) {
r = Parse(L, p, e, ARRAY_VAL, depth - 1);
if (UNLIKELY(r.rc == -1)) { if (UNLIKELY(r.rc == -1)) {
lua_pop(L, 1); lua_pop(L, 1);
return r; return r;
@ -162,6 +168,7 @@ static struct DecodeJson Parse(struct lua_State *L, const char *p,
break; break;
} }
lua_rawseti(L, -2, i++ + 1); lua_rawseti(L, -2, i++ + 1);
r = Parse(L, p, e, AFTER_VALUE | ARRAY_END, depth - 1);
} }
if (!i) { if (!i) {
// we need this kludge so `[]` won't round-trip as `{}` // we need this kludge so `[]` won't round-trip as `{}`
@ -171,24 +178,24 @@ static struct DecodeJson Parse(struct lua_State *L, const char *p,
return (struct DecodeJson){1, p}; return (struct DecodeJson){1, p};
case ']': case ']':
if (context == ARRAY_VAL) { if (0 != (context & ARRAY_END)) {
return (struct DecodeJson){0, p}; return (struct DecodeJson){0, p};
} else { } else {
return (struct DecodeJson){-1, "unexpected ']'"}; return (struct DecodeJson){-1, "unexpected ']'"};
} }
case '}': case '}':
if (context == OBJECT_KEY) { if (0 != (context & OBJECT_END)) {
return (struct DecodeJson){0, p}; return (struct DecodeJson){0, p};
} else { } else {
return (struct DecodeJson){-1, "unexpected '}'"}; return (struct DecodeJson){-1, "unexpected '}'"};
} }
case '{': // Object case '{': // Object
if (UNLIKELY(context == OBJECT_KEY)) goto BadObjectKey; if (UNLIKELY(0 != (context & OBJECT_KEY))) goto BadObjectKey;
lua_newtable(L); lua_newtable(L);
r = Parse(L, p, e, OBJECT_KEY | OBJECT_END, depth - 1);
for (;;) { for (;;) {
r = Parse(L, p, e, OBJECT_KEY, depth - 1);
if (r.rc == -1) { if (r.rc == -1) {
lua_pop(L, 1); lua_pop(L, 1);
return r; return r;
@ -208,6 +215,7 @@ static struct DecodeJson Parse(struct lua_State *L, const char *p,
} }
p = r.p; p = r.p;
lua_settable(L, -3); lua_settable(L, -3);
r = Parse(L, p, e, OBJECT_KEY | AFTER_VALUE | OBJECT_END, depth - 1);
} }
return (struct DecodeJson){1, p}; return (struct DecodeJson){1, p};
@ -355,7 +363,7 @@ static struct DecodeJson Parse(struct lua_State *L, const char *p,
return (struct DecodeJson){-1, "illegal character"}; return (struct DecodeJson){-1, "illegal character"};
} }
} }
if (UNLIKELY(context == TOP_LEVEL)) { if (UNLIKELY(depth == TOP_LEVEL)) {
return (struct DecodeJson){0, 0}; return (struct DecodeJson){0, 0};
} else { } else {
return (struct DecodeJson){-1, "unexpected eof"}; return (struct DecodeJson){-1, "unexpected eof"};
@ -386,10 +394,10 @@ static struct DecodeJson Parse(struct lua_State *L, const char *p,
* @return r.p is string describing error if `rc < 0` * @return r.p is string describing error if `rc < 0`
*/ */
struct DecodeJson DecodeJson(struct lua_State *L, const char *p, size_t n) { struct DecodeJson DecodeJson(struct lua_State *L, const char *p, size_t n) {
int depth = 64; int depth = TOP_LEVEL;
if (n == -1) n = p ? strlen(p) : 0; if (n == -1) n = p ? strlen(p) : 0;
if (lua_checkstack(L, depth * 4)) { if (lua_checkstack(L, depth * 4)) {
return Parse(L, p, p + n, TOP_LEVEL, depth); return Parse(L, p, p + n, 0, depth);
} else { } else {
return (struct DecodeJson){-1, "can't set stack depth"}; return (struct DecodeJson){-1, "can't set stack depth"};
} }