From 1f4765cca4874fa92fcfad888fbe6d5f2fbf74d1 Mon Sep 17 00:00:00 2001 From: Sverker Eriksson Date: Tue, 22 Jan 2013 19:25:36 +0100 Subject: erl_interface: even more utf8 atom stuff --- lib/erl_interface/src/connect/ei_connect.c | 2 +- lib/erl_interface/src/encode/encode_atom.c | 59 +++++++++++++++++++++++++++++- lib/erl_interface/src/legacy/erl_connect.c | 2 +- lib/erl_interface/src/misc/ei_format.c | 4 +- lib/erl_interface/src/misc/ei_printterm.c | 2 +- lib/erl_interface/src/misc/show_msg.c | 2 +- lib/erl_interface/src/prog/ei_fake_prog.c | 6 +++ 7 files changed, 69 insertions(+), 8 deletions(-) (limited to 'lib/erl_interface/src') diff --git a/lib/erl_interface/src/connect/ei_connect.c b/lib/erl_interface/src/connect/ei_connect.c index a17257795e..4421bbb7fe 100644 --- a/lib/erl_interface/src/connect/ei_connect.c +++ b/lib/erl_interface/src/connect/ei_connect.c @@ -1071,7 +1071,7 @@ int ei_rpc(ei_cnode* ec, int fd, char *mod, char *fun, int i, index; ei_term t; erlang_msg msg; - char rex[MAXATOMLEN+1]; + char rex[MAXATOMLEN]; if (ei_rpc_to(ec, fd, mod, fun, inbuf, inbuflen) < 0) { return -1; diff --git a/lib/erl_interface/src/encode/encode_atom.c b/lib/erl_interface/src/encode/encode_atom.c index a3d7c4c759..8bbe962396 100644 --- a/lib/erl_interface/src/encode/encode_atom.c +++ b/lib/erl_interface/src/encode/encode_atom.c @@ -22,6 +22,11 @@ #include "eiext.h" #include "putget.h" + +static int copy_ascii_atom(char* dst, const char* src, int slen); +static int copy_utf8_atom(char* dst, const char* src, int slen); + + int ei_encode_atom(char *buf, int *index, const char *p) { size_t len = strlen(p); @@ -54,7 +59,8 @@ int ei_encode_atom_len_as(char *buf, int *index, const char *p, int len, char *s0 = s; int offs; - if (from_enc == ERLANG_LATIN1 && len >= MAXATOMLEN) { + if (len >= MAXATOMLEN && (from_enc == ERLANG_LATIN1 || + from_enc == ERLANG_ASCII)) { return -1; } @@ -68,6 +74,8 @@ int ei_encode_atom_len_as(char *buf, int *index, const char *p, int len, if (len < 0) return -1; break; case ERLANG_ASCII: + if (copy_ascii_atom(s+2, p, len) < 0) return -1; + break; case ERLANG_LATIN1: memcpy(s+2, p, len); break; @@ -93,9 +101,11 @@ int ei_encode_atom_len_as(char *buf, int *index, const char *p, int len, len = latin1_to_utf8((buf ? s+offs : NULL), p, len, MAXATOMLEN_UTF8-1, NULL); break; case ERLANG_ASCII: + if (buf && copy_ascii_atom(s+offs, p, len) < 0) return -1; + break; case ERLANG_UTF8: if (len >= 256) offs++; - if (buf) memcpy(s+offs, p, len); + if (buf && copy_utf8_atom(s+offs, p, len) < 0) return -1; break; default: return -1; @@ -133,3 +143,48 @@ ei_internal_put_atom(char** bufp, const char* p, int slen, *bufp += ix; return 0; } + + +int copy_ascii_atom(char* dst, const char* src, int slen) +{ + while (slen > 0) { + if ((src[0] & 0x80) != 0) return -1; + *dst++ = *src++; + slen--; + } + return 0; +} + +int copy_utf8_atom(char* dst, const char* src, int slen) +{ + int num_chars = 0; + + while (slen > 0) { + if (++num_chars >= MAXATOMLEN) return -1; + if ((src[0] & 0x80) != 0) { + if ((src[0] & 0xE0) == 0xC0) { + if (slen < 2 || (src[1] & 0xC0) != 0x80) return -1; + *dst++ = *src++; + slen--; + } + else if ((src[0] & 0xF0) == 0xE0) { + if (slen < 3 || (src[1] & 0xC0) != 0x80 || (src[2] & 0xC0) != 0x80) return -1; + *dst++ = *src++; + *dst++ = *src++; + slen -= 2; + } + else if ((src[0] & 0xF8) == 0xF0) { + if (slen < 4 || (src[1] & 0xC0) != 0x80 || (src[2] & 0xC0) != 0x80 || (src[3] & 0xC0) != 0x80) return -1; + *dst++ = *src++; + *dst++ = *src++; + *dst++ = *src++; + slen -= 3; + } + else return -1; + } + *dst++ = *src++; + slen--; + } + return 0; +} + diff --git a/lib/erl_interface/src/legacy/erl_connect.c b/lib/erl_interface/src/legacy/erl_connect.c index be83fa8469..f82704ea8b 100644 --- a/lib/erl_interface/src/legacy/erl_connect.c +++ b/lib/erl_interface/src/legacy/erl_connect.c @@ -125,7 +125,7 @@ static ei_cnode erl_if_ec; int erl_connect_init(int this_node_number, char *cookie, short creation) { - char nn[MAXATOMLEN+1]; + char nn[MAXATOMLEN]; sprintf(nn, "c%d", this_node_number); diff --git a/lib/erl_interface/src/misc/ei_format.c b/lib/erl_interface/src/misc/ei_format.c index 281a192535..b5f11e618e 100644 --- a/lib/erl_interface/src/misc/ei_format.c +++ b/lib/erl_interface/src/misc/ei_format.c @@ -139,8 +139,8 @@ static int patom(const char** fmt, ei_x_buff* x) --(*fmt); len = *fmt - start; /* FIXME why truncate atom name and not fail?! */ - if (len > MAXATOMLEN) - len = MAXATOMLEN; + if (len >= MAXATOMLEN) + len = MAXATOMLEN-1; return ei_x_encode_atom_len(x, start, len); } diff --git a/lib/erl_interface/src/misc/ei_printterm.c b/lib/erl_interface/src/misc/ei_printterm.c index 91fe73e68c..f3003a6172 100644 --- a/lib/erl_interface/src/misc/ei_printterm.c +++ b/lib/erl_interface/src/misc/ei_printterm.c @@ -115,7 +115,7 @@ static int print_term(FILE* fp, ei_x_buff* x, const char* buf, int* index) { int i, doquote, n, m, ty, r; - char a[MAXATOMLEN+1], *p; + char a[MAXATOMLEN], *p; int ch_written = 0; /* counter of written chars */ erlang_pid pid; erlang_port port; diff --git a/lib/erl_interface/src/misc/show_msg.c b/lib/erl_interface/src/misc/show_msg.c index ca46b15aff..33b09643ca 100644 --- a/lib/erl_interface/src/misc/show_msg.c +++ b/lib/erl_interface/src/misc/show_msg.c @@ -457,7 +457,7 @@ static void show_term(const char *termbuf, int *index, FILE *stream) break; case ERL_FUN_EXT: { - char atom[MAXATOMLEN+1]; + char atom[MAXATOMLEN]; long idx; long uniq; const char* s = termbuf + *index, * s0 = s; diff --git a/lib/erl_interface/src/prog/ei_fake_prog.c b/lib/erl_interface/src/prog/ei_fake_prog.c index 68eb537211..34101a2851 100644 --- a/lib/erl_interface/src/prog/ei_fake_prog.c +++ b/lib/erl_interface/src/prog/ei_fake_prog.c @@ -96,6 +96,7 @@ int main(void) EI_ULONGLONG *ulonglongp = (EI_ULONGLONG*)NULL; EI_ULONGLONG ulonglongx = 0; #endif + enum erlang_char_encoding enc; intx = erl_errno; @@ -148,9 +149,13 @@ int main(void) ei_x_encode_string(&eix, charp); ei_x_encode_string_len(&eix, charp, intx); ei_encode_atom(charp, intp, charp); + ei_encode_atom_as(charp, intp, charp, ERLANG_LATIN1, ERLANG_UTF8); ei_encode_atom_len(charp, intp, charp, intx); + ei_encode_atom_len_as(charp, intp, charp, intx, ERLANG_ASCII, ERLANG_LATIN1); ei_x_encode_atom(&eix, charp); + ei_x_encode_atom_as(&eix, charp, ERLANG_LATIN1, ERLANG_UTF8); ei_x_encode_atom_len(&eix, charp, intx); + ei_x_encode_atom_len_as(&eix, charp, intx, ERLANG_LATIN1, ERLANG_UTF8); ei_encode_binary(charp, intp, (void *)0, longx); ei_x_encode_binary(&eix, (void*)0, intx); ei_encode_pid(charp, intp, &epid); @@ -181,6 +186,7 @@ int main(void) ei_decode_char(charp, intp, charp); ei_decode_string(charp, intp, charp); ei_decode_atom(charp, intp, charp); + ei_decode_atom_as(charp, intp, charp, MAXATOMLEN_UTF8, ERLANG_WHATEVER, &enc, &enc); ei_decode_binary(charp, intp, (void *)0, longp); ei_decode_fun(charp, intp, &efun); free_fun(&efun); -- cgit v1.2.3