[ruby-dev:22431] Re: Dir.glob と Shjift_JIS について
From:
"H.Yamamoto" <ocean@...2.ccsnet.ne.jp>
Date:
2003-12-26 09:33:15 UTC
List:
ruby-dev #22431
こんにちは、山本です。
ruby1.8.1のリリースおめでとうございます。
最新のdir.cに合わせて、パッチを作ってみました。
ruby1.8.1との差異は、
1. マルチバイト対応
2. puts Dir.glob('a.cpp/')
1.8.1 #=> 'a.cpp/' （ファイルなのに、/で終わるパターンにマッチする)
patch #=> ''
3. glob_helper()でstat()に渡すまえに末尾の'/'（ルート以外）を必ずしも除去していなかった点を改善した。
また、glob_helper()を単純にするためいくつかの仮定をしました。問題があったら戻します。
1. stat("e:")が成功すると仮定。マクロNEED_DOT()を除去した。
bcb付属stat()ではstat("e:.")とする必要がある。
2. opendir()をディレクトリ以外のエントリに対して呼ぶとENOTDIRを返すと仮定。stat + opendir の部分を opendir 単独に変更した。（やりすぎ？）
bcb付属opendir()はEINVALを返す。
bcbの場合はruby組み込みのstat() & opendir()を使うので、結局問題ないのですが・・・
--- dir.c-1.8.1 Sat Dec 20 02:20:58 2003
+++ dir.c Fri Dec 26 17:53:32 2003
@@ -78,11 +78,61 @@ char *strchr _((char*,char));
#define downcase(c) (nocase && ISUPPER(c) ? tolower(c) : (c))
+#define compare(c1, c2) (((unsigned char)(c1)) - ((unsigned char)(c2)))
-#ifndef CharNext /* defined as CharNext[AW] on Windows. */
-# if defined(DJGPP)
-# define CharNext(p) ((p) + mblen(p, MB_CUR_MAX))
-# else
-# define CharNext(p) ((p) + 1)
-# endif
+/* caution: in case *p == '\0'
+ Next(p) == p + 1 in single byte environment
+ Next(p) == p in multi byte environment
+*/
+#if defined(CharNext)
+# define Next(p) CharNext(p)
+#elif defined(DJGPP)
+# define Next(p) ((p) + mblen(p, MB_CUR_MAX))
+#elif defined(__EMX__)
+# define Next(p) ((p) + emx_mblen(p))
+static inline int
+emx_mblen(p)
+ const char *p;
+{
+ int n = mblen(p, INT_MAX);
+ return (n < 0) ? 1 : n;
+}
#endif
+#ifndef Next /* single byte environment */
+# define Next(p) ((p) + 1)
+# define Inc(p) (++(p))
+# define Compare(p1, p2) (compare(downcase(*(p1)), downcase(*(p2))))
+#else /* multi byte environment */
+# define Inc(p) ((p) = Next(p))
+# define Compare(p1, p2) (CompareImpl(p1, p2, nocase))
+static int
+CompareImpl(p1, p2, nocase)
+ const char *p1;
+ const char *p2;
+ int nocase;
+{
+ const int len1 = Next(p1) - p1;
+ const int len2 = Next(p2) - p2;
+
+ if (len1 == 0) return len2;
+ if (len2 == 0) return -len1;
+
+ if (len1 == 1)
+ if (len2 == 1)
+ return compare(downcase(*p1), downcase(*p2));
+ else {
+ const int ret = compare(downcase(*p1), *p2);
+ return ret ? ret : -1;
+ }
+ else
+ if (len2 == 1) {
+ const int ret = compare(*p1, downcase(*p2));
+ return ret ? ret : 1;
+ }
+ else {
+ const int ret = memcmp(p1, p2, len1 < len2 ? len1 : len2);
+ return ret ? ret : len1 - len2;
+ }
+}
+#endif /* environment */
+
#if defined DOSISH
@@ -96,3 +146,3 @@ range(pat, test, flags)
char *pat;
- char test;
+ char *test;
int flags;
@@ -107,11 +157,10 @@ range(pat, test, flags)
- test = downcase(test);
-
while (*pat) {
- int cstart, cend;
- cstart = cend = *pat++;
- if (cstart == ']')
- return ok == not ? 0 : pat;
- else if (escape && cstart == '\\')
- cstart = cend = *pat++;
+ char *pstart, *pend;
+ pstart = pend = pat;
+ if (*pstart == ']')
+ return ok == not ? 0 : ++pat;
+ else if (escape && *pstart == '\\')
+ pstart = pend = ++pat;
+ Inc(pat);
if (*pat == '-' && pat[1] != ']') {
@@ -119,8 +168,8 @@ range(pat, test, flags)
pat++;
- cend = pat[1];
- if (!cend)
+ pend = pat+1;
+ if (!*pend)
return 0;
- pat += 2;
+ pat = Next(pend);
}
- if (downcase(cstart) <= test && test <= downcase(cend))
+ if (Compare(pstart, test) <= 0 && Compare(test, pend) <= 0)
ok = 1;
@@ -131,4 +180,5 @@ range(pat, test, flags)
#define ISDIRSEP(c) (pathname && isdirsep(c))
-#define PERIOD(s) (period && *(s) == '.' && \
- ((s) == string || ISDIRSEP((s)[-1])))
+#define PERIOD_S() (period && *s == '.' && \
+ (!s_prev || ISDIRSEP(*s_prev)))
+#define INC_S() (s = Next(s_prev = s))
static int
@@ -140,4 +190,4 @@ fnmatch(pat, string, flags)
int c;
- int test;
- const char *s = string;
+ const char *test;
+ const char *s = string, *s_prev = 0;
int escape = !(flags & FNM_NOESCAPE);
@@ -147,14 +197,15 @@ fnmatch(pat, string, flags)
- while (c = *pat++) {
+ while (c = *pat) {
switch (c) {
case '?':
- if (!*s || ISDIRSEP(*s) || PERIOD(s))
+ if (!*s || ISDIRSEP(*s) || PERIOD_S())
return FNM_NOMATCH;
- s++;
+ INC_S();
+ ++pat;
break;
case '*':
- while ((c = *pat++) == '*')
+ while ((c = *++pat) == '*')
;
- if (PERIOD(s))
+ if (PERIOD_S())
return FNM_NOMATCH;
@@ -170,3 +221,3 @@ fnmatch(pat, string, flags)
if (*s) {
- s++;
+ INC_S();
break;
@@ -176,7 +227,5 @@ fnmatch(pat, string, flags)
- test = escape && c == '\\' ? *pat : c;
- test = downcase(test);
- pat--;
+ test = escape && c == '\\' ? pat+1 : pat;
while (*s) {
- if ((c == '[' || downcase(*s) == test) &&
+ if ((c == '[' || Compare(s, test) == 0) &&
!fnmatch(pat, s, flags | FNM_DOTMATCH))
@@ -185,3 +234,3 @@ fnmatch(pat, string, flags)
break;
- s++;
+ INC_S();
}
@@ -190,8 +239,8 @@ fnmatch(pat, string, flags)
case '[':
- if (!*s || ISDIRSEP(*s) || PERIOD(s))
+ if (!*s || ISDIRSEP(*s) || PERIOD_S())
return FNM_NOMATCH;
- pat = range(pat, *s, flags);
+ pat = range(pat+1, s, flags);
if (!pat)
return FNM_NOMATCH;
- s++;
+ INC_S();
break;
@@ -199,12 +248,8 @@ fnmatch(pat, string, flags)
case '\\':
- if (escape
+ if (escape && pat[1]
#if defined DOSISH
- && *pat && strchr("*?[\\", *pat)
+ && strchr("*?[\\", pat[1])
#endif
) {
- c = *pat;
- if (!c)
- c = '\\';
- else
- pat++;
+ c = *++pat;
}
@@ -218,5 +263,6 @@ fnmatch(pat, string, flags)
#endif
- if(downcase(c) != downcase(*s))
+ if(Compare(pat, s) != 0)
return FNM_NOMATCH;
- s++;
+ INC_S();
+ Inc(pat);
break;
@@ -770,7 +816,7 @@ dir_s_rmdir(obj, dir)
static int
-has_magic(s, send, flags)
- char *s, *send;
+has_magic(p, m, flags)
+ register char *p;
+ char **m;
int flags;
{
- register char *p = s;
register char c;
@@ -779,3 +825,3 @@ has_magic(s, send, flags)
- while ((c = *p++) != '\0') {
+ while (c = *p++, c != '\0' && c != '/') {
switch (c) {
@@ -783,3 +829,3 @@ has_magic(s, send, flags)
case '*':
- return Qtrue;
+ goto found;
@@ -790,3 +836,3 @@ has_magic(s, send, flags)
if (open)
- return Qtrue;
+ goto found;
continue;
@@ -794,61 +840,68 @@ has_magic(s, send, flags)
case '\\':
- if (escape && *p++ == '\0')
- return Qfalse;
+ if (escape && (c = *p++, c == '\0' || c == '/'))
+ goto miss;
+ continue;
}
- if (send && p >= send) break;
+ p = Next(p-1);
}
- return Qfalse;
+
+ miss:
+ *m = p-1;
+ return 0;
+
+ found:
+ while (*p != '\0' && *p != '/')
+ Inc(p);
+ *m = p;
+ return 1;
}
-static char*
-extract_path(p, pend)
- char *p, *pend;
+static int
+remove_backslashes(p, pend)
+ char *p;
+ char *pend;
{
- char *alloc;
- int len;
+ char *t = p;
+ char *s = p;
+ int n = 0;
- len = pend - p;
- alloc = ALLOC_N(char, len+1);
- memcpy(alloc, p, len);
- if (len > 1 && pend[-1] == '/'
-#if defined DOSISH_DRIVE_LETTER
- && pend[-2] != ':'
-#endif
- ) {
- alloc[len-1] = 0;
+ while (*p && p < pend) {
+ if (*p == '\\') {
+ if (t != s) {
+ memmove(t, s, p - s);
+ n++;
}
- else {
- alloc[len] = 0;
+ t += p - s;
+ s = ++p;
+ if (!(*p && p < pend)) break;
}
-
- return alloc;
+ Inc(p);
}
-static char*
-extract_elem(path)
- char *path;
-{
- char *pend;
+ while (*p++);
- pend = strchr(path, '/');
- if (!pend) pend = path + strlen(path);
+ if (t != s) {
+ memmove(t, s, p - s); /* move '\0' too */
+ n++;
+ }
- return extract_path(path, pend);
+ return n;
}
-static void
-remove_backslashes(p)
+static int
+fnmatch_for_substr(p, pend, string, flags)
char *p;
+ char *pend;
+ const char *string;
+ int flags;
{
- char *pend = p + strlen(p);
- char *t = p;
+ int ret;
+ char c;
- while (p < pend) {
- if (*p == '\\') {
- if (++p == pend) break;
- }
- *t++ = *p++;
- }
- *t = '\0';
+ c = *pend;
+ *pend = '\0'; /* should I allocate new string? */
+ ret = fnmatch(p, string, flags);
+ *pend = c;
+ return ret;
}
@@ -894,5 +947,6 @@ glob_call_func(func, path, arg)
static int
-glob_helper(path, sub, flags, func, arg)
+glob_helper(path, sub, separator, flags, func, arg) /* if separator p[-1] is removable '/' */
char *path;
char *sub;
+ int separator;
int flags;
@@ -902,29 +956,4 @@ glob_helper(path, sub, flags, func, arg)
struct stat st;
- char *p, *m;
int status = 0;
-
- p = sub ? sub : path;
- if (!has_magic(p, 0, flags)) {
-#if defined DOSISH
- remove_backslashes(path);
-#else
- if (!(flags & FNM_NOESCAPE)) remove_backslashes(p);
-#endif
- if (lstat(path, &st) == 0) {
- status = glob_call_func(func, path, arg);
- if (status) return status;
- }
- else if (errno != ENOENT) {
- /* In case stat error is other than ENOENT and
- we may want to know what is wrong. */
- rb_sys_warning(path);
- }
- return 0;
- }
-
- while (p && !status) {
- if (*p == '/') p++;
- m = strchr(p, '/');
- if (has_magic(p, m, flags)) {
- char *dir, *base, *magic, *buf;
+ char *p = sub, *m, *buf;
DIR *dirp;
@@ -938,22 +967,68 @@ glob_helper(path, sub, flags, func, arg)
- base = extract_path(path, p);
- if (path == p) dir = ".";
- else dir = base;
-
- magic = extract_elem(p);
- if (stat(dir, &st) < 0) {
- if (errno != ENOENT) rb_sys_warning(dir);
- free(base);
- free(magic);
- break;
+ while (*p && !has_magic(p, &m, flags)) {
+ if (*m == '/') {
+ separator = 1;
+ p = m + 1;
}
- if (S_ISDIR(st.st_mode)) {
- if (m && strcmp(magic, "**") == 0) {
- int n = strlen(base);
+ else {
+ separator = 0;
+ p = m;
+ }
+ }
+
+ if (!(flags & FNM_NOESCAPE)) {
+ int n = remove_backslashes(sub, p);
+ p -= n;
+ m -= n;
+ }
+
+ if (*p == '\0') { /* magic not found */
+ if (!separator) {
+ if (lstat(path, &st) < 0) {
+ /* In case stat error is other than ENOENT and
+ we may want to know what is wrong. */
+ if (errno != ENOENT) rb_sys_warning(path);
+ }
+ else {
+ status = glob_call_func(func, path, arg);
+ }
+ }
+ else {
+ const char c = p[-1];
+ p[-1] = '\0';
+ if (lstat(path, &st) < 0) {
+ if (errno != ENOENT) rb_sys_warning(path);
+ p[-1] = c;
+ }
+ else if (S_ISDIR(st.st_mode)) {
+ p[-1] = c;
+ status = glob_call_func(func, path, arg);
+ }
+ }
+ return status;
+ }
+
+ if (p[0] == '*' && p[1] == '*' && p[2] == '/') {
+ const int n = p - path;
recursive = 1;
- buf = ALLOC_N(char, n+strlen(m)+3);
- sprintf(buf, "%s%s", base, *base ? m : m+1);
- status = glob_helper(buf, buf+n, flags, func, arg);
+ buf = ALLOC_N(char, n+strlen(p+3)+1);
+ memcpy(buf, path, n);
+ strcpy(buf+n, p+3);
+ status = glob_helper(buf, buf+n, separator, flags, func, arg);
free(buf);
- if (status) goto finalize;
+ if (status) return status;
+ }
+
+ {
+ char *dir;
+ if (path == p) {
+ dir = ALLOC_N(char, 2);
+ dir[0] = '.';
+ dir[1] = '\0';
+ }
+ else {
+ const int n = separator ? (p - path) - 1 : (p - path);
+ dir = ALLOC_N(char, n+1);
+ memcpy(dir, path, n);
+ dir[n] = '\0';
}
@@ -961,21 +1036,11 @@ glob_helper(path, sub, flags, func, arg)
if (dirp == NULL) {
- rb_sys_warning(dir);
- free(base);
- free(magic);
- break;
- }
+ if (errno != ENOENT && errno != ENOTDIR) rb_sys_warning(dir);
+ free(dir);
+ return 0;
}
- else {
- free(base);
- free(magic);
- break;
+ free(dir);
}
-#if defined DOSISH_DRIVE_LETTER
-#define BASE (*base && !((isdirsep(*base) && !base[1]) || (base[1] == ':' && isdirsep(base[2]) && !base[3])))
-#else
-#define BASE (*base && !(isdirsep(*base) && !base[1]))
-#endif
-
for (dp = readdir(dirp); dp != NULL; dp = readdir(dirp)) {
+ const int n = p - path;
if (recursive) {
@@ -983,4 +1048,5 @@ glob_helper(path, sub, flags, func, arg)
continue;
- buf = ALLOC_N(char, strlen(base)+NAMLEN(dp)+strlen(m)+6);
- sprintf(buf, "%s%s%s", base, (BASE) ? "/" : "", dp->d_name);
+ buf = ALLOC_N(char, n+NAMLEN(dp)+strlen(m)+3+1);
+ memcpy(buf, path, n);
+ strcpy(buf+n, dp->d_name);
if (lstat(buf, &st) < 0) {
@@ -992,5 +1058,5 @@ glob_helper(path, sub, flags, func, arg)
char *t = buf+strlen(buf);
- strcpy(t, "/**");
+ memcpy(t, "/**", 3);
strcpy(t+3, m);
- status = glob_helper(buf, t, flags, func, arg);
+ status = glob_helper(buf, t+1, 1, flags, func, arg);
free(buf);
@@ -1002,6 +1068,7 @@ glob_helper(path, sub, flags, func, arg)
}
- if (fnmatch(magic, dp->d_name, flags) == 0) {
- buf = ALLOC_N(char, strlen(base)+NAMLEN(dp)+2);
- sprintf(buf, "%s%s%s", base, (BASE) ? "/" : "", dp->d_name);
- if (!m) {
+ if (fnmatch_for_substr(p, m, dp->d_name, flags) == 0) {
+ buf = ALLOC_N(char, n+NAMLEN(dp)+1);
+ memcpy(buf, path, n);
+ strcpy(buf+n, dp->d_name);
+ if (*m == '\0') {
status = glob_call_func(func, buf, arg);
@@ -1018,7 +1085,3 @@ glob_helper(path, sub, flags, func, arg)
closedir(dirp);
- finalize:
*tail = 0;
- free(base);
- free(magic);
- if (link) {
while (link) {
@@ -1027,9 +1090,8 @@ glob_helper(path, sub, flags, func, arg)
if (S_ISDIR(st.st_mode)) {
- int len = strlen(link->path);
- int mlen = strlen(m);
- char *t = ALLOC_N(char, len+mlen+1);
-
- sprintf(t, "%s%s", link->path, m);
- status = glob_helper(t, t+len, flags, func, arg);
- free(t);
+ const int len = strlen(link->path);
+ buf = ALLOC_N(char, len+strlen(m)+1);
+ memcpy(buf, link->path, len);
+ strcpy(buf+len, m);
+ status = glob_helper(buf, buf+len+1, 1, flags, func, arg);
+ free(buf);
}
@@ -1045,7 +1107,2 @@ glob_helper(path, sub, flags, func, arg)
}
- break;
- }
- }
- p = m;
- }
return status;
@@ -1060,3 +1117,13 @@ rb_glob2(path, flags, func, arg)
{
- int status = glob_helper(path, 0, flags, func, arg);
+ char *root = path;
+ int status;
+
+#if defined DOSISH
+ flags |= FNM_CASEFOLD;
+ root = rb_path_skip_prefix(root);
+#endif
+
+ if (*root == '/') root++;
+
+ status = glob_helper(path, root, 0, flags, func, arg);
if (status) rb_jump_tag(status);
@@ -1124,3 +1191,3 @@ push_braces(ary, s, flags)
}
- p++;
+ Inc(p);
}
@@ -1132,3 +1199,3 @@ push_braces(ary, s, flags)
}
- p++;
+ Inc(p);
}
@@ -1142,9 +1209,9 @@ push_braces(ary, s, flags)
while (*p != '}') {
- t = p + 1;
- for (p = t; *p!='}' && *p!=','; p++) {
+ t = Next(p);
+ for (p = t; *p!='}' && *p!=','; Inc(p)) {
/* skip inner braces */
- if (*p == '{') while (*p!='}') p++;
+ if (*p == '{') while (*p!='}') Inc(p);
}
memcpy(b, t, p-t);
- strcpy(b+(p-t), rbrace+1);
+ strcpy(b+(p-t), Next(rbrace));
push_braces(ary, buf, flags);
@@ -1159,3 +1226,2 @@ push_braces(ary, s, flags)
#define isdelim(c) ((c)=='\0')
-
static VALUE
@@ -1184,5 +1250,5 @@ rb_push_glob(str, flags)
while (p < pend) {
- t = buf;
nest = maxnest = 0;
while (p < pend && isdelim(*p)) p++;
+ t = p;
while (p < pend && !isdelim(*p)) {
@@ -1191,8 +1257,9 @@ rb_push_glob(str, flags)
if (!noescape && *p == '\\') {
- *t++ = *p++;
- if (p == pend) break;
+ p++;
+ if (p == pend || isdelim(*p)) break;
}
- *t++ = *p++;
+ p = Next(p);
}
- *t = '\0';
+ memcpy(buf, t, p - t);
+ buf[p - t] = '\0';
if (maxnest == 0) {