[Python-checkins] r82313 - in python/branches/py3k: Doc/library/importlib.rst Lib/importlib/abc.py Lib/importlib/test/source/test_abc_loader.py Misc/NEWS

brett.cannon python-checkins at python.org
Mon Jun 28 01:57:46 CEST 2010


Author: brett.cannon
Date: Mon Jun 28 01:57:46 2010
New Revision: 82313

Log:
Implement importlib.abc.SourceLoader and deprecate PyLoader and PyPycLoader.

SourceLoader is a simplification of both PyLoader and PyPycLoader. If one only
wants to use source, then they need to only implement get_data and
get_filename. To also use bytecode -- sourceless loading is not supported --
then two abstract methods -- path_mtime and set_data -- need to be implemented.
Compared to PyLoader and PyPycLoader, there are less abstract methods
introduced and bytecode files become an optimization controlled by the ABC and
hidden from the user (this need came about as PEP 3147 showed that not treating
bytecode as an optimization can cause problems for compatibility).

PyLoader is deprecated in favor of SourceLoader. To be compatible from Python
3.1 onwards, a subclass need only use simple methods for source_path and
is_package. Otherwise conditional subclassing based on whether Python 3.1 or
Python 3.2 is being is the only change. The documentation and docstring for
PyLoader explain what is exactly needed.

PyPycLoader is deprecated also in favor of SourceLoader. Because PEP 3147
shifted bytecode path details so much, there is no foolproof way to provide
backwards-compatibility with SourceLoader. Because of this the class is simply
deprecated and users should move to SourceLoader (and optionally PyLoader for
Python 3.1). This does lead to a loss of support for sourceless loading
unfortunately.

At some point before Python 3.2 is released, SourceLoader will be moved over to
importlib._bootstrap so that the core code of importlib relies on the new code
instead of the old PyPycLoader code. This commit is being done now so that
there is no issue in having the API in Python 3.1a1.


Modified:
   python/branches/py3k/Doc/library/importlib.rst
   python/branches/py3k/Lib/importlib/abc.py
   python/branches/py3k/Lib/importlib/test/source/test_abc_loader.py
   python/branches/py3k/Misc/NEWS

Modified: python/branches/py3k/Doc/library/importlib.rst
==============================================================================
--- python/branches/py3k/Doc/library/importlib.rst	(original)
+++ python/branches/py3k/Doc/library/importlib.rst	Mon Jun 28 01:57:46 2010
@@ -18,12 +18,12 @@
 :func:`__import__` function) in Python source code. This provides an
 implementation of :keyword:`import` which is portable to any Python
 interpreter. This also provides a reference implementation which is easier to
-comprehend than one in a programming language other than Python.
+comprehend than one implemented in a programming language other than Python.
 
-Two, the components to implement :keyword:`import` can be exposed in this
+Two, the components to implement :keyword:`import` are exposed in this
 package, making it easier for users to create their own custom objects (known
 generically as an :term:`importer`) to participate in the import process.
-Details on providing custom importers can be found in :pep:`302`.
+Details on custom importers can be found in :pep:`302`.
 
 .. seealso::
 
@@ -37,7 +37,7 @@
 
     The :func:`.__import__` function
         The built-in function for which the :keyword:`import` statement is
-        syntactic sugar.
+        syntactic sugar for.
 
     :pep:`235`
         Import on Case-Insensitive Platforms
@@ -46,7 +46,7 @@
         Defining Python Source Code Encodings
 
     :pep:`302`
-        New Import Hooks.
+        New Import Hooks
 
     :pep:`328`
         Imports: Multi-Line and Absolute/Relative
@@ -66,8 +66,7 @@
 
 .. function:: __import__(name, globals={}, locals={}, fromlist=list(), level=0)
 
-    An implementation of the built-in :func:`__import__` function. See the
-    built-in function's documentation for usage instructions.
+    An implementation of the built-in :func:`__import__` function.
 
 .. function:: import_module(name, package=None)
 
@@ -213,22 +212,108 @@
 
     .. method:: get_filename(fullname)
 
-        An abstract method that is to return the value for :attr:`__file__` for
+        An abstract method that is to return the value of :attr:`__file__` for
         the specified module. If no path is available, :exc:`ImportError` is
         raised.
 
+        If source code is available, then the method should return the path to
+        the source file, regardless of whether a bytecode was used to load the
+        module.
+
+
+.. class:: SourceLoader
+
+    An abstract base class for implementing source (and optionally bytecode)
+    file loading. The class inherits from both :class:`ResourceLoader` and
+    :class:`ExecutionLoader`, requiring the implementation of:
+
+    * :meth:`ResourceLoader.get_data`
+    * :meth:`ExecutionLoader.get_filename`
+          Implement to only return the path to the source file; sourceless
+          loading is not supported.
+
+    The abstract methods defined by this class are to add optional bytecode
+    file support. Not implementing these optional methods causes the loader to
+    only work with source code. Implementing the methods allows the loader to
+    work with source *and* bytecode files; it does not allow for *sourceless*
+    loading where only bytecode is provided.  Bytecode files are an
+    optimization to speed up loading by removing the parsing step of Python's
+    compiler, and so no bytecode-specific API is exposed.
+
+    .. method:: path_mtime(self, path)
+
+        Optional abstract method which returns the modification time for the
+        specified path.
+
+    .. method:: set_data(self, path, data)
+
+        Optional abstract method which writes the specified bytes to a file
+        path.
+
+    .. method:: get_code(self, fullname)
+
+        Concrete implementation of :meth:`InspectLoader.get_code`.
+
+    .. method:: load_module(self, fullname)
+
+        Concrete implementation of :meth:`Loader.load_module`.
+
+    .. method:: get_source(self, fullname)
+
+        Concrete implementation of :meth:`InspectLoader.get_source`.
+
+    .. method:: is_package(self, fullname)
+
+        Concrete implementation of :meth:`InspectLoader.is_package`. A module
+        is determined to be a package if its file path is a file named
+        ``__init__`` when the file extension is removed.
+
 
 .. class:: PyLoader
 
     An abstract base class inheriting from
-    :class:`importlib.abc.ExecutionLoader` and
-    :class:`importlib.abc.ResourceLoader` designed to ease the loading of
+    :class:`ExecutionLoader` and
+    :class:`ResourceLoader` designed to ease the loading of
     Python source modules (bytecode is not handled; see
-    :class:`importlib.abc.PyPycLoader` for a source/bytecode ABC). A subclass
+    :class:`SourceLoader` for a source/bytecode ABC). A subclass
     implementing this ABC will only need to worry about exposing how the source
     code is stored; all other details for loading Python source code will be
     handled by the concrete implementations of key methods.
 
+    .. deprecated:: 3.2
+        This class has been deprecated in favor of :class:`SourceLoader` and is
+        slated for removal in Python 3.4. See below for how to create a
+        subclass that is compatbile with Python 3.1 onwards.
+
+    If compatibility with Python 3.1 is required, then use the following idiom
+    to implement a subclass that will work with Python 3.1 onwards (make sure
+    to implement :meth:`ExecutionLoader.get_filename`)::
+
+        try:
+            from importlib.abc import SourceLoader
+        except ImportError:
+            from importlib.abc import PyLoader as SourceLoader
+
+
+        class CustomLoader(SourceLoader):
+            def get_filename(self, fullname):
+                """Return the path to the source file."""
+                # Implement ...
+
+            def source_path(self, fullname):
+                """Implement source_path in terms of get_filename."""
+                try:
+                    return self.get_filename(fullname)
+                except ImportError:
+                    return None
+
+            def is_package(self, fullname):
+                """Implement is_package by looking for an __init__ file
+                name as returned by get_filename."""
+                filename = os.path.basename(self.get_filename(fullname))
+                return os.path.splitext(filename)[0] == '__init__'
+
+
     .. method:: source_path(fullname)
 
         An abstract method that returns the path to the source code for a
@@ -270,10 +355,18 @@
 
 .. class:: PyPycLoader
 
-    An abstract base class inheriting from :class:`importlib.abc.PyLoader`.
+    An abstract base class inheriting from :class:`PyLoader`.
     This ABC is meant to help in creating loaders that support both Python
     source and bytecode.
 
+    .. deprecated:: 3.2
+        This class has been deprecated in favor of :class:`SourceLoader` and to
+        properly support :pep:`3147`. If compatibility is required with
+        Python 3.1, implement both :class:`SourceLoader` and :class:`PyLoader`;
+        instructions on how to do so are included in the documentation for
+        :class:`PyLoader`. Do note that this solution will not support
+        sourceless/bytecode-only loading; only source *and* bytecode loading.
+
     .. method:: source_mtime(fullname)
 
         An abstract method which returns the modification time for the source
@@ -292,8 +385,8 @@
     .. method:: get_filename(fullname)
 
         A concrete implementation of
-        :meth:`importlib.abc.ExecutionLoader.get_filename` that relies on
-        :meth:`importlib.abc.PyLoader.source_path` and :meth:`bytecode_path`.
+        :meth:`ExecutionLoader.get_filename` that relies on
+        :meth:`PyLoader.source_path` and :meth:`bytecode_path`.
         If :meth:`source_path` returns a path, then that value is returned.
         Else if :meth:`bytecode_path` returns a path, that path will be
         returned. If a path is not available from both methods,
@@ -420,100 +513,3 @@
     attribute to be used at the global level of the module during
     initialization.
 
-
-Example
--------
-
-Below is an example meta path importer that uses a dict for back-end storage
-for source code. While not an optimal solution -- manipulations of
-:attr:`__path__` on packages does not influence import -- it does illustrate
-what little is required to implement an importer.
-
-.. testcode::
-
-    """An importer where source is stored in a dict."""
-    from importlib import abc
-
-
-    class DictImporter(abc.Finder, abc.PyLoader):
-
-        """A meta path importer that stores source code in a dict.
-
-        The keys are the module names -- packages must end in ``.__init__``.
-        The values must be something that can be passed to 'bytes'.
-
-        """
-
-        def __init__(self, memory):
-            """Store the dict."""
-            self.memory = memory
-
-        def contains(self, name):
-            """See if a module or package is in the dict."""
-            if name in self.memory:
-                return name
-            package_name = '{}.__init__'.format(name)
-            if  package_name in self.memory:
-                return package_name
-            return False
-
-        __contains__ = contains  # Convenience.
-
-        def find_module(self, fullname, path=None):
-            """Find the module in the dict."""
-            if fullname in self:
-                return self
-            return None
-
-        def source_path(self, fullname):
-            """Return the module name if the module is in the dict."""
-            if not fullname in self:
-                raise ImportError
-            return fullname
-
-        def get_data(self, path):
-            """Return the bytes for the source.
-
-            The value found in the dict is passed through 'bytes' before being
-            returned.
-
-            """
-            name = self.contains(path)
-            if not name:
-                raise IOError
-            return bytes(self.memory[name])
-
-        def is_package(self, fullname):
-            """Tell if module is a package based on whether the dict contains the
-            name with ``.__init__`` appended to it."""
-            if fullname not in self:
-                raise ImportError
-            if fullname in self.memory:
-                return False
-            # If name is in this importer but not as it is then it must end in
-            # ``__init__``.
-            else:
-                return True
-
-.. testcode::
-    :hide:
-
-    import importlib
-    import sys
-
-
-    # Build the dict; keys of name, value of __package__.
-    names = {'_top_level': '', '_pkg.__init__': '_pkg', '_pkg.mod': '_pkg'}
-    source = {name: "name = {!r}".format(name).encode() for name in names}
-
-    # Register the meta path importer.
-    importer = DictImporter(source)
-    sys.meta_path.append(importer)
-
-    # Sanity check.
-    for name in names:
-        module = importlib.import_module(name)
-        assert module.__name__ == name
-        assert getattr(module, 'name') == name
-        assert module.__loader__ is importer
-        assert module.__package__ == names[name]

Modified: python/branches/py3k/Lib/importlib/abc.py
==============================================================================
--- python/branches/py3k/Lib/importlib/abc.py	(original)
+++ python/branches/py3k/Lib/importlib/abc.py	Mon Jun 28 01:57:46 2010
@@ -1,8 +1,16 @@
 """Abstract base classes related to import."""
 from . import _bootstrap
 from . import machinery
+from . import util
 import abc
+import imp
+import io
+import marshal
+import os.path
+import sys
+import tokenize
 import types
+import warnings
 
 
 class Loader(metaclass=abc.ABCMeta):
@@ -58,19 +66,19 @@
     def is_package(self, fullname:str) -> bool:
         """Abstract method which when implemented should return whether the
         module is a package."""
-        return NotImplementedError
+        raise NotImplementedError
 
     @abc.abstractmethod
     def get_code(self, fullname:str) -> types.CodeType:
         """Abstract method which when implemented should return the code object
         for the module"""
-        return NotImplementedError
+        raise NotImplementedError
 
     @abc.abstractmethod
     def get_source(self, fullname:str) -> str:
         """Abstract method which should return the source code for the
         module."""
-        return NotImplementedError
+        raise NotImplementedError
 
 InspectLoader.register(machinery.BuiltinImporter)
 InspectLoader.register(machinery.FrozenImporter)
@@ -92,33 +100,273 @@
         raise NotImplementedError
 
 
-class PyLoader(_bootstrap.PyLoader, ResourceLoader, ExecutionLoader):
+class SourceLoader(ResourceLoader, ExecutionLoader):
 
-    """Abstract base class to assist in loading source code by requiring only
-    back-end storage methods to be implemented.
+    """Abstract base class for loading source code (and optionally any
+    corresponding bytecode).
 
-    The methods get_code, get_source, and load_module are implemented for the
-    user.
+    To support loading from source code, the abstractmethods inherited from
+    ResourceLoader and ExecutionLoader need to be implemented. To also support
+    loading from bytecode, the optional methods specified directly by this ABC
+    is required.
+
+    Inherited abstractmethods not implemented in this ABC:
+
+        * ResourceLoader.get_data
+        * ExecutionLoader.get_filename
+
+    """
+
+    def path_mtime(self, path:str) -> int:
+        """Optional method that returns the modification time for the specified
+        path.
+
+        Implementing this method allows the loader to read bytecode files.
+
+        """
+        raise NotImplementedError
+
+    def set_data(self, path:str, data:bytes) -> None:
+        """Optional method which writes data to a file path.
+
+        Implementing this method allows for the writing of bytecode files.
+
+        """
+        raise NotImplementedError
+
+    def is_package(self, fullname):
+        """Concrete implementation of InspectLoader.is_package by checking if
+        the path returned by get_filename has a filename of '__init__.py'."""
+        filename = os.path.basename(self.get_filename(fullname))
+        return os.path.splitext(filename)[0] == '__init__'
+
+    def get_source(self, fullname):
+        """Concrete implementation of InspectLoader.get_source."""
+        path = self.get_filename(fullname)
+        try:
+            source_bytes = self.get_data(path)
+        except IOError:
+            raise ImportError("source not available through get_data()")
+        encoding = tokenize.detect_encoding(io.BytesIO(source_bytes).readline)
+        return source_bytes.decode(encoding[0])
+
+    def get_code(self, fullname):
+        """Concrete implementation of InspectLoader.get_code.
+
+        Reading of bytecode requires path_mtime to be implemented. To write
+        bytecode, set_data must also be implemented.
+
+        """
+        source_path = self.get_filename(fullname)
+        bytecode_path = imp.cache_from_source(source_path)
+        source_mtime = None
+        if bytecode_path is not None:
+            try:
+                source_mtime = self.path_mtime(source_path)
+            except NotImplementedError:
+                pass
+            else:
+                try:
+                    data = self.get_data(bytecode_path)
+                except IOError:
+                    pass
+                else:
+                    magic = data[:4]
+                    raw_timestamp = data[4:8]
+                    if (len(magic) == 4 and len(raw_timestamp) == 4 and
+                            magic == imp.get_magic() and
+                            marshal._r_long(raw_timestamp) == source_mtime):
+                        return marshal.loads(data[8:])
+        source_bytes = self.get_data(source_path)
+        code_object = compile(source_bytes, source_path, 'exec',
+                                dont_inherit=True)
+        if (not sys.dont_write_bytecode and bytecode_path is not None and
+                source_mtime is not None):
+            # If e.g. Jython ever implements imp.cache_from_source to have
+            # their own cached file format, this block of code will most likely
+            # throw an exception.
+            data = bytearray(imp.get_magic())
+            data.extend(marshal._w_long(source_mtime))
+            data.extend(marshal.dumps(code_object))
+            try:
+                self.set_data(bytecode_path, data)
+            except (NotImplementedError, IOError):
+                pass
+        return code_object
+
+    @util.module_for_loader
+    def load_module(self, module):
+        """Concrete implementation of Loader.load_module.
+
+        Requires ExecutionLoader.get_filename and ResourceLoader.get_data to be
+        implemented to load source code. Use of bytecode is dictated by whether
+        get_code uses/writes bytecode.
+
+        """
+        name = module.__name__
+        code_object = self.get_code(name)
+        module.__file__ = self.get_filename(name)
+        module.__cached__ = imp.cache_from_source(module.__file__)
+        module.__package__ = name
+        is_package = self.is_package(name)
+        if is_package:
+            module.__path__  = [os.path.dirname(module.__file__)]
+        else:
+            module.__package__ = module.__package__.rpartition('.')[0]
+        module.__loader__ = self
+        exec(code_object, module.__dict__)
+        return module
+
+
+class PyLoader(SourceLoader):
+
+    """Implement the deprecated PyLoader ABC in terms of SourceLoader.
+
+    This class has been deprecated! It is slated for removal in Python 3.4.
+    If compatibility with Python 3.1 is not needed then implement the
+    SourceLoader ABC instead of this class. If Python 3.1 compatibility is
+    needed, then use the following idiom to have a single class that is
+    compatible with Python 3.1 onwards::
+
+        try:
+            from importlib.abc import SourceLoader
+        except ImportError:
+            from importlib.abc import PyLoader as SourceLoader
+
+
+        class CustomLoader(SourceLoader):
+            def get_filename(self, fullname):
+                # Implement ...
+
+            def source_path(self, fullname):
+                '''Implement source_path in terms of get_filename.'''
+                try:
+                    return self.get_filename(fullname)
+                except ImportError:
+                    return None
+
+            def is_package(self, fullname):
+                filename = os.path.basename(self.get_filename(fullname))
+                return os.path.splitext(filename)[0] == '__init__'
 
     """
 
     @abc.abstractmethod
+    def is_package(self, fullname):
+        raise NotImplementedError
+
+    @abc.abstractmethod
     def source_path(self, fullname:str) -> object:
         """Abstract method which when implemented should return the path to the
-        sourced code for the module."""
+        source code for the module."""
         raise NotImplementedError
 
+    def get_filename(self, fullname):
+        """Implement get_filename in terms of source_path.
 
-class PyPycLoader(_bootstrap.PyPycLoader, PyLoader):
+        As get_filename should only return a source file path there is no
+        chance of the path not existing but loading still being possible, so
+        ImportError should propagate instead of being turned into returning
+        None.
+
+        """
+        warnings.warn("importlib.abc.PyLoader is deprecated and is "
+                            "slated for removal in Python 3.4; "
+                            "use SourceLoader instead. "
+                            "See the importlib documentation on how to be "
+                            "compatible with Python 3.1 onwards.",
+                        PendingDeprecationWarning)
+        path = self.source_path(fullname)
+        if path is None:
+            raise ImportError
+        else:
+            return path
+
+PyLoader.register(_bootstrap.PyLoader)
+
+
+class PyPycLoader(PyLoader):
 
     """Abstract base class to assist in loading source and bytecode by
     requiring only back-end storage methods to be implemented.
 
+    This class has been deprecated! Removal is slated for Python 3.4. Implement
+    the SourceLoader ABC instead. If Python 3.1 compatibility is needed, see
+    PyLoader.
+
     The methods get_code, get_source, and load_module are implemented for the
     user.
 
     """
 
+    def get_filename(self, fullname):
+        """Return the source or bytecode file path."""
+        path = self.source_path(fullname)
+        if path is not None:
+            return path
+        path = self.bytecode_path(fullname)
+        if path is not None:
+            return path
+        raise ImportError("no source or bytecode path available for "
+                            "{0!r}".format(fullname))
+
+    def get_code(self, fullname):
+        """Get a code object from source or bytecode."""
+        warnings.warn("importlib.abc.PyPycLoader is deprecated and slated for "
+                            "removal in Python 3.4; use SourceLoader instead. "
+                            "If Python 3.1 compatibility is required, see the "
+                            "latest documentation for PyLoader.",
+                        PendingDeprecationWarning)
+        source_timestamp = self.source_mtime(fullname)
+        # Try to use bytecode if it is available.
+        bytecode_path = self.bytecode_path(fullname)
+        if bytecode_path:
+            data = self.get_data(bytecode_path)
+            try:
+                magic = data[:4]
+                if len(magic) < 4:
+                    raise ImportError("bad magic number in {}".format(fullname))
+                raw_timestamp = data[4:8]
+                if len(raw_timestamp) < 4:
+                    raise EOFError("bad timestamp in {}".format(fullname))
+                pyc_timestamp = marshal._r_long(raw_timestamp)
+                bytecode = data[8:]
+                # Verify that the magic number is valid.
+                if imp.get_magic() != magic:
+                    raise ImportError("bad magic number in {}".format(fullname))
+                # Verify that the bytecode is not stale (only matters when
+                # there is source to fall back on.
+                if source_timestamp:
+                    if pyc_timestamp < source_timestamp:
+                        raise ImportError("bytecode is stale")
+            except (ImportError, EOFError):
+                # If source is available give it a shot.
+                if source_timestamp is not None:
+                    pass
+                else:
+                    raise
+            else:
+                # Bytecode seems fine, so try to use it.
+                return marshal.loads(bytecode)
+        elif source_timestamp is None:
+            raise ImportError("no source or bytecode available to create code "
+                                "object for {0!r}".format(fullname))
+        # Use the source.
+        source_path = self.source_path(fullname)
+        if source_path is None:
+            message = "a source path must exist to load {0}".format(fullname)
+            raise ImportError(message)
+        source = self.get_data(source_path)
+        code_object = compile(source, source_path, 'exec', dont_inherit=True)
+        # Generate bytecode and write it out.
+        if not sys.dont_write_bytecode:
+            data = bytearray(imp.get_magic())
+            data.extend(marshal._w_long(source_timestamp))
+            data.extend(marshal.dumps(code_object))
+            self.write_bytecode(fullname, data)
+        return code_object
+
+
     @abc.abstractmethod
     def source_mtime(self, fullname:str) -> int:
         """Abstract method which when implemented should return the
@@ -137,3 +385,5 @@
         bytecode for the module, returning a boolean representing whether the
         bytecode was written or not."""
         raise NotImplementedError
+
+PyPycLoader.register(_bootstrap.PyPycLoader)

Modified: python/branches/py3k/Lib/importlib/test/source/test_abc_loader.py
==============================================================================
--- python/branches/py3k/Lib/importlib/test/source/test_abc_loader.py	(original)
+++ python/branches/py3k/Lib/importlib/test/source/test_abc_loader.py	Mon Jun 28 01:57:46 2010
@@ -1,14 +1,67 @@
 import importlib
 from importlib import abc
+
 from .. import abc as testing_abc
 from .. import util
 from . import util as source_util
+
 import imp
+import inspect
 import marshal
 import os
 import sys
 import types
 import unittest
+import warnings
+
+
+class SourceOnlyLoaderMock(abc.SourceLoader):
+
+    # Globals that should be defined for all modules.
+    source = (b"_ = '::'.join([__name__, __file__, __cached__, __package__, "
+              b"repr(__loader__)])")
+
+    def __init__(self, path):
+        self.path = path
+
+    def get_data(self, path):
+        assert self.path == path
+        return self.source
+
+    def get_filename(self, fullname):
+        return self.path
+
+
+class SourceLoaderMock(SourceOnlyLoaderMock):
+
+    source_mtime = 1
+
+    def __init__(self, path, magic=imp.get_magic()):
+        super().__init__(path)
+        self.bytecode_path = imp.cache_from_source(self.path)
+        data = bytearray(magic)
+        data.extend(marshal._w_long(self.source_mtime))
+        code_object = compile(self.source, self.path, 'exec',
+                                dont_inherit=True)
+        data.extend(marshal.dumps(code_object))
+        self.bytecode = bytes(data)
+        self.written = {}
+
+    def get_data(self, path):
+        if path == self.path:
+            return super().get_data(path)
+        elif path == self.bytecode_path:
+            return self.bytecode
+        else:
+            raise IOError
+
+    def path_mtime(self, path):
+        assert path == self.path
+        return self.source_mtime
+
+    def set_data(self, path, data):
+        self.written[path] = bytes(data)
+        return path == self.bytecode_path
 
 
 class PyLoaderMock(abc.PyLoader):
@@ -33,17 +86,42 @@
         return self.source
 
     def is_package(self, name):
+        filename = os.path.basename(self.get_filename(name))
+        return os.path.splitext(filename)[0] == '__init__'
+
+    def source_path(self, name):
         try:
-            return '__init__' in self.module_paths[name]
+            return self.module_paths[name]
         except KeyError:
             raise ImportError
 
-    def source_path(self, name):
+    def get_filename(self, name):
+        """Silence deprecation warning."""
+        with warnings.catch_warnings(record=True) as w:
+            warnings.simplefilter("always")
+            path = super().get_filename(name)
+            assert len(w) == 1
+            assert issubclass(w[0].category, PendingDeprecationWarning)
+            return path
+
+
+class PyLoaderCompatMock(PyLoaderMock):
+
+    """Mock that matches what is suggested to have a loader that is compatible
+    from Python 3.1 onwards."""
+
+    def get_filename(self, fullname):
         try:
-            return self.module_paths[name]
+            return self.module_paths[fullname]
         except KeyError:
             raise ImportError
 
+    def source_path(self, fullname):
+        try:
+            return self.get_filename(fullname)
+        except ImportError:
+            return None
+
 
 class PyPycLoaderMock(abc.PyPycLoader, PyLoaderMock):
 
@@ -114,6 +192,13 @@
         except TypeError:
             return '__init__' in self.bytecode_to_path[name]
 
+    def get_code(self, name):
+        with warnings.catch_warnings(record=True) as w:
+            warnings.simplefilter("always")
+            code_object = super().get_code(name)
+            assert len(w) == 1
+            assert issubclass(w[0].category, PendingDeprecationWarning)
+            return code_object
 
 class PyLoaderTests(testing_abc.LoaderTests):
 
@@ -200,6 +285,14 @@
         return mock
 
 
+class PyLoaderCompatTests(PyLoaderTests):
+
+    """Test that the suggested code to make a loader that is compatible from
+    Python 3.1 forward works."""
+
+    mocker = PyLoaderCompatMock
+
+
 class PyLoaderInterfaceTests(unittest.TestCase):
 
     """Tests for importlib.abc.PyLoader to make sure that when source_path()
@@ -413,7 +506,7 @@
     def test_bad_bytecode(self):
         # Malformed code object bytecode should lead to a ValueError.
         name = 'mod'
-        bc = {name: {'path': os.path.join('path', 'to', 'mod'), 'bc': b'XXX'}}
+        bc = {name: {'path': os.path.join('path', 'to', 'mod'), 'bc': b'NNN'}}
         mock = PyPycLoaderMock({name: None}, bc)
         with util.uncache(name), self.assertRaises(ValueError):
             mock.load_module(name)
@@ -465,12 +558,307 @@
             mock.load_module(name)
 
 
+class SourceLoaderTestHarness(unittest.TestCase):
+
+    def setUp(self, *, is_package=True, **kwargs):
+        self.package = 'pkg'
+        if is_package:
+            self.path = os.path.join(self.package, '__init__.py')
+            self.name = self.package
+        else:
+            module_name = 'mod'
+            self.path = os.path.join(self.package, '.'.join(['mod', 'py']))
+            self.name = '.'.join([self.package, module_name])
+        self.cached = imp.cache_from_source(self.path)
+        self.loader = self.loader_mock(self.path, **kwargs)
+
+    def verify_module(self, module):
+        self.assertEqual(module.__name__, self.name)
+        self.assertEqual(module.__file__, self.path)
+        self.assertEqual(module.__cached__, self.cached)
+        self.assertEqual(module.__package__, self.package)
+        self.assertEqual(module.__loader__, self.loader)
+        values = module._.split('::')
+        self.assertEqual(values[0], self.name)
+        self.assertEqual(values[1], self.path)
+        self.assertEqual(values[2], self.cached)
+        self.assertEqual(values[3], self.package)
+        self.assertEqual(values[4], repr(self.loader))
+
+    def verify_code(self, code_object):
+        module = imp.new_module(self.name)
+        module.__file__ = self.path
+        module.__cached__ = self.cached
+        module.__package__ = self.package
+        module.__loader__ = self.loader
+        module.__path__ = []
+        exec(code_object, module.__dict__)
+        self.verify_module(module)
+
+
+class SourceOnlyLoaderTests(SourceLoaderTestHarness):
+
+    """Test importlib.abc.SourceLoader for source-only loading.
+
+    Reload testing is subsumed by the tests for
+    importlib.util.module_for_loader.
+
+    """
+
+    loader_mock = SourceOnlyLoaderMock
+
+    def test_get_source(self):
+        # Verify the source code is returned as a string.
+        # If an IOError is raised by get_data then raise ImportError.
+        expected_source = self.loader.source.decode('utf-8')
+        self.assertEqual(self.loader.get_source(self.name), expected_source)
+        def raise_IOError(path):
+            raise IOError
+        self.loader.get_data = raise_IOError
+        with self.assertRaises(ImportError):
+            self.loader.get_source(self.name)
+
+    def test_is_package(self):
+        # Properly detect when loading a package.
+        self.setUp(is_package=True)
+        self.assertTrue(self.loader.is_package(self.name))
+        self.setUp(is_package=False)
+        self.assertFalse(self.loader.is_package(self.name))
+
+    def test_get_code(self):
+        # Verify the code object is created.
+        code_object = self.loader.get_code(self.name)
+        self.verify_code(code_object)
+
+    def test_load_module(self):
+        # Loading a module should set __name__, __loader__, __package__,
+        # __path__ (for packages), __file__, and __cached__.
+        # The module should also be put into sys.modules.
+        with util.uncache(self.name):
+            module = self.loader.load_module(self.name)
+            self.verify_module(module)
+            self.assertEqual(module.__path__, [os.path.dirname(self.path)])
+            self.assertTrue(self.name in sys.modules)
+
+    def test_package_settings(self):
+        # __package__ needs to be set, while __path__ is set on if the module
+        # is a package.
+        # Testing the values for a package are covered by test_load_module.
+        self.setUp(is_package=False)
+        with util.uncache(self.name):
+            module = self.loader.load_module(self.name)
+            self.verify_module(module)
+            self.assertTrue(not hasattr(module, '__path__'))
+
+    def test_get_source_encoding(self):
+        # Source is considered encoded in UTF-8 by default unless otherwise
+        # specified by an encoding line.
+        source = "_ = 'ü'"
+        self.loader.source = source.encode('utf-8')
+        returned_source = self.loader.get_source(self.name)
+        self.assertEqual(returned_source, source)
+        source = "# coding: latin-1\n_ = ü"
+        self.loader.source = source.encode('latin-1')
+        returned_source = self.loader.get_source(self.name)
+        self.assertEqual(returned_source, source)
+
+
+ at unittest.skipIf(sys.dont_write_bytecode, "sys.dont_write_bytecode is true")
+class SourceLoaderBytecodeTests(SourceLoaderTestHarness):
+
+    """Test importlib.abc.SourceLoader's use of bytecode.
+
+    Source-only testing handled by SourceOnlyLoaderTests.
+
+    """
+
+    loader_mock = SourceLoaderMock
+
+    def verify_code(self, code_object, *, bytecode_written=False):
+        super().verify_code(code_object)
+        if bytecode_written:
+            self.assertIn(self.cached, self.loader.written)
+            data = bytearray(imp.get_magic())
+            data.extend(marshal._w_long(self.loader.source_mtime))
+            data.extend(marshal.dumps(code_object))
+            self.assertEqual(self.loader.written[self.cached], bytes(data))
+
+    def test_code_with_everything(self):
+        # When everything should work.
+        code_object = self.loader.get_code(self.name)
+        self.verify_code(code_object)
+
+    def test_no_bytecode(self):
+        # If no bytecode exists then move on to the source.
+        self.loader.bytecode_path = "<does not exist>"
+        # Sanity check
+        with self.assertRaises(IOError):
+            bytecode_path = imp.cache_from_source(self.path)
+            self.loader.get_data(bytecode_path)
+        code_object = self.loader.get_code(self.name)
+        self.verify_code(code_object, bytecode_written=True)
+
+    def test_code_bad_timestamp(self):
+        # Bytecode is only used when the timestamp matches the source EXACTLY.
+        for source_mtime in (0, 2):
+            assert source_mtime != self.loader.source_mtime
+            original = self.loader.source_mtime
+            self.loader.source_mtime = source_mtime
+            # If bytecode is used then EOFError would be raised by marshal.
+            self.loader.bytecode = self.loader.bytecode[8:]
+            code_object = self.loader.get_code(self.name)
+            self.verify_code(code_object, bytecode_written=True)
+            self.loader.source_mtime = original
+
+    def test_code_bad_magic(self):
+        # Skip over bytecode with a bad magic number.
+        self.setUp(magic=b'0000')
+        # If bytecode is used then EOFError would be raised by marshal.
+        self.loader.bytecode = self.loader.bytecode[8:]
+        code_object = self.loader.get_code(self.name)
+        self.verify_code(code_object, bytecode_written=True)
+
+    def test_dont_write_bytecode(self):
+        # Bytecode is not written if sys.dont_write_bytecode is true.
+        # Can assume it is false already thanks to the skipIf class decorator.
+        try:
+            sys.dont_write_bytecode = True
+            self.loader.bytecode_path = "<does not exist>"
+            code_object = self.loader.get_code(self.name)
+            self.assertNotIn(self.cached, self.loader.written)
+        finally:
+            sys.dont_write_bytecode = False
+
+    def test_no_set_data(self):
+        # If set_data is not defined, one can still read bytecode.
+        self.setUp(magic=b'0000')
+        original_set_data = self.loader.__class__.set_data
+        try:
+            del self.loader.__class__.set_data
+            code_object = self.loader.get_code(self.name)
+            self.verify_code(code_object)
+        finally:
+            self.loader.__class__.set_data = original_set_data
+
+    def test_set_data_raises_exceptions(self):
+        # Raising NotImplementedError or IOError is okay for set_data.
+        def raise_exception(exc):
+            def closure(*args, **kwargs):
+                raise exc
+            return closure
+
+        self.setUp(magic=b'0000')
+        for exc in (NotImplementedError, IOError):
+            self.loader.set_data = raise_exception(exc)
+            code_object = self.loader.get_code(self.name)
+            self.verify_code(code_object)
+
+class AbstractMethodImplTests(unittest.TestCase):
+
+    """Test the concrete abstractmethod implementations."""
+
+    class Loader(abc.Loader):
+        def load_module(self, fullname):
+            super().load_module(fullname)
+
+    class Finder(abc.Finder):
+        def find_module(self, _):
+            super().find_module(_)
+
+    class ResourceLoader(Loader, abc.ResourceLoader):
+        def get_data(self, _):
+            super().get_data(_)
+
+    class InspectLoader(Loader, abc.InspectLoader):
+        def is_package(self, _):
+            super().is_package(_)
+
+        def get_code(self, _):
+            super().get_code(_)
+
+        def get_source(self, _):
+            super().get_source(_)
+
+    class ExecutionLoader(InspectLoader, abc.ExecutionLoader):
+        def get_filename(self, _):
+            super().get_filename(_)
+
+    class SourceLoader(ResourceLoader, ExecutionLoader, abc.SourceLoader):
+        pass
+
+    class PyLoader(ResourceLoader, InspectLoader, abc.PyLoader):
+        def source_path(self, _):
+            super().source_path(_)
+
+    class PyPycLoader(PyLoader, abc.PyPycLoader):
+        def bytecode_path(self, _):
+            super().bytecode_path(_)
+
+        def source_mtime(self, _):
+            super().source_mtime(_)
+
+        def write_bytecode(self, _, _2):
+            super().write_bytecode(_, _2)
+
+    def raises_NotImplementedError(self, ins, *args):
+        for method_name in args:
+            method = getattr(ins, method_name)
+            arg_count = len(inspect.getfullargspec(method)[0]) - 1
+            args = [''] * arg_count
+            try:
+                method(*args)
+            except NotImplementedError:
+                pass
+            else:
+                msg = "{}.{} did not raise NotImplementedError"
+                self.fail(msg.format(ins.__class__.__name__, method_name))
+
+    def test_Loader(self):
+        self.raises_NotImplementedError(self.Loader(), 'load_module')
+
+    def test_Finder(self):
+        self.raises_NotImplementedError(self.Finder(), 'find_module')
+
+    def test_ResourceLoader(self):
+        self.raises_NotImplementedError(self.ResourceLoader(), 'load_module',
+                                        'get_data')
+
+    def test_InspectLoader(self):
+        self.raises_NotImplementedError(self.InspectLoader(), 'load_module',
+                                        'is_package', 'get_code', 'get_source')
+
+    def test_ExecutionLoader(self):
+        self.raises_NotImplementedError(self.ExecutionLoader(), 'load_module',
+                                        'is_package', 'get_code', 'get_source',
+                                        'get_filename')
+
+    def test_SourceLoader(self):
+        ins = self.SourceLoader()
+        # Required abstractmethods.
+        self.raises_NotImplementedError(ins, 'get_filename', 'get_data')
+        # Optional abstractmethods.
+        self.raises_NotImplementedError(ins,'path_mtime', 'set_data')
+
+    def test_PyLoader(self):
+        self.raises_NotImplementedError(self.PyLoader(), 'source_path',
+                                        'get_data', 'is_package')
+
+    def test_PyPycLoader(self):
+        self.raises_NotImplementedError(self.PyPycLoader(), 'source_path',
+                                        'source_mtime', 'bytecode_path',
+                                        'write_bytecode')
+
+
 def test_main():
     from test.support import run_unittest
-    run_unittest(PyLoaderTests, PyLoaderInterfaceTests, PyLoaderGetSourceTests,
+    run_unittest(PyLoaderTests, PyLoaderCompatTests,
+                    PyLoaderInterfaceTests, PyLoaderGetSourceTests,
                     PyPycLoaderTests, PyPycLoaderInterfaceTests,
                     SkipWritingBytecodeTests, RegeneratedBytecodeTests,
-                    BadBytecodeFailureTests, MissingPathsTests)
+                    BadBytecodeFailureTests, MissingPathsTests,
+                    SourceOnlyLoaderTests,
+                    SourceLoaderBytecodeTests,
+                    AbstractMethodImplTests)
 
 
 if __name__ == '__main__':

Modified: python/branches/py3k/Misc/NEWS
==============================================================================
--- python/branches/py3k/Misc/NEWS	(original)
+++ python/branches/py3k/Misc/NEWS	Mon Jun 28 01:57:46 2010
@@ -456,6 +456,9 @@
 Library
 -------
 
+- Implement importlib.abc.SourceLoader and deprecate PyLoader and PyPycLoader
+  for removal in Python 3.4.
+
 - Issue #9064: pdb's "up" and "down" commands now accept an optional argument.
 
 - Issue #9018: os.path.normcase() now raises a TypeError if the argument is


More information about the Python-checkins mailing list