summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorJeremy Allison <jra@samba.org>2002-11-27 21:01:26 +0000
committerJeremy Allison <jra@samba.org>2002-11-27 21:01:26 +0000
commit5767dbbdcde61dd2e49f4138ce096827c8ac948a (patch)
tree23e5654a6b42383cc808ad9a6285c7751e94b2fa
parent296c93923c643fbbf1720af2f647f29aea708bd5 (diff)
downloadsamba-5767dbbdcde61dd2e49f4138ce096827c8ac948a.tar.gz
samba-5767dbbdcde61dd2e49f4138ce096827c8ac948a.tar.bz2
samba-5767dbbdcde61dd2e49f4138ce096827c8ac948a.zip
Change default codepage to UNIX.
Jeremy. (This used to be commit 537cd78268abae8005de73c7bfd9ec24865da654)
-rw-r--r--source3/python/py_tdbpack.c88
1 files changed, 54 insertions, 34 deletions
diff --git a/source3/python/py_tdbpack.c b/source3/python/py_tdbpack.c
index d8c3d46773..d70937e8e0 100644
--- a/source3/python/py_tdbpack.c
+++ b/source3/python/py_tdbpack.c
@@ -28,7 +28,9 @@
#include "Python.h"
static PyObject * pytdbpack_number(char ch, PyObject *val_iter, PyObject *packed_list);
-static PyObject * pytdbpack_str_850(PyObject *val_iter, PyObject *packed_list);
+static PyObject * pytdbpack_str(char ch,
+ PyObject *val_iter, PyObject *packed_list,
+ const char *encoding);
static PyObject * pytdbpack_buffer(PyObject *val_iter, PyObject *packed_list);
static PyObject *pytdbunpack_item(char, char **pbuf, int *plen, PyObject *);
@@ -37,6 +39,9 @@ static PyObject *pytdbpack_data(const char *format_str,
PyObject *val_seq,
PyObject *val_list);
+static PyObject *
+pytdbunpack_string(char **pbuf, int *plen, const char *encoding);
+
static void pack_le_uint32(unsigned long val_long, unsigned char *pbuf);
@@ -58,19 +63,17 @@ tdb/tdbutil module, with appropriate adjustments for Python datatypes.
Python strings are used to specify the format of data to be packed or
unpacked.
-Strings are always stored in codepage 850. Unicode objects are translated
-to cp850; plain strings are assumed to be in latin-1 and are also
-translated.
-
-This may be a problem in the future if it is different to the Samba codepage.
-It might be better to have the caller do the conversion, but that would conflict
-with existing CMI code.
+String encodings are implied by the database format: they may be either DOS
+codepage (currently hardcoded to 850), or Unix codepage (currently hardcoded
+to be the same as the default Python encoding).
tdbpack format strings:
- 'f': NULL-terminated string in codepage 850
+ 'f': NUL-terminated string in codepage iso8859-1
+
+ 'P': same as 'f'
- 'P': same as 'f'
+ 'F': NUL-terminated string in iso-8859-1
'd': 4 byte little-endian unsigned number
@@ -145,7 +148,11 @@ notes:
";
-const char *pytdb_string_encoding = "cp850";
+const char *pytdb_dos_encoding = "cp850";
+
+/* NULL, meaning that the Samba default encoding *must* be the same as the
+ Python default encoding. */
+const char *pytdb_unix_encoding = NULL;
/*
@@ -228,7 +235,7 @@ pytdbpack_data(const char *format_str,
case 'f':
case 'P':
- if (!(packed_list = pytdbpack_str_850(val_iter, packed_list)))
+ if (!(packed_list = pytdbpack_str(ch, val_iter, packed_list, pytdb_unix_encoding)))
return NULL;
break;
@@ -287,27 +294,29 @@ pytdbpack_number(char ch, PyObject *val_iter, PyObject *packed_list)
/*
- * Take one string from the iterator val_iter, convert it to 8-bit CP850, and
- * return it.
+ * Take one string from the iterator val_iter, convert it to 8-bit, and return
+ * it.
*
* If the input is neither a string nor Unicode, an exception is raised.
*
- * If the input is Unicode, then it is converted to CP850.
+ * If the input is Unicode, then it is converted to the appropriate encoding.
*
- * If the input is a String, then it is converted to Unicode using the default
- * decoding method, and then converted to CP850. This in effect gives
- * conversion from latin-1 (currently the PSA's default) to CP850, without
- * needing a custom translation table.
+ * If the input is a String, and encoding is not null, then it is converted to
+ * Unicode using the default decoding method, and then converted to the
+ * encoding. If the encoding is NULL, then the string is written out as-is --
+ * this is used when the default Python encoding is the same as the Samba
+ * encoding.
*
* I hope this approach avoids being too fragile w.r.t. being passed either
* Unicode or String objects.
*/
static PyObject *
-pytdbpack_str_850(PyObject *val_iter, PyObject *packed_list)
+pytdbpack_str(char ch,
+ PyObject *val_iter, PyObject *packed_list, const char *encoding)
{
PyObject *val_obj = NULL;
PyObject *unicode_obj = NULL;
- PyObject *cp850_str = NULL;
+ PyObject *coded_str = NULL;
PyObject *nul_str = NULL;
PyObject *new_list = NULL;
@@ -315,31 +324,41 @@ pytdbpack_str_850(PyObject *val_iter, PyObject *packed_list)
goto out;
if (PyUnicode_Check(val_obj)) {
- unicode_obj = val_obj;
+ if (!(coded_str = PyUnicode_AsEncodedString(val_obj, encoding, NULL)))
+ goto out;
}
- else {
- /* string */
+ else if (PyString_Check(val_obj) && !encoding) {
+ /* For efficiency, we assume that the Python interpreter has
+ the same default string encoding as Samba's native string
+ encoding. On the PSA, both are always 8859-1. */
+ coded_str = val_obj;
+ Py_INCREF(coded_str);
+ }
+ else if (PyString_Check(val_obj)) {
+ /* String, but needs to be converted */
if (!(unicode_obj = PyString_AsDecodedObject(val_obj, NULL, NULL)))
goto out;
- Py_XDECREF(val_obj);
- val_obj = NULL;
+ if (!(coded_str = PyUnicode_AsEncodedString(unicode_obj, encoding, NULL)))
+ goto out;
}
-
- if (!(cp850_str = PyUnicode_AsEncodedString(unicode_obj, pytdb_string_encoding, NULL)))
+ else {
+ pytdbpack_bad_type(ch, "String or Unicode", val_obj);
goto out;
+ }
if (!nul_str)
/* this is constant and often-used; hold it forever */
if (!(nul_str = PyString_FromStringAndSize("", 1)))
goto out;
- if ((PyList_Append(packed_list, cp850_str) != -1)
+ if ((PyList_Append(packed_list, coded_str) != -1)
&& (PyList_Append(packed_list, nul_str) != -1))
new_list = packed_list;
out:
+ Py_XDECREF(val_obj);
Py_XDECREF(unicode_obj);
- Py_XDECREF(cp850_str);
+ Py_XDECREF(coded_str);
return new_list;
}
@@ -361,7 +380,8 @@ pytdbpack_buffer(PyObject *val_iter, PyObject *packed_list)
if (!(packed_list = pytdbpack_number('d', val_iter, packed_list)))
return NULL;
- /* this assumes that the string is the right length; the old code did the same. */
+ /* this assumes that the string is the right length; the old code did
+ the same. */
if (!(val_obj = PyIter_Next(val_iter)))
return NULL;
@@ -537,7 +557,7 @@ static PyObject *pytdbunpack_int16(char **pbuf, int *plen)
static PyObject *
-pytdbunpack_string(char **pbuf, int *plen)
+pytdbunpack_string(char **pbuf, int *plen, const char *encoding)
{
int len;
char *nul_ptr, *start;
@@ -555,7 +575,7 @@ pytdbunpack_string(char **pbuf, int *plen)
*pbuf += len + 1; /* skip \0 */
*plen -= len + 1;
- return PyString_Decode(start, len, pytdb_string_encoding, NULL);
+ return PyString_Decode(start, len, encoding, NULL);
}
@@ -641,7 +661,7 @@ static PyObject *pytdbunpack_item(char ch,
result = pytdbunpack_uint32(pbuf, plen);
}
else if (ch == 'f' || ch == 'P') { /* nul-term string */
- result = pytdbunpack_string(pbuf, plen);
+ result = pytdbunpack_string(pbuf, plen, pytdb_unix_encoding);
}
else if (ch == 'B') { /* length, buffer */
return pytdbunpack_buffer(pbuf, plen, val_list);