/*
Return code page number of the encoding.
Cache code page into a hash for performance since finding the code page in
Encoding#names is slow.
*/
static UINT
fenix_code_page(rb_encoding *enc)
{
VALUE code_page_value, name_key;
VALUE encoding, names_ary = Qundef, name;
char *enc_name;
struct RString fake_str;
ID names;
long i;
if (!enc)
return system_code_page();
enc_name = (char *)rb_enc_name(enc);
fake_str.basic.flags = T_STRING|RSTRING_NOEMBED;
fake_str.basic.klass = rb_cString;
fake_str.as.heap.len = strlen(enc_name);
fake_str.as.heap.ptr = enc_name;
fake_str.as.heap.aux.capa = fake_str.as.heap.len;
name_key = (VALUE)&fake_str;
ENCODING_CODERANGE_SET(name_key, rb_usascii_encindex(), ENC_CODERANGE_7BIT);
OBJ_FREEZE(name_key);
code_page_value = rb_hash_lookup(rb_code_page, name_key);
if (code_page_value != Qnil) {
// printf("cached code page: %i\n", FIX2INT(code_page_value));
if (FIX2INT(code_page_value) == -1) {
return system_code_page();
} else {
return (UINT)FIX2INT(code_page_value);
}
}
name_key = rb_usascii_str_new2(enc_name);
encoding = rb_enc_from_encoding(enc);
if (!NIL_P(encoding)) {
CONST_ID(names, "names");
names_ary = rb_funcall(encoding, names, 0);
}
if (names_ary != Qundef) {
for (i = 0; i < RARRAY_LEN(names_ary); i++) {
name = RARRAY_PTR(names_ary)[i];
if (strncmp("CP", RSTRING_PTR(name), 2) == 0) {
int code_page = atoi(RSTRING_PTR(name) + 2);
rb_hash_aset(rb_code_page, name_key, INT2FIX(code_page));
return (UINT)code_page;
}
}
}
rb_hash_aset(rb_code_page, name_key, INT2FIX(-1));
return system_code_page();
}
/*
Return code page number of the encoding.
Cache code page into a hash for performance since finding the code page in
Encoding#names is slow.
*/
static UINT
code_page(rb_encoding *enc)
{
VALUE code_page_value, name_key;
VALUE encoding, names_ary = Qundef, name;
char *enc_name;
struct RString fake_str;
ID names;
long i;
if (!enc)
return system_code_page();
enc_name = (char *)rb_enc_name(enc);
fake_str.basic.flags = T_STRING|RSTRING_NOEMBED;
fake_str.basic.klass = rb_cString;
fake_str.as.heap.len = strlen(enc_name);
fake_str.as.heap.ptr = enc_name;
fake_str.as.heap.aux.capa = fake_str.as.heap.len;
name_key = (VALUE)&fake_str;
ENCODING_CODERANGE_SET(name_key, rb_usascii_encindex(), ENC_CODERANGE_7BIT);
code_page_value = rb_hash_lookup(rb_code_page, name_key);
if (code_page_value != Qnil)
return (UINT)FIX2INT(code_page_value);
name_key = rb_usascii_str_new2(enc_name);
encoding = rb_enc_from_encoding(enc);
if (!NIL_P(encoding)) {
CONST_ID(names, "names");
names_ary = rb_funcall(encoding, names, 0);
}
/* map US-ASCII and ASCII-8bit as code page 1252 (us-ascii) */
if (enc == rb_usascii_encoding() || enc == rb_ascii8bit_encoding()) {
UINT code_page = 1252;
rb_hash_aset(rb_code_page, name_key, INT2FIX(code_page));
return code_page;
}
if (names_ary != Qundef) {
for (i = 0; i < RARRAY_LEN(names_ary); i++) {
name = RARRAY_PTR(names_ary)[i];
if (strncmp("CP", RSTRING_PTR(name), 2) == 0) {
int code_page = atoi(RSTRING_PTR(name) + 2);
if (code_page != 0) {
rb_hash_aset(rb_code_page, name_key, INT2FIX(code_page));
return (UINT)code_page;
}
}
}
}
rb_hash_aset(rb_code_page, name_key, INT2FIX(INVALID_CODE_PAGE));
return INVALID_CODE_PAGE;
}
/*
Return code page number of the encoding.
Cache code page into a hash for performance since finding the code page in
Encoding#names is slow.
*/
static UINT
code_page(rb_encoding *enc)
{
int enc_idx;
if (!enc)
return system_code_page();
enc_idx = rb_enc_to_index(enc);
/* map US-ASCII and ASCII-8bit as code page 1252 (us-ascii) */
if (enc_idx == rb_usascii_encindex() || enc_idx == rb_ascii8bit_encindex()) {
return 1252;
}
if (enc_idx == rb_utf8_encindex()) {
return CP_UTF8;
}
if (0 <= enc_idx && (unsigned int)enc_idx < rb_code_page.count)
return rb_code_page.table[enc_idx];
return INVALID_CODE_PAGE;
}
VALUE
rb_file_expand_path_internal(VALUE fname, VALUE dname, int abs_mode, int long_name, VALUE result)
{
size_t size = 0, whome_len = 0;
size_t buffer_len = 0;
long wpath_len = 0, wdir_len = 0;
char *fullpath = NULL;
wchar_t *wfullpath = NULL, *wpath = NULL, *wpath_pos = NULL;
wchar_t *wdir = NULL, *wdir_pos = NULL;
wchar_t *whome = NULL, *buffer = NULL, *buffer_pos = NULL;
UINT path_cp, cp;
VALUE path = fname, dir = dname;
wchar_t wfullpath_buffer[PATH_BUFFER_SIZE];
wchar_t path_drive = L'\0', dir_drive = L'\0';
int ignore_dir = 0;
rb_encoding *path_encoding;
int tainted = 0;
/* tainted if path is tainted */
tainted = OBJ_TAINTED(path);
/* get path encoding */
if (NIL_P(dir)) {
path_encoding = rb_enc_get(path);
}
else {
path_encoding = rb_enc_check(path, dir);
}
cp = path_cp = code_page(path_encoding);
/* workaround invalid codepage */
if (path_cp == INVALID_CODE_PAGE) {
cp = CP_UTF8;
if (!NIL_P(path)) {
path = fix_string_encoding(path, path_encoding);
}
}
/* convert char * to wchar_t */
if (!NIL_P(path)) {
wpath = mbstr_to_wstr(cp, RSTRING_PTR(path), (int)RSTRING_LEN(path), &wpath_len);
wpath_pos = wpath;
}
/* determine if we need the user's home directory */
/* expand '~' only if NOT rb_file_absolute_path() where `abs_mode` is 1 */
if (abs_mode == 0 && wpath_len > 0 && wpath_pos[0] == L'~' &&
(wpath_len == 1 || IS_DIR_SEPARATOR_P(wpath_pos[1]))) {
/* tainted if expanding '~' */
tainted = 1;
whome = home_dir();
if (whome == NULL) {
xfree(wpath);
rb_raise(rb_eArgError, "couldn't find HOME environment -- expanding `~'");
}
whome_len = wcslen(whome);
if (PathIsRelativeW(whome) && !(whome_len >= 2 && IS_DIR_UNC_P(whome))) {
xfree(wpath);
xfree(whome);
rb_raise(rb_eArgError, "non-absolute home");
}
if (path_cp == INVALID_CODE_PAGE || rb_enc_str_asciionly_p(path)) {
/* use filesystem encoding if expanding home dir */
path_encoding = rb_filesystem_encoding();
cp = path_cp = system_code_page();
}
/* ignores dir since we are expanding home */
ignore_dir = 1;
/* exclude ~ from the result */
wpath_pos++;
wpath_len--;
/* exclude separator if present */
if (wpath_len && IS_DIR_SEPARATOR_P(wpath_pos[0])) {
wpath_pos++;
wpath_len--;
}
}
else if (wpath_len >= 2 && wpath_pos[1] == L':') {
if (wpath_len >= 3 && IS_DIR_SEPARATOR_P(wpath_pos[2])) {
/* ignore dir since path contains a drive letter and a root slash */
ignore_dir = 1;
}
else {
/* determine if we ignore dir or not later */
path_drive = wpath_pos[0];
wpath_pos += 2;
wpath_len -= 2;
}
}
else if (abs_mode == 0 && wpath_len >= 2 && wpath_pos[0] == L'~') {
result = rb_str_new_cstr("can't find user ");
result = append_wstr(result, wpath_pos + 1, user_length_in_path(wpath_pos + 1, wpath_len - 1),
cp, path_cp, path_encoding);
if (wpath)
xfree(wpath);
rb_exc_raise(rb_exc_new_str(rb_eArgError, result));
}
/* convert dir */
if (!ignore_dir && !NIL_P(dir)) {
/* fix string encoding */
if (path_cp == INVALID_CODE_PAGE) {
dir = fix_string_encoding(dir, path_encoding);
}
/* convert char * to wchar_t */
if (!NIL_P(dir)) {
wdir = mbstr_to_wstr(cp, RSTRING_PTR(dir), (int)RSTRING_LEN(dir), &wdir_len);
wdir_pos = wdir;
}
if (abs_mode == 0 && wdir_len > 0 && wdir_pos[0] == L'~' &&
(wdir_len == 1 || IS_DIR_SEPARATOR_P(wdir_pos[1]))) {
/* tainted if expanding '~' */
tainted = 1;
whome = home_dir();
if (whome == NULL) {
free(wpath);
free(wdir);
rb_raise(rb_eArgError, "couldn't find HOME environment -- expanding `~'");
}
whome_len = wcslen(whome);
if (PathIsRelativeW(whome) && !(whome_len >= 2 && IS_DIR_UNC_P(whome))) {
free(wpath);
free(wdir);
xfree(whome);
rb_raise(rb_eArgError, "non-absolute home");
}
/* exclude ~ from the result */
wdir_pos++;
wdir_len--;
/* exclude separator if present */
if (wdir_len && IS_DIR_SEPARATOR_P(wdir_pos[0])) {
wdir_pos++;
wdir_len--;
}
}
else if (wdir_len >= 2 && wdir[1] == L':') {
dir_drive = wdir[0];
if (wpath_len && IS_DIR_SEPARATOR_P(wpath_pos[0])) {
wdir_len = 2;
}
}
else if (wdir_len >= 2 && IS_DIR_UNC_P(wdir)) {
/* UNC path */
if (wpath_len && IS_DIR_SEPARATOR_P(wpath_pos[0])) {
/* cut the UNC path tail to '//host/share' */
long separators = 0;
long pos = 2;
while (pos < wdir_len && separators < 2) {
if (IS_DIR_SEPARATOR_P(wdir[pos])) {
separators++;
}
pos++;
}
if (separators == 2)
wdir_len = pos - 1;
}
}
else if (abs_mode == 0 && wdir_len >= 2 && wdir_pos[0] == L'~') {
result = rb_str_new_cstr("can't find user ");
result = append_wstr(result, wdir_pos + 1, user_length_in_path(wdir_pos + 1, wdir_len - 1),
cp, path_cp, path_encoding);
if (wpath)
free(wpath);
if (wdir)
free(wdir);
rb_exc_raise(rb_exc_new_str(rb_eArgError, result));
}
}
/* determine if we ignore dir or not */
if (!ignore_dir && path_drive && dir_drive) {
if (towupper(path_drive) != towupper(dir_drive)) {
/* ignore dir since path drive is different from dir drive */
ignore_dir = 1;
wdir_len = 0;
dir_drive = 0;
}
}
if (!ignore_dir && wpath_len >= 2 && IS_DIR_UNC_P(wpath)) {
/* ignore dir since path has UNC root */
ignore_dir = 1;
wdir_len = 0;
}
else if (!ignore_dir && wpath_len >= 1 && IS_DIR_SEPARATOR_P(wpath[0]) &&
!dir_drive && !(wdir_len >= 2 && IS_DIR_UNC_P(wdir))) {
/* ignore dir since path has root slash and dir doesn't have drive or UNC root */
ignore_dir = 1;
wdir_len = 0;
}
buffer_len = wpath_len + 1 + wdir_len + 1 + whome_len + 1;
buffer = buffer_pos = (wchar_t *)xmalloc((buffer_len + 1) * sizeof(wchar_t));
/* add home */
if (whome_len) {
wcsncpy(buffer_pos, whome, whome_len);
buffer_pos += whome_len;
}
/* Add separator if required */
if (whome_len && wcsrchr(L"\\/:", buffer_pos[-1]) == NULL) {
buffer_pos[0] = L'\\';
buffer_pos++;
}
else if (!dir_drive && path_drive) {
*buffer_pos++ = path_drive;
*buffer_pos++ = L':';
}
if (wdir_len) {
/* tainted if dir is used and dir is tainted */
if (!tainted && OBJ_TAINTED(dir))
tainted = 1;
wcsncpy(buffer_pos, wdir_pos, wdir_len);
buffer_pos += wdir_len;
}
/* add separator if required */
if (wdir_len && wcsrchr(L"\\/:", buffer_pos[-1]) == NULL) {
buffer_pos[0] = L'\\';
buffer_pos++;
}
/* now deal with path */
if (wpath_len) {
wcsncpy(buffer_pos, wpath_pos, wpath_len);
buffer_pos += wpath_len;
}
/* GetFullPathNameW requires at least "." to determine current directory */
if (wpath_len == 0) {
buffer_pos[0] = L'.';
buffer_pos++;
}
/* Ensure buffer is NULL terminated */
buffer_pos[0] = L'\0';
/* tainted if path is relative */
if (!tainted && PathIsRelativeW(buffer) && !(buffer_len >= 2 && IS_DIR_UNC_P(buffer)))
tainted = 1;
/* FIXME: Make this more robust */
/* Determine require buffer size */
size = GetFullPathNameW(buffer, PATH_BUFFER_SIZE, wfullpath_buffer, NULL);
if (size > PATH_BUFFER_SIZE) {
/* allocate more memory than alloted originally by PATH_BUFFER_SIZE */
wfullpath = (wchar_t *)xmalloc(size * sizeof(wchar_t));
size = GetFullPathNameW(buffer, size, wfullpath, NULL);
}
else {
wfullpath = wfullpath_buffer;
}
/* Remove any trailing slashes */
if (IS_DIR_SEPARATOR_P(wfullpath[size - 1]) &&
wfullpath[size - 2] != L':' &&
!(size == 2 && IS_DIR_UNC_P(wfullpath))) {
size -= 1;
wfullpath[size] = L'\0';
}
/* Remove any trailing dot */
if (wfullpath[size - 1] == L'.') {
size -= 1;
wfullpath[size] = L'\0';
}
/* removes trailing invalid ':$DATA' */
size = remove_invalid_alternative_data(wfullpath, size);
/* Replace the trailing path to long name */
if (long_name)
size = replace_to_long_name(&wfullpath, size, (wfullpath != wfullpath_buffer));
/* sanitize backslashes with forwardslashes */
replace_wchar(wfullpath, L'\\', L'/');
/* convert to VALUE and set the path encoding */
rb_str_set_len(result, 0);
result = append_wstr(result, wfullpath, size, cp, path_cp, path_encoding);
/* makes the result object tainted if expanding tainted strings or returning modified path */
if (tainted)
OBJ_TAINT(result);
/* TODO: better cleanup */
if (buffer)
xfree(buffer);
if (wpath)
free(wpath);
if (wdir)
free(wdir);
if (whome)
xfree(whome);
if (wfullpath != wfullpath_buffer)
xfree(wfullpath);
if (fullpath)
xfree(fullpath);
rb_enc_associate(result, path_encoding);
return result;
}
