Only one import of the low-level C/C++ module from the pure Python module is
attempted now. Previously a second import of the low-level C/C++ module was attempted
after an ImportError occurred and was done to support 'split modules'. A 'split module' is
a configuration where the pure Python module is a module within a Python package and the
low-level C/C++ module is a global Python module. Now a 'split module' configuration is
no longer supported by default. This configuration can be supported with a simple
customization, such as:
%module(package="mypackage", moduleimport="import $module") foo
or if using -builtin:
%module(package="mypackage", moduleimport="from $module import *") foo
instead of
%module(package="mypackage") foo
See the updated Python chapter titled "Location of modules" in the documentation.
Closes#848#1343
* adr26-master:
Cleanup accessing/decref of globals, to avoid code bloat in init function.
Fix ISOC build errors.
Fix unused variable warning.
#1360: Leak of SWIG var link object
Conflicts:
CHANGES.current
When using -builtin, the two step C-extension module import is now
one step and the wrapped API is only available once and not in an underlying
module attribute like it is without -builtin. To understand this, consider a
module named 'example' (using: %module example). The C-extension is compiled into
a Python module called '_example' and a pure Python module provides the actual
API from the module called 'example'. It was previously possible to additionally
access the API from the module attribute 'example._example'. The latter was an
implementation detail and is no longer available. It shouldn't have been used, but
if necessary it can be resurrected using the moduleimport attribute described in the
Python chapter of the documentation. If both modules are provided in a Python
package, try:
%module(moduleimport="from . import _example\nfrom ._example import *") example
or more generically:
%module(moduleimport="from . import $module\nfrom .$module import *") example
and if both are provided as global modules, try:
%module(moduleimport="import _example\nfrom _example import *") example
or more generically:
%module(moduleimport="import $module\nfrom $module import *") example
The module import code shown will appear in the example.py file.
The implementation for a renamed constructor for -builtin no longer
uses the same implementation for non-builtin mode which makes a call
into the compiled C module, eg _constructor_rename for a module called
constructor_rename. The output in the constructor_rename.py file is now:
RenamedConstructor = new_RenamedConstructor
instead of
def RenamedConstructor():
val = _constructor_rename.new_RenamedConstructor()
return val
when wrapping:
struct Foo {
%rename(RenamedConstructor) Foo();
Foo() {}
};
See the constructor_rename.i testcase.
This change has been made to make the wrapper more efficient and is a
step towards the next commit which will remove duplicating the symbols
in both the pure Python module and implementation C module
(constructor_rename and _constructor_rename respectively in this case).
When using the moduleimport option, such as:
%module(moduleimport="import $module") example
the minimum Python version check disappeared from the generated Python
file. The code has been refactored and _swig_python_version_info is
no longer deleted after initial use as it can be used in a few places,
in particular, when -builtin is used.
This option turned off the insertion of Python import statements
derived from a %import directive. For example given:
%module module_b
%import "module_a.i"
then module_b.py will contain:
import module_a
This option was originally added in 658add5, apparently to fix some sort
of circular import problem of which there are no details. It is
undocumented and is now removed as part of the simplification of the
SWIG Python command line options for SWIG 4.
Issue #1340.
Both the command line and %module options of the same name have been
removed. These were undocumented. The -outputtuple option returned a
Python tuple instead of a list, mostly typically in the OUTPUT
typemap implementations.
It unclear why a tuple instead of a list return type is needed and
hence this option has been removed as part of the simplification of
the SWIG Python command line options for SWIG 4.
Issue #1340.
The -cppcast option is still turned on by default. The -nocppcast option
to turn off the use of c++ casts (const_cast, static_cast etc) has been
removed. However, defining SWIG_NO_CPLUSPLUS_CAST will still generate C casts
instead of C++ casts for C++ wrappers.
This a revert of commit fc79264a48f186f8bbd367e91fa9dbf9758aa092:
"Revert "Remove -cppcast and -nocppcast command line options""
The Scilab and Javascript casting problems are now fixed, so -cppcast
is now switched on as default.
The -cppcast option is still turned on by default. The -nocppcast option
to turn off the use of c++ casts (const_cast, static_cast etc) has been
removed. However, defining SWIG_NO_CPLUSPLUS_CAST will still generate C casts
instead of C++ casts for C++ wrappers.
This option controlled the __repr__ proxy class method.
Also removes:
-old_repr
-oldrepr
-new_repr
-newrepr
Note: -newrepr was and remains on by default. When -oldrepr/-old_repr
option resulted in code that did not run.
This option was for backwards compatibility with very old versions of
SWIG that generated a shadow 'XPtr' class derived from proxy class X.
If anyone is still using these, they are best off using the actual proxy
class, or otherwise add them in manually using %pythoncode.
Issue #1340.
This option was originally added to help users switch to using
-fastunpack so that typemaps using obj0 did not need changing by
providing an alias for swig_obj[0] to obj0. The correct solution was
always to replace obj0 with $self in the typemap. This is now mandatory.
However, the -nofastunpack option can still be used to circumvent using
the mandatory typemap change.
The -nofastunpack option has been left in as there isn't much code
simplification in removing the nofastunpack code. This is because the
nofastunpack code is still sometimes needed (eg varags and overloaded
functions).
Python fixed many APIs to use const char * instead of char * at around
Python 2.4. As we support 2.7 and later, we can now remove the non-const
string usage.
Types changed:
PyArg_ParseTuple
PyArg_ParseTupleAndKeywords
PyArg_UnpackTuple
PyDict_SetItemString
PyMethodDef
PyModuleDef
SWIG_Python_UnpackTuple
SWIG_Python_str_FromChar
SWIG_addvarlink
swig_const_info
This option, if used, has not had any effect on Python 3 code since commit a863d3 9 years ago.
I think we can assume that it is not needed for Python 3.
Running the examples and test-suite (Python 2) doesn't change the code
paths with and without -safecstrings because only SWIG_OLDOBJ and SWIG_NEWOBJ
are used in the typemaps and the following code is thus unaltered by -safecstrings
(which sets SWIG_PYTHON_SAFE_CSTRINGS):
%#if defined(SWIG_PYTHON_SAFE_CSTRINGS)
if (*alloc != SWIG_OLDOBJ)
%#else
if (*alloc == SWIG_NEWOBJ)
%#endif
{
*cptr = %new_copy_array(cstr, len + 1, char);
*alloc = SWIG_NEWOBJ;
printf("safe strings: %s\n", *cptr ? *cptr : "NULLSTRING");
} else {
*cptr = cstr;
*alloc = SWIG_OLDOBJ;
}
Note: nosafecstrings was also the default and -O didn't actually change this.
A custom implementation for Py_None was implemented in SWIG_Py_None().
This was used by default on Windows only. It isn't clear why this
was done just for Windows. Now Py_None is the real Py_None on all
operating systems.
This option runs the equivalent init code in Python C code instead of pure Python code.
The init code adds the C++ pointer into the 'this' variable.
If the variable already exists, it appends it. This
seems to only happen in directors and multiple inheritance, see
PyMulti in director_basic_runme.py.
We only still default to generate a no-op __del__ method for theoretical
compatibility with old code, and 4.0.0 is a good time to make a cut-off
should such code really still exist.
On the flip-side, the presence of a __del__ method seems to be able to
prevent garbage collection from kicking in (see #1215), so it's
definitely desirable to get rid of it when there's nothing for it to do.
Conflicts:
Source/Modules/python.cxx
This is a cherry-pick and merge from the patch in #1261
What SWIG calls "modern" classes are supported by Python 2.3 and up
which means they're supported by all the Python versions we aim to
support in 4.0.0.
Conflicts:
Source/Modules/python.cxx
This is a cherry-pick and merge from the patch in #1261
There were only needed to support Python < 2.2, and we now require at
least Python 2.6.
Conflicts:
.travis.yml
Examples/test-suite/python/autodoc_runme.py
Source/Modules/python.cxx
This is a cherry-pick and merge from patch in #1261
Fix when using -builtin and wrapping std::map, std::set, std::unordered_map or
std::unordered_set to ensure __contains__ is called. This is a wrapper for the STL
container's find method. Without it, Python will do its own slower sequence search.
Found via `codespell -q 3 -L "uint,bae,objext,cmo,goin,struc,ois,upto"`
whitespaces were unintentionally fixed due to my editors settings.
Rebased patch #1327
Output methods in the same place that they would be generated in the
proxy class when not using -fastproxy. When using autodoc, the two
definitions of the method are then right after each other.
The previous implementation failed with Python 3 and abstract base clases.
The new implementation replaces the Python 2 implementation using new.instancemethod with C API PyMethod_New to match the equivalent Python 3 implementation which uses PyInstanceMethod_New.
Closes#1310
* python-type-error:
Python fastdispatch error message improvements
Correct PyErr_Fetch/PyErr_Restore variable names
Overloaded C++ function Python wrappers now raise a TypeError instead of NotImplementedError
When overloaded C++ methods are called, the fastdispatch mode sometimes
optimises out the generation of the typecheck typemap code, resulting in
the 'Possible C/C++ prototypes' information not being in the error message.
Now when a TypeError is thrown, the exception message is modified to contain the
'Possible C/C++ prototypes'.
See issue #1293
Occurs when the types passed are incorrect. This change means
there is now consistency with non-overloaded function wrappers which have always
raised TypeError when the incorrect types are passed.
See issue #1293
Corrects adding metaclass to the nondynamic class so that the
implementation works. Uses metaclass syntax that will work for both
python 2 and 3 or proper Python 3 syntax when using -py3.
Fixes nondynamic mode when an instance variable is set with the same
name as a class variable in a class derived from a SWIG proxy class.
The python_nondynamic testcase now works using -modern and testcase
enhanced slightly too.
Fixes#718
Python modules can be run as main from the command line with the -c flag
Currently this does not work as the python wrapper module does not correctly import the extension module. This commit makes the generated code consistent with PEP 366.
Avoid casts between incompatible function types where possible (when
keyword args are in use, it is not possible to avoid such warnings as
they are inherent in the design of Python's C API in that particular
case). Fixes#1259.
This is the Doxygen work begun in Google Summer of Code projects 2008
and 2012 and subsequently improved by numerous contributors.
* vadz-doxygen: (314 commits)
Add changes entry for Doxygen support
Add some missing doctype tyemaps
Doxygen warnings cleanup
Move doxygen warning numbers
Add Python doxygen example
Doxygen example
Add Doxygen to include paths
Doxygen source rename
More merge fixes from doxygen branches
Correct python example headers
Correct source code headers
Another merge fix from doxygen branches
Java enums output format fixes
Add omitted doxygen_parsing_enums testcase
PEP8 conformance for comment verifier module
Clean up merge problem
Doxygen html tweaks
Update html chapter numbering for added Doxygen chapter
Fixes to makechap.py to detect ill-formed headers
html fixes for Doxygen
Add missing CPlusPlus17.html file
Format files to unix format
Doxygen testcase tweak to match that in the html docs
Doxygen html documentation updates and corrections
Remove doxygen Examples subdirectory
Beautify doxygen source code
Code formatting fixes in doxygen code
Remove unused doxygen code
new_node refactor
Various merge fixes in doxygen branches
Unused variable warning fix
Fix wrongly resetting indent after formulae in Doxygen comments
Add support for doxygen:alias feature
Get rid of meaningless return type of DoxygenParser methods
Return enum, not untyped int, when classifying Doxygen commands
Get rid of unnecessary "typedef enum" in C++ code
Use slash, not backslash, in "C/C++" in the documentation
Replace literal "<" with "<" in HTML documentation
Fix broken link to java.sun.com in Doxygen documentation
Fix using com.sun.tools.javadoc package under macOS
Fix error reporting for special characters in Doxygen parsing code
Switch Python Doxygen unit tests to use inspect.getdoc()
Use correct separator in Java class path under Windows.
Remove executable permission from appveyor.yml.
Use JAVA_HOME value in configure to detect Java.
Display JAVA_HOME value in "make java_version".
Fix harmless MSVC warning in DoxygenTranslator code.
Reset "_last" for all but first enum elements.
Don't duplicate Javadoc from global enum Doxygen comments twice.
Move Doxygen comments concatenation from the parser to the lexer.
Fix shift/reduce conflicts in Doxygen pre/post comment parsing.
Rewrote part of the grammar dealing with Doxygen comments for enums.
No changes, just remove spurious white space only differences.
Move Doxygen comment mangling from the parser to the lexer.
Merge "-builtin" autodoc bugs workarounds from master into test.
Quote JAVA_HOME variable value in Java test suite makefile.
Remove unused C_COMMENT_STRING terminal from the grammar.
Fix missing returns in the Doxygen test suite code.
Fix trimming whitespace from Doxygen comments.
Remove code not doing anything from PyDocConverter.
Remove unused <sstream> header.
Remove unreferenced struct declaration.
Remove unused Swig_warn() function.
Remove any whitespace before ignored Doxygen commands.
Remove trailing space from one of Doxygen tests.
Fix autodoc strings generated in Python builtin case and the test.
Fix Doxygen unit test in Python "-builtin" case.
Use class docstrings in "-builtin" Python case.
Don't indent Doxygen doc strings in generated Python code.
Add a possibility to flexibly ignore custom Doxygen tags.
Stop completely ignoring many Doxygen comments.
Fix structural Doxygen comment recognition in the parser.
No changes, just make checking for Doxygen structural tags more sane.
Use "//", not "#", for comments in SWIG input.
Allow upper case letters and digits in Doxygen words.
Pass the node the Doxygen comment is attached to to DoxygenParser.
Get rid of findCommand() which duplicaed commandBelongs().
Recognize unknown Doxygen tags correctly.
No real changes, just pass original command to commandBelongs().
Describe Doxygen-specific %features in a single place.
Give warnings for unknown Doxygen commands in Doxygen parser.
Document the return type when translating Doxygen @return to Python.
Fix translated Doxygen comments for overloaded functions in Python.
Also merge Doxygen comments for overloaded constructors in Python.
Allow using enum elements as default values for Python functions.
Don't always use "*args" for all Python wrapper functions.
No real changes, just make PYTHON::check_kwargs() const.
Refactor: move makeParameterName() to common Language base class.
Remove long line wrapping from Python parameter list generation code.
Simplify and make more efficient building Python docstrings.
Translate Doxygen code blocks to Sphinx code blocks.
Add a simple test of multiple parameters to Doxygen test suite.
Make Python parameters types hyperlinks in the doc strings.
Make Language::classLookup() and enumLookup() static.
Fix arguments of @param, @return etc translations to Python.
Remove unused method from PyDocConverter.
No real changes, just remove an unnecessary variable.
Preserve relative indentation when parsing Doxygen comments.
Use Sphinx-friendly formatting for overloaded functions documentation.
Add poor man trailing white space detection to Doxygen Python tests.
...
Source/DoxygenTranslator/src directory is renamed Source/Doxygen
Renamed files in this directory to short names using lowercase as is the
convention for the rest of the SWIG source.
C++ extension is also .cxx like other SWIG source code.
I used doxy as the prefix for most file renames because without this
Doxygen/parser.* would be easily confused with CParse/parser.* so
Doxygen/doxyparser.* is renamed from DoxygenTranslator/src/DoxygenParser.*
* master:
Add Octave 4.4 to Travis allow_failures
Fixes for appveyor image changes
Javascript test-suite Makefile parallel jobs
Add changes entry for csconstruct, dconstruct and javaconstruct fix
Fix lookup of csconstruct, dconstructor and javaconstruct typemaps
Javascript %nspace fix in generated C++ code
Add C++17 documentation chapter
Add changes notes for C++17 nested namespaces support
Test for invalid C++17 nested namespace aliases
Test c++17 nested namespaces and %nspace
Add c++17 nested namespaces runtime test for C#
Add c++17 nested namespaces runtime test for Python
Add support for c++17 nested namespaces
Update CHANGES.current
.travis.yml: test against Octave 4.4
Examples/test-suite/register_par.i: rename 'tree' to 'swig_tree'
Examples/octave/module_load/runme.m: update 'exist()' statements for Octave >= 4.4
Examples/octave/module_load/runme.m: do not use duplicate function names
Examples/Makefile.in: unset OCTAVE_PATH when running Octave for tests
Lib/octave: fix getting/setting global variables for Octave >= 4.4
Lib/octave: use new class for function member dereference with Octave >= 4.4
Lib/octave: fix operator installation for Octave >= 4.4
Lib/Octave: in Octave >= 4.4, assign reference to base class in subclass
Lib/octave: fix call to mlock() for Octave >= 4.4
Lib/octave: fix call to octave::call_stack::current() for Octave >= 4.4
Lib/octave: 'octave_exit' not longer exists in Octave >= 4.4
Lib/octave: replace is_bool_type() with islogical() for Octave >= 4.4
Lib/octave: replace is_numeric_type() with isnumeric() for Octave >= 4.4
Lib/octave: replace is_cell() with iscell() for Octave >= 4.4
Lib/octave: call octave::feval() instead of feval() for Octave >= 4.4
Lib/octave: fix function name passed to unwind_protect::begin_frame()
C#, D, Java methodmodifiers on destructors
Javascript assert.h - move to header section
Appveyor cl compiler warning fixes during configure
Java vector wrappers cast correction
test-suite fixes (Java directors) for compilers that don't support varargs
Go - use director.swg like other languages
test-suite fixes (2) for compilers that don't support varargs
Consistent spacing in generated exception specifications
test-suite fixes for compilers that don't support vararg macros
Enhance Travis testing to use gcc 8 and test C++17 and C17
Enhance SWIG_isfinite for older standards: C++03/C++98/C89
test-suite support for gcc-8 targeting C++11 and C++14
Scilab portability fixes - remove use of strdup
Scilab array overbounds fix handling char type exceptions
test-suite fix for c++17 and throw macro
Remove use of 'register' in C source
test-suite support for C++17: switch testing of the deprecated C++17 'register' keyword from C++ to C
Examples update to support C++17: exception specification throw removal
Cosmetic syntax tweak using throw in Octave directors
test-suite support for C++17 (Java): exception specification throw removal
test-suite support for C++17: exception specification throw removal
__cplusplus macro usage tweak
Improve detection of Python's 2to3 tool
Correct C shared library creation when specifing CC to configure
Remove superfluous parens in generated Python scripts.
[ci] guile 2.2 build no longer expected to fail
guile - resstructure some configure tests
Disable guile configuration if guile-config and guile report a different version
Fix guile executable detection on early 2.0.x guile versions
guile - drop GDB_INTERFACE related stuff
guile - replace obsolete scm_listify with scm_list_n
guile - use more reliable method of finding guile executable based on guile-config
Fix go version matching in configure for go1.10
[Python] Suppress new pycodestyle warning
Add if-no-present action for jsv8inc arg
Fix typo in help --with-jscoreinc and --with-jscorelib
Fix off-by-one error
* Makefile.in (configfiles): Update URLs for latest configfiles.
Add changes entry for Ruby %alias fix for global functions
[Ruby] Pass Qnil instead of NULL to rb_funcall()
Fix typo
Fix ruby %alias directive for native c functions
Stop testing Python on Appveyor msys/mingw
Fix -Wimplicit-fallthrough gcc-7.3 warning
* shared_ptr_directors:
Add director typemaps for pointer const ref types
Generation of director method declarations fixes
Add director shared_ptr typemaps for D
Add director shared_ptr typemaps for C#
Typo correction in changes file
Enhancements for directorin typemaps
Add director shared_ptr typemaps for Java
Director shared_ptr typemaps for scripting languages
R memory handling standardisation
Scilab, R and Octave shared_ptr typemaps null fix
Octave and R shared_ptr typemaps update
Remove duplicate director shared_ptr pointer reference typemaps
For shared_ptr directorin, make copy of shared_ptr in all cases.
Add directorin typemap for Python and Ruby shared_ptr.
- Fixes generation of director method declarations containing C++11 ref-qualifiers.
- Fixes generation of director method declarations returning more complex types such
as const ref pointers.
- Rewrite Swig_method_call to use more up to date code in the core.
The directorin typemaps will now generate a temporary variable
(specified after the type), such as:
%typemap(directorin) MyType (MyType *temp) { ... use temp ... }
The shared_ptr director typemaps have been fixed for use in functions
that take more than one parameter.
32 bit and 64 bit compiled versions of SWIG generated different Python files
when default arguments were outside the range of 32 bit signed integers.
The default arguments specified in Python are now only those that are in the
range of a 32 bit signed integer, otherwise the default is obtained from C/C++ code.
Closes#1108
errno needs setting otherwise a failed strol may result in subsequent
unnecessary code generation to use *args instead of actual default
argument values. Related to issue #1108.
Choose compile time failures over runtime errors.
This change highlighted a problem in the constructor wrappers for tp_init
where the kwargs parameter was missing in the generated wrapper function.
Leave casts in for 0 initialization.
1. Fix negative octals. Currently not handled correctly by `-py3`
(unusual case, but incorrect).
2. Fix arguments of type "octal + something" (e.g. `0640 | 04`).
Currently drops everything after the first octal. Nasty!
3. Fix bool arguments "0 + something" (e.g. `0 | 1`) are always
"False" (unusual case, but incorrect).
4. Remove special handling of "TRUE" and "FALSE" from
`convertValue` since there's no reason these have to match
"true" and "false".
5. Remove the Python 2 vs. Python 3 distinction based on the
`-py3` flag. Now the same python code is produced for default
arguments for Python 2 and Python 3. For this, octal default
arguments, e.g. 0644, are now wrapped as `int('644', 8)`. This
is required, as Python 2 and Python 3 have incompatible syntax
for octal literals.
Fixes#707
This ensures NotImplemented is returned on error so that the Python
interpreter will handle the operators correctly instead of throwing an
exception. NotImplemented was not being returned for non-builtin wrappers
when the operator overload did not have a function overload.
See PEP 207 and https://docs.python.org/3/library/constants.html#NotImplemented
Mentioned in SF patch #303 and SF bug #1208.
Fix Coverity issue reported for wrapper argument checking:
"Null-checking args suggests that it may be null, but it has already
been dereferenced on all paths leading to the check."
So 'args' is null checked, but after dereferencing it with
PyTuple_Check(args).
The way Python docstrings are indented has changed on master, so use the
standard inspect module in Python autodoc unit test to ignore the differences
in their indentation level between -builtin and non-builtin cases to make the
test still pass with the branch version, which avoids the use of different
(but almost identical) values in the test itself.
Add optional moduleimport attribute to %module so that the
default module import code can be overridden. See the
"Searching for the wrapper module" documentation in Python.html.
Example:
%module(moduleimport="import _foo") foo
$module also expands to the low-level C/C++ module name, so
the following is the same as above
%module(moduleimport="import $module") foo
Issue https://github.com/swig/swig/issues/769
The closure names used for builtin slots are mangled with their functype so
that overloaded C++ method names can be used for multiple slots.
For example:
%feature("python:slot", "mp_subscript", functype="binaryfunc") SimpleArray::__getitem__;
%feature("python:slot", "sq_item", functype="ssizeargfunc") SimpleArray::__getitem__(Py_ssize_t n);
will generate closures:
SWIGPY_SSIZEARGFUNC_CLOSURE(_wrap_SimpleArray___getitem__) /* defines _wrap_SimpleArray___getitem___ssizeargfunc_closure */
SWIGPY_BINARYFUNC_CLOSURE(_wrap_SimpleArray___getitem__) /* defines _wrap_SimpleArray___getitem___binaryfunc_closure */
Default hash is the underlying C/C++ pointer.
This matches up with testing for equivalence (Py_EQ in SwigPyObject_richcompare)
which compares the pointers.
William S Fulton