diff options
Diffstat (limited to 'erts/emulator/beam/erl_unicode.c')
-rw-r--r-- | erts/emulator/beam/erl_unicode.c | 168 |
1 files changed, 114 insertions, 54 deletions
diff --git a/erts/emulator/beam/erl_unicode.c b/erts/emulator/beam/erl_unicode.c index e00440b905..3a968594f3 100644 --- a/erts/emulator/beam/erl_unicode.c +++ b/erts/emulator/beam/erl_unicode.c @@ -1476,6 +1476,9 @@ static Eterm do_utf8_to_list_normalize(Process *p, Uint num, byte *bytes, Uint s Uint16 savepoints[4]; int numpoints = 0; + if (num == 0) + return NIL; + ASSERT(num > 0); hp = HAlloc(p,num * 2); /* May be to much */ @@ -1981,9 +1984,21 @@ BIF_RETTYPE binary_to_existing_atom_2(BIF_ALIST_2) * string routines, that will certainly fail on some OS. */ -char *erts_convert_filename_to_native(Eterm name, char *statbuf, size_t statbuf_size, ErtsAlcType_t alloc_type, int allow_empty, int allow_atom, Sint *used) +char *erts_convert_filename_to_native(Eterm name, char *statbuf, size_t statbuf_size, + ErtsAlcType_t alloc_type, int allow_empty, + int allow_atom, Sint *used) { int encoding = erts_get_native_filename_encoding(); + return erts_convert_filename_to_encoding(name, statbuf, statbuf_size, alloc_type, + allow_empty, allow_atom, encoding, + used, 0); +} + +char *erts_convert_filename_to_encoding(Eterm name, char *statbuf, size_t statbuf_size, + ErtsAlcType_t alloc_type, int allow_empty, + int allow_atom, int encoding, Sint *used, + Uint extra) +{ char* name_buf = NULL; if ((allow_atom && is_atom(name)) || @@ -1995,13 +2010,14 @@ char *erts_convert_filename_to_native(Eterm name, char *statbuf, size_t statbuf_ } if (encoding == ERL_FILENAME_WIN_WCHAR) { need += 2; + extra *= 2; } else { ++need; } if (used) *used = (Sint) need; - if (need > statbuf_size) { - name_buf = (char *) erts_alloc(alloc_type, need); + if (need+extra > statbuf_size) { + name_buf = (char *) erts_alloc(alloc_type, need+extra); } else { name_buf = statbuf; } @@ -2013,52 +2029,27 @@ char *erts_convert_filename_to_native(Eterm name, char *statbuf, size_t statbuf_ } else if (is_binary(name)) { byte *temp_alloc = NULL; byte *bytes; - byte *err_pos; - Uint size,num_chars; + Uint size; size = binary_size(name); bytes = erts_get_aligned_binary_bytes(name, &temp_alloc); + if (encoding != ERL_FILENAME_WIN_WCHAR) { /*Add 0 termination only*/ if (used) *used = (Sint) size+1; - if (size+1 > statbuf_size) { - name_buf = (char *) erts_alloc(alloc_type, size+1); + if (size+1+extra > statbuf_size) { + name_buf = (char *) erts_alloc(alloc_type, size+1+extra); } else { name_buf = statbuf; } memcpy(name_buf,bytes,size); name_buf[size]=0; - } else if (erts_analyze_utf8(bytes,size,&err_pos,&num_chars,NULL) != ERTS_UTF8_OK || - erts_get_user_requested_filename_encoding() == ERL_FILENAME_LATIN1) { - byte *p; - /* What to do now? Maybe latin1, so just take byte for byte instead */ - if (used) - *used = (Sint) (size+1)*2; - if ((size+1)*2 > statbuf_size) { - name_buf = (char *) erts_alloc(alloc_type, (size+1)*2); - } else { - name_buf = statbuf; - } - p = (byte *) name_buf; - while (size--) { - *p++ = *bytes++; - *p++ = 0; - } - *p++ = 0; - *p++ = 0; - } else { /* WIN_WCHAR and valid UTF8 */ - if (used) - *used = (Sint) (num_chars+1)*2; - if ((num_chars+1)*2 > statbuf_size) { - name_buf = (char *) erts_alloc(alloc_type, (num_chars+1)*2); - } else { - name_buf = statbuf; - } - erts_copy_utf8_to_utf16_little((byte *) name_buf, bytes, num_chars); - name_buf[num_chars*2] = 0; - name_buf[num_chars*2+1] = 0; - } + } else { + name_buf = erts_convert_filename_to_wchar(bytes, size, + statbuf, statbuf_size, + alloc_type, used, extra); + } erts_free_aligned_binary_bytes(temp_alloc); } else { return NULL; @@ -2066,6 +2057,50 @@ char *erts_convert_filename_to_native(Eterm name, char *statbuf, size_t statbuf_ return name_buf; } +char* erts_convert_filename_to_wchar(byte* bytes, Uint size, + char *statbuf, size_t statbuf_size, + ErtsAlcType_t alloc_type, Sint* used, + Uint extra_wchars) +{ + byte *err_pos; + Uint num_chars; + char* name_buf = NULL; + Sint need; + char *p; + + if (erts_analyze_utf8(bytes,size,&err_pos,&num_chars,NULL) != ERTS_UTF8_OK || + erts_get_user_requested_filename_encoding() == ERL_FILENAME_LATIN1) { + + /* What to do now? Maybe latin1, so just take byte for byte instead */ + need = (Sint) (size + extra_wchars + 1) * 2; + if (need > statbuf_size) { + name_buf = (char *) erts_alloc(alloc_type, need); + } else { + name_buf = statbuf; + } + p = name_buf; + while (size--) { + *p++ = *bytes++; + *p++ = 0; + } + } else { /* WIN_WCHAR and valid UTF8 */ + need = (Sint) (num_chars + extra_wchars + 1) * 2; + if (need > statbuf_size) { + name_buf = (char *) erts_alloc(alloc_type, need); + } else { + name_buf = statbuf; + } + erts_copy_utf8_to_utf16_little((byte *) name_buf, bytes, num_chars); + p = name_buf + num_chars*2; + } + *p++ = 0; + *p++ = 0; + if (used) + *used = p - name_buf; + return name_buf; +} + + static int filename_len_16bit(byte *str) { byte *p = str; @@ -2145,16 +2180,31 @@ Sint erts_native_filename_need(Eterm ioterm, int encoding) ap = atom_tab(atom_val(ioterm)); switch (encoding) { case ERL_FILENAME_LATIN1: - need = ap->len; + need = ap->latin1_chars; /* May be -1 */ break; case ERL_FILENAME_UTF8_MAC: case ERL_FILENAME_UTF8: - for (i = 0; i < ap->len; i++) { - need += (ap->name[i] >= 0x80) ? 2 : 1; - } + need = ap->len; break; case ERL_FILENAME_WIN_WCHAR: - need = 2*(ap->len); + if (ap->latin1_chars >= 0) { + need = 2* ap->latin1_chars; + } + else { + for (i = 0; i < ap->len; ) { + if (ap->name[i] < 0x80) { + i++; + } else if (ap->name[i] < 0xE0) { + i += 2; + } else if (ap->name[i] < 0xF0) { + i += 3; + } else { + need = -1; + break; + } + need += 2; + } + } break; default: need = -1; @@ -2284,26 +2334,36 @@ void erts_native_filename_put(Eterm ioterm, int encoding, byte *p) switch (encoding) { case ERL_FILENAME_LATIN1: for (i = 0; i < ap->len; i++) { - *p++ = ap->name[i]; - } - break; - case ERL_FILENAME_UTF8_MAC: - case ERL_FILENAME_UTF8: - for (i = 0; i < ap->len; i++) { - if(ap->name[i] < 0x80) { + if (ap->name[i] < 0x80) { *p++ = ap->name[i]; } else { - *p++ = (((ap->name[i]) >> 6) | ((byte) 0xC0)); - *p++ = (((ap->name[i]) & 0x3F) | ((byte) 0x80)); + ASSERT(ap->name[i] < 0xC4); + *p++ = ((ap->name[i] & 3) << 6) | (ap->name[i+1] & 0x3F); + i++; } } break; + case ERL_FILENAME_UTF8_MAC: + case ERL_FILENAME_UTF8: + sys_memcpy(p, ap->name, ap->len); + break; case ERL_FILENAME_WIN_WCHAR: for (i = 0; i < ap->len; i++) { /* Little endian */ - *p++ = ap->name[i]; - *p++ = 0; - } + if (ap->name[i] < 0x80) { + *p++ = ap->name[i]; + *p++ = 0; + } else if (ap->name[i] < 0xE0) { + *p++ = ((ap->name[i] & 3) << 6) | (ap->name[i+1] & 0x3F); + *p++ = ((ap->name[i] & 0x1C) >> 2); + i++; + } else { + ASSERT(ap->name[i] < 0xF0); + *p++ = ((ap->name[i+1] & 3) << 6) | (ap->name[i+2] & 0x3C); + *p++ = ((ap->name[i] & 0xF) << 4) | ((ap->name[i+1] & 0x3C) >> 2); + i += 2; + } + } break; default: ASSERT(0); |