static SgObject strip_trailing_slash(SgObject path)
{
size_t s = SG_STRING_SIZE(path);
if (s == 1) return path;
if (SG_STRING_VALUE_AT(path, --s) == '/') {
while (SG_STRING_VALUE_AT(path, s-1) == '/') s--;
SG_STRING_SIZE(path) = s;
}
return path;
}
/* TODO maybe try not to use string port? */
static SgObject normalise_path(SgObject fullpath)
{
SgStringPort sp;
SgObject out = Sg_InitStringOutputPort(&sp, SG_STRING_SIZE(fullpath));
int64_t pos = 0;
int i = 0;
while (i < SG_STRING_SIZE(fullpath)) {
SgChar c = SG_STRING_VALUE_AT(fullpath, i++);
if (c == '.') {
if (i != SG_STRING_SIZE(fullpath)) {
SgChar c2 = SG_STRING_VALUE_AT(fullpath, i++);
if (c2 == '.') {
if (SG_STRING_VALUE_AT(fullpath, i) == '/' ||
SG_STRING_SIZE(fullpath) == i) {
if (pos-2 > 0) {
SgObject tmp = Sg_GetStringFromStringPort(out);
/* skip previous '.' and '/' */
pos = search_separator(tmp, pos-2);
} else {
pos = 1; /* root */
}
if (pos <= 0) pos = 1; /* root */
Sg_SetPortPosition(out, pos, SG_BEGIN);
if (pos == 1) i++;
} else {
/* ok just a file named '..?' or longer*/
Sg_PutcUnsafe(out, '.');
Sg_PutcUnsafe(out, '.');
pos += 2;
for (; i < SG_STRING_SIZE(fullpath); i++) {
SgChar c3 = SG_STRING_VALUE_AT(fullpath, i);
if (c3 != '.') break;
Sg_PutcUnsafe(out, c3);
pos++;
}
}
} else if (c2 != '/') {
Sg_PutcUnsafe(out, '.');
Sg_PutcUnsafe(out, c2);
pos += 2;
}
}
} else {
Sg_PutcUnsafe(out, c);
pos++;
}
}
return strip_trailing_slash(Sg_GetStringFromStringPort(out));
}
static int glob_match1(SgObject pat, SgObject path_element, int flags)
{
const int period = !(flags & SG_DOTMATCH);
/* Flags are taken from Ruby but I don't know what FNM_PATHNAME does on glob.
so ignore.*/
/* const int pathname = flags & SG_PATHNAME; */
int pos = 0;
SgObject cp;
if (period) {
if (SG_STRING_VALUE_AT(path_element, 0) == '.' &&
/* leading period */
!(SG_STRINGP(SG_CAR(pat)) &&
SG_STRING_VALUE_AT(SG_STRING(SG_CAR(pat)), 0) == '.')) {
return FALSE;
}
}
SG_FOR_EACH(cp, pat) {
/* the matching is pretty much simple, a rule may contain the followings:
- string
- charset
- pattern (regular expression)
these are resolved by prefix match, one char match or regex match,
respectively. */
SgObject p = SG_CAR(cp);
if (pos >= SG_STRING_SIZE(path_element)) return FALSE;
if (SG_STRINGP(p)) {
int i;
for (i = 0; i < SG_STRING_SIZE(p); i++) {
if (!SG_EQ(SG_STRING_VALUE_AT(p, i),
SG_STRING_VALUE_AT(path_element, pos++))) {
return FALSE;
}
}
} else if (SG_CHAR_SET_P(p)) {
if (!Sg_CharSetContains(p, SG_STRING_VALUE_AT(path_element, pos++))) {
return FALSE;
}
} else if (SG_PATTERNP(p)) {
SgMatcher *m = Sg_RegexTextMatcher(SG_PATTERN(p), path_element, pos,
SG_STRING_SIZE(path_element));
return Sg_RegexTextMatches(SG_TEXT_MATCHER(m));
} else {
Sg_Error(UC("[Internal] Unknown glob rule '%S' in '%S'"), p, pat);
return FALSE; /* dummy */
}
}
/* from mosh */
static const wchar_t* utf32ToUtf16(SgString *path)
{
int size = SG_STRING_SIZE(path);
SgCodec *codec = Sg_MakeUtf16Codec(UTF_16LE);
SgTranscoder *tcoder = Sg_MakeTranscoder(codec, LF, SG_REPLACE_ERROR);
SgPort *out, *tout;
SgBytePort bp;
SgTranscodedPort tp;
out = Sg_InitByteArrayOutputPort(&bp, sizeof(wchar_t) * (size + 1));
tout = Sg_InitTranscodedPort(&tp, out, tcoder, SG_OUTPUT_PORT);
Sg_TranscoderWrite(tcoder, tout, SG_STRING_VALUE(path),
SG_STRING_SIZE(path));
Sg_TranscoderPutc(tcoder, tout, '\0');
return (const wchar_t*)Sg_GetByteArrayFromBinaryPort(&bp);
}
static int next_dirsep(SgObject path, int skipped)
{
while (skipped < SG_STRING_SIZE(path) && !dirsep_p(S(path, skipped))) {
skipped++;
}
return skipped;
}
SgObject Sg_BuildPath(SgString *path, SgString *file)
{
int psize = SG_STRING_SIZE(path), fsize = SG_STRING_SIZE(file);
int i, j, offset = 1;
SgObject ret;
if (SG_STRING_VALUE_AT(path, psize-1) == '/') offset--;
ret = Sg_ReserveString(psize + fsize + offset, 0);
for (i = 0; i < psize; i++) {
SG_STRING_VALUE_AT(ret, i) = SG_STRING_VALUE_AT(path, i);
}
if (offset) {
SG_STRING_VALUE_AT(ret, i++) = '/';
}
for (j = 0; j < fsize; i++, j++) {
SG_STRING_VALUE_AT(ret, i) = SG_STRING_VALUE_AT(file, j);
}
return ret;
}
SgObject Sg_DirectoryName(SgString *path)
{
int size = SG_STRING_SIZE(path), i;
for (i = size-1; i >= 0; i--) {
if (SG_STRING_VALUE_AT(path, i) == '/') break;
}
if (i <= 0) return SG_FALSE;
return Sg_Substring(path, 0, i);
}
static SgObject remove_backslashes(SgObject path)
{
int i, j, count = 0;
SgObject r;
for (i = 0; i < SG_STRING_SIZE(path); i++) {
if (SG_STRING_VALUE_AT(path, i) != '\\') count++;
}
/* no backslash */
if (SG_STRING_SIZE(path) == count) return path;
r = Sg_ReserveString(count, '\0');
for (i = 0, j = 0; i < SG_STRING_SIZE(path); i++) {
if (SG_STRING_VALUE_AT(path, i) != '\\') {
SG_STRING_VALUE_AT(r, j++) = SG_STRING_VALUE_AT(path, i);
}
}
return r;
}
/*
TODO Should we skip?
*/
static int detect_prefix(SgObject path)
{
/* network address or so e.g. \\foo\bar */
if (dirsep_p(S(path,0)) && dirsep_p(S(path,1))) {
int skipped = 2;
while (dirsep_p(S(path, skipped))) {
skipped++;
}
if ((skipped = next_dirsep(path, skipped)) < SG_STRING_SIZE(path) &&
skipped+1 < SG_STRING_SIZE(path) && !dirsep_p(S(path, skipped+1))) {
skipped = next_dirsep(path, skipped+1);
}
return skipped;
}
if (has_drive_letter(path)) {
return 2;
}
return 0;
}
/*
converts given path template to pattern
e.g.)
- "foo/bar/\*" -> (("foo") ("bar") (ANY))
- "foo/bar/buz*" -> (("foo") ("bar") ("buz" ANY))
- "foo/bar/[b][u]z*" -> (("foo") ("bar") ([b] [u] "z" ANY))
each element of the list represents a matching rule of path element.
*/
static int find_close_bracket(SgString *path, int start, int flags)
{
const int escape = !(flags & SG_NOESCAPE);
int i;
for (i = start; i < SG_STRING_SIZE(path); i++) {
switch (SG_STRING_VALUE_AT(path, i)) {
case ']': return i;
case '\\':
if (escape) i++;
break;
}
}
return start;
}
static SgObject glob_make_pattern(SgString *path, int flags)
{
const int escape = !(flags & SG_NOESCAPE);
SgObject h = SG_NIL, t = SG_NIL, h1 = SG_NIL, t1 = SG_NIL;
int i, start;
#define emit() \
do { \
if (start != i) { \
SgObject tmp = Sg_Substring(path, start, i); \
if (escape) tmp = remove_backslashes(tmp); \
SG_APPEND1(h1, t1, tmp); \
} \
start = i+1; \
} while (0)
for (i = 0, start = 0; i < SG_STRING_SIZE(path);) {
SgChar c = SG_STRING_VALUE_AT(path, i);
switch (c) {
case '[': {
int s = i, e;
e = find_close_bracket(path, start, flags);
if (s != e) {
emit();
SG_APPEND1(h1, t1, Sg_ParseCharSetString(path, FALSE, s, i=++e));
start = i;
}
i++;
} break;
case '/':
/* next */
emit();
/* if the path starts with '/', then this can be null */
if (!SG_NULLP(h1)) {
SG_APPEND1(h, t, convert_star(h1));
}
h1 = t1 = SG_NIL; /* reset it */
/* this need to be updated */
start = ++i;
break;
case '*': {
int has = (start != i);
emit();
/* merge it if it's there */
if (!has && SG_STRING_SIZE(path) - i >= 3 &&
SG_STRING_VALUE_AT(path, i+1) == '*' &&
SG_STRING_VALUE_AT(path, i+2) == '/') {
do {
i += 3;
/* skip '/' */
while (SG_STRING_VALUE_AT(path, i) == '/') i++;
} while (SG_STRING_VALUE_AT(path, i) == '*' &&
SG_STRING_VALUE_AT(path, i+1) == '*' &&
SG_STRING_VALUE_AT(path, i+2) == '/');
SG_APPEND1(h1, t1, STAR_SLASH);
SG_APPEND1(h, t, h1);
h1 = t1 = SG_NIL; /* reset it */
start = i;
} else {
SG_APPEND1(h1, t1, STAR);
while (SG_STRING_VALUE_AT(path, i) == '*') i++;
}
break;
}
case '?':
emit();
SG_APPEND1(h1, t1, FULL_CHARSET);
i++;
break;
default:
i++;
break;
}
}
emit();
if (!SG_NULLP(h1)) {
SG_APPEND1(h, t, convert_star(h1));
SG_APPEND1(h, t, SG_LIST1(ANY));
} else {
SG_APPEND1(h, t, SG_LIST1(DIR));
}
#undef emit
return h;
}
static SgObject brace_expand(SgString *str, int flags)
{
const int escape = !(flags & SG_NOESCAPE);
int lbrace = 0, rbrace = 0, nest = 0, i;
int haslb = FALSE, hasrb = FALSE;
/* find { and }*/
for (i = 0; i < SG_STRING_SIZE(str); i++) {
if (SG_STRING_VALUE_AT(str, i) == '{' && nest++ == 0) {
lbrace = i;
haslb = TRUE;
}
if (SG_STRING_VALUE_AT(str, i) == '}' && --nest == 0) {
rbrace = i;
hasrb = TRUE;
break;
}
if (SG_STRING_VALUE_AT(str, i) == '\\' && escape) {
if (++i == SG_STRING_SIZE(str)) break;
}
}
/* make "foo/{a,b}" to ("foo/a" "foo/b") */
if (haslb && hasrb) {
SgObject h = SG_NIL, t = SG_NIL;
SgPort *out;
SgStringPort tp;
int i;
/* copy value until the first '{' */
out = Sg_InitStringOutputPort(&tp, 255);
for (i = 0; i < lbrace; i++) {
Sg_PutcUnsafe(out, SG_STRING_VALUE_AT(str, i));
}
/* skip '{' */
i++;
while (i < rbrace) {
/* now we need to copy one by one */
int nest = 0, j;
SgObject tmp;
for (;SG_STRING_VALUE_AT(str, i) != ',' || nest != 0; i++) {
if (i >= rbrace) break;
if (SG_STRING_VALUE_AT(str, i) == '{') nest++;
if (SG_STRING_VALUE_AT(str, i) == '}') nest--;
if (SG_STRING_VALUE_AT(str, i) == '\\' && escape) {
if (++i == rbrace) break;
}
Sg_PutcUnsafe(out, SG_STRING_VALUE_AT(str, i));
}
/* skip ',' */
i++;
/* copy after the '}' */
for (j = rbrace+1; j < SG_STRING_SIZE(str); j++) {
Sg_PutcUnsafe(out, SG_STRING_VALUE_AT(str, j));
}
tmp = Sg_GetStringFromStringPort(&tp);
SG_APPEND(h, t, brace_expand(tmp, flags));
/* back to the starting position */
Sg_SetPortPosition(out, lbrace, SG_BEGIN);
}
SG_CLEAN_STRING_PORT(&tp);
return h;
} else {
return SG_LIST1(str);
}
}
