diff options
author | Sverker Eriksson <[email protected]> | 2013-01-23 18:09:35 +0100 |
---|---|---|
committer | Sverker Eriksson <[email protected]> | 2013-01-23 18:09:35 +0100 |
commit | b8e623410d1c22fe6d5fdeb8ccb0b2305533f033 (patch) | |
tree | 708d64e36e18b61ae1801c02ec3aeef42a697be3 /lib/erl_interface/src/legacy/erl_marshal.c | |
parent | e99df74bee7c245ec76678e336fcd09d4b51a089 (diff) | |
parent | d6e3e256b850050b7a86323b2948009d5fcc30a9 (diff) | |
download | otp-b8e623410d1c22fe6d5fdeb8ccb0b2305533f033.tar.gz otp-b8e623410d1c22fe6d5fdeb8ccb0b2305533f033.tar.bz2 otp-b8e623410d1c22fe6d5fdeb8ccb0b2305533f033.zip |
Merge branch 'sverk/r16/utf8-atoms'
* sverk/r16/utf8-atoms:
erl_interface: Fix bug when transcoding atoms from and to UTF8
erl_interface: Changed erlang_char_encoding interface
erts: Testcase doing unicode atom printout with ~w
erl_interface: even more utf8 atom stuff
erts: Fix bug in analyze_utf8 causing faulty latin1 detection
Add UTF-8 node name support for epmd
workaround...
Fix merge conflict with hasse
UTF-8 atom documentation
test case
erl_interface: utf8 atoms continued
Add utf8 atom distribution test cases
atom fixes for NIFs and atom_to_binary
UTF-8 support for distribution
Implement UTF-8 atom support for jinterface
erl_interface: Enable decode of unicode atoms
stdlib: Fix printing of unicode atoms
erts: Change internal representation of atoms to utf8
erts: Refactor rename DFLAG(S)_INTERNAL_TAGS for conformity
Conflicts:
erts/emulator/beam/io.c
OTP-10753
Diffstat (limited to 'lib/erl_interface/src/legacy/erl_marshal.c')
-rw-r--r-- | lib/erl_interface/src/legacy/erl_marshal.c | 413 |
1 files changed, 233 insertions, 180 deletions
diff --git a/lib/erl_interface/src/legacy/erl_marshal.c b/lib/erl_interface/src/legacy/erl_marshal.c index dad715c762..4c45cebb02 100644 --- a/lib/erl_interface/src/legacy/erl_marshal.c +++ b/lib/erl_interface/src/legacy/erl_marshal.c @@ -44,6 +44,9 @@ int erl_fp_compare(unsigned *a, unsigned *b); static void erl_long_to_fp(long l, unsigned *d); #endif +static int cmpbytes(unsigned char* s1,int l1,unsigned char* s2,int l2); +static int cmpatoms(unsigned char* s1, int l1, unsigned char tag1, unsigned char* s2, int l2, unsigned char tag2); + /* Used when comparing two encoded byte arrays */ /* this global data is ok (from threading point of view) since it is * initialized once and never changed @@ -51,7 +54,13 @@ static void erl_long_to_fp(long l, unsigned *d); #define CMP_ARRAY_SIZE 256 /* FIXME problem for threaded ? */ -static char cmp_array[CMP_ARRAY_SIZE]; + +static enum +{ + ERL_NUM_CMP=1, ERL_ATOM_CMP, ERL_REF_CMP, ERL_FUN_CMP, ERL_PORT_CMP, + ERL_PID_CMP, ERL_TUPLE_CMP, ERL_NIL_CMP, ERL_LIST_CMP, ERL_BIN_CMP +}cmp_array[CMP_ARRAY_SIZE]; + static int init_cmp_array_p=1; /* initialize array, the first time */ #if defined(VXWORKS) && CPU == PPC860 @@ -69,10 +78,8 @@ static int init_cmp_array_p=1; /* initialize array, the first time */ static int cmp_floats(double f1, double f2); static INLINE double to_float(long l); -#define ERL_NUM_CMP 1 -#define ERL_REF_CMP 3 - #define IS_ERL_NUM(t) (cmp_array[t]==ERL_NUM_CMP) +#define IS_ERL_ATOM(t) (cmp_array[t]==ERL_ATOM_CMP) #define CMP_NUM_CLASS_SIZE 256 static unsigned char cmp_num_class[CMP_NUM_CLASS_SIZE]; @@ -100,25 +107,28 @@ void erl_init_marshal(void) { if (init_cmp_array_p) { memset(cmp_array, 0, CMP_ARRAY_SIZE); - cmp_array[ERL_SMALL_INTEGER_EXT] = 1; - cmp_array[ERL_INTEGER_EXT] = 1; - cmp_array[ERL_FLOAT_EXT] = 1; - cmp_array[NEW_FLOAT_EXT] = 1; - cmp_array[ERL_SMALL_BIG_EXT] = 1; - cmp_array[ERL_LARGE_BIG_EXT] = 1; - cmp_array[ERL_ATOM_EXT] = 2; - cmp_array[ERL_REFERENCE_EXT] = 3; - cmp_array[ERL_NEW_REFERENCE_EXT] = 3; - cmp_array[ERL_FUN_EXT] = 4; - cmp_array[ERL_NEW_FUN_EXT] = 4; - cmp_array[ERL_PORT_EXT] = 5; - cmp_array[ERL_PID_EXT] = 6; - cmp_array[ERL_SMALL_TUPLE_EXT] = 7; - cmp_array[ERL_LARGE_TUPLE_EXT] = 7; - cmp_array[ERL_NIL_EXT] = 8; - cmp_array[ERL_STRING_EXT] = 9; - cmp_array[ERL_LIST_EXT] = 9; - cmp_array[ERL_BINARY_EXT] = 10; + cmp_array[ERL_SMALL_INTEGER_EXT] = ERL_NUM_CMP; + cmp_array[ERL_INTEGER_EXT] = ERL_NUM_CMP; + cmp_array[ERL_FLOAT_EXT] = ERL_NUM_CMP; + cmp_array[NEW_FLOAT_EXT] = ERL_NUM_CMP; + cmp_array[ERL_SMALL_BIG_EXT] = ERL_NUM_CMP; + cmp_array[ERL_LARGE_BIG_EXT] = ERL_NUM_CMP; + cmp_array[ERL_ATOM_EXT] = ERL_ATOM_CMP; + cmp_array[ERL_ATOM_UTF8_EXT] = ERL_ATOM_CMP; + cmp_array[ERL_SMALL_ATOM_EXT] = ERL_ATOM_CMP; + cmp_array[ERL_SMALL_ATOM_UTF8_EXT] = ERL_ATOM_CMP; + cmp_array[ERL_REFERENCE_EXT] = ERL_REF_CMP; + cmp_array[ERL_NEW_REFERENCE_EXT] = ERL_REF_CMP; + cmp_array[ERL_FUN_EXT] = ERL_FUN_CMP; + cmp_array[ERL_NEW_FUN_EXT] = ERL_FUN_CMP; + cmp_array[ERL_PORT_EXT] = ERL_PORT_CMP; + cmp_array[ERL_PID_EXT] = ERL_PID_CMP; + cmp_array[ERL_SMALL_TUPLE_EXT] = ERL_TUPLE_CMP; + cmp_array[ERL_LARGE_TUPLE_EXT] = ERL_TUPLE_CMP; + cmp_array[ERL_NIL_EXT] = ERL_NIL_CMP; + cmp_array[ERL_STRING_EXT] = ERL_LIST_CMP; + cmp_array[ERL_LIST_EXT] = ERL_LIST_CMP; + cmp_array[ERL_BINARY_EXT] = ERL_BIN_CMP; init_cmp_array_p = 0; } if (init_cmp_num_class_p) { @@ -156,6 +166,21 @@ static int erl_length_x(const ETERM *ep) { *============================================================== */ +static void encode_atom(Erl_Atom_data* a, unsigned char **ext) +{ + int ix = 0; + if (a->latin1) { + ei_encode_atom_len_as((char*)*ext, &ix, a->latin1, a->lenL, + ERLANG_LATIN1, ERLANG_LATIN1); + } + else if (ei_encode_atom_len_as((char*)*ext, &ix, a->utf8, a->lenU, + ERLANG_UTF8, ERLANG_LATIN1) < 0) { + ei_encode_atom_len_as((char*)*ext, &ix, a->utf8, a->lenU, + ERLANG_UTF8, ERLANG_UTF8); + } + *ext += ix; +} + /* * The actual ENCODE engine. * Returns 0 on success, otherwise 1. @@ -170,12 +195,7 @@ int erl_encode_it(ETERM *ep, unsigned char **ext, int dist) switch(ERL_TYPE(ep)) { case ERL_ATOM: - i = ep->uval.aval.len; - *(*ext)++ = ERL_ATOM_EXT; - *(*ext)++ = (i >>8) &0xff; - *(*ext)++ = i &0xff; - memcpy((void *) *ext, (const void *) ep->uval.aval.a, i); - *ext += i; + encode_atom(&ep->uval.aval.d, ext); return 0; case ERL_INTEGER: @@ -286,12 +306,7 @@ int erl_encode_it(ETERM *ep, unsigned char **ext, int dist) case ERL_PID: *(*ext)++ = ERL_PID_EXT; /* First poke in node as an atom */ - i = strlen((char *)ERL_PID_NODE(ep)); - *(*ext)++ = ERL_ATOM_EXT; - *(*ext)++ = (i >>8) &0xff; - *(*ext)++ = i &0xff; - memcpy(*ext, ERL_PID_NODE(ep), i); - *ext += i; + encode_atom(&ep->uval.pidval.node, ext); /* And then fill in the integer fields */ i = ERL_PID_NUMBER(ep); *(*ext)++ = (i >> 24) &0xff; @@ -319,11 +334,8 @@ int erl_encode_it(ETERM *ep, unsigned char **ext, int dist) *(*ext)++ = (len >> 8) &0xff; *(*ext)++ = len &0xff; - *(*ext)++ = ERL_ATOM_EXT; - *(*ext)++ = (i >> 8) &0xff; - *(*ext)++ = i &0xff; - memcpy(*ext, ERL_REF_NODE(ep), i); - *ext += i; + encode_atom(&ep->uval.refval.node, ext); + *(*ext)++ = ERL_REF_CREATION(ep); /* Then the integer fields */ for (j = 0; j < ERL_REF_LEN(ep); j++) { @@ -338,12 +350,7 @@ int erl_encode_it(ETERM *ep, unsigned char **ext, int dist) case ERL_PORT: *(*ext)++ = ERL_PORT_EXT; /* First poke in node as an atom */ - i = strlen((char *)ERL_PORT_NODE(ep)); - *(*ext)++ = ERL_ATOM_EXT; - *(*ext)++ = (i >>8) &0xff; - *(*ext)++ = i &0xff; - memcpy(*ext, ERL_PORT_NODE(ep), i); - *ext += i; + encode_atom(&ep->uval.portval.node, ext); /* Then the integer fields */ i = ERL_PORT_NUMBER(ep); *(*ext)++ = (i >> 24) &0xff; @@ -494,6 +501,16 @@ int erl_term_len(ETERM *ep) return 1+erl_term_len_helper(ep, 4); } +static int atom_len_helper(Erl_Atom_data* a) +{ + if (erl_atom_ptr_latin1(a)) { + return 1 + 2 + a->lenL; /* ERL_ATOM_EXT */ + } + else { + return 1 + 1 + (a->lenU > 255) + a->lenU; + } +} + static int erl_term_len_helper(ETERM *ep, int dist) { int len = 0; @@ -505,8 +522,7 @@ static int erl_term_len_helper(ETERM *ep, int dist) if (ep) { switch (ERL_TYPE(ep)) { case ERL_ATOM: - i = ep->uval.aval.len; - len = i + 3; + len = atom_len_helper(&ep->uval.aval.d); break; case ERL_INTEGER: @@ -538,20 +554,15 @@ static int erl_term_len_helper(ETERM *ep, int dist) break; case ERL_PID: - /* 1 + N + 4 + 4 + 1 where N = 3 + strlen */ - i = strlen((char *)ERL_PID_NODE(ep)); - len = 13 + i; + len = 1 + atom_len_helper(&ep->uval.pidval.node) + 4 + 4 + 1; break; case ERL_REF: - i = strlen((char *)ERL_REF_NODE(ep)); - len = 1 + 2 + (i+3) + 1 + ERL_REF_LEN(ep) * 4; + len = 1 + 2 + atom_len_helper(&ep->uval.refval.node) + 1 + ERL_REF_LEN(ep) * 4; break; case ERL_PORT: - /* 1 + N + 4 + 1 where N = 3 + strlen */ - i = strlen((char *)ERL_PORT_NODE(ep)); - len = 9 + i; + len = 1 + atom_len_helper(&ep->uval.portval.node) + 4 + 1; break; case ERL_EMPTY_LIST: @@ -644,31 +655,36 @@ int erl_encode_buf(ETERM *ep, unsigned char **ext) } /* erl_encode_buf */ -/* - * A nice macro to make it look cleaner in the - * cases of PID's,PORT's and REF's below. - * It reads the NODE name from a buffer. - */ -#define READ_THE_NODE(ext,cp,len,i) \ -/* eat first atom, repr. the node */ \ -if (**ext != ERL_ATOM_EXT) \ - return (ETERM *) NULL; \ -*ext += 1; \ -i = (**ext << 8) | (*ext)[1]; \ -cp = (char *) *(ext) + 2; \ -*ext += (i + 2); \ -len = i - -#define STATIC_NODE_BUF_SZ 30 - -#define SET_NODE(node,node_buf,cp,len) \ -if (len >= STATIC_NODE_BUF_SZ) node = erl_malloc(len+1); \ -else node = node_buf; \ -memcpy(node, cp, len); \ -node[len] = '\0' - -#define RESET_NODE(node,len) \ -if (len >= STATIC_NODE_BUF_SZ) free(node) + +static int read_atom(unsigned char** ext, Erl_Atom_data* a) +{ + char buf[MAXATOMLEN_UTF8]; + int offs = 0; + enum erlang_char_encoding enc; + int ret = ei_decode_atom_as((char*)*ext, &offs, buf, MAXATOMLEN_UTF8, + ERLANG_LATIN1|ERLANG_UTF8, NULL, &enc); + *ext += offs; + + if (ret == 0) { + int i = strlen(buf); + char* clone = erl_malloc(i+1); + memcpy(clone, buf, i+1); + + a->latin1 = NULL; + a->lenL = 0; + a->utf8 = NULL; + a->lenU = 0; + if (enc & (ERLANG_LATIN1 | ERLANG_ASCII)) { + a->latin1 = clone; + a->lenL = i; + } + if (enc & (ERLANG_UTF8 | ERLANG_ASCII)) { + a->utf8 = clone; + a->lenU = i; + } + } + return ret; +} /* * The actual DECODE engine. @@ -679,13 +695,13 @@ static ETERM *erl_decode_it(unsigned char **ext) char *cp; ETERM *ep,*tp,*np; unsigned int u,sign; - int i,j,len,arity; + int i,j,arity; double ff; /* Assume we are going to decode an integer */ ep = erl_alloc_eterm(ERL_INTEGER); ERL_COUNT(ep) = 1; - + switch (*(*ext)++) { case ERL_INTEGER_EXT: @@ -774,138 +790,90 @@ static ETERM *erl_decode_it(unsigned char **ext) return ep; case ERL_ATOM_EXT: + case ERL_SMALL_ATOM_EXT: + case ERL_ATOM_UTF8_EXT: + case ERL_SMALL_ATOM_UTF8_EXT: + ERL_TYPE(ep) = ERL_ATOM; - i = (**ext << 8) | (*ext)[1]; - cp = (char *) *(ext) + 2; - *ext += (i + 2); - ep->uval.aval.len = i; - ep->uval.aval.a = (char *) erl_malloc(i+1); - memcpy(ep->uval.aval.a, cp, i); - ep->uval.aval.a[i]='\0'; + --(*ext); + if (read_atom(ext, &ep->uval.aval.d) < 0) return NULL; return ep; case ERL_PID_EXT: - erl_free_term(ep); - { /* Why not use the constructors? */ - char *node; - char node_buf[STATIC_NODE_BUF_SZ]; + { unsigned int number, serial; unsigned char creation; - ETERM *eterm_p; - READ_THE_NODE(ext,cp,len,i); - SET_NODE(node,node_buf,cp,len); + ERL_TYPE(ep) = ERL_PID; + if (read_atom(ext, &ep->uval.pidval.node) < 0) return NULL; /* get the integers */ -#if 0 - /* FIXME: Remove code or whatever.... - Ints on the wire are big-endian (== network byte order) - so use ntoh[sl]. (But some are little-endian! Arrrgh!) - Also, the libc authors can be expected to optimize them - heavily. However, the marshalling makes no guarantees - about alignments -- so it won't work at all. */ - number = ntohl(*((unsigned int *)*ext)++); - serial = ntohl(*((unsigned int *)*ext)++); -#else number = ((*ext)[0] << 24) | ((*ext)[1]) << 16 | ((*ext)[2]) << 8 | ((*ext)[3]); *ext += 4; serial = ((*ext)[0] << 24) | ((*ext)[1]) << 16 | ((*ext)[2]) << 8 | ((*ext)[3]); *ext += 4; -#endif creation = *(*ext)++; - eterm_p = erl_mk_pid(node, number, serial, creation); - RESET_NODE(node,len); - return eterm_p; + erl_mk_pid_helper(ep, number, serial, creation); + return ep; } case ERL_REFERENCE_EXT: - erl_free_term(ep); { - char *node; - char node_buf[STATIC_NODE_BUF_SZ]; - unsigned int number; + unsigned int n[3] = {0, 0, 0}; unsigned char creation; - ETERM *eterm_p; - READ_THE_NODE(ext,cp,len,i); - SET_NODE(node,node_buf,cp,len); + ERL_TYPE(ep) = ERL_REF; + if (read_atom(ext, &ep->uval.refval.node) < 0) return NULL; /* get the integers */ -#if 0 - number = ntohl(*((unsigned int *)*ext)++); -#else - number = ((*ext)[0] << 24) | ((*ext)[1]) << 16 | + n[0] = ((*ext)[0] << 24) | ((*ext)[1]) << 16 | ((*ext)[2]) << 8 | ((*ext)[3]); *ext += 4; -#endif creation = *(*ext)++; - eterm_p = erl_mk_ref(node, number, creation); - RESET_NODE(node,len); - return eterm_p; + __erl_mk_reference(ep, NULL, 1, n, creation); + return ep; } case ERL_NEW_REFERENCE_EXT: - erl_free_term(ep); { - char *node; - char node_buf[STATIC_NODE_BUF_SZ]; size_t cnt, i; unsigned int n[3]; unsigned char creation; - ETERM *eterm_p; -#if 0 - cnt = ntohs(*((unsigned short *)*ext)++); -#else + ERL_TYPE(ep) = ERL_REF; cnt = ((*ext)[0] << 8) | (*ext)[1]; *ext += 2; -#endif - READ_THE_NODE(ext,cp,len,i); - SET_NODE(node,node_buf,cp,len); + if (read_atom(ext, &ep->uval.refval.node) < 0) return NULL; /* get the integers */ creation = *(*ext)++; for(i = 0; i < cnt; i++) { -#if 0 - n[i] = ntohl(*((unsigned int *)*ext)++); -#else n[i] = ((*ext)[0] << 24) | ((*ext)[1]) << 16 | ((*ext)[2]) << 8 | ((*ext)[3]); *ext += 4; -#endif } - eterm_p = __erl_mk_reference(node, cnt, n, creation); - RESET_NODE(node,len); - return eterm_p; + __erl_mk_reference(ep, NULL, cnt, n, creation); + return ep; } case ERL_PORT_EXT: - erl_free_term(ep); { - char *node; - char node_buf[STATIC_NODE_BUF_SZ]; unsigned int number; unsigned char creation; - ETERM *eterm_p; - READ_THE_NODE(ext,cp,len,i); - SET_NODE(node,node_buf,cp,len); + ERL_TYPE(ep) = ERL_PORT; + if (read_atom(ext, &ep->uval.portval.node) < 0) return NULL; /* get the integers */ -#if 0 - number = ntohl(*((unsigned int *)*ext)++); -#else number = ((*ext)[0] << 24) | ((*ext)[1]) << 16 | ((*ext)[2]) << 8 | ((*ext)[3]); *ext += 4; -#endif creation = *(*ext)++; - eterm_p = erl_mk_port(node, number, creation); - RESET_NODE(node,len); - return eterm_p; + erl_mk_port_helper(ep, number, creation); + return ep; } case ERL_NIL_EXT: @@ -1140,6 +1108,9 @@ unsigned char erl_ext_type(unsigned char *ext) case ERL_INTEGER_EXT: return ERL_INTEGER; case ERL_ATOM_EXT: + case ERL_ATOM_UTF8_EXT: + case ERL_SMALL_ATOM_EXT: + case ERL_SMALL_ATOM_UTF8_EXT: return ERL_ATOM; case ERL_PID_EXT: return ERL_PID; @@ -1191,6 +1162,9 @@ int erl_ext_size(unsigned char *t) case ERL_SMALL_INTEGER_EXT: case ERL_INTEGER_EXT: case ERL_ATOM_EXT: + case ERL_ATOM_UTF8_EXT: + case ERL_SMALL_ATOM_EXT: + case ERL_SMALL_ATOM_UTF8_EXT: case ERL_PID_EXT: case ERL_PORT_EXT: case ERL_REFERENCE_EXT: @@ -1229,15 +1203,32 @@ int erl_ext_size(unsigned char *t) } /* ext_size */ -/* - * A nice macro that eats up the atom pointed to. - */ -#define JUMP_ATOM(ext,i) \ -if (**ext != ERL_ATOM_EXT) \ - return 0; \ -*ext += 1; \ -i = (**ext << 8) | (*ext)[1]; \ -*ext += (i + 2) + +static int jump_atom(unsigned char** ext) +{ + unsigned char* e = *ext; + int len; + + switch (*e++) { + case ERL_ATOM_EXT: + case ERL_ATOM_UTF8_EXT: + len = (e[0] << 8) | e[1]; + e += (len + 2); + break; + + case ERL_SMALL_ATOM_EXT: + case ERL_SMALL_ATOM_UTF8_EXT: + len = e[0]; + e += (len + 1); + break; + + default: + return 0; + } + *ext = e; + return 1; +} + /* * MOVE the POINTER PAST the ENCODED ETERM we @@ -1259,25 +1250,27 @@ static int jump(unsigned char **ext) *ext += 1; break; case ERL_ATOM_EXT: - i = (**ext << 8) | (*ext)[1]; - *ext += (i + 2); + case ERL_ATOM_UTF8_EXT: + case ERL_SMALL_ATOM_EXT: + case ERL_SMALL_ATOM_UTF8_EXT: + jump_atom(ext); break; case ERL_PID_EXT: /* eat first atom */ - JUMP_ATOM(ext,i); + if (!jump_atom(ext)) return 0; *ext += 9; /* Two int's and the creation field */ break; case ERL_REFERENCE_EXT: case ERL_PORT_EXT: /* first field is an atom */ - JUMP_ATOM(ext,i); + if (!jump_atom(ext)) return 0; *ext += 5; /* One int and the creation field */ break; case ERL_NEW_REFERENCE_EXT: n = (**ext << 8) | (*ext)[1]; *ext += 2; /* first field is an atom */ - JUMP_ATOM(ext,i); + if (!jump_atom(ext)) return 0; *ext += 4*n+1; break; case ERL_NIL_EXT: @@ -1425,6 +1418,58 @@ static int cmpbytes(unsigned char* s1,int l1,unsigned char* s2,int l2) } /* cmpbytes */ +#define tag2enc(T) ((T)==ERL_ATOM_EXT || (T)==ERL_SMALL_ATOM_EXT ? ERLANG_LATIN1 : ERLANG_UTF8) + +static int cmpatoms(unsigned char* s1, int l1, unsigned char tag1, + unsigned char* s2, int l2, unsigned char tag2) +{ + enum erlang_char_encoding enc1 = tag2enc(tag1); + enum erlang_char_encoding enc2 = tag2enc(tag2); + + if (enc1 == enc2) { + return cmpbytes(s1, l1,s2,l2); + } + + if (enc1 == ERLANG_LATIN1) { + return cmp_latin1_vs_utf8((char*)s1, l1, (char*)s2, l2); + } + else { + return -cmp_latin1_vs_utf8((char*)s2, l2, (char*)s1, l1); + } +} + +int cmp_latin1_vs_utf8(const char* strL, int lenL, const char* strU, int lenU) +{ + unsigned char* sL = (unsigned char*)strL; + unsigned char* sU = (unsigned char*)strU; + unsigned char* sL_end = sL + lenL; + unsigned char* sU_end = sU + lenU; + + while(sL < sL_end && sU < sU_end) { + unsigned char UasL; + if (*sL >= 0x80) { + if (*sU < 0xC4 && (sU+1) < sU_end) { + UasL = ((sU[0] & 0x3) << 6) | (sU[1] & 0x3F); + } + else return -1; + } + else { + UasL = *sU; + } + if (*sL < UasL) return -1; + if (*sL > UasL) return 1; + + sL++; + if (*sU < 0x80) sU++; + else if (*sU < 0xE0) sU += 2; + else if (*sU < 0xF0) sU += 3; + else /*if (*sU < 0xF8)*/ sU += 4; + } + + return (sU >= sU_end) - (sL >= sL_end); /* -1, 0 or 1 */ +} + + #define CMP_EXT_ERROR_CODE 4711 #define CMP_EXT_INT32_BE(AP, BP) \ @@ -1437,9 +1482,8 @@ do { \ #define CMP_EXT_SKIP_ATOM(EP) \ do { \ - if ((EP)[0] != ERL_ATOM_EXT) \ + if (!jump_atom(&(EP))) \ return CMP_EXT_ERROR_CODE; \ - (EP) += 3 + ((EP)[1] << 8 | (EP)[2]); \ } while (0) /* @@ -1561,6 +1605,7 @@ static int cmp_exe2(unsigned char **e1, unsigned char **e2) int min, ret,i,j,k; double ff1, ff2; unsigned char *tmp1, *tmp2; + unsigned char tag1, tag2; if ( ((*e1)[0] == ERL_STRING_EXT) && ((*e2)[0] == ERL_LIST_EXT) ) { return cmp_string_list(e1, e2); @@ -1568,8 +1613,10 @@ static int cmp_exe2(unsigned char **e1, unsigned char **e2) return -cmp_string_list(e2, e1); } - *e2 += 1; - switch (*(*e1)++) + tag1 = *(*e1)++; + tag2 = *(*e2)++; + i = j = 0; + switch (tag1) { case ERL_SMALL_INTEGER_EXT: if (**e1 < **e2) ret = -1; @@ -1589,11 +1636,17 @@ static int cmp_exe2(unsigned char **e1, unsigned char **e2) *e1 += 4; *e2 += 4; return ret; case ERL_ATOM_EXT: - i = (**e1 << 8) | (*e1)[1]; - j = (**e2 << 8) | (*e2)[1]; - ret = cmpbytes(*e1 +2, i, *e2 +2, j); - *e1 += (i + 2); - *e2 += (j + 2); + case ERL_ATOM_UTF8_EXT: + i = (**e1) << 8; (*e1)++; + j = (**e2) << 8; (*e2)++; + /*fall through*/ + case ERL_SMALL_ATOM_EXT: + case ERL_SMALL_ATOM_UTF8_EXT: + i |= (**e1); (*e1)++; + j |= (**e2); (*e2)++; + ret = cmpatoms(*e1, i, tag1, *e2, j, tag2); + *e1 += i; + *e2 += j; return ret; case ERL_PID_EXT: { unsigned char *n1 = *e1; @@ -1622,7 +1675,7 @@ static int cmp_exe2(unsigned char **e1, unsigned char **e2) } case ERL_PORT_EXT: /* First compare node names ... */ - if (**e1 != ERL_ATOM_EXT || **e2 != ERL_ATOM_EXT) + if (!IS_ERL_ATOM(**e1) || !IS_ERL_ATOM(**e2)) return CMP_EXT_ERROR_CODE; ret = cmp_exe2(e1, e2); *e1 += 5; *e2 += 5; |