diff --git a/Include/unicodeobject.h b/Include/unicodeobject.h index 363776bdf0a..a70585c1e69 100644 --- a/Include/unicodeobject.h +++ b/Include/unicodeobject.h @@ -1742,7 +1742,7 @@ PyAPI_FUNC(PyObject*) PyUnicode_DecodeLocale( /* Encode a Unicode object to the current locale encoding. The encoder is strict is *surrogateescape* is equal to zero, otherwise the "surrogateescape" error handler is used. Return a bytes object. The string - cannot contain embedded null characters.. */ + cannot contain embedded null characters. */ PyAPI_FUNC(PyObject*) PyUnicode_EncodeLocale( PyObject *unicode, diff --git a/Lib/test/test_cmd_line_script.py b/Lib/test/test_cmd_line_script.py index f066204393a..6051e18eddc 100644 --- a/Lib/test/test_cmd_line_script.py +++ b/Lib/test/test_cmd_line_script.py @@ -367,11 +367,10 @@ class CmdLineTest(unittest.TestCase): # Mac OS X denies the creation of a file with an invalid UTF-8 name. # Windows allows to create a name with an arbitrary bytes name, but # Python cannot a undecodable bytes argument to a subprocess. - #if (support.TESTFN_UNDECODABLE - #and sys.platform not in ('win32', 'darwin')): - # name = os.fsdecode(support.TESTFN_UNDECODABLE) - #elif support.TESTFN_NONASCII: - if support.TESTFN_NONASCII: + if (support.TESTFN_UNDECODABLE + and sys.platform not in ('win32', 'darwin')): + name = os.fsdecode(support.TESTFN_UNDECODABLE) + elif support.TESTFN_NONASCII: name = support.TESTFN_NONASCII else: self.skipTest("need support.TESTFN_NONASCII") diff --git a/Misc/NEWS b/Misc/NEWS index f66354447df..da9233715f9 100644 --- a/Misc/NEWS +++ b/Misc/NEWS @@ -10,6 +10,12 @@ What's New in Python 3.4.0 Alpha 1? Core and Builtins ----------------- +- Issue #16455: On FreeBSD and Solaris, if the locale is C, the + ASCII/surrogateescape codec is now used, instead of the locale encoding, to + decode the command line arguments. This change fixes inconsistencies with + os.fsencode() and os.fsdecode() because these operating systems announces an + ASCII locale encoding, whereas the ISO-8859-1 encoding is used in practice. + - Issue #16562: Optimize dict equality testing. Patch by Serhiy Storchaka. - Issue #16588: Silence unused-but-set warnings in Python/thread_pthread diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c index dd8d7b22323..6491fdc3169 100644 --- a/Objects/unicodeobject.c +++ b/Objects/unicodeobject.c @@ -2863,10 +2863,10 @@ PyUnicode_FromEncodedObject(register PyObject *obj, /* Convert encoding to lower case and replace '_' with '-' in order to catch e.g. UTF_8. Return 0 on error (encoding is longer than lower_len-1), 1 on success. */ -static int -normalize_encoding(const char *encoding, - char *lower, - size_t lower_len) +int +_Py_normalize_encoding(const char *encoding, + char *lower, + size_t lower_len) { const char *e; char *l; @@ -2908,7 +2908,7 @@ PyUnicode_Decode(const char *s, char lower[11]; /* Enough for any encoding shortcut */ /* Shortcuts for common default encodings */ - if (normalize_encoding(encoding, lower, sizeof(lower))) { + if (_Py_normalize_encoding(encoding, lower, sizeof(lower))) { if ((strcmp(lower, "utf-8") == 0) || (strcmp(lower, "utf8") == 0)) return PyUnicode_DecodeUTF8Stateful(s, size, errors, NULL); @@ -3110,7 +3110,8 @@ locale_error_handler(const char *errors, int *surrogateescape) *surrogateescape = 0; return 0; } - if (strcmp(errors, "surrogateescape") == 0) { + if (errors == "surrogateescape" + || strcmp(errors, "surrogateescape") == 0) { *surrogateescape = 1; return 0; } @@ -3148,7 +3149,7 @@ PyUnicode_EncodeLocale(PyObject *unicode, const char *errors) } if (surrogateescape) { - /* locale encoding with surrogateescape */ + /* "surrogateescape" error handler */ char *str; str = _Py_wchar2char(wstr, &error_pos); @@ -3168,6 +3169,7 @@ PyUnicode_EncodeLocale(PyObject *unicode, const char *errors) PyMem_Free(str); } else { + /* strict mode */ size_t len, len2; len = wcstombs(NULL, wstr, 0); @@ -3273,7 +3275,7 @@ PyUnicode_AsEncodedString(PyObject *unicode, } /* Shortcuts for common default encodings */ - if (normalize_encoding(encoding, lower, sizeof(lower))) { + if (_Py_normalize_encoding(encoding, lower, sizeof(lower))) { if ((strcmp(lower, "utf-8") == 0) || (strcmp(lower, "utf8") == 0)) { @@ -3413,8 +3415,8 @@ PyUnicode_DecodeLocaleAndSize(const char *str, Py_ssize_t len, return NULL; } - if (surrogateescape) - { + if (surrogateescape) { + /* "surrogateescape" error handler */ wstr = _Py_char2wchar(str, &wlen); if (wstr == NULL) { if (wlen == (size_t)-1) @@ -3428,6 +3430,7 @@ PyUnicode_DecodeLocaleAndSize(const char *str, Py_ssize_t len, PyMem_Free(wstr); } else { + /* strict mode */ #ifndef HAVE_BROKEN_MBSTOWCS wlen = mbstowcs(NULL, str, 0); #else @@ -3447,7 +3450,6 @@ PyUnicode_DecodeLocaleAndSize(const char *str, Py_ssize_t len, return PyErr_NoMemory(); } - /* This shouldn't fail now */ wlen2 = mbstowcs(wstr, str, wlen+1); if (wlen2 == (size_t)-1) { if (wstr != smallbuf) diff --git a/Python/fileutils.c b/Python/fileutils.c index 2cd75ce1636..9c0833e003e 100644 --- a/Python/fileutils.c +++ b/Python/fileutils.c @@ -5,6 +5,7 @@ #endif #ifdef HAVE_LANGINFO_H +#include #include #endif @@ -42,7 +43,182 @@ _Py_device_encoding(int fd) Py_RETURN_NONE; } -#ifdef HAVE_STAT +#if !defined(__APPLE__) && !defined(MS_WINDOWS) +extern int _Py_normalize_encoding(const char *, char *, size_t); + +/* Workaround FreeBSD and OpenIndiana locale encoding issue with the C locale. + On these operating systems, nl_langinfo(CODESET) announces an alias of the + ASCII encoding, whereas mbstowcs() and wcstombs() functions use the + ISO-8859-1 encoding. The problem is that os.fsencode() and os.fsdecode() use + locale.getpreferredencoding() codec. For example, if command line arguments + are decoded by mbstowcs() and encoded back by os.fsencode(), we get a + UnicodeEncodeError instead of retrieving the original byte string. + + The workaround is enabled if setlocale(LC_CTYPE, NULL) returns "C", + nl_langinfo(CODESET) announces "ascii" (or an alias to ASCII), and at least + one byte in range 0x80-0xff can be decoded from the locale encoding. The + workaround is also enabled on error, for example if getting the locale + failed. + + Values of locale_is_ascii: + + 1: the workaround is used: _Py_wchar2char() uses + encode_ascii_surrogateescape() and _Py_char2wchar() uses + decode_ascii_surrogateescape() + 0: the workaround is not used: _Py_wchar2char() uses wcstombs() and + _Py_char2wchar() uses mbstowcs() + -1: unknown, need to call check_force_ascii() to get the value +*/ +static int force_ascii = -1; + +static int +check_force_ascii(void) +{ + char *loc; +#if defined(HAVE_LANGINFO_H) && defined(CODESET) + char *codeset, **alias; + char encoding[100]; + int is_ascii; + unsigned int i; + char* ascii_aliases[] = { + "ascii", + "646", + "ansi-x3.4-1968", + "ansi-x3-4-1968", + "ansi-x3.4-1986", + "cp367", + "csascii", + "ibm367", + "iso646-us", + "iso-646.irv-1991", + "iso-ir-6", + "us", + "us-ascii", + NULL + }; +#endif + + loc = setlocale(LC_CTYPE, NULL); + if (loc == NULL) + goto error; + if (strcmp(loc, "C") != 0) { + /* the LC_CTYPE locale is different than C */ + return 0; + } + +#if defined(HAVE_LANGINFO_H) && defined(CODESET) + codeset = nl_langinfo(CODESET); + if (!codeset || codeset[0] == '\0') { + /* CODESET is not set or empty */ + goto error; + } + if (!_Py_normalize_encoding(codeset, encoding, sizeof(encoding))) + goto error; + + is_ascii = 0; + for (alias=ascii_aliases; *alias != NULL; alias++) { + if (strcmp(encoding, *alias) == 0) { + is_ascii = 1; + break; + } + } + if (!is_ascii) { + /* nl_langinfo(CODESET) is not "ascii" or an alias of ASCII */ + return 0; + } + + for (i=0x80; i<0xff; i++) { + unsigned char ch; + wchar_t wch; + size_t res; + + ch = (unsigned char)i; + res = mbstowcs(&wch, (char*)&ch, 1); + if (res != (size_t)-1) { + /* decoding a non-ASCII character from the locale encoding succeed: + the locale encoding is not ASCII, force ASCII */ + return 1; + } + } + /* None of the bytes in the range 0x80-0xff can be decoded from the locale + encoding: the locale encoding is really ASCII */ + return 0; +#else + /* nl_langinfo(CODESET) is not available: always force ASCII */ + return 1; +#endif + +error: + /* if an error occured, force the ASCII encoding */ + return 1; +} + +static char* +encode_ascii_surrogateescape(const wchar_t *text, size_t *error_pos) +{ + char *result = NULL, *out; + size_t len, i; + wchar_t ch; + + if (error_pos != NULL) + *error_pos = (size_t)-1; + + len = wcslen(text); + + result = PyMem_Malloc(len + 1); /* +1 for NUL byte */ + if (result == NULL) + return NULL; + + out = result; + for (i=0; i 128. This will still roundtrip correctly in the locale's charset, which must be an ASCII superset. */ - res = PyMem_Malloc((strlen(arg)+1)*sizeof(wchar_t)); - if (!res) + res = decode_ascii_surrogateescape(arg, size); + if (res == NULL) goto oom; - in = (unsigned char*)arg; - out = res; - while(*in) - if(*in < 128) - *out++ = *in++; - else - *out++ = 0xdc00 + *in++; - *out = 0; #endif /* HAVE_MBRTOWC */ - if (size != NULL) - *size = out - res; return res; oom: if (size != NULL) @@ -236,6 +419,14 @@ _Py_wchar2char(const wchar_t *text, size_t *error_pos) size_t i, size, converted; wchar_t c, buf[2]; +#ifndef MS_WINDOWS + if (force_ascii == -1) + force_ascii = check_force_ascii(); + + if (force_ascii) + return encode_ascii_surrogateescape(text, error_pos); +#endif + /* The function works in two steps: 1. compute the length of the output buffer in bytes (size) 2. outputs the bytes */ @@ -276,7 +467,7 @@ _Py_wchar2char(const wchar_t *text, size_t *error_pos) } } if (result != NULL) { - *bytes = 0; + *bytes = '\0'; break; } @@ -320,6 +511,8 @@ _Py_wstat(const wchar_t* path, struct stat *buf) } #endif +#ifdef HAVE_STAT + /* Call _wstat() on Windows, or encode the path to the filesystem encoding and call stat() otherwise. Only fill st_mode attribute on Windows. @@ -352,6 +545,8 @@ _Py_stat(PyObject *path, struct stat *statbuf) #endif } +#endif + /* Open a file. Use _wfopen() on Windows, encode the path to the locale encoding and use fopen() otherwise. */ @@ -533,4 +728,3 @@ _Py_wgetcwd(wchar_t *buf, size_t size) #endif } -#endif