1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
|
2009-01-16 Thomas Heller <theller@python.net>
* Add the ability to generate code for variables exported from
dlls. Based on a patch from Carlos Scheidegger.
Add the c_bool type.
* In Python 2.6 the md5 module is deprecated, use hashlib (when
available) instead.
2009-01-08 Thomas Heller <theller@ctypes.org>
* The codegenerator now uses ctypes sized int and uint types
for the sized integers <from stdint.h>.
2009-01-08 Thomas Heller <theller@python.net>
* setup.py: Correct the version number. Remove the 'zip_safe'
argument in the setup call; it was left over from setuptools
experiments.
* On Windows, the codegenerator uses WinDLL instead of CDLL so
that functions with stdcall mangled names can be loaded.
Use the correct registry key to find gccxml.
2008-10-20 Thomas Heller <theller@ctypes.org>
* Remove the import of setuptools in the setup.py script.
* Add support for float and double constants and variables.
* Fix structure support. Patch from Shalabh Chaturvedi.
2008-02-01 Thomas Heller <theller@ctypes.org>
* Include the variable initializers in the generated code (in a
comment).
* Change version number to 0.5.6. Also, the version number is now
available as 'ctypeslib.__version__'.
* Make the codegenerator compatible with gccxml 0.9.0.
* Move the calculation of initializers into a separate class; this
makes it easier to customize the behaviour depending on the
initializer type.
2008-01-24 Thomas Heller <theller@python.net>
* Fix the codegenerator to handle structures without fields; it
did fail to calculate the packing.
2007-04-11 Thomas Heller <theller@python.net>
* Add a 'compilerflags' optional argument to the include()
function. This allows to specify the compilerflags for h2xml when
the code is 'compiled'. Default is 'compilerflags=["-c"]' which
will include #define constants.
2007-03-01 Thomas Heller <theller@python.net>
* Applied a patch from Michal J. Gajda: This patch adds the
--preload option to xml2py.py to resolve missing references from
DLLs. See
http://sourceforge.net/tracker/?func=detail&aid=1670758&group_id=71702&atid=532156
* Added ctypeslib.contrib.pythonhdr module, by Lenard Lindstrom:
The is a python module, pythonhdr.py, which declares several,
hopefully useful, Python C API declarations and functions. Also
includes is a unit test.
2007-02-16 Thomas Heller <theller@python.net>
* The xml2py script now also accepts full pathnames to shared
libraries in the -l command line switch.
* Fix failing tests on linux.
* The code generator will now happily generate Python code for the
C 'long double' type. This type is not currently supported by
ctypes; so the generated code must be fixed manually. Still
better than to let the code generation crash.
2007-02-15 Thomas Heller <theller@python.net>
* In ctypeslib.dynamic_module the function update_from() has been
removed. Use the include() function instead, which will - if
needed - recompile the C code passed to it and adds generated
Python code to the calling module.
* The xml2py script can now add comments for functions and
structures into the generated code. The comments include the name
and linenumber of the header file where the function/structure was
defined, plus the C function argument names.
2006-12-08 Thomas Heller <theller@python.net>
* Created new module ctypeslib.util. This module contains small
helper functions. When using these functions it *may* be better
to copy them into the own project instead of having ctypeslib as
dependency. YMMV.
* Moved byref_offset from the ctypeslib.test package into
ctypeslib.util, and renaming the function to byref_at.
2006-11-23 Thomas Heller <theller@python.net>
* Fixed a bug in the codegenerator with unsized arrays.
Thanks to Danielle Varrazzo for the report.
2006-11-17 Thomas Heller <theller@python.net>
* In ctypeslib.test, add a byref_offset module that contains a
byref_offset function; with tests.
Experimental code, currently.
2006-11-14 Thomas Heller <theller@python.net>
* Added ctypeslib.contrib package which contains contributed code
that does not yet fit elsewehere. ctypeslib.contrib.get_exports,
contributed by leppton, lists the exports in a dll.
2006-11-10 Thomas Heller <theller@python.net>
* ctypeslib/h2xml.py: Refactored a function compile_to_xml() out
of main(), CompilerErrors are catched in main() but not in
compile_to_xml().
Added test that an exception is raised in
dynamic_module.include() when a CompilerError occurs.
* ctypeslib/codegen/cparser.py: No configuration file
"~/ctypes_codegen/cparser.cfg" is used (or needed) anymore.
2006-11-09 Thomas Heller <theller@python.net>
* Inmplemented this change in a different way again.
* Reverted this change because it does not (yet) work:
* ctypeslib/codegen/cparser.py: The final xml-file is now created
before searching for preprocessor definitions (if no output file
is requested it will be deleted afterwards). The advantage is
that compilation errors are detected early and don't disturb the
cpp magic.
2006-11-08 Thomas Heller <theller@python.net>
* ctypeslib\dynamic_module.y: Generate the .h, .xml, and
.typedesc.bz2 files in a temporary directory. Should that
directory contain the GCCXML version number? Or the ctypeslib
version number?
2006-11-07 Thomas Heller <theller@python.net>
* ctypeslib\codegen\gccxmlparser.py: Patch from Giovanni Bajo: Use
cElementtree, if installed, instead of xml.sax for XML parsing.
This gives a significant speedup.
2006-11-03 Thomas Heller <theller@python.net>
* ctypeslib\experimental\*.py contains experimental stuff. This
directory is *not* package because it doesn't contain an
__init__.py file.
* ctypeslib\dynamic_module.py: A module implementing incremental,
dynamic code generation. Removed the former ctypeslib\dynmod.py.
* Renamed restructured the ctypes_codegen package. It is now
named ctypeslib, with a subpackage ctypeslib.codegen. The
ctypeslib package now contains the h2xml and xml2py scripts.
2006-10-31 Thomas Heller <theller@python.net>
* ctypes_codegen\codegenerator.py (Generator.init_value): Reworked
the init function to use eval, the output should be correct now.
* Cast #define'd symbols to the type found, to avoid compilation
errors because of 'const'ness mismatch.
2006-10-20 Thomas Heller <theller@python.net>
* Added ctypes_codegen.test package for codegenerator tests.
* Renamed modules from ctypes_codegen.h2xml_main to
ctypes_codegen.h2xml and ctypes_codegen.xml2py to
ctypes_codegen.xml2py.
So, with Python2.5 it is possible to execute
'python -m ctypes_codegen.h2xml ...' and
'python -m ctypes_codegen.xml2py ...'.
2006-10-05 Thomas Heller <theller@python.net>
* Added '#!/usr/bin/env python' to the h2xml.py and xml2py.py
scripts. Patch by Victor Stinner.
2006-09-19 Thomas Heller <theller@python.net>
* cparser.py: Lots of changes. The summary: xml2py doesn't use a
config file for #define symbols to exclude, instead it loops
several times to find the symbols that don't compile.
* scripts/xml2py_main.py: Change the generated code that loads
libraries. On windows, the absolute pathname is no longer used.
* codegenerator.py: Rework the .init_value() method so that
decode_value() is used in all cases. Correct the decoding of
integer values - it seems decimal numbers are used except when
they start with a '0' which means hex (previously I assumed hex
always).
2006-09-15 Thomas Heller <theller@python.net>
* codegenerator.py: Refactored a method 'generate_items' out of
'generate_code'. This allows better incremental code generation.
* os.name is 'win32', not 'nt', on windows. This prevented that
definitions from ctypes.wintypes to be used in generated code.
2006-09-06 Thomas Heller <theller@python.net>
* codegenerator.py: '_anonymous_' declarations are only generated
for unnamed Structure/Union fields if the type of the field is a
Structure or Union. (Windows header files contain unnamed
structure fields of type 'int', probably for padding.)
2006-08-11 Thomas Heller <theller@python.net>
* codegenerator.py: Empty _field_ attributes are now generated for
Structure/Union (for weird reasons having to do with COM
interfaces NO _fields_ attribute was generated before).
2006-06-23 Thomas Heller <theller@python.net>
* xml2py.py: Remove the use_generators argument and the
corresponding command line option for xml2py.py, since the code
generator does not create decorators any more.
Add a '-c' command line option which will insert comments in the
generated code that will point the the location in the include
file where the data types are defined. This replaces the
hardcoded USE_COMMENTS variable.
2006-06-13 Thomas Heller <theller@python.net>
* xml2py.py: The '-l' option for the xml2py.py script now takes
the linker library name instead of the library pathname. I still
have to find out which library (on posix systems) really defines a
function. The generated code that loads the libraries has been
reworked.
2006-06-14 Thomas Heller <theller@python.net>
* h2xml.py: The subprocess module is now required, although the
code is still compatible with Python 2.3.
* The '#define' symbols that cause compilation errors are now
determined at runtime:
If the '-c' option is given, compiler error messages are parsed to
find those '#define ' symbols that are causing them. These
symbols are then appended to the file
~/.ctypes_codegen/cparser_excludes. If these errors occur, it is
required to restart h2xml.py, but chances are good that the next
run works better (it may be required to repeat this process
several times until all errors have vanished).
The cparser_config.py module has been removed because it does no
longer list symbols that have to be excluded.
2006-06-13 Thomas Heller <theller@python.net>
* Fixed a problem in gccxmlparser.py when 'Constructor'
descriptions created by gccxml gave no 'name' attribute.
2006-06-09 Thomas Heller <theller@python.net>
* ctypes_codegen\codegenerator.py: Added a USE_COMMENTS variable
which allows to turn comment generation off. Generated
'_anonymous_' declarations for unnamed structure fields.
Changed the way libraries are loaded, the cdecl and stdcall
decorators are no longer used. Functions are generated with
assignments of .restype and .argtypes.:
2006-06-08 Thomas Heller <theller@python.net>
* ctypes_codegen\codegenerator.py: When generating the '__all__'
list, don't break long strings.
* h2xml does no longer create a file named 'None' when no output
file was given on the command line.
* Added a TODO list.
* Initial import of the current ctypes codegenerator version.
|
