Make redbean serialization deterministic

This commit is contained in:
Justine Tunney 2022-07-09 04:09:51 -07:00
parent 85aecbda67
commit c9e68b0ebc
15 changed files with 452 additions and 150 deletions

View file

@ -14,7 +14,7 @@ int LuaEncodeUrl(lua_State *);
int LuaParseUrl(lua_State *);
int LuaPushHeader(lua_State *, struct HttpMessage *, char *, int);
int LuaPushHeaders(lua_State *, struct HttpMessage *, const char *);
void EscapeLuaString(char *, size_t, char **);
int EscapeLuaString(char *, size_t, char **);
void LuaPrintStack(lua_State *);
void LuaPushLatin1(lua_State *, const char *, size_t);
void LuaPushUrlParams(lua_State *, struct UrlParams *);

View file

@ -17,25 +17,32 @@
PERFORMANCE OF THIS SOFTWARE.
*/
#include "libc/bits/bits.h"
#include "libc/log/rop.h"
#include "libc/stdio/append.internal.h"
#include "libc/str/str.h"
#include "third_party/lua/cosmo.h"
#include "third_party/lua/lua.h"
void EscapeLuaString(char *s, size_t len, char **buf) {
appendw(buf, '"');
for (size_t i = 0; i < len; i++) {
int EscapeLuaString(char *s, size_t len, char **buf) {
int rc;
size_t i;
RETURN_ON_ERROR(appendw(buf, '"'));
for (i = 0; i < len; i++) {
if (' ' <= s[i] && s[i] <= 0x7e) {
appendw(buf, s[i]);
RETURN_ON_ERROR(appendw(buf, s[i]));
} else if (s[i] == '\n') {
appendw(buf, '\\' | 'n' << 8);
RETURN_ON_ERROR(appendw(buf, '\\' | 'n' << 8));
} else if (s[i] == '\\' || s[i] == '\'' || s[i] == '\"') {
appendw(buf, '\\' | s[i] << 8);
RETURN_ON_ERROR(appendw(buf, '\\' | s[i] << 8));
} else {
appendw(buf, '\\' | 'x' << 010 |
"0123456789abcdef"[(s[i] & 0xF0) >> 4] << 020 |
"0123456789abcdef"[(s[i] & 0x0F) >> 0] << 030);
RETURN_ON_ERROR(
appendw(buf, '\\' | 'x' << 010 |
"0123456789abcdef"[(s[i] & 0xF0) >> 4] << 020 |
"0123456789abcdef"[(s[i] & 0x0F) >> 0] << 030));
}
}
appendw(buf, '"');
RETURN_ON_ERROR(appendw(buf, '"'));
return 0;
OnError:
return -1;
}

View file

@ -19,9 +19,11 @@
#include "libc/assert.h"
#include "libc/bits/bits.h"
#include "libc/fmt/itoa.h"
#include "libc/log/rop.h"
#include "libc/mem/mem.h"
#include "libc/runtime/gc.internal.h"
#include "libc/stdio/append.internal.h"
#include "libc/stdio/strlist.internal.h"
#include "net/http/escape.h"
#include "third_party/lua/cosmo.h"
#include "third_party/lua/lauxlib.h"
@ -32,34 +34,37 @@ static int LuaEncodeJsonDataImpl(lua_State *L, char **buf, int level,
char *numformat, int idx,
struct LuaVisited *visited) {
char *s;
int sli, rc;
bool isarray;
size_t tbllen, i, z;
struct StrList sl = {0};
char ibuf[21], fmt[] = "%.14g";
if (level > 0) {
switch (lua_type(L, idx)) {
case LUA_TNIL:
appendw(buf, READ32LE("null"));
RETURN_ON_ERROR(appendw(buf, READ32LE("null")));
return 0;
case LUA_TBOOLEAN:
appendw(buf, lua_toboolean(L, idx) ? READ32LE("true")
: READ64LE("false\0\0"));
RETURN_ON_ERROR(appendw(buf, lua_toboolean(L, idx)
? READ32LE("true")
: READ64LE("false\0\0")));
return 0;
case LUA_TSTRING:
s = lua_tolstring(L, idx, &z);
s = EscapeJsStringLiteral(s, z, &z);
appendw(buf, '"');
appendd(buf, s, z);
appendw(buf, '"');
if (!(s = EscapeJsStringLiteral(s, z, &z))) goto OnError;
RETURN_ON_ERROR(appendw(buf, '"'));
RETURN_ON_ERROR(appendd(buf, s, z));
RETURN_ON_ERROR(appendw(buf, '"'));
free(s);
return 0;
case LUA_TNUMBER:
if (lua_isinteger(L, idx)) {
appendd(buf, ibuf,
FormatInt64(ibuf, luaL_checkinteger(L, idx)) - ibuf);
RETURN_ON_ERROR(appendd(
buf, ibuf, FormatInt64(ibuf, luaL_checkinteger(L, idx)) - ibuf));
} else {
// TODO(jart): replace this api
while (*numformat == '%' || *numformat == '.' ||
@ -73,16 +78,16 @@ static int LuaEncodeJsonDataImpl(lua_State *L, char **buf, int level,
fmt[4] = *numformat;
break;
default:
free(visited->p);
luaL_error(L, "numformat string not allowed");
unreachable;
// prevent format string hacking
goto OnError;
}
appendf(buf, fmt, lua_tonumber(L, idx));
RETURN_ON_ERROR(appendf(buf, fmt, lua_tonumber(L, idx)));
}
return 0;
case LUA_TTABLE:
if (LuaPushVisit(visited, lua_topointer(L, idx))) {
RETURN_ON_ERROR(rc = LuaPushVisit(visited, lua_topointer(L, idx)));
if (!rc) {
lua_pushvalue(L, idx); // table ref
tbllen = lua_rawlen(L, -1);
// encode tables with numeric indices and empty tables as arrays
@ -90,12 +95,12 @@ static int LuaEncodeJsonDataImpl(lua_State *L, char **buf, int level,
tbllen > 0 || // integer keys present
(lua_pushnil(L), !lua_next(L, -2)) || // no non-integer keys
(lua_pop(L, 2), false); // pop key/value pushed by lua_next
appendw(buf, isarray ? '[' : '{');
if (isarray) {
for (i = 1; i <= tbllen; i++) {
if (i > 1) appendw(buf, ',');
RETURN_ON_ERROR(sli = AppendStrList(&sl));
lua_rawgeti(L, -1, i); // table/-2, value/-1
LuaEncodeJsonDataImpl(L, buf, level - 1, numformat, -1, visited);
RETURN_ON_ERROR(LuaEncodeJsonDataImpl(L, &sl.p[sli], level - 1,
numformat, -1, visited));
lua_pop(L, 1);
}
} else {
@ -103,45 +108,58 @@ static int LuaEncodeJsonDataImpl(lua_State *L, char **buf, int level,
lua_pushnil(L); // push the first key
while (lua_next(L, -2)) {
if (lua_type(L, -2) != LUA_TSTRING) {
free(visited->p);
luaL_error(L, "json tables must be arrays or use string keys");
unreachable;
// json tables must be arrays or use string keys
goto OnError;
}
if (i++ > 1) appendw(buf, ',');
LuaEncodeJsonDataImpl(L, buf, level - 1, numformat, -2, visited);
appendw(buf, ':');
LuaEncodeJsonDataImpl(L, buf, level - 1, numformat, -1, visited);
RETURN_ON_ERROR(sli = AppendStrList(&sl));
RETURN_ON_ERROR(LuaEncodeJsonDataImpl(L, &sl.p[sli], level - 1,
numformat, -2, visited));
RETURN_ON_ERROR(appendw(&sl.p[sli], ':'));
RETURN_ON_ERROR(LuaEncodeJsonDataImpl(L, &sl.p[sli], level - 1,
numformat, -1, visited));
lua_pop(L, 1); // table/-2, key/-1
}
// stack: table/-1, as the key was popped by lua_next
}
appendw(buf, isarray ? ']' : '}');
RETURN_ON_ERROR(appendw(buf, isarray ? '[' : '{'));
SortStrList(&sl);
RETURN_ON_ERROR(JoinStrList(&sl, buf, ','));
FreeStrList(&sl);
RETURN_ON_ERROR(appendw(buf, isarray ? ']' : '}'));
LuaPopVisit(visited);
lua_pop(L, 1); // table ref
return 0;
} else {
free(visited->p);
luaL_error(L, "can't serialize cyclic data structure to json");
unreachable;
// cyclic data structure
goto OnError;
}
default:
free(visited->p);
luaL_error(L, "won't serialize %s to json", luaL_typename(L, idx));
unreachable;
// unsupported lua type
goto OnError;
}
} else {
free(visited->p);
luaL_error(L, "too many nested tables");
unreachable;
// too much depth
goto OnError;
}
OnError:
FreeStrList(&sl);
return -1;
}
/**
* Encodes Lua data structure as JSON.
*
* @param L is Lua interpreter state
* @param buf receives encoded output string
* @param numformat controls double formatting
* @param idx is index of item on Lua stack
* @return 0 on success, or -1 on error
*/
int LuaEncodeJsonData(lua_State *L, char **buf, char *numformat, int idx) {
int rc;
struct LuaVisited visited = {0};
rc = LuaEncodeJsonDataImpl(L, buf, 64, numformat, idx, &visited);
assert(!visited.n);
free(visited.p);
return rc;
}

View file

@ -19,9 +19,11 @@
#include "libc/assert.h"
#include "libc/bits/bits.h"
#include "libc/fmt/itoa.h"
#include "libc/log/rop.h"
#include "libc/math.h"
#include "libc/mem/mem.h"
#include "libc/stdio/append.internal.h"
#include "libc/stdio/strlist.internal.h"
#include "libc/x/x.h"
#include "third_party/lua/cosmo.h"
#include "third_party/lua/lauxlib.h"
@ -60,41 +62,45 @@ static int LuaEncodeLuaDataImpl(lua_State *L, char **buf, int level,
char *s;
bool isarray;
lua_Integer i;
int ktype, vtype;
struct StrList sl = {0};
int ktype, vtype, sli, rc;
size_t tbllen, buflen, slen;
char ibuf[24], fmt[] = "%.14g";
if (level > 0) {
switch (lua_type(L, idx)) {
case LUA_TNIL:
appendw(buf, READ32LE("nil"));
RETURN_ON_ERROR(appendw(buf, READ32LE("nil")));
return 0;
case LUA_TSTRING:
s = lua_tolstring(L, idx, &slen);
EscapeLuaString(s, slen, buf);
RETURN_ON_ERROR(EscapeLuaString(s, slen, buf));
return 0;
case LUA_TFUNCTION:
appendf(buf, "\"func@%p\"", lua_topointer(L, idx));
RETURN_ON_ERROR(
appendf(buf, "\"%s@%p\"", "func", lua_topointer(L, idx)));
return 0;
case LUA_TLIGHTUSERDATA:
appendf(buf, "\"light@%p\"", lua_topointer(L, idx));
RETURN_ON_ERROR(
appendf(buf, "\"%s@%p\"", "light", lua_topointer(L, idx)));
return 0;
case LUA_TTHREAD:
appendf(buf, "\"thread@%p\"", lua_topointer(L, idx));
RETURN_ON_ERROR(
appendf(buf, "\"%s@%p\"", "thread", lua_topointer(L, idx)));
return 0;
case LUA_TUSERDATA:
if (luaL_callmeta(L, idx, "__repr")) {
if (lua_type(L, -1) == LUA_TSTRING) {
s = lua_tolstring(L, -1, &slen);
appendd(buf, s, slen);
RETURN_ON_ERROR(appendd(buf, s, slen));
} else {
appendf(buf, "[[error %s returned a %s value]]", "__repr",
luaL_typename(L, -1));
RETURN_ON_ERROR(appendf(buf, "[[error %s returned a %s value]]",
"__repr", luaL_typename(L, -1)));
}
lua_pop(L, 1);
return 0;
@ -102,21 +108,23 @@ static int LuaEncodeLuaDataImpl(lua_State *L, char **buf, int level,
if (luaL_callmeta(L, idx, "__tostring")) {
if (lua_type(L, -1) == LUA_TSTRING) {
s = lua_tolstring(L, -1, &slen);
EscapeLuaString(s, slen, buf);
RETURN_ON_ERROR(EscapeLuaString(s, slen, buf));
} else {
appendf(buf, "[[error %s returned a %s value]]", "__tostring",
luaL_typename(L, -1));
RETURN_ON_ERROR(appendf(buf, "[[error %s returned a %s value]]",
"__tostring", luaL_typename(L, -1)));
}
lua_pop(L, 1);
return 0;
}
appendf(buf, "\"udata@%p\"", lua_touserdata(L, idx));
RETURN_ON_ERROR(
appendf(buf, "\"%s@%p\"", "udata", lua_touserdata(L, idx)));
return 0;
case LUA_TNUMBER:
if (lua_isinteger(L, idx)) {
appendd(buf, ibuf,
FormatFlex64(ibuf, luaL_checkinteger(L, idx), 2) - ibuf);
RETURN_ON_ERROR(
appendd(buf, ibuf,
FormatFlex64(ibuf, luaL_checkinteger(L, idx), 2) - ibuf));
} else {
// TODO(jart): replace this api
while (*numformat == '%' || *numformat == '.' ||
@ -130,71 +138,87 @@ static int LuaEncodeLuaDataImpl(lua_State *L, char **buf, int level,
fmt[4] = *numformat;
break;
default:
free(visited->p);
luaL_error(L, "numformat string not allowed");
unreachable;
// prevent format string hacking
goto OnError;
}
appendf(buf, fmt, lua_tonumber(L, idx));
RETURN_ON_ERROR(appendf(buf, fmt, lua_tonumber(L, idx)));
}
return 0;
case LUA_TBOOLEAN:
appendw(buf, lua_toboolean(L, idx) ? READ32LE("true")
: READ64LE("false\0\0"));
RETURN_ON_ERROR(appendw(buf, lua_toboolean(L, idx)
? READ32LE("true")
: READ64LE("false\0\0")));
return 0;
case LUA_TTABLE:
i = 0;
if (LuaPushVisit(visited, lua_topointer(L, idx))) {
appendw(buf, '{');
RETURN_ON_ERROR(rc = LuaPushVisit(visited, lua_topointer(L, idx)));
if (!rc) {
lua_pushvalue(L, idx); // idx becomes invalid once we change stack
isarray = IsLuaArray(L);
lua_pushnil(L); // push the first key
while (lua_next(L, -2)) {
ktype = lua_type(L, -2);
vtype = lua_type(L, -1);
if (i++ > 0) appendw(buf, ',' | ' ' << 8);
RETURN_ON_ERROR(sli = AppendStrList(&sl));
if (isarray) {
// use {v₁,v₂,...} for lua-normal integer keys
} else if (ktype == LUA_TSTRING && IsLuaIdentifier(L, -2)) {
// use {𝑘=𝑣} syntax when 𝑘 is legal as a lua identifier
s = lua_tolstring(L, -2, &slen);
appendd(buf, s, slen);
appendw(buf, '=');
RETURN_ON_ERROR(appendd(&sl.p[sli], s, slen));
RETURN_ON_ERROR(appendw(&sl.p[sli], '='));
} else {
// use {[𝑘]=𝑣} otherwise
appendw(buf, '[');
LuaEncodeLuaDataImpl(L, buf, level - 1, numformat, -2, visited);
appendw(buf, ']' | '=' << 010);
RETURN_ON_ERROR(appendw(&sl.p[sli], '['));
RETURN_ON_ERROR(LuaEncodeLuaDataImpl(L, &sl.p[sli], level - 1,
numformat, -2, visited));
RETURN_ON_ERROR(appendw(&sl.p[sli], ']' | '=' << 010));
}
LuaEncodeLuaDataImpl(L, buf, level - 1, numformat, -1, visited);
RETURN_ON_ERROR(LuaEncodeLuaDataImpl(L, &sl.p[sli], level - 1,
numformat, -1, visited));
lua_pop(L, 1); // table/-2, key/-1
}
lua_pop(L, 1); // table ref
RETURN_ON_ERROR(appendw(buf, '{'));
SortStrList(&sl);
RETURN_ON_ERROR(JoinStrList(&sl, buf, READ16LE(", ")));
RETURN_ON_ERROR(appendw(buf, '}'));
FreeStrList(&sl);
LuaPopVisit(visited);
appendw(buf, '}');
} else {
appendf(buf, "\"cyclic@%p\"", lua_topointer(L, idx));
RETURN_ON_ERROR(
appendf(buf, "\"%s@%p\"", "cyclic", lua_topointer(L, idx)));
}
return 0;
default:
free(visited->p);
luaL_error(L, "can't serialize value of this type");
unreachable;
// unsupported lua type
goto OnError;
}
} else {
free(visited->p);
luaL_error(L, "too many nested tables");
unreachable;
// too much depth
goto OnError;
}
OnError:
FreeStrList(&sl);
return -1;
}
/**
* Encodes Lua data structure as Lua code string.
*
* @param L is Lua interpreter state
* @param buf receives encoded output string
* @param numformat controls double formatting
* @param idx is index of item on Lua stack
* @return 0 on success, or -1 on error
*/
int LuaEncodeLuaData(lua_State *L, char **buf, char *numformat, int idx) {
int rc;
struct LuaVisited visited = {0};
rc = LuaEncodeLuaDataImpl(L, buf, 64, numformat, idx, &visited);
assert(!visited.n);
free(visited.p);
return rc;
}

View file

@ -20,16 +20,23 @@
#include "libc/x/x.h"
#include "third_party/lua/visitor.h"
bool LuaPushVisit(struct LuaVisited *visited, const void *p) {
int i;
int LuaPushVisit(struct LuaVisited *visited, const void *p) {
int i, n2;
const void **p2;
for (i = 0; i < visited->n; ++i) {
if (visited->p[i] == p) {
return false;
return 1;
}
}
visited->p = xrealloc(visited->p, ++visited->n * sizeof(*visited->p));
n2 = visited->n;
if ((p2 = realloc(visited->p, ++n2 * sizeof(*visited->p)))) {
visited->p = p2;
visited->n = n2;
} else {
return -1;
}
visited->p[visited->n - 1] = p;
return true;
return 0;
}
void LuaPopVisit(struct LuaVisited *visited) {

View file

@ -8,7 +8,7 @@ struct LuaVisited {
const void **p;
};
bool LuaPushVisit(struct LuaVisited *, const void *);
int LuaPushVisit(struct LuaVisited *, const void *);
void LuaPopVisit(struct LuaVisited *);
COSMOPOLITAN_C_END_