1 |
|
/* Copyright (c) 2001, 2002, 2007 by Intevation GmbH |
2 |
|
* Authors: |
3 |
|
* Bram de Greve <[email protected]> |
4 |
|
* Bernhard Herzog <[email protected]> |
5 |
|
* |
6 |
|
* This program is free software under the GPL (>=v2) |
7 |
|
* Read the file COPYING coming with Thuban for details. |
8 |
|
*/ |
9 |
|
|
10 |
#include "pyshapelib_common.h" |
#include "pyshapelib_common.h" |
11 |
|
|
12 |
|
/* UNICODE & LANGUAGE DRIVER SUPPORT FOR DBFLIB |
13 |
|
* |
14 |
|
* When writing Unicode objects to a dbflib database, the unicode has to be |
15 |
|
* encoded in 8-bit characters using a code page. This code page is indentified |
16 |
|
* by the Language Driver ID (LDID) field in the database header. |
17 |
|
* |
18 |
|
* At this moment, this need unofficial modifications of the maptools shapelib |
19 |
|
* library because they do not read the LDID. No patch has been submitted yet, |
20 |
|
* but the version contained in the Thuban source tree incorporates the required |
21 |
|
* modifications. |
22 |
|
* |
23 |
|
* pyshapelib is designed to compile with either the patched or unpatched shapelib |
24 |
|
* by defining HAVE_LANGUAGE_DRIVER as true or false respectively. In the latter |
25 |
|
* case, a Windows ANSI code page (cp1252) is assumed |
26 |
|
*/ |
27 |
|
#if HAVE_LANGUAGE_DRIVER |
28 |
|
|
29 |
|
#define PYSHAPELIB_NUM_LANGUAGE_DRIVERS 256 |
30 |
|
|
31 |
|
#define PYSHAPELIB_ADD_LANGUAGE_DRIVER(ldid, codec, name)\ |
32 |
|
codecs[ldid] = codec;\ |
33 |
|
drivers[ldid] = "LDID_" name;\ |
34 |
|
PyModule_AddIntConstant(module, "LDID_" name, ldid) |
35 |
|
|
36 |
|
static char* codecs[PYSHAPELIB_NUM_LANGUAGE_DRIVERS]; |
37 |
|
static char* drivers[PYSHAPELIB_NUM_LANGUAGE_DRIVERS]; |
38 |
|
|
39 |
|
#endif |
40 |
|
|
41 |
|
|
42 |
|
|
43 |
|
/** Determine name of Python's built-in codec |
44 |
|
*/ |
45 |
|
static char* get_codec(DBFHandle handle) |
46 |
|
{ |
47 |
|
#if HAVE_LANGUAGE_DRIVER |
48 |
|
if (!codecs[handle->nLanguageDriver]) |
49 |
|
{ |
50 |
|
PyErr_Format(PyExc_ValueError, "Language Driver ID %d not recognized", handle->nLanguageDriver); |
51 |
|
} |
52 |
|
return codecs[handle->nLanguageDriver]; |
53 |
|
#else |
54 |
|
return "cp1252"; |
55 |
|
#endif |
56 |
|
} |
57 |
|
|
58 |
|
|
59 |
|
|
60 |
|
/** decode to unicode object |
61 |
|
*/ |
62 |
|
static PyObject* decode_string(DBFHandle handle, const char* string) |
63 |
|
{ |
64 |
|
char* codec = get_codec(handle); |
65 |
|
if (!codec) return NULL; |
66 |
|
return PyUnicode_Decode(string, strlen(string), codec, NULL); |
67 |
|
} |
68 |
|
|
69 |
|
/** encode unicode object to normal Python string object |
70 |
|
*/ |
71 |
|
static PyObject* encode_string(DBFHandle handle, PyObject* string) |
72 |
|
{ |
73 |
|
char* codec = get_codec(handle); |
74 |
|
if (!codec) return NULL; |
75 |
|
|
76 |
|
if (PyString_Check(string)) |
77 |
|
{ |
78 |
|
return PyString_AsEncodedObject(string, codec, NULL); |
79 |
|
} |
80 |
|
if (PyUnicode_Check(string)) |
81 |
|
{ |
82 |
|
return PyUnicode_AsEncodedString(string, codec, NULL); |
83 |
|
} |
84 |
|
|
85 |
|
PyErr_SetString(PyExc_TypeError, "value is neither a string or unicode object"); |
86 |
|
return NULL; |
87 |
|
} |
88 |
|
|
89 |
|
|
90 |
|
|
91 |
/* --- DBFFile ------------------------------------------------------------------------------------------------------- */ |
/* --- DBFFile ------------------------------------------------------------------------------------------------------- */ |
92 |
|
|
93 |
typedef struct { |
typedef struct { |
246 |
static PyObject* do_read_attribute(DBFHandle handle, int record, int field, char * name) |
static PyObject* do_read_attribute(DBFHandle handle, int record, int field, char * name) |
247 |
{ |
{ |
248 |
int type, width; |
int type, width; |
249 |
const char* temp; |
const char* string; |
250 |
type = DBFGetFieldInfo(handle, field, name, &width, NULL); |
type = DBFGetFieldInfo(handle, field, name, &width, NULL); |
251 |
|
|
252 |
/* For strings NULL and the empty string are indistinguishable |
/* For strings NULL and the empty string are indistinguishable |
263 |
switch (type) |
switch (type) |
264 |
{ |
{ |
265 |
case FTString: |
case FTString: |
266 |
temp = DBFReadStringAttribute(handle, record, field); |
string = DBFReadStringAttribute(handle, record, field); |
267 |
if (temp) return PyString_FromString(temp); |
if (string) return decode_string(handle, string); |
268 |
|
|
269 |
case FTInteger: |
case FTInteger: |
270 |
return PyInt_FromLong((long)DBFReadIntegerAttribute(handle, record, field)); |
return PyInt_FromLong((long)DBFReadIntegerAttribute(handle, record, field)); |
273 |
return PyFloat_FromDouble(DBFReadDoubleAttribute(handle, record, field)); |
return PyFloat_FromDouble(DBFReadDoubleAttribute(handle, record, field)); |
274 |
|
|
275 |
case FTLogical: |
case FTLogical: |
276 |
temp = DBFReadLogicalAttribute(handle, record, field); |
string = DBFReadLogicalAttribute(handle, record, field); |
277 |
if (temp) |
if (string) |
278 |
{ |
{ |
279 |
switch (temp[0]) |
switch (string[0]) |
280 |
{ |
{ |
281 |
case 'F': |
case 'F': |
282 |
case 'N': |
case 'N': |
378 |
/* write a single field of a record. */ |
/* write a single field of a record. */ |
379 |
static int do_write_field(DBFHandle handle, int record, int field, int type, PyObject* value) |
static int do_write_field(DBFHandle handle, int record, int field, int type, PyObject* value) |
380 |
{ |
{ |
381 |
|
PyObject* encoded_string = NULL; |
382 |
char * string_value; |
char * string_value; |
383 |
int int_value; |
int int_value; |
384 |
double double_value; |
double double_value; |
393 |
switch (type) |
switch (type) |
394 |
{ |
{ |
395 |
case FTString: |
case FTString: |
396 |
string_value = PyString_AsString(value); |
encoded_string = encode_string(handle, value); |
397 |
if (!string_value) return 0; |
if (!encoded_string) return 0; |
398 |
if (DBFWriteStringAttribute(handle, record, field, string_value)) return 1; |
string_value = PyString_AsString(encoded_string); |
399 |
|
if (!string_value) |
400 |
|
{ |
401 |
|
Py_DECREF(encoded_string); |
402 |
|
return 0; |
403 |
|
} |
404 |
|
if (DBFWriteStringAttribute(handle, record, field, string_value)) |
405 |
|
{ |
406 |
|
Py_DECREF(encoded_string); |
407 |
|
return 1; |
408 |
|
} |
409 |
|
Py_DECREF(encoded_string); |
410 |
break; |
break; |
411 |
|
|
412 |
case FTInteger: |
case FTInteger: |
548 |
#endif |
#endif |
549 |
|
|
550 |
|
|
551 |
|
#if HAVE_LANGUAGE_DRIVER |
552 |
|
|
553 |
|
static PyObject* dbffile_language_driver(DBFFileObject* self, void* closure) |
554 |
|
{ |
555 |
|
return PyInt_FromLong((long)self->handle->nLanguageDriver); |
556 |
|
} |
557 |
|
|
558 |
|
#endif |
559 |
|
|
560 |
|
|
561 |
static struct PyMethodDef dbffile_methods[] = |
static struct PyMethodDef dbffile_methods[] = |
562 |
{ |
{ |
610 |
|
|
611 |
static struct PyGetSetDef dbffile_getsetters[] = |
static struct PyGetSetDef dbffile_getsetters[] = |
612 |
{ |
{ |
613 |
|
#if HAVE_LANGUAGE_DRIVER |
614 |
|
{"language_driver", (getter)dbffile_language_driver, NULL, "Language Driver ID (read-only)" }, |
615 |
|
#endif |
616 |
{NULL} |
{NULL} |
617 |
}; |
}; |
618 |
|
|
686 |
|
|
687 |
|
|
688 |
|
|
689 |
|
#if HAVE_LANGUAGE_DRIVER |
690 |
|
|
691 |
|
/** translate a numeric Language Driver ID to the name of Python's codec. |
692 |
|
*/ |
693 |
|
static PyObject* dbflib_language_driver_codec(PyObject* module, PyObject* args) |
694 |
|
{ |
695 |
|
int ldid; |
696 |
|
if (!PyArg_ParseTuple(args, "i:language_driver_name", &ldid)) return NULL; |
697 |
|
if (ldid < 0 || ldid >= PYSHAPELIB_NUM_LANGUAGE_DRIVERS || !codecs[ldid]) |
698 |
|
{ |
699 |
|
PyErr_SetString(PyExc_ValueError, "invalid driver id"); |
700 |
|
return NULL; |
701 |
|
} |
702 |
|
return PyString_FromString(codecs[ldid]); |
703 |
|
} |
704 |
|
|
705 |
|
/** translate a numeric Language Driver ID to a string represting its constant. |
706 |
|
*/ |
707 |
|
static PyObject* dbflib_language_driver_name(PyObject* module, PyObject* args) |
708 |
|
{ |
709 |
|
int ldid; |
710 |
|
if (!PyArg_ParseTuple(args, "i:language_driver_name", &ldid)) return NULL; |
711 |
|
if (ldid < 0 || ldid >= PYSHAPELIB_NUM_LANGUAGE_DRIVERS || !drivers[ldid]) |
712 |
|
{ |
713 |
|
PyErr_SetString(PyExc_ValueError, "invalid driver id"); |
714 |
|
return NULL; |
715 |
|
} |
716 |
|
return PyString_FromString(drivers[ldid]); |
717 |
|
} |
718 |
|
|
719 |
|
#endif |
720 |
|
|
721 |
|
|
722 |
|
|
723 |
static struct PyMethodDef dbflib_methods[] = |
static struct PyMethodDef dbflib_methods[] = |
724 |
{ |
{ |
725 |
{"open", (PyCFunction)dbflib_open, METH_VARARGS, |
{"open", (PyCFunction)dbflib_open, METH_VARARGS, |
726 |
"open(name [, mode]) -> DBFFile\n\n" |
"open(name [, mode]) -> DBFFile\n\n" |
727 |
"opens a DBFFile" }, |
"opens a DBFFile" }, |
728 |
{"create", (PyCFunction)dbflib_create, METH_VARARGS, |
{"create", (PyCFunction)dbflib_create, METH_VARARGS, |
729 |
"create(name) -> DBFFile\n\n" |
"create(name [, language_driver]) -> DBFFile\n\n" |
730 |
"create a DBFFile" }, |
"create a DBFFile " }, |
731 |
|
#if HAVE_LANGUAGE_DRIVER |
732 |
|
{"language_driver_codec", (PyCFunction)dbflib_language_driver_codec, METH_VARARGS, |
733 |
|
"language_driver_codec(driver_id) -> string\n\n" |
734 |
|
"translate language driver id into the name of the Python's codec used as code page." }, |
735 |
|
{"language_driver_name", (PyCFunction)dbflib_language_driver_name, METH_VARARGS, |
736 |
|
"language_driver_name(driver_id) -> string\n\n" |
737 |
|
"translate language driver id into a string." }, |
738 |
|
#endif |
739 |
{NULL} |
{NULL} |
740 |
}; |
}; |
741 |
|
|
742 |
|
|
|
|
|
743 |
PyMODINIT_FUNC initdbflib(void) |
PyMODINIT_FUNC initdbflib(void) |
744 |
{ |
{ |
745 |
|
int i; |
746 |
|
|
747 |
PyObject* module = Py_InitModule("dbflib", dbflib_methods); |
PyObject* module = Py_InitModule("dbflib", dbflib_methods); |
748 |
if (!module) return; |
if (!module) return; |
749 |
|
|
755 |
PYSHAPELIB_ADD_CONSTANT(FTLogical); |
PYSHAPELIB_ADD_CONSTANT(FTLogical); |
756 |
PYSHAPELIB_ADD_CONSTANT(FTInvalid); |
PYSHAPELIB_ADD_CONSTANT(FTInvalid); |
757 |
PyModule_AddIntConstant(module, "_have_commit", HAVE_UPDATE_HEADER); |
PyModule_AddIntConstant(module, "_have_commit", HAVE_UPDATE_HEADER); |
758 |
|
|
759 |
|
#if HAVE_LANGUAGE_DRIVER |
760 |
|
/* table compiled from these resources: |
761 |
|
* http://www.clicketyclick.dk/databases/xbase/format/dbf.html |
762 |
|
* http://www.esrinl.com/content/file.asp?id=307 |
763 |
|
* http://msdn2.microsoft.com/en-us/library/aa975345(VS.71).aspx |
764 |
|
*/ |
765 |
|
for (i = 0; i < PYSHAPELIB_NUM_LANGUAGE_DRIVERS; ++i) |
766 |
|
{ |
767 |
|
codecs[i] = NULL; |
768 |
|
drivers[i] = NULL; |
769 |
|
} |
770 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x00, "cp1252", "NOT_SET"); |
771 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x01, "cp437", "DOS_USA"); |
772 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x02, "cp850", "DOS_INTERNATIONAL"); |
773 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x03, "cp1252", "WINDOWS_ANSI"); |
774 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x04, "mac_roman", "STANDARD_MACINTOSH"); |
775 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x08, "cp865", "DANISH_OEM"); |
776 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x09, "cp437", "DUTCH_OEM"); |
777 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x0a, "cp850", "DUTCH_OEM_2"); |
778 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x0b, "cp437", "FINNISH_OEM"); |
779 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x0d, "cp437", "FRENCH_OEM"); |
780 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x0e, "cp850", "FRENCH_OEM_2"); |
781 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x0f, "cp437", "GERMAN_OEM"); |
782 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x10, "cp850", "GERMAN_OEM_2"); |
783 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x11, "cp437", "ITALIAN_OEM"); |
784 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x12, "cp850", "ITALIAN_OEM_2"); |
785 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x13, "cp932", "JAPANESE_SHIFT_JIS"); |
786 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x14, "cp850", "SPANISH_OEM_2"); |
787 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x15, "cp437", "SWEDISH_OEM"); |
788 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x16, "cp850", "SWEDISH_OEM_2"); |
789 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x17, "cp865", "NORWEGIAN_OEM"); |
790 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x18, "cp437", "SPANISH_OEM"); |
791 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x19, "cp437", "ENGLISH_BRITAIN_OEM"); |
792 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x1a, "cp850", "ENGLISH_BRITAIN_OEM_2"); |
793 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x0b, "cp437", "ENGLISH_US_OEM"); |
794 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x1c, "cp863", "FRENCH_CANADA_OEM"); |
795 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x1d, "cp850", "FRENCH_OEM_2"); |
796 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x1f, "cp852", "CZECH_OEM"); |
797 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x22, "cp852", "HUNGARIAN_OEM"); |
798 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x23, "cp852", "POLISH_OEM"); |
799 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x24, "cp860", "PORTUGUESE_OEM"); |
800 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x25, "cp850", "PORTUGUESE_OEM_2"); |
801 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x26, "cp866", "RUSSIAN_OEM"); |
802 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x37, "cp850", "ENGLISH_US_OEM_2"); |
803 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x40, "cp852", "ROMANIAN_OEM"); |
804 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x4d, "cp936", "CHINESE_GBK_PRC"); |
805 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x4e, "cp949", "KOREAN_ANSI_OEM);"); |
806 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x4f, "cp950", "CHINESE_BIG5_TAIWAN"); |
807 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x50, "cp874", "THAI_ANSI_OEM"); |
808 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x57, "cp1252", "ESRI_ANSI"); |
809 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x58, "cp1252", "WESTERN_EUROPEAN_ANSI"); |
810 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x59, "cp1252", "SPANISH_ANSI"); |
811 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x64, "cp852", "EASTERN_EUROPEAN_MSDOS"); |
812 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x65, "cp866", "RUSSIAN_MSDOS"); |
813 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x66, "cp865", "NORDIC_MSDOS"); |
814 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x67, "cp861", "ICELANDIC_MSDOS"); |
815 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x68, "cp895", "CZECH_MSDOS"); |
816 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x69, "cp620", "POLISH_MSDOS"); |
817 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x6a, "cp737", "GREEK_MSDOS"); |
818 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x6b, "cp857", "TURKISH_MSDOS"); |
819 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x6c, "cp863", "FRENCH_CANADA_MSDOS"); |
820 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x78, "cp950", "TAIWAN_BIG5"); |
821 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x79, "cp949", "HANGUL_WANSUG"); |
822 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x7a, "cp936", "PRC_GBK"); |
823 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x7b, "cp932", "JAPANESE_SHIFT_JIS"); |
824 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x7c, "cp874", "THAI_WINDOWS_MSDOS"); |
825 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x7d, "cp1255", "HEBREW_WINDOWS"); |
826 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x7e, "cp1256", "ARABIC_WINDOWS"); |
827 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x86, "cp737", "GREEK_OEM"); |
828 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x87, "cp852", "SLOVENIAN_OEM"); |
829 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x88, "cp857", "TURKISH_OEM"); |
830 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x96, "mac_cyrillic", "RUSSIAN_MACINTOSH"); |
831 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x97, "mac_latin2", "EASTERN_EUROPEAN_MACINTOSH"); |
832 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0x98, "mac_greek", "GREEK_MACINTOSH"); |
833 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0xc8, "cp1250", "EASTERN_EUROPEAN_WINDOWS"); |
834 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0xc9, "cp1251", "RUSSIAN_WINDOWS"); |
835 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0xca, "cp1254", "TURKISH_WINDOWS"); |
836 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0xcb, "cp1253", "GREEK_WINDOWS"); |
837 |
|
PYSHAPELIB_ADD_LANGUAGE_DRIVER(0xcc, "cp1257", "BALTIC_WINDOWS"); |
838 |
|
#endif |
839 |
|
|
840 |
} |
} |