Massive API refactoring; see setuptools.txt changelog for details. Also,

add ``#egg=project-version`` link support, and docs on how to make your
package available for EasyInstall to find.

--HG--
branch : setuptools
extra : convert_revision : svn%3A6015fed2-1504-0410-9fe1-9d1591cc4771/sandbox/trunk/setuptools%4041135

10 files changed, 3420 insertions, 250 deletions

diff --git a/EasyInstall.txt b/EasyInstall.txt
index fb72b14c..33e23f09 100755
--- a/EasyInstall.txt
+++ b/EasyInstall.txt
@@ -623,6 +623,22 @@ Known Issues
    Made ``easy-install.pth`` work in platform-specific alternate site
    directories (e.g. ``~/Library/Python/2.x/site-packages``).
 
+ * If you manually delete the current version of a package, the next run of
+   EasyInstall against the target directory will now remove the stray entry
+   from the ``easy-install.pth``file.
+
+ * EasyInstall now recognizes URLs with a ``#egg=project_name`` fragment ID
+   as pointing to the named project's source checkout.  Such URLs have a lower
+   match precedence than any other kind of distribution, so they'll only be
+   used if they have a higher version number than any other available
+   distribution.  (Future versions may allow you to specify that you want to
+   use source checkouts instead of other kinds of distributions.)  The ``#egg``
+   fragment can contain a version if it's formatted as ``#egg=proj-ver``,
+   where ``proj`` is the project name, and ``ver`` is the version number.  You
+   *must* use the format for these values that the ``bdist_egg`` command uses;
+   i.e., all non-alphanumeric runs must be condensed to single underscore
+   characters.
+
 0.5a12
  * Fix ``python -m easy_install`` not working due to setuptools being installed
    as a zipfile.  Update safety scanner to check for modules that might be used
diff --git a/api_tests.txt b/api_tests.txt
new file mode 100755
index 00000000..0c39bcd7
--- /dev/null
+++ b/api_tests.txt
@@ -0,0 +1,247 @@
+Pluggable Distributions of Python Software
+==========================================
+
+Distributions
+-------------
+
+A "Distribution" is a collection of files that represent a "Release" of a
+"Project" as of a particular point in time, denoted by a
+"Version"::
+
+    >>> import sys, pkg_resources
+    >>> from pkg_resources import Distribution
+    >>> Distribution(project_name="Foo", version="1.2")
+    Foo 1.2
+
+Distributions have a location, which can be a filename, URL, or really anything
+else you care to use::
+
+    >>> dist = Distribution(
+    ...     location="http://example.com/something",
+    ...     project_name="Bar", version="0.9"
+    ... )
+
+    >>> dist
+    Bar 0.9 (http://example.com/something)
+
+
+Distributions have various introspectable attributes::
+
+    >>> dist.location
+    'http://example.com/something'
+
+    >>> dist.project_name
+    'Bar'
+
+    >>> dist.version
+    '0.9'
+
+    >>> dist.py_version == sys.version[:3]
+    True
+
+    >>> print dist.platform
+    None
+
+Including various computed attributes::
+
+    >>> from pkg_resources import parse_version
+    >>> dist.parsed_version == parse_version(dist.version)
+    True
+
+    >>> dist.key    # case-insensitive form of the project name
+    'bar'
+
+Distributions are compared (and hashed) by version first::
+
+    >>> Distribution(version='1.0') == Distribution(version='1.0')
+    True
+    >>> Distribution(version='1.0') == Distribution(version='1.1')
+    False
+    >>> Distribution(version='1.0') <  Distribution(version='1.1')
+    True
+
+but also by project name (case-insensitive), platform, Python version,
+location, etc.::
+
+    >>> Distribution(project_name="Foo",version="1.0") == \
+    ... Distribution(project_name="Foo",version="1.0")
+    True
+
+    >>> Distribution(project_name="Foo",version="1.0") == \
+    ... Distribution(project_name="foo",version="1.0")
+    True
+
+    >>> Distribution(project_name="Foo",version="1.0") == \
+    ... Distribution(project_name="Foo",version="1.1")
+    False
+
+    >>> Distribution(project_name="Foo",py_version="2.3",version="1.0") == \
+    ... Distribution(project_name="Foo",py_version="2.4",version="1.0")
+    False
+
+    >>> Distribution(location="spam",version="1.0") == \
+    ... Distribution(location="spam",version="1.0")
+    True
+
+    >>> Distribution(location="spam",version="1.0") == \
+    ... Distribution(location="baz",version="1.0")
+    False
+
+
+
+Hash and compare distribution by prio/plat
+
+Get version from metadata
+provider capabilities
+egg_name()
+as_requirement()
+from_location, from_filename (w/path normalization)
+
+Releases may have zero or more "Requirements", which indicate
+what releases of another project the release requires in order to
+function.  A Requirement names the other project, expresses some criteria
+as to what releases of that project are acceptable, and lists any "Extras"
+that the requiring release may need from that project.  (An Extra is an
+optional feature of a Release, that can only be used if its additional
+Requirements are satisfied.)
+
+
+
+The Working Set
+---------------
+
+A collection of active distributions is called a Working Set.  Note that a
+Working Set can contain any importable distribution, not just pluggable ones.
+For example, the Python standard library is an importable distribution that
+will usually be part of the Working Set, even though it is not pluggable.
+Similarly, when you are doing development work on a project, the files you are
+editing are also a Distribution.  (And, with a little attention to the
+directory names used,  and including some additional metadata, such a
+"development distribution" can be made pluggable as well.)
+
+    >>> from pkg_resources import WorkingSet
+
+A working set's entries are the sys.path entries that correspond to the active
+distributions.  By default, the working set's entries are the items on
+``sys.path``::
+
+    >>> ws = WorkingSet()
+    >>> ws.entries == sys.path
+    True
+
+But you can also create an empty working set explicitly, and add distributions
+to it::
+
+    >>> ws = WorkingSet([])
+    >>> ws.add(dist)
+    >>> ws.entries
+    ['http://example.com/something']
+    >>> dist in ws
+    True
+    >>> Distribution('foo') in ws
+    False
+
+And you can iterate over its distributions::
+
+    >>> list(ws)
+    [Bar 0.9 (http://example.com/something)]
+
+Adding the same distribution more than once is a no-op::
+
+    >>> ws.add(dist)
+    >>> list(ws)
+    [Bar 0.9 (http://example.com/something)]
+
+For that matter, adding multiple distributions for the same project also does
+nothing, because a working set can only hold one active distribution per
+project -- the first one added to it::
+
+    >>> ws.add(
+    ...     Distribution(
+    ...         'http://example.com/something', project_name="Bar",
+    ...         version="7.2"
+    ...     )
+    ... )
+    >>> list(ws)
+    [Bar 0.9 (http://example.com/something)]
+
+You can append a path entry to a working set using ``add_entry()``::
+
+    >>> ws.entries
+    ['http://example.com/something']
+    >>> ws.add_entry(pkg_resources.__file__)
+    >>> ws.entries
+    ['http://example.com/something', '...pkg_resources.py...']
+
+Multiple additions result in multiple entries, even if the entry is already in
+the working set (because ``sys.path`` can contain the same entry more than
+once)::
+
+    >>> ws.add_entry(pkg_resources.__file__)
+    >>> ws.entries
+    ['...example.com...', '...pkg_resources...', '...pkg_resources...']
+
+And you can specify the path entry a distribution was found under, using the
+optional second parameter to ``add()``
+
+    >>> ws.add(dist,"foo")
+    >>> ws.add(dist,"bar")
+    >>> ws.entries
+    ['http://example.com/something', ..., 'foo', 'bar']
+
+But even if a distribution is found under multiple path entries, it still only
+shows up once when iterating the working set:
+
+    >>> list(ws)
+    [Bar 0.9 (http://example.com/something)]
+
+You can ask a WorkingSet to ``find()`` a distribution matching a requirement::
+
+    >>> from pkg_resources import Requirement
+    >>> print ws.find(Requirement.parse("Foo==1.0"))    # no match, return None
+    None
+
+    >>> ws.find(Requirement.parse("Bar==0.9"))  # match, return distribution
+    Bar 0.9 (http://example.com/something)
+
+Note that asking for a conflicting version of a distribution already in a
+working set triggers a ``pkg_resources.VersionConflict`` error:
+
+    >>> ws.find(Requirement.parse("Bar==1.0")) # doctest: +NORMALIZE_WHITESPACE
+    Traceback (most recent call last):
+      ...
+    VersionConflict: (Bar 0.9 (http://example.com/something),
+                      Requirement.parse('Bar==1.0'))
+
+You can subscribe a callback function to receive notifications whenever a new
+distribution is added to a working set.  The callback is immediately invoked
+once for each existing distribution in the working set, and then is called
+again for new distributions added thereafter::
+
+    >>> def added(dist): print "Added", dist
+    >>> ws.subscribe(added)
+    Added Bar 0.9
+    >>> foo12 = Distribution(project_name="Foo", version="1.2") 
+    >>> ws.add(foo12)
+    Added Foo 1.2
+
+Note, however, that only the first distribution added for a given project name
+will trigger a callback, even during the initial ``subscribe()`` callback::
+
+    >>> foo14 = Distribution(project_name="Foo", version="1.4") 
+    >>> ws.add(foo14)   # no callback, because Foo 1.2 is already active
+
+    >>> ws = WorkingSet([])
+    >>> ws.add(foo12)
+    >>> ws.add(foo14)
+    >>> ws.subscribe(added)
+    Added Foo 1.2
+    
+And adding a callback more than once has no effect, either::
+
+    >>> ws.subscribe(added)     # no callbacks
+
+    # and no double-callbacks on subsequent additions, either
+    >>> ws.add(Distribution(project_name="JustATest", version="0.99"))
+    Added JustATest 0.99
+
diff --git a/pkg_resources.py b/pkg_resources.py
index 1acb1a44..92c7b14f 100644
--- a/pkg_resources.py
+++ b/pkg_resources.py
@@ -23,7 +23,8 @@ __all__ = [
     'get_importer', 'find_distributions', 'find_on_path', 'register_finder',
     'split_sections', 'declare_namespace', 'register_namespace_handler',
     'safe_name', 'safe_version', 'run_main', 'BINARY_DIST', 'run_script',
-    'get_default_cache', 'EmptyProvider', 'empty_provider',
+    'get_default_cache', 'EmptyProvider', 'empty_provider', 'normalize_path',
+    'WorkingSet', 'working_set', 'add_activation_listener', 'CHECKOUT_DIST',
 ]
 
 import sys, os, zipimport, time, re, imp
@@ -38,7 +39,6 @@ from sets import ImmutableSet
 
 
 
-
 class ResolutionError(Exception):
     """Abstract base for dependency resolution errors"""
 
@@ -57,6 +57,7 @@ PY_MAJOR = sys.version[:3]
 EGG_DIST    = 3
 BINARY_DIST = 2
 SOURCE_DIST = 1
+CHECKOUT_DIST = 0
 
 def register_loader_type(loader_type, provider_factory):
     """Register `provider_factory` to make providers for `loader_type`
@@ -79,7 +80,6 @@ def get_provider(moduleName):
 
 
 
-
 def get_platform():
     """Return this platform's string for platform-specific distributions
 
@@ -203,6 +203,211 @@ class IResourceProvider(IMetadataProvider):
 
 
 
+class WorkingSet(object):
+    """A collection of active distributions on sys.path (or a similar list)"""
+
+    def __init__(self, entries=None):
+        """Create working set from list of path entries (default=sys.path)"""
+        self.entries = []
+        self.entry_keys = {}
+        self.by_key = {}
+        self.callbacks = []
+
+        if entries is None:
+            entries = sys.path
+
+        for entry in entries:
+            self.add_entry(entry)
+
+
+    def add_entry(self, entry):
+        """Add a path item to ``.entries``, finding any distributions on it
+
+        ``find_distributions(entry,False)`` is used to find distributions
+        corresponding to the path entry, and they are added.  `entry` is
+        always appended to ``.entries``, even if it is already present.
+        (This is because ``sys.path`` can contain the same value more than
+        once, and the ``.entries`` of the ``sys.path`` WorkingSet should always
+        equal ``sys.path``.)
+        """
+        self.entry_keys.setdefault(entry, [])
+        self.entries.append(entry)
+        for dist in find_distributions(entry, False):
+            self.add(dist, entry)
+
+
+    def __contains__(self,dist):
+        """True if `dist` is the active distribution for its project"""
+        return self.by_key.get(dist.key) == dist
+
+
+
+
+
+    def __iter__(self):
+        """Yield distributions for non-duplicate projects in the working set
+
+        The yield order is the order in which the items' path entries were
+        added to the working set.
+        """
+        for item in self.entries:
+            for key in self.entry_keys[item]:
+                yield self.by_key[key]
+
+
+    def find(self, req):
+        """Find a distribution matching requirement `req`
+
+        If there is an active distribution for the requested project, this
+        returns it as long as it meets the version requirement specified by
+        `req`.  But, if there is an active distribution for the project and it
+        does *not* meet the `req` requirement, ``VersionConflict`` is raised.
+        If there is no active distribution for the requested project, ``None``
+        is returned.
+        """
+        dist = self.by_key.get(req.key)
+        if dist is not None and dist not in req:
+            raise VersionConflict(dist,req)     # XXX add more info
+        else:
+            return dist
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+    def add(self, dist, entry=None):
+        """Add `dist` to working set, associated with `entry`
+
+        If `entry` is unspecified, it defaults to the ``.location`` of `dist`.
+        On exit from this routine, `entry` is added to the end of the working
+        set's ``.entries`` (if it wasn't already present).
+
+        `dist` is only added to the working set if it's for a project that
+        doesn't already have a distribution in the set.  If it's added, any
+        callbacks registered with the ``subscribe()`` method will be called.
+        """
+        if entry is None:
+            entry = dist.location
+
+        if entry not in self.entry_keys:
+            self.entries.append(entry)
+            self.entry_keys[entry] = []
+
+        if dist.key in self.by_key:
+            return      # ignore hidden distros
+
+        self.by_key[dist.key] = dist
+        keys = self.entry_keys[entry]
+
+        if dist.key not in keys:
+            keys.append(dist.key)
+
+        self._added_new(dist)
+
+
+
+
+
+
+
+
+
+
+
+
+
+    def resolve(self, requirements, env=None, installer=None):
+        """List all distributions needed to (recursively) meet `requirements`
+
+        `requirements` must be a sequence of ``Requirement`` objects.  `env`,
+        if supplied, should be an ``AvailableDistributions`` instance.  If
+        not supplied, it defaults to all distributions available within any
+        entry or distribution in the working set.  `installer`, if supplied,
+        will be invoked with each requirement that cannot be met by an
+        already-installed distribution; it should return a ``Distribution`` or
+        ``None``.
+        """
+        if env is None:
+            env = AvailableDistributions(self.entries)
+
+        requirements = list(requirements)[::-1]  # set up the stack
+        processed = {}  # set of processed requirements
+        best = {}       # key -> dist
+        to_activate = []
+
+        while requirements:
+            req = requirements.pop()
+            if req in processed:
+                # Ignore cyclic or redundant dependencies
+                continue
+
+            dist = best.get(req.key)
+            if dist is None:
+                # Find the best distribution and add it to the map
+                dist = best[req.key] = env.best_match(req, self, installer)
+                if dist is None:
+                    raise DistributionNotFound(req)  # XXX put more info here
+                to_activate.append(dist)
+            elif dist not in req:
+                # Oops, the "best" so far conflicts with a dependency
+                raise VersionConflict(dist,req) # XXX put more info here
+
+            requirements.extend(dist.depends(req.extras)[::-1])
+            processed[req] = True
+
+        return to_activate    # return list of distros to activate
+
+    def require(self, *requirements):
+        """Ensure that distributions matching `requirements` are activated
+
+        `requirements` must be a string or a (possibly-nested) sequence
+        thereof, specifying the distributions and versions required.  The
+        return value is a sequence of the distributions that needed to be
+        activated to fulfill the requirements; all relevant distributions are
+        included, even if they were already activated in this working set.
+        """
+
+        needed = self.resolve(parse_requirements(requirements))
+
+        for dist in needed:
+            self.add(dist)
+
+        return needed
+
+
+    def subscribe(self, callback):
+        """Invoke `callback` for all distributions (including existing ones)"""
+        if callback in self.callbacks:
+            return
+        self.callbacks.append(callback)
+        for dist in self:
+            callback(dist)
+
+
+    def _added_new(self, dist):
+        for callback in self.callbacks:
+            callback(dist)
+
+
+
+
+
+
+
+
+
+
+
 class AvailableDistributions(object):
     """Searchable snapshot of distributions on a search path"""
 
@@ -296,69 +501,27 @@ class AvailableDistributions(object):
         """Remove `dist` from the distribution map"""
         self._distmap[dist.key].remove(dist)
 
-    def best_match(self, requirement, path=None, installer=None):
-        """Find distribution best matching `requirement` and usable on `path`
+    def best_match(self, req, working_set, installer=None):
+        """Find distribution best matching `req` and usable on `working_set`
 
-        If a distribution that's already installed on `path` is unsuitable,
+        If a distribution that's already active in `working_set` is unsuitable,
         a VersionConflict is raised.  If one or more suitable distributions are
-        already installed, the leftmost distribution (i.e., the one first in
+        already active, the leftmost distribution (i.e., the one first in
         the search path) is returned.  Otherwise, the available distribution
-        with the highest version number is returned, or a deferred distribution
-        object is returned if a suitable ``obtain()`` method exists.  If there
-        is no way to meet the requirement, None is returned.
+        with the highest version number is returned.  If nothing is available,
+        returns ``obtain(req,installer)`` or ``None`` if no distribution can
+        be obtained.
         """
-        if path is None:
-            path = sys.path
-
-        distros = self.get(requirement.key, ())
-        find = dict([(dist.location,dist) for dist in distros]).get
-
-        for item in path:
-            dist = find(item)
-            if dist is not None:
-                if dist in requirement:
-                    return dist
-                else:
-                    raise VersionConflict(dist,requirement) # XXX add more info
-
-        for dist in distros:
-            if dist in requirement:
-                return dist
-        return self.obtain(requirement, installer) # try and download/install
-
-    def resolve(self, requirements, path=None, installer=None):
-        """List all distributions needed to (recursively) meet requirements"""
-
-        if path is None:
-            path = sys.path
+        dist = working_set.find(req)
+        if dist is not None:
+            return dist
 
-        requirements = list(requirements)[::-1]  # set up the stack
-        processed = {}  # set of processed requirements
-        best = {}       # key -> dist
-        to_install = []
-
-        while requirements:
-            req = requirements.pop()
-            if req in processed:
-                # Ignore cyclic or redundant dependencies
-                continue
-
-            dist = best.get(req.key)
-            if dist is None:
-                # Find the best distribution and add it to the map
-                dist = best[req.key] = self.best_match(req, path, installer)
-                if dist is None:
-                    raise DistributionNotFound(req)  # XXX put more info here
-                to_install.append(dist)
-
-            elif dist not in req:
-                # Oops, the "best" so far conflicts with a dependency
-                raise VersionConflict(dist,req) # XXX put more info here
+        for dist in self.get(req.key, ()):
+            if dist in req:
+                return dist
 
-            requirements.extend(dist.depends(req.extras)[::-1])
-            processed[req] = True
+        return self.obtain(req, installer) # try and download/install
 
-        return to_install    # return list of distros to install
 
     def obtain(self, requirement, installer=None):
         """Obtain a distro that matches requirement (e.g. via download)"""
@@ -367,6 +530,7 @@ class AvailableDistributions(object):
 
     def __len__(self): return len(self._distmap)
 
+
 class ResourceManager:
     """Manage resource extraction and packages"""
 
@@ -531,23 +695,6 @@ def get_default_cache():
             "Please set the PYTHON_EGG_CACHE enviroment variable"
         )
 
-def require(*requirements):
-    """Ensure that distributions matching `requirements` are on ``sys.path``
-
-    `requirements` must be a string or a (possibly-nested) sequence
-    thereof, specifying the distributions and versions required.
-
-    XXX This doesn't support arbitrary PEP 302 sys.path items yet, because
-    ``find_distributions()`` is hardcoded at the moment.
-    """
-
-    requirements = parse_requirements(requirements)
-    to_install = AvailableDistributions().resolve(requirements)
-    for dist in to_install:
-        dist.install_on(sys.path)
-    return to_install
-
-
 def safe_name(name):
     """Convert an arbitrary string to a standard distribution name
 
@@ -572,6 +719,23 @@ def safe_version(version):
 
 
 
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
 class NullProvider:
     """Try to implement resources and metadata for arbitrary PEP 302 loaders"""
 
@@ -1035,24 +1199,25 @@ def register_finder(importer_type, distribution_finder):
     _distribution_finders[importer_type] = distribution_finder
 
 
-def find_distributions(path_item):
+def find_distributions(path_item, only=False):
     """Yield distributions accessible via `path_item`"""
     importer = get_importer(path_item)
     finder = _find_adapter(_distribution_finders, importer)
-    return finder(importer,path_item)
+    return finder(importer, path_item, only)
 
-def find_in_zip(importer,path_item):
+def find_in_zip(importer, path_item, only=False):
     metadata = EggMetadata(importer)
     if metadata.has_metadata('PKG-INFO'):
         yield Distribution.from_filename(path_item, metadata=metadata)
+    if only:
+        return  # don't yield nested distros
     for subitem in metadata.resource_listdir('/'):
         if subitem.endswith('.egg'):
             subpath = os.path.join(path_item, subitem)
             for dist in find_in_zip(zipimport.zipimporter(subpath), subpath):
                 yield dist
 
-register_finder(zipimport.zipimporter,find_in_zip)
-
+register_finder(zipimport.zipimporter, find_in_zip)
 
 def StringIO(*args, **kw):
     """Thunk to load the real StringIO on demand"""
@@ -1063,22 +1228,21 @@ def StringIO(*args, **kw):
         from StringIO import StringIO
     return StringIO(*args,**kw)
 
-
-def find_nothing(importer,path_item):
+def find_nothing(importer, path_item, only=False):
     return ()
-
 register_finder(object,find_nothing)
 
-def find_on_path(importer,path_item):
+def find_on_path(importer, path_item, only=False):
     """Yield distributions accessible on a sys.path directory"""
     if not os.path.exists(path_item):
         return
-    elif os.path.isdir(path_item):
+    path_item = normalize_path(path_item)
+    if os.path.isdir(path_item):
         if path_item.lower().endswith('.egg'):
             # unpacked egg
             yield Distribution.from_filename(
                 path_item, metadata=PathMetadata(
-                    path_item,os.path.join(path_item,'EGG-INFO')
+                    path_item, os.path.join(path_item,'EGG-INFO')
                 )
             )
         else:
@@ -1086,16 +1250,18 @@ def find_on_path(importer,path_item):
             for entry in os.listdir(path_item):
                 fullpath = os.path.join(path_item, entry)
                 lower = entry.lower()
-                if lower.endswith('.egg'):
-                    for dist in find_distributions(fullpath):
-                        yield dist
-                elif lower.endswith('.egg-info'):
+                if lower.endswith('.egg-info'):
                     if os.path.isdir(fullpath):
                         # development egg
                         metadata = PathMetadata(path_item, fullpath)
                         dist_name = os.path.splitext(entry)[0]
-                        yield Distribution(path_item,metadata,project_name=dist_name)
-                elif lower.endswith('.egg-link'):
+                        yield Distribution(
+                            path_item, metadata, project_name=dist_name
+                        )
+                elif not only and lower.endswith('.egg'):
+                    for dist in find_distributions(fullpath):
+                        yield dist
+                elif not only and lower.endswith('.egg-link'):
                     for line in file(fullpath):
                         if not line.strip(): continue
                         for item in find_distributions(line.rstrip()):
@@ -1103,8 +1269,6 @@ def find_on_path(importer,path_item):
 
 register_finder(ImpWrapper,find_on_path)
 
-
-
 _namespace_handlers = {}
 _namespace_packages = {}
 
@@ -1209,9 +1373,9 @@ def null_ns_handler(importer, path_item, packageName, module):
 register_namespace_handler(object,null_ns_handler)
 
 
-
-
-
+def normalize_path(filename):
+    """Normalize a file/dir name for comparison purposes"""
+    return os.path.normcase(os.path.realpath(filename))
 
 
 
@@ -1312,10 +1476,9 @@ def parse_version(s):
 
 class Distribution(object):
     """Wrap an actual or potential sys.path entry w/metadata"""
-
     def __init__(self,
-        location, metadata=None, project_name=None, version=None,
-        py_version=PY_MAJOR, platform=None, distro_type = EGG_DIST
+        location=None, metadata=None, project_name=None, version=None,
+        py_version=PY_MAJOR, platform=None, precedence = EGG_DIST
     ):
         self.project_name = safe_name(project_name or 'Unknown')
         if version is not None:
@@ -1323,15 +1486,9 @@ class Distribution(object):
         self.py_version = py_version
         self.platform = platform
         self.location = location
-        self.distro_type = distro_type
+        self.precedence = precedence
         self._provider = metadata or empty_provider
 
-    def installed_on(self,path=None):
-        """Is this distro installed on `path`? (defaults to ``sys.path``)"""
-        if path is None:
-            path = sys.path
-        return self.location in path
-
     #@classmethod
     def from_location(cls,location,basename,metadata=None):
         project_name, version, py_version, platform = [None]*4
@@ -1348,7 +1505,14 @@ class Distribution(object):
         )
     from_location = classmethod(from_location)
 
-
+    hashcmp = property(
+        lambda self: (
+            getattr(self,'parsed_version',()), self.precedence, self.key,
+            self.location, self.py_version, self.platform
+        )
+    )
+    def __cmp__(self, other): return cmp(self.hashcmp, other)
+    def __hash__(self): return hash(self.hashcmp)
 
 
     # These properties have to be lazy so that we don't have to load any
@@ -1389,7 +1553,7 @@ class Distribution(object):
                 )
     version = property(version)
 
-        
+
 
 
     #@property
@@ -1424,7 +1588,7 @@ class Distribution(object):
             for line in self.get_metadata_lines(name):
                 yield line
 
-    def install_on(self,path=None):
+    def activate(self,path=None):
         """Ensure distribution is importable on `path` (default=sys.path)"""
         if path is None: path = sys.path
         if self.location not in path:
@@ -1446,7 +1610,10 @@ class Distribution(object):
         return filename
 
     def __repr__(self):
-        return "%s (%s)" % (self,self.location)
+        if self.location:
+            return "%s (%s)" % (self,self.location)
+        else:
+            return str(self)
 
     def __str__(self):
         version = getattr(self,'version',None) or "[unknown version]"
@@ -1460,19 +1627,13 @@ class Distribution(object):
 
     #@classmethod
     def from_filename(cls,filename,metadata=None):
-        return cls.from_location(filename, os.path.basename(filename), metadata)
+        return cls.from_location(
+            normalize_path(filename), os.path.basename(filename), metadata
+        )
     from_filename = classmethod(from_filename)
 
     def as_requirement(self):
-        return Requirement.parse('%s==%s' % (dist.project_name, dist.version))
-
-
-
-
-
-
-
-
+        return Requirement.parse('%s==%s' % (self.project_name, self.version))
 
 
 
@@ -1538,9 +1699,9 @@ def parse_requirements(strs):
 
 
 def _sort_dists(dists):
-    tmp = [(dist.parsed_version,dist.distro_type,dist) for dist in dists]
+    tmp = [(dist.hashcmp,dist) for dist in dists]
     tmp.sort()
-    dists[::-1] = [d for v,t,d in tmp]
+    dists[::-1] = [d for hc,d in tmp]
 
 
 
@@ -1560,7 +1721,6 @@ def _sort_dists(dists):
 
 
 class Requirement:
-
     def __init__(self, project_name, specs=(), extras=()):
         self.project_name = project_name
         self.key = project_name.lower()
@@ -1575,11 +1735,12 @@ class Requirement:
         self.__hash = hash(self.hashCmp)
 
     def __str__(self):
-        return self.project_name + ','.join([''.join(s) for s in self.specs])
+        specs = ','.join([''.join(s) for s in self.specs])
+        extras = ','.join(self.extras)
+        if extras: extras = '[%s]' % extras
+        return '%s%s%s' % (self.project_name, extras, specs)
 
-    def __repr__(self):
-        return "Requirement(%r, %r, %r)" % \
-            (self.project_name,self.specs,self.extras)
+    def __repr__(self): return "Requirement.parse(%r)" % str(self)
 
     def __eq__(self,other):
         return isinstance(other,Requirement) and self.hashCmp==other.hashCmp
@@ -1693,17 +1854,17 @@ def _initialize(g):
 _initialize(globals())
 
 
+# Prepare the master working set and make the ``require()`` API available
 
+working_set = WorkingSet()
+require = working_set.require
+add_activation_listener = working_set.subscribe
 
-
-
-
-
-
-
-
-
-
+# Activate all distributions already on sys.path, and ensure that
+# all distributions added to the working set in the future (e.g. by
+# calling ``require()``) will get activated as well.
+#
+add_activation_listener(lambda dist: dist.activate())
 
 
 
diff --git a/setuptools.txt b/setuptools.txt
index d0c57525..6834615c 100755
--- a/setuptools.txt
+++ b/setuptools.txt
@@ -732,6 +732,58 @@ might consider submitting a patch to the ``setuptools.command.sdist`` module
 so we can include support for it, too.)
 
 
+Making your package available for EasyInstall
+---------------------------------------------
+
+If you use the ``register`` command (``setup.py register``) to register your
+package with PyPI, that's most of the battle right there.  (See the
+`docs for the register command`_ for more details.)
+
+.. _docs for the register command: http://docs.python.org/dist/package-index.html
+
+If you also use the `upload`_ command to upload actual distributions of your
+package, that's even better, because EasyInstall will be able to find and
+download them directly from your project's PyPI page.
+
+However, there may be reasons why you don't want to upload distributions to
+PyPI, and just want your existing distributions (or perhaps a Subversion
+checkout) to be used instead.
+
+So here's what you need to do before running the ``register`` command.  There
+are three ``setup()`` arguments that affect EasyInstall:
+
+``url`` and ``download_url``
+   These become links on your project's PyPI page.  EasyInstall will examine
+   them to see if they link to a package ("primary links"), or whether they are
+   HTML pages.  If they're HTML pages, EasyInstall scans all HREF's on the
+   page for primary links
+
+``long_description``
+   EasyInstall will check any URLs contained in this argument to see if they
+   are primary links.
+
+A URL is considered a "primary link" if it is a link to a .tar.gz, .tgz, .zip,
+.egg, .egg.zip, .tar.bz2, or .exe file, or if it has an ``#egg=project`` or
+``#egg=project-version`` fragment identifier attached to it.  EasyInstall
+attempts to determine a project name and optional version number from the text
+of a primary link *without* downloading it.  When it has found all the primary
+links, EasyInstall will select the best match based on requested version,
+platform compatibility, and other criteria.
+
+So, if your ``url`` or ``download_url`` point either directly to a downloadable
+source distribution, or to HTML page(s) that have direct links to such, then
+EasyInstall will be able to locate downloads automatically.  If you want to
+make Subversion checkouts available, then you should create links with either
+``#egg=project`` or ``#egg=project-version`` added to the URL (replacing
+``project`` and ``version`` with appropriate values).
+
+Note that Subversion checkout links are of lower precedence than other kinds
+of distributions, so EasyInstall will not select a Subversion checkout for
+downloading unless it has a version included in the ``#egg=`` suffix, and
+it's a higher version than EasyInstall has seen in any other links for your
+project.
+
+
 Distributing Extensions compiled with Pyrex
 -------------------------------------------
 
@@ -1342,7 +1394,7 @@ Release Notes/Change History
    
  * Fixed ``pkg_resources.resource_exists()`` not working correctly.
 
- * Many ``pkg_resources`` API changes:
+ * Many ``pkg_resources`` API changes and enhancements:
 
    * ``Distribution`` objects now implement the ``IResourceProvider`` and
      ``IMetadataProvider`` interfaces, so you don't need to reference the (no
@@ -1354,11 +1406,35 @@ Release Notes/Change History
 
    * The ``path`` attribute of ``Distribution`` objects is now ``location``,
      because it isn't necessarily a filesystem path (and hasn't been for some
-     time now).
+     time now).  The ``location`` of ``Distribution`` objects in the filesystem
+     should always be normalized using ``pkg_resources.normalize_path()``; all
+     of the setuptools and EasyInstall code that generates distributions from
+     the filesystem (including ``Distribution.from_filename()``) ensure this
+     invariant, but if you use a more generic API like ``Distribution()`` or
+     ``Distribution.from_location()`` you should take care that you don't
+     create a distribution with an un-normalized filesystem path.
 
    * ``Distribution`` objects now have an ``as_requirement()`` method that
      returns a ``Requirement`` for the distribution's project name and version.
 
+   * Distribution objects no longer have an ``installed_on()`` method, and the
+     ``install_on()`` method is now ``activate()`` (but may go away altogether
+     soon).
+
+   * ``find_distributions()`` now takes an additional argument called ``only``,
+     that tells it to only yield distributions whose location is the passed-in
+     path.  (It defaults to False, so that the default behavior is unchanged.)
+
+   * The ``resolve()`` method of ``AvailableDistributions`` is now a method of
+     ``WorkingSet`` instead, and the ``best_match()`` method now uses a working
+     set instead of a path list as its second argument.
+
+   * There is a new ``pkg_resources.add_activation_listener()`` API that lets
+     you register a callback for notifications about distributions added to
+     ``sys.path`` (including the distributions already on it).  This is
+     basically a hook for extensible applications and frameworks to be able to
+     search for plugin metadata in distributions added at runtime.
+     
 0.5a13
  * Fixed a bug in resource extraction from nested packages in a zipped egg.
 
diff --git a/setuptools/command/develop.py b/setuptools/command/develop.py
index 40d22b28..1eb8bf6b 100755
--- a/setuptools/command/develop.py
+++ b/setuptools/command/develop.py
@@ -1,6 +1,6 @@
 from setuptools.command.easy_install import easy_install
 from distutils.util import convert_path
-from pkg_resources import Distribution, PathMetadata
+from pkg_resources import Distribution, PathMetadata, normalize_path
 from distutils import log
 import sys, os
 
@@ -49,7 +49,7 @@ class develop(easy_install):
 
         # Make a distribution for the package's source
         self.dist = Distribution(
-            self.egg_path,
+            normalize_path(self.egg_path),
             PathMetadata(self.egg_path, os.path.abspath(ei.egg_info)),
             project_name = ei.egg_name
         )
diff --git a/setuptools/command/easy_install.py b/setuptools/command/easy_install.py
index a9b769e3..e3b40a24 100755
--- a/setuptools/command/easy_install.py
+++ b/setuptools/command/easy_install.py
@@ -139,7 +139,7 @@ class easy_install(Command):
         )
         # default --record from the install command
         self.set_undefined_options('install', ('record', 'record'))
-        normpath = map(normalize,sys.path)
+        normpath = map(normalize_path, sys.path)
         self.all_site_dirs = get_site_dirs()
         if self.site_dirs is not None:
             site_dirs = [
@@ -148,15 +148,15 @@ class easy_install(Command):
             for d in site_dirs:
                 if not os.path.isdir(d):
                     log.warn("%s (in --site-dirs) does not exist", d)
-                elif normalize(d) not in normpath:
+                elif normalize_path(d) not in normpath:
                     raise DistutilsOptionError(
                         d+" (in --site-dirs) is not on sys.path"
                     )
                 else:
-                    self.all_site_dirs.append(normalize(d))
+                    self.all_site_dirs.append(normalize_path(d))
 
-        instdir = self.install_dir or self.all_site_dirs[-1]
-        if normalize(instdir) in self.all_site_dirs:
+        instdir = normalize_path(self.install_dir or self.all_site_dirs[-1])
+        if instdir in self.all_site_dirs:
             if self.pth_file is None:
                 self.pth_file = PthDistributions(
                     os.path.join(instdir,'easy-install.pth')
@@ -171,12 +171,12 @@ class easy_install(Command):
                 "Can't do single-version installs outside 'site-package' dirs"
             )
 
-        self.install_dir = normalize(instdir)
+        self.install_dir = instdir
         self.index_url = self.index_url or "http://www.python.org/pypi"
-        self.shadow_path = map(normalize,sys.path)
-        for path_item in self.install_dir, self.script_dir:
-            if normalize(path_item) not in self.shadow_path:
-                self.shadow_path.insert(0, normalize(path_item))
+        self.shadow_path = self.all_site_dirs[:]
+        for path_item in self.install_dir, normalize_path(self.script_dir):
+            if path_item not in self.shadow_path:
+                self.shadow_path.insert(0, path_item)
         if self.package_index is None:
             self.package_index = self.create_index(
                 self.index_url, search_path = self.shadow_path
@@ -354,8 +354,8 @@ class easy_install(Command):
         if dist in requirement:
             log.info("Processing dependencies for %s", requirement)
             try:
-                self.local_index.resolve(
-                    [requirement], self.shadow_path, self.easy_install
+                WorkingSet(self.shadow_path).resolve(
+                    [requirement], self.local_index, self.easy_install
                 )
             except DistributionNotFound, e:
                 raise DistutilsError(
@@ -711,7 +711,7 @@ similar to one of these examples, in order to select the desired version:
     pkg_resources.require("%(name)s==%(version)s")  # this exact version
     pkg_resources.require("%(name)s>=%(version)s")  # this version or higher
 """
-        if self.install_dir not in map(normalize,sys.path):
+        if self.install_dir not in map(normalize_path,sys.path):
             msg += """
 
 Note also that the installation directory must be on sys.path at runtime for
@@ -782,14 +782,14 @@ PYTHONPATH, or by being added to sys.path by your code.)
             return
 
         for d in self.pth_file.get(dist.key,()):    # drop old entries
-            if self.multi_version or normalize(d.location) != normalize(dist.location):
+            if self.multi_version or d.location != dist.location:
                 log.info("Removing %s from easy-install.pth file", d)
                 self.pth_file.remove(d)
-                if normalize(d.location) in self.shadow_path:
+                if d.location in self.shadow_path:
                     self.shadow_path.remove(d.location)
 
         if not self.multi_version:
-            if normalize(dist.location) in map(normalize,self.pth_file.paths):
+            if dist.location in map(normalize_path,self.pth_file.paths):
                 log.info(
                     "%s is already the active version in easy-install.pth",
                     dist
@@ -797,8 +797,8 @@ PYTHONPATH, or by being added to sys.path by your code.)
             else:
                 log.info("Adding %s to easy-install.pth file", dist)
                 self.pth_file.add(dist) # add new entry
-                if normalize(dist.location) not in self.shadow_path:
-                    self.shadow_path.append(normalize(dist.location))
+                if dist.location not in self.shadow_path:
+                    self.shadow_path.append(dist.location)
 
         self.pth_file.save()
 
@@ -894,8 +894,8 @@ def get_site_dirs():
                                          'site-packages'))
 
     sitedirs = filter(os.path.isdir, sitedirs)
-    sitedirs = map(normalize, sitedirs)
-    return sitedirs or [normalize(get_python_lib())]    # ensure at least one
+    sitedirs = map(normalize_path, sitedirs)
+    return sitedirs or [normalize_path(get_python_lib())] # ensure at least one
 
 
 
@@ -906,7 +906,7 @@ def expand_paths(inputs):
     seen = {}
 
     for dirname in inputs:
-        dirname = normalize(dirname)
+        dirname = normalize_path(dirname)
         if dirname in seen:
             continue
 
@@ -933,7 +933,7 @@ def expand_paths(inputs):
             # Yield existing non-dupe, non-import directory lines from it
             for line in lines:
                 if not line.startswith("import"):
-                    line = normalize(line.rstrip())
+                    line = normalize_path(line.rstrip())
                     if line not in seen:
                         seen[line] = 1
                         if not os.path.isdir(line):
@@ -1027,7 +1027,6 @@ class PthDistributions(AvailableDistributions):
     """A .pth file with Distribution paths in it"""
 
     dirty = False
-
     def __init__(self, filename):
         self.filename = filename; self._load()
         AvailableDistributions.__init__(
@@ -1039,14 +1038,17 @@ class PthDistributions(AvailableDistributions):
         if os.path.isfile(self.filename):
             self.paths = [line.rstrip() for line in open(self.filename,'rt')]
             while self.paths and not self.paths[-1].strip(): self.paths.pop()
-
+        # delete non-existent paths, in case somebody deleted a package
+        # manually:
+        for line in list(yield_lines(self.paths)):
+            if not os.path.exists(line):
+                self.paths.remove(line); self.dirty = True
+                
     def save(self):
         """Write changed .pth file back to disk"""
         if self.dirty:
             data = '\n'.join(self.paths+[''])
-            f = open(self.filename,'wt')
-            f.write(data)
-            f.close()
+            f = open(self.filename,'wt'); f.write(data); f.close()
             self.dirty = False
 
     def add(self,dist):
@@ -1062,15 +1064,13 @@ class PthDistributions(AvailableDistributions):
         AvailableDistributions.remove(self,dist)
 
 
-
-
 def main(argv, **kw):
     from setuptools import setup
     setup(script_args = ['-q','easy_install', '-v']+argv, **kw)
 
 
-def normalize(path):
-    return os.path.normcase(os.path.realpath(path))
+
+
 
 
 
diff --git a/setuptools/package_index.py b/setuptools/package_index.py
index 4c2d7616..7c4fa49d 100755
--- a/setuptools/package_index.py
+++ b/setuptools/package_index.py
@@ -5,7 +5,8 @@ from pkg_resources import *
 from distutils import log
 from distutils.errors import DistutilsError
 
-HREF = re.compile(r"""href\s*=\s*['"]?([^'"> ]+)""", re.I)
+EGG_FRAGMENT = re.compile('^egg=(\\w+(-\\w+)?)$')
+HREF = re.compile("""href\\s*=\\s*['"]?([^'"> ]+)""", re.I)
 URL_SCHEME = re.compile('([-+.a-z0-9]{2,}):',re.I).match
 EXTENSIONS = ".tar.gz .tar.bz2 .tar .zip .tgz".split()
 
@@ -38,28 +39,34 @@ def parse_bdist_wininst(name):
 
 
 
-
 def distros_for_url(url, metadata=None):
     """Yield egg or source distribution objects that might be found at a URL"""
 
-    path = urlparse.urlparse(url)[2]
+    scheme, server, path, parameters, query, fragment = urlparse.urlparse(url)
     base = urllib2.unquote(path.split('/')[-1])
-    return distros_for_filename(url, base, metadata)
+    dists = distros_for_location(url, base, metadata) 
+    if fragment and not dists:
+        match = EGG_FRAGMENT.match(fragment)
+        if match:
+            return interpret_distro_name(
+                url, match.group(1), metadata, precedence = CHECKOUT_DIST
+            )
+    return dists
 
 
-def distros_for_filename(url_or_path, basename, metadata=None):
+def distros_for_location(location, basename, metadata=None):
     """Yield egg or source distribution objects based on basename"""
     if basename.endswith('.egg.zip'):
         basename = basename[:-4]    # strip the .zip
 
     if basename.endswith('.egg'):   # only one, unambiguous interpretation       
-        return [Distribution.from_location(url_or_path, basename, metadata)]
+        return [Distribution.from_location(location, basename, metadata)]
 
     if basename.endswith('.exe'):
         win_base, py_ver = parse_bdist_wininst(basename)
         if win_base is not None:
             return interpret_distro_name(
-                url_or_path, win_base, metadata, py_ver, BINARY_DIST, "win32"
+                location, win_base, metadata, py_ver, BINARY_DIST, "win32"
             )
 
     # Try source distro extensions (.zip, .tgz, etc.)
@@ -67,22 +74,28 @@ def distros_for_filename(url_or_path, basename, metadata=None):
     for ext in EXTENSIONS:
         if basename.endswith(ext):
             basename = basename[:-len(ext)]
-            return interpret_distro_name(url_or_path, basename, metadata)
+            return interpret_distro_name(location, basename, metadata)
 
     return []  # no extension matched
 
 
 
+def distros_for_filename(filename, metadata=None):
+    """Yield possible egg or source distribution objects based on a filename"""
+    return distros_for_location(
+        normalize_path(filename), os.path.basename(filename), metadata
+    )
 
 
-
-
-
-
-
-def interpret_distro_name(url_or_path, basename, metadata,
-    py_version=None, distro_type=SOURCE_DIST, platform=None
+def interpret_distro_name(location, basename, metadata,
+    py_version=None, precedence=SOURCE_DIST, platform=None
 ):
+    """Generate alternative interpretations of a source distro name
+
+    Note: if `location` is a filesystem filename, you should call
+    ``pkg_resources.normalize_path()`` on it before passing it to this
+    routine!
+    """
 
     # Generate alternative interpretations of a source distro name
     # Because some packages are ambiguous as to name/versions split
@@ -99,8 +112,8 @@ def interpret_distro_name(url_or_path, basename, metadata,
     parts = basename.split('-')
     for p in range(1,len(parts)+1):
         yield Distribution(
-            url_or_path, metadata, '-'.join(parts[:p]), '-'.join(parts[p:]),
-            py_version=py_version, distro_type = distro_type,
+            location, metadata, '-'.join(parts[:p]), '-'.join(parts[p:]),
+            py_version=py_version, precedence = precedence,
             platform = platform
         )
 
@@ -108,19 +121,6 @@ def interpret_distro_name(url_or_path, basename, metadata,
 
 
 
-
-
-
-
-
-
-
-
-
-
-
-
-
 class PackageIndex(AvailableDistributions):
     """A distribution index that scans web pages for download URLs"""
 
@@ -142,11 +142,7 @@ class PackageIndex(AvailableDistributions):
         if not URL_SCHEME(url):
             # process filenames or directories
             if os.path.isfile(url):
-                dists = list(
-                    distros_for_filename(
-                        os.path.realpath(url), os.path.basename(url)
-                    )
-                )
+                dists = list(distros_for_filename(url))
             elif os.path.isdir(url):
                 url = os.path.realpath(url)
                 for item in os.listdir(url):
@@ -160,8 +156,6 @@ class PackageIndex(AvailableDistributions):
 
         if dists:
             self.debug("Found link: %s", url)
-
-
         if dists or not retrieve or url in self.fetched_urls:
             for dist in dists:
                 self.add(dist)
@@ -203,6 +197,12 @@ class PackageIndex(AvailableDistributions):
 
 
 
+
+
+
+
+
+
     def process_index(self,url,page):
         """Process the contents of a PyPI page"""
 
@@ -344,11 +344,11 @@ class PackageIndex(AvailableDistributions):
         if force_scan:
             self.find_packages(requirement)
 
-        dist = self.best_match(requirement, [])     # XXX
+        dist = self.best_match(requirement, WorkingSet([]))     # XXX
 
         if dist is not None:
             self.info("Best match: %s", dist)
-            return self.download(dist.path, tmpdir)
+            return self.download(dist.location, tmpdir)
 
         self.warn(
             "No local packages or download links found for %s", requirement
@@ -475,8 +475,8 @@ class PackageIndex(AvailableDistributions):
         file.close()
         raise DistutilsError("Unexpected HTML page found at "+url)
 
-
     def _download_svn(self, url, filename):
+        url = url.split('#',1)[0]   # remove any fragment for svn's sake
         self.info("Doing subversion checkout from %s to %s", url, filename)
         os.system("svn checkout -q %s %s" % (url, filename))
         return filename
diff --git a/setuptools/tests/__init__.py b/setuptools/tests/__init__.py
index 06ca095d..97be5626 100644
--- a/setuptools/tests/__init__.py
+++ b/setuptools/tests/__init__.py
@@ -408,8 +408,15 @@ class TestCommandTests(TestCase):
 
 
 
+def test_api():
+    import doctest
+    return doctest.DocFileSuite(
+        'api_tests.txt', optionflags=doctest.ELLIPSIS, package='pkg_resources',
+    )
+
+
 testClasses = (DependsTests, DistroTests, FeatureTests, TestCommandTests)
-testNames = ["setuptools.tests.test_resources"]
+testNames = ["setuptools.tests.test_resources", "setuptools.tests.test_api"]
 
 def test_suite():
     return TestSuite(
@@ -442,10 +449,3 @@ def test_suite():
 
 
 
-
-
-
-
-
-
-
diff --git a/setuptools/tests/doctest.py b/setuptools/tests/doctest.py
new file mode 100644
index 00000000..fcaf3a4d
--- /dev/null
+++ b/setuptools/tests/doctest.py
@@ -0,0 +1,2677 @@
+# Module doctest.
+# Released to the public domain 16-Jan-2001, by Tim Peters (tim@python.org).
+# Major enhancements and refactoring by:
+#     Jim Fulton
+#     Edward Loper
+
+# Provided as-is; use at your own risk; no warranty; no promises; enjoy!
+
+try:
+    basestring
+except NameError:
+    basestring = str,unicode
+
+try:
+    enumerate
+except NameError:
+    def enumerate(seq):
+        return zip(range(len(seq)),seq)
+
+r"""Module doctest -- a framework for running examples in docstrings.
+
+In simplest use, end each module M to be tested with:
+
+def _test():
+    import doctest
+    doctest.testmod()
+
+if __name__ == "__main__":
+    _test()
+
+Then running the module as a script will cause the examples in the
+docstrings to get executed and verified:
+
+python M.py
+
+This won't display anything unless an example fails, in which case the
+failing example(s) and the cause(s) of the failure(s) are printed to stdout
+(why not stderr? because stderr is a lame hack <0.2 wink>), and the final
+line of output is "Test failed.".
+
+Run it with the -v switch instead:
+
+python M.py -v
+
+and a detailed report of all examples tried is printed to stdout, along
+with assorted summaries at the end.
+
+You can force verbose mode by passing "verbose=True" to testmod, or prohibit
+it by passing "verbose=False".  In either of those cases, sys.argv is not
+examined by testmod.
+
+There are a variety of other ways to run doctests, including integration
+with the unittest framework, and support for running non-Python text
+files containing doctests.  There are also many ways to override parts
+of doctest's default behaviors.  See the Library Reference Manual for
+details.
+"""
+
+__docformat__ = 'reStructuredText en'
+
+__all__ = [
+    # 0, Option Flags
+    'register_optionflag',
+    'DONT_ACCEPT_TRUE_FOR_1',
+    'DONT_ACCEPT_BLANKLINE',
+    'NORMALIZE_WHITESPACE',
+    'ELLIPSIS',
+    'IGNORE_EXCEPTION_DETAIL',
+    'COMPARISON_FLAGS',
+    'REPORT_UDIFF',
+    'REPORT_CDIFF',
+    'REPORT_NDIFF',
+    'REPORT_ONLY_FIRST_FAILURE',
+    'REPORTING_FLAGS',
+    # 1. Utility Functions
+    'is_private',
+    # 2. Example & DocTest
+    'Example',
+    'DocTest',
+    # 3. Doctest Parser
+    'DocTestParser',
+    # 4. Doctest Finder
+    'DocTestFinder',
+    # 5. Doctest Runner
+    'DocTestRunner',
+    'OutputChecker',
+    'DocTestFailure',
+    'UnexpectedException',
+    'DebugRunner',
+    # 6. Test Functions
+    'testmod',
+    'testfile',
+    'run_docstring_examples',
+    # 7. Tester
+    'Tester',
+    # 8. Unittest Support
+    'DocTestSuite',
+    'DocFileSuite',
+    'set_unittest_reportflags',
+    # 9. Debugging Support
+    'script_from_examples',
+    'testsource',
+    'debug_src',
+    'debug',
+]
+
+import __future__
+
+import sys, traceback, inspect, linecache, os, re, types
+import unittest, difflib, pdb, tempfile
+import warnings
+from StringIO import StringIO
+
+# Don't whine about the deprecated is_private function in this
+# module's tests.
+warnings.filterwarnings("ignore", "is_private", DeprecationWarning,
+                        __name__, 0)
+
+# There are 4 basic classes:
+#  - Example: a <source, want> pair, plus an intra-docstring line number.
+#  - DocTest: a collection of examples, parsed from a docstring, plus
+#    info about where the docstring came from (name, filename, lineno).
+#  - DocTestFinder: extracts DocTests from a given object's docstring and
+#    its contained objects' docstrings.
+#  - DocTestRunner: runs DocTest cases, and accumulates statistics.
+#
+# So the basic picture is:
+#
+#                             list of:
+# +------+                   +---------+                   +-------+
+# |object| --DocTestFinder-> | DocTest | --DocTestRunner-> |results|
+# +------+                   +---------+                   +-------+
+#                            | Example |
+#                            |   ...   |
+#                            | Example |
+#                            +---------+
+
+# Option constants.
+
+OPTIONFLAGS_BY_NAME = {}
+def register_optionflag(name):
+    flag = 1 << len(OPTIONFLAGS_BY_NAME)
+    OPTIONFLAGS_BY_NAME[name] = flag
+    return flag
+
+DONT_ACCEPT_TRUE_FOR_1 = register_optionflag('DONT_ACCEPT_TRUE_FOR_1')
+DONT_ACCEPT_BLANKLINE = register_optionflag('DONT_ACCEPT_BLANKLINE')
+NORMALIZE_WHITESPACE = register_optionflag('NORMALIZE_WHITESPACE')
+ELLIPSIS = register_optionflag('ELLIPSIS')
+IGNORE_EXCEPTION_DETAIL = register_optionflag('IGNORE_EXCEPTION_DETAIL')
+
+COMPARISON_FLAGS = (DONT_ACCEPT_TRUE_FOR_1 |
+                    DONT_ACCEPT_BLANKLINE |
+                    NORMALIZE_WHITESPACE |
+                    ELLIPSIS |
+                    IGNORE_EXCEPTION_DETAIL)
+
+REPORT_UDIFF = register_optionflag('REPORT_UDIFF')
+REPORT_CDIFF = register_optionflag('REPORT_CDIFF')
+REPORT_NDIFF = register_optionflag('REPORT_NDIFF')
+REPORT_ONLY_FIRST_FAILURE = register_optionflag('REPORT_ONLY_FIRST_FAILURE')
+
+REPORTING_FLAGS = (REPORT_UDIFF |
+                   REPORT_CDIFF |
+                   REPORT_NDIFF |
+                   REPORT_ONLY_FIRST_FAILURE)
+
+# Special string markers for use in `want` strings:
+BLANKLINE_MARKER = '<BLANKLINE>'
+ELLIPSIS_MARKER = '...'
+
+######################################################################
+## Table of Contents
+######################################################################
+#  1. Utility Functions
+#  2. Example & DocTest -- store test cases
+#  3. DocTest Parser -- extracts examples from strings
+#  4. DocTest Finder -- extracts test cases from objects
+#  5. DocTest Runner -- runs test cases
+#  6. Test Functions -- convenient wrappers for testing
+#  7. Tester Class -- for backwards compatibility
+#  8. Unittest Support
+#  9. Debugging Support
+# 10. Example Usage
+
+######################################################################
+## 1. Utility Functions
+######################################################################
+
+def is_private(prefix, base):
+    """prefix, base -> true iff name prefix + "." + base is "private".
+
+    Prefix may be an empty string, and base does not contain a period.
+    Prefix is ignored (although functions you write conforming to this
+    protocol may make use of it).
+    Return true iff base begins with an (at least one) underscore, but
+    does not both begin and end with (at least) two underscores.
+
+    >>> is_private("a.b", "my_func")
+    False
+    >>> is_private("____", "_my_func")
+    True
+    >>> is_private("someclass", "__init__")
+    False
+    >>> is_private("sometypo", "__init_")
+    True
+    >>> is_private("x.y.z", "_")
+    True
+    >>> is_private("_x.y.z", "__")
+    False
+    >>> is_private("", "")  # senseless but consistent
+    False
+    """
+    warnings.warn("is_private is deprecated; it wasn't useful; "
+                  "examine DocTestFinder.find() lists instead",
+                  DeprecationWarning, stacklevel=2)
+    return base[:1] == "_" and not base[:2] == "__" == base[-2:]
+
+def _extract_future_flags(globs):
+    """
+    Return the compiler-flags associated with the future features that
+    have been imported into the given namespace (globs).
+    """
+    flags = 0
+    for fname in __future__.all_feature_names:
+        feature = globs.get(fname, None)
+        if feature is getattr(__future__, fname):
+            flags |= feature.compiler_flag
+    return flags
+
+def _normalize_module(module, depth=2):
+    """
+    Return the module specified by `module`.  In particular:
+      - If `module` is a module, then return module.
+      - If `module` is a string, then import and return the
+        module with that name.
+      - If `module` is None, then return the calling module.
+        The calling module is assumed to be the module of
+        the stack frame at the given depth in the call stack.
+    """
+    if inspect.ismodule(module):
+        return module
+    elif isinstance(module, (str, unicode)):
+        return __import__(module, globals(), locals(), ["*"])
+    elif module is None:
+        return sys.modules[sys._getframe(depth).f_globals['__name__']]
+    else:
+        raise TypeError("Expected a module, string, or None")
+
+def _indent(s, indent=4):
+    """
+    Add the given number of space characters to the beginning every
+    non-blank line in `s`, and return the result.
+    """
+    # This regexp matches the start of non-blank lines:
+    return re.sub('(?m)^(?!$)', indent*' ', s)
+
+def _exception_traceback(exc_info):
+    """
+    Return a string containing a traceback message for the given
+    exc_info tuple (as returned by sys.exc_info()).
+    """
+    # Get a traceback message.
+    excout = StringIO()
+    exc_type, exc_val, exc_tb = exc_info
+    traceback.print_exception(exc_type, exc_val, exc_tb, file=excout)
+    return excout.getvalue()
+
+# Override some StringIO methods.
+class _SpoofOut(StringIO):
+    def getvalue(self):
+        result = StringIO.getvalue(self)
+        # If anything at all was written, make sure there's a trailing
+        # newline.  There's no way for the expected output to indicate
+        # that a trailing newline is missing.
+        if result and not result.endswith("\n"):
+            result += "\n"
+        # Prevent softspace from screwing up the next test case, in
+        # case they used print with a trailing comma in an example.
+        if hasattr(self, "softspace"):
+            del self.softspace
+        return result
+
+    def truncate(self,   size=None):
+        StringIO.truncate(self, size)
+        if hasattr(self, "softspace"):
+            del self.softspace
+
+# Worst-case linear-time ellipsis matching.
+def _ellipsis_match(want, got):
+    """
+    Essentially the only subtle case:
+    >>> _ellipsis_match('aa...aa', 'aaa')
+    False
+    """
+    if want.find(ELLIPSIS_MARKER)==-1:
+        return want == got
+
+    # Find "the real" strings.
+    ws = want.split(ELLIPSIS_MARKER)
+    assert len(ws) >= 2
+
+    # Deal with exact matches possibly needed at one or both ends.
+    startpos, endpos = 0, len(got)
+    w = ws[0]
+    if w:   # starts with exact match
+        if got.startswith(w):
+            startpos = len(w)
+            del ws[0]
+        else:
+            return False
+    w = ws[-1]
+    if w:   # ends with exact match
+        if got.endswith(w):
+            endpos -= len(w)
+            del ws[-1]
+        else:
+            return False
+
+    if startpos > endpos:
+        # Exact end matches required more characters than we have, as in
+        # _ellipsis_match('aa...aa', 'aaa')
+        return False
+
+    # For the rest, we only need to find the leftmost non-overlapping
+    # match for each piece.  If there's no overall match that way alone,
+    # there's no overall match period.
+    for w in ws:
+        # w may be '' at times, if there are consecutive ellipses, or
+        # due to an ellipsis at the start or end of `want`.  That's OK.
+        # Search for an empty string succeeds, and doesn't change startpos.
+        startpos = got.find(w, startpos, endpos)
+        if startpos < 0:
+            return False
+        startpos += len(w)
+
+    return True
+
+def _comment_line(line):
+    "Return a commented form of the given line"
+    line = line.rstrip()
+    if line:
+        return '# '+line
+    else:
+        return '#'
+
+class _OutputRedirectingPdb(pdb.Pdb):
+    """
+    A specialized version of the python debugger that redirects stdout
+    to a given stream when interacting with the user.  Stdout is *not*
+    redirected when traced code is executed.
+    """
+    def __init__(self, out):
+        self.__out = out
+        pdb.Pdb.__init__(self)
+
+    def trace_dispatch(self, *args):
+        # Redirect stdout to the given stream.
+        save_stdout = sys.stdout
+        sys.stdout = self.__out
+        # Call Pdb's trace dispatch method.
+        try:
+            return pdb.Pdb.trace_dispatch(self, *args)
+        finally:
+            sys.stdout = save_stdout
+
+# [XX] Normalize with respect to os.path.pardir?
+def _module_relative_path(module, path):
+    if not inspect.ismodule(module):
+        raise TypeError, 'Expected a module: %r' % module
+    if path.startswith('/'):
+        raise ValueError, 'Module-relative files may not have absolute paths'
+
+    # Find the base directory for the path.
+    if hasattr(module, '__file__'):
+        # A normal module/package
+        basedir = os.path.split(module.__file__)[0]
+    elif module.__name__ == '__main__':
+        # An interactive session.
+        if len(sys.argv)>0 and sys.argv[0] != '':
+            basedir = os.path.split(sys.argv[0])[0]
+        else:
+            basedir = os.curdir
+    else:
+        # A module w/o __file__ (this includes builtins)
+        raise ValueError("Can't resolve paths relative to the module " +
+                         module + " (it has no __file__)")
+
+    # Combine the base directory and the path.
+    return os.path.join(basedir, *(path.split('/')))
+
+######################################################################
+## 2. Example & DocTest
+######################################################################
+## - An "example" is a <source, want> pair, where "source" is a
+##   fragment of source code, and "want" is the expected output for
+##   "source."  The Example class also includes information about
+##   where the example was extracted from.
+##
+## - A "doctest" is a collection of examples, typically extracted from
+##   a string (such as an object's docstring).  The DocTest class also
+##   includes information about where the string was extracted from.
+
+class Example:
+    """
+    A single doctest example, consisting of source code and expected
+    output.  `Example` defines the following attributes:
+
+      - source: A single Python statement, always ending with a newline.
+        The constructor adds a newline if needed.
+
+      - want: The expected output from running the source code (either
+        from stdout, or a traceback in case of exception).  `want` ends
+        with a newline unless it's empty, in which case it's an empty
+        string.  The constructor adds a newline if needed.
+
+      - exc_msg: The exception message generated by the example, if
+        the example is expected to generate an exception; or `None` if
+        it is not expected to generate an exception.  This exception
+        message is compared against the return value of
+        `traceback.format_exception_only()`.  `exc_msg` ends with a
+        newline unless it's `None`.  The constructor adds a newline
+        if needed.
+
+      - lineno: The line number within the DocTest string containing
+        this Example where the Example begins.  This line number is
+        zero-based, with respect to the beginning of the DocTest.
+
+      - indent: The example's indentation in the DocTest string.
+        I.e., the number of space characters that preceed the
+        example's first prompt.
+
+      - options: A dictionary mapping from option flags to True or
+        False, which is used to override default options for this
+        example.  Any option flags not contained in this dictionary
+        are left at their default value (as specified by the
+        DocTestRunner's optionflags).  By default, no options are set.
+    """
+    def __init__(self, source, want, exc_msg=None, lineno=0, indent=0,
+                 options=None):
+        # Normalize inputs.
+        if not source.endswith('\n'):
+            source += '\n'
+        if want and not want.endswith('\n'):
+            want += '\n'
+        if exc_msg is not None and not exc_msg.endswith('\n'):
+            exc_msg += '\n'
+        # Store properties.
+        self.source = source
+        self.want = want
+        self.lineno = lineno
+        self.indent = indent
+        if options is None: options = {}
+        self.options = options
+        self.exc_msg = exc_msg
+
+class DocTest:
+    """
+    A collection of doctest examples that should be run in a single
+    namespace.  Each `DocTest` defines the following attributes:
+
+      - examples: the list of examples.
+
+      - globs: The namespace (aka globals) that the examples should
+        be run in.
+
+      - name: A name identifying the DocTest (typically, the name of
+        the object whose docstring this DocTest was extracted from).
+
+      - filename: The name of the file that this DocTest was extracted
+        from, or `None` if the filename is unknown.
+
+      - lineno: The line number within filename where this DocTest
+        begins, or `None` if the line number is unavailable.  This
+        line number is zero-based, with respect to the beginning of
+        the file.
+
+      - docstring: The string that the examples were extracted from,
+        or `None` if the string is unavailable.
+    """
+    def __init__(self, examples, globs, name, filename, lineno, docstring):
+        """
+        Create a new DocTest containing the given examples.  The
+        DocTest's globals are initialized with a copy of `globs`.
+        """
+        assert not isinstance(examples, basestring), \
+               "DocTest no longer accepts str; use DocTestParser instead"
+        self.examples = examples
+        self.docstring = docstring
+        self.globs = globs.copy()
+        self.name = name
+        self.filename = filename
+        self.lineno = lineno
+
+    def __repr__(self):
+        if len(self.examples) == 0:
+            examples = 'no examples'
+        elif len(self.examples) == 1:
+            examples = '1 example'
+        else:
+            examples = '%d examples' % len(self.examples)
+        return ('<DocTest %s from %s:%s (%s)>' %
+                (self.name, self.filename, self.lineno, examples))
+
+
+    # This lets us sort tests by name:
+    def __cmp__(self, other):
+        if not isinstance(other, DocTest):
+            return -1
+        return cmp((self.name, self.filename, self.lineno, id(self)),
+                   (other.name, other.filename, other.lineno, id(other)))
+
+######################################################################
+## 3. DocTestParser
+######################################################################
+
+class DocTestParser:
+    """
+    A class used to parse strings containing doctest examples.
+    """
+    # This regular expression is used to find doctest examples in a
+    # string.  It defines three groups: `source` is the source code
+    # (including leading indentation and prompts); `indent` is the
+    # indentation of the first (PS1) line of the source code; and
+    # `want` is the expected output (including leading indentation).
+    _EXAMPLE_RE = re.compile(r'''
+        # Source consists of a PS1 line followed by zero or more PS2 lines.
+        (?P<source>
+            (?:^(?P<indent> [ ]*) >>>    .*)    # PS1 line
+            (?:\n           [ ]*  \.\.\. .*)*)  # PS2 lines
+        \n?
+        # Want consists of any non-blank lines that do not start with PS1.
+        (?P<want> (?:(?![ ]*$)    # Not a blank line
+                     (?![ ]*>>>)  # Not a line starting with PS1
+                     .*$\n?       # But any other line
+                  )*)
+        ''', re.MULTILINE | re.VERBOSE)
+
+    # A regular expression for handling `want` strings that contain
+    # expected exceptions.  It divides `want` into three pieces:
+    #    - the traceback header line (`hdr`)
+    #    - the traceback stack (`stack`)
+    #    - the exception message (`msg`), as generated by
+    #      traceback.format_exception_only()
+    # `msg` may have multiple lines.  We assume/require that the
+    # exception message is the first non-indented line starting with a word
+    # character following the traceback header line.
+    _EXCEPTION_RE = re.compile(r"""
+        # Grab the traceback header.  Different versions of Python have
+        # said different things on the first traceback line.
+        ^(?P<hdr> Traceback\ \(
+            (?: most\ recent\ call\ last
+            |   innermost\ last
+            ) \) :
+        )
+        \s* $                # toss trailing whitespace on the header.
+        (?P<stack> .*?)      # don't blink: absorb stuff until...
+        ^ (?P<msg> \w+ .*)   #     a line *starts* with alphanum.
+        """, re.VERBOSE | re.MULTILINE | re.DOTALL)
+
+    # A callable returning a true value iff its argument is a blank line
+    # or contains a single comment.
+    _IS_BLANK_OR_COMMENT = re.compile(r'^[ ]*(#.*)?$').match
+
+    def parse(self, string, name='<string>'):
+        """
+        Divide the given string into examples and intervening text,
+        and return them as a list of alternating Examples and strings.
+        Line numbers for the Examples are 0-based.  The optional
+        argument `name` is a name identifying this string, and is only
+        used for error messages.
+        """
+        string = string.expandtabs()
+        # If all lines begin with the same indentation, then strip it.
+        min_indent = self._min_indent(string)
+        if min_indent > 0:
+            string = '\n'.join([l[min_indent:] for l in string.split('\n')])
+
+        output = []
+        charno, lineno = 0, 0
+        # Find all doctest examples in the string:
+        for m in self._EXAMPLE_RE.finditer(string):
+            # Add the pre-example text to `output`.
+            output.append(string[charno:m.start()])
+            # Update lineno (lines before this example)
+            lineno += string.count('\n', charno, m.start())
+            # Extract info from the regexp match.
+            (source, options, want, exc_msg) = \
+                     self._parse_example(m, name, lineno)
+            # Create an Example, and add it to the list.
+            if not self._IS_BLANK_OR_COMMENT(source):
+                output.append( Example(source, want, exc_msg,
+                                    lineno=lineno,
+                                    indent=min_indent+len(m.group('indent')),
+                                    options=options) )
+            # Update lineno (lines inside this example)
+            lineno += string.count('\n', m.start(), m.end())
+            # Update charno.
+            charno = m.end()
+        # Add any remaining post-example text to `output`.
+        output.append(string[charno:])
+        return output
+
+    def get_doctest(self, string, globs, name, filename, lineno):
+        """
+        Extract all doctest examples from the given string, and
+        collect them into a `DocTest` object.
+
+        `globs`, `name`, `filename`, and `lineno` are attributes for
+        the new `DocTest` object.  See the documentation for `DocTest`
+        for more information.
+        """
+        return DocTest(self.get_examples(string, name), globs,
+                       name, filename, lineno, string)
+
+    def get_examples(self, string, name='<string>'):
+        """
+        Extract all doctest examples from the given string, and return
+        them as a list of `Example` objects.  Line numbers are
+        0-based, because it's most common in doctests that nothing
+        interesting appears on the same line as opening triple-quote,
+        and so the first interesting line is called \"line 1\" then.
+
+        The optional argument `name` is a name identifying this
+        string, and is only used for error messages.
+        """
+        return [x for x in self.parse(string, name)
+                if isinstance(x, Example)]
+
+    def _parse_example(self, m, name, lineno):
+        """
+        Given a regular expression match from `_EXAMPLE_RE` (`m`),
+        return a pair `(source, want)`, where `source` is the matched
+        example's source code (with prompts and indentation stripped);
+        and `want` is the example's expected output (with indentation
+        stripped).
+
+        `name` is the string's name, and `lineno` is the line number
+        where the example starts; both are used for error messages.
+        """
+        # Get the example's indentation level.
+        indent = len(m.group('indent'))
+
+        # Divide source into lines; check that they're properly
+        # indented; and then strip their indentation & prompts.
+        source_lines = m.group('source').split('\n')
+        self._check_prompt_blank(source_lines, indent, name, lineno)
+        self._check_prefix(source_lines[1:], ' '*indent + '.', name, lineno)
+        source = '\n'.join([sl[indent+4:] for sl in source_lines])
+
+        # Divide want into lines; check that it's properly indented; and
+        # then strip the indentation.  Spaces before the last newline should
+        # be preserved, so plain rstrip() isn't good enough.
+        want = m.group('want')
+        want_lines = want.split('\n')
+        if len(want_lines) > 1 and re.match(r' *$', want_lines[-1]):
+            del want_lines[-1]  # forget final newline & spaces after it
+        self._check_prefix(want_lines, ' '*indent, name,
+                           lineno + len(source_lines))
+        want = '\n'.join([wl[indent:] for wl in want_lines])
+
+        # If `want` contains a traceback message, then extract it.
+        m = self._EXCEPTION_RE.match(want)
+        if m:
+            exc_msg = m.group('msg')
+        else:
+            exc_msg = None
+
+        # Extract options from the source.
+        options = self._find_options(source, name, lineno)
+
+        return source, options, want, exc_msg
+
+    # This regular expression looks for option directives in the
+    # source code of an example.  Option directives are comments
+    # starting with "doctest:".  Warning: this may give false
+    # positives for string-literals that contain the string
+    # "#doctest:".  Eliminating these false positives would require
+    # actually parsing the string; but we limit them by ignoring any
+    # line containing "#doctest:" that is *followed* by a quote mark.
+    _OPTION_DIRECTIVE_RE = re.compile(r'#\s*doctest:\s*([^\n\'"]*)$',
+                                      re.MULTILINE)
+
+    def _find_options(self, source, name, lineno):
+        """
+        Return a dictionary containing option overrides extracted from
+        option directives in the given source string.
+
+        `name` is the string's name, and `lineno` is the line number
+        where the example starts; both are used for error messages.
+        """
+        options = {}
+        # (note: with the current regexp, this will match at most once:)
+        for m in self._OPTION_DIRECTIVE_RE.finditer(source):
+            option_strings = m.group(1).replace(',', ' ').split()
+            for option in option_strings:
+                if (option[0] not in '+-' or
+                    option[1:] not in OPTIONFLAGS_BY_NAME):
+                    raise ValueError('line %r of the doctest for %s '
+                                     'has an invalid option: %r' %
+                                     (lineno+1, name, option))
+                flag = OPTIONFLAGS_BY_NAME[option[1:]]
+                options[flag] = (option[0] == '+')
+        if options and self._IS_BLANK_OR_COMMENT(source):
+            raise ValueError('line %r of the doctest for %s has an option '
+                             'directive on a line with no example: %r' %
+                             (lineno, name, source))
+        return options
+
+    # This regular expression finds the indentation of every non-blank
+    # line in a string.
+    _INDENT_RE = re.compile('^([ ]*)(?=\S)', re.MULTILINE)
+
+    def _min_indent(self, s):
+        "Return the minimum indentation of any non-blank line in `s`"
+        indents = [len(indent) for indent in self._INDENT_RE.findall(s)]
+        if len(indents) > 0:
+            return min(indents)
+        else:
+            return 0
+
+    def _check_prompt_blank(self, lines, indent, name, lineno):
+        """
+        Given the lines of a source string (including prompts and
+        leading indentation), check to make sure that every prompt is
+        followed by a space character.  If any line is not followed by
+        a space character, then raise ValueError.
+        """
+        for i, line in enumerate(lines):
+            if len(line) >= indent+4 and line[indent+3] != ' ':
+                raise ValueError('line %r of the docstring for %s '
+                                 'lacks blank after %s: %r' %
+                                 (lineno+i+1, name,
+                                  line[indent:indent+3], line))
+
+    def _check_prefix(self, lines, prefix, name, lineno):
+        """
+        Check that every line in the given list starts with the given
+        prefix; if any line does not, then raise a ValueError.
+        """
+        for i, line in enumerate(lines):
+            if line and not line.startswith(prefix):
+                raise ValueError('line %r of the docstring for %s has '
+                                 'inconsistent leading whitespace: %r' %
+                                 (lineno+i+1, name, line))
+
+
+######################################################################
+## 4. DocTest Finder
+######################################################################
+
+class DocTestFinder:
+    """
+    A class used to extract the DocTests that are relevant to a given
+    object, from its docstring and the docstrings of its contained
+    objects.  Doctests can currently be extracted from the following
+    object types: modules, functions, classes, methods, staticmethods,
+    classmethods, and properties.
+    """
+
+    def __init__(self, verbose=False, parser=DocTestParser(),
+                 recurse=True, _namefilter=None, exclude_empty=True):
+        """
+        Create a new doctest finder.
+
+        The optional argument `parser` specifies a class or
+        function that should be used to create new DocTest objects (or
+        objects that implement the same interface as DocTest).  The
+        signature for this factory function should match the signature
+        of the DocTest constructor.
+
+        If the optional argument `recurse` is false, then `find` will
+        only examine the given object, and not any contained objects.
+
+        If the optional argument `exclude_empty` is false, then `find`
+        will include tests for objects with empty docstrings.
+        """
+        self._parser = parser
+        self._verbose = verbose
+        self._recurse = recurse
+        self._exclude_empty = exclude_empty
+        # _namefilter is undocumented, and exists only for temporary backward-
+        # compatibility support of testmod's deprecated isprivate mess.
+        self._namefilter = _namefilter
+
+    def find(self, obj, name=None, module=None, globs=None,
+             extraglobs=None):
+        """
+        Return a list of the DocTests that are defined by the given
+        object's docstring, or by any of its contained objects'
+        docstrings.
+
+        The optional parameter `module` is the module that contains
+        the given object.  If the module is not specified or is None, then
+        the test finder will attempt to automatically determine the
+        correct module.  The object's module is used:
+
+            - As a default namespace, if `globs` is not specified.
+            - To prevent the DocTestFinder from extracting DocTests
+              from objects that are imported from other modules.
+            - To find the name of the file containing the object.
+            - To help find the line number of the object within its
+              file.
+
+        Contained objects whose module does not match `module` are ignored.
+
+        If `module` is False, no attempt to find the module will be made.
+        This is obscure, of use mostly in tests:  if `module` is False, or
+        is None but cannot be found automatically, then all objects are
+        considered to belong to the (non-existent) module, so all contained
+        objects will (recursively) be searched for doctests.
+
+        The globals for each DocTest is formed by combining `globs`
+        and `extraglobs` (bindings in `extraglobs` override bindings
+        in `globs`).  A new copy of the globals dictionary is created
+        for each DocTest.  If `globs` is not specified, then it
+        defaults to the module's `__dict__`, if specified, or {}
+        otherwise.  If `extraglobs` is not specified, then it defaults
+        to {}.
+
+        """
+        # If name was not specified, then extract it from the object.
+        if name is None:
+            name = getattr(obj, '__name__', None)
+            if name is None:
+                raise ValueError("DocTestFinder.find: name must be given "
+                        "when obj.__name__ doesn't exist: %r" %
+                                 (type(obj),))
+
+        # Find the module that contains the given object (if obj is
+        # a module, then module=obj.).  Note: this may fail, in which
+        # case module will be None.
+        if module is False:
+            module = None
+        elif module is None:
+            module = inspect.getmodule(obj)
+
+        # Read the module's source code.  This is used by
+        # DocTestFinder._find_lineno to find the line number for a
+        # given object's docstring.
+        try:
+            file = inspect.getsourcefile(obj) or inspect.getfile(obj)
+            source_lines = linecache.getlines(file)
+            if not source_lines:
+                source_lines = None
+        except TypeError:
+            source_lines = None
+
+        # Initialize globals, and merge in extraglobs.
+        if globs is None:
+            if module is None:
+                globs = {}
+            else:
+                globs = module.__dict__.copy()
+        else:
+            globs = globs.copy()
+        if extraglobs is not None:
+            globs.update(extraglobs)
+
+        # Recursively expore `obj`, extracting DocTests.
+        tests = []
+        self._find(tests, obj, name, module, source_lines, globs, {})
+        return tests
+
+    def _filter(self, obj, prefix, base):
+        """
+        Return true if the given object should not be examined.
+        """
+        return (self._namefilter is not None and
+                self._namefilter(prefix, base))
+
+    def _from_module(self, module, object):
+        """
+        Return true if the given object is defined in the given
+        module.
+        """
+        if module is None:
+            return True
+        elif inspect.isfunction(object):
+            return module.__dict__ is object.func_globals
+        elif inspect.isclass(object):
+            return module.__name__ == object.__module__
+        elif inspect.getmodule(object) is not None:
+            return module is inspect.getmodule(object)
+        elif hasattr(object, '__module__'):
+            return module.__name__ == object.__module__
+        elif isinstance(object, property):
+            return True # [XX] no way not be sure.
+        else:
+            raise ValueError("object must be a class or function")
+
+    def _find(self, tests, obj, name, module, source_lines, globs, seen):
+        """
+        Find tests for the given object and any contained objects, and
+        add them to `tests`.
+        """
+        if self._verbose:
+            print 'Finding tests in %s' % name
+
+        # If we've already processed this object, then ignore it.
+        if id(obj) in seen:
+            return
+        seen[id(obj)] = 1
+
+        # Find a test for this object, and add it to the list of tests.
+        test = self._get_test(obj, name, module, globs, source_lines)
+        if test is not None:
+            tests.append(test)
+
+        # Look for tests in a module's contained objects.
+        if inspect.ismodule(obj) and self._recurse:
+            for valname, val in obj.__dict__.items():
+                # Check if this contained object should be ignored.
+                if self._filter(val, name, valname):
+                    continue
+                valname = '%s.%s' % (name, valname)
+                # Recurse to functions & classes.
+                if ((inspect.isfunction(val) or inspect.isclass(val)) and
+                    self._from_module(module, val)):
+                    self._find(tests, val, valname, module, source_lines,
+                               globs, seen)
+
+        # Look for tests in a module's __test__ dictionary.
+        if inspect.ismodule(obj) and self._recurse:
+            for valname, val in getattr(obj, '__test__', {}).items():
+                if not isinstance(valname, basestring):
+                    raise ValueError("DocTestFinder.find: __test__ keys "
+                                     "must be strings: %r" %
+                                     (type(valname),))
+                if not (inspect.isfunction(val) or inspect.isclass(val) or
+                        inspect.ismethod(val) or inspect.ismodule(val) or
+                        isinstance(val, basestring)):
+                    raise ValueError("DocTestFinder.find: __test__ values "
+                                     "must be strings, functions, methods, "
+                                     "classes, or modules: %r" %
+                                     (type(val),))
+                valname = '%s.__test__.%s' % (name, valname)
+                self._find(tests, val, valname, module, source_lines,
+                           globs, seen)
+
+        # Look for tests in a class's contained objects.
+        if inspect.isclass(obj) and self._recurse:
+            for valname, val in obj.__dict__.items():
+                # Check if this contained object should be ignored.
+                if self._filter(val, name, valname):
+                    continue
+                # Special handling for staticmethod/classmethod.
+                if isinstance(val, staticmethod):
+                    val = getattr(obj, valname)
+                if isinstance(val, classmethod):
+                    val = getattr(obj, valname).im_func
+
+                # Recurse to methods, properties, and nested classes.
+                if ((inspect.isfunction(val) or inspect.isclass(val) or
+                      isinstance(val, property)) and
+                      self._from_module(module, val)):
+                    valname = '%s.%s' % (name, valname)
+                    self._find(tests, val, valname, module, source_lines,
+                               globs, seen)
+
+    def _get_test(self, obj, name, module, globs, source_lines):
+        """
+        Return a DocTest for the given object, if it defines a docstring;
+        otherwise, return None.
+        """
+        # Extract the object's docstring.  If it doesn't have one,
+        # then return None (no test for this object).
+        if isinstance(obj, basestring):
+            docstring = obj
+        else:
+            try:
+                if obj.__doc__ is None:
+                    docstring = ''
+                else:
+                    docstring = obj.__doc__
+                    if not isinstance(docstring, basestring):
+                        docstring = str(docstring)
+            except (TypeError, AttributeError):
+                docstring = ''
+
+        # Find the docstring's location in the file.
+        lineno = self._find_lineno(obj, source_lines)
+
+        # Don't bother if the docstring is empty.
+        if self._exclude_empty and not docstring:
+            return None
+
+        # Return a DocTest for this object.
+        if module is None:
+            filename = None
+        else:
+            filename = getattr(module, '__file__', module.__name__)
+            if filename[-4:] in (".pyc", ".pyo"):
+                filename = filename[:-1]
+        return self._parser.get_doctest(docstring, globs, name,
+                                        filename, lineno)
+
+    def _find_lineno(self, obj, source_lines):
+        """
+        Return a line number of the given object's docstring.  Note:
+        this method assumes that the object has a docstring.
+        """
+        lineno = None
+
+        # Find the line number for modules.
+        if inspect.ismodule(obj):
+            lineno = 0
+
+        # Find the line number for classes.
+        # Note: this could be fooled if a class is defined multiple
+        # times in a single file.
+        if inspect.isclass(obj):
+            if source_lines is None:
+                return None
+            pat = re.compile(r'^\s*class\s*%s\b' %
+                             getattr(obj, '__name__', '-'))
+            for i, line in enumerate(source_lines):
+                if pat.match(line):
+                    lineno = i
+                    break
+
+        # Find the line number for functions & methods.
+        if inspect.ismethod(obj): obj = obj.im_func
+        if inspect.isfunction(obj): obj = obj.func_code
+        if inspect.istraceback(obj): obj = obj.tb_frame
+        if inspect.isframe(obj): obj = obj.f_code
+        if inspect.iscode(obj):
+            lineno = getattr(obj, 'co_firstlineno', None)-1
+
+        # Find the line number where the docstring starts.  Assume
+        # that it's the first line that begins with a quote mark.
+        # Note: this could be fooled by a multiline function
+        # signature, where a continuation line begins with a quote
+        # mark.
+        if lineno is not None:
+            if source_lines is None:
+                return lineno+1
+            pat = re.compile('(^|.*:)\s*\w*("|\')')
+            for lineno in range(lineno, len(source_lines)):
+                if pat.match(source_lines[lineno]):
+                    return lineno
+
+        # We couldn't find the line number.
+        return None
+
+######################################################################
+## 5. DocTest Runner
+######################################################################
+
+class DocTestRunner:
+    """
+    A class used to run DocTest test cases, and accumulate statistics.
+    The `run` method is used to process a single DocTest case.  It
+    returns a tuple `(f, t)`, where `t` is the number of test cases
+    tried, and `f` is the number of test cases that failed.
+
+        >>> tests = DocTestFinder().find(_TestClass)
+        >>> runner = DocTestRunner(verbose=False)
+        >>> for test in tests:
+        ...     print runner.run(test)
+        (0, 2)
+        (0, 1)
+        (0, 2)
+        (0, 2)
+
+    The `summarize` method prints a summary of all the test cases that
+    have been run by the runner, and returns an aggregated `(f, t)`
+    tuple:
+
+        >>> runner.summarize(verbose=1)
+        4 items passed all tests:
+           2 tests in _TestClass
+           2 tests in _TestClass.__init__
+           2 tests in _TestClass.get
+           1 tests in _TestClass.square
+        7 tests in 4 items.
+        7 passed and 0 failed.
+        Test passed.
+        (0, 7)
+
+    The aggregated number of tried examples and failed examples is
+    also available via the `tries` and `failures` attributes:
+
+        >>> runner.tries
+        7
+        >>> runner.failures
+        0
+
+    The comparison between expected outputs and actual outputs is done
+    by an `OutputChecker`.  This comparison may be customized with a
+    number of option flags; see the documentation for `testmod` for
+    more information.  If the option flags are insufficient, then the
+    comparison may also be customized by passing a subclass of
+    `OutputChecker` to the constructor.
+
+    The test runner's display output can be controlled in two ways.
+    First, an output function (`out) can be passed to
+    `TestRunner.run`; this function will be called with strings that
+    should be displayed.  It defaults to `sys.stdout.write`.  If
+    capturing the output is not sufficient, then the display output
+    can be also customized by subclassing DocTestRunner, and
+    overriding the methods `report_start`, `report_success`,
+    `report_unexpected_exception`, and `report_failure`.
+    """
+    # This divider string is used to separate failure messages, and to
+    # separate sections of the summary.
+    DIVIDER = "*" * 70
+
+    def __init__(self, checker=None, verbose=None, optionflags=0):
+        """
+        Create a new test runner.
+
+        Optional keyword arg `checker` is the `OutputChecker` that
+        should be used to compare the expected outputs and actual
+        outputs of doctest examples.
+
+        Optional keyword arg 'verbose' prints lots of stuff if true,
+        only failures if false; by default, it's true iff '-v' is in
+        sys.argv.
+
+        Optional argument `optionflags` can be used to control how the
+        test runner compares expected output to actual output, and how
+        it displays failures.  See the documentation for `testmod` for
+        more information.
+        """
+        self._checker = checker or OutputChecker()
+        if verbose is None:
+            verbose = '-v' in sys.argv
+        self._verbose = verbose
+        self.optionflags = optionflags
+        self.original_optionflags = optionflags
+
+        # Keep track of the examples we've run.
+        self.tries = 0
+        self.failures = 0
+        self._name2ft = {}
+
+        # Create a fake output target for capturing doctest output.
+        self._fakeout = _SpoofOut()
+
+    #/////////////////////////////////////////////////////////////////
+    # Reporting methods
+    #/////////////////////////////////////////////////////////////////
+
+    def report_start(self, out, test, example):
+        """
+        Report that the test runner is about to process the given
+        example.  (Only displays a message if verbose=True)
+        """
+        if self._verbose:
+            if example.want:
+                out('Trying:\n' + _indent(example.source) +
+                    'Expecting:\n' + _indent(example.want))
+            else:
+                out('Trying:\n' + _indent(example.source) +
+                    'Expecting nothing\n')
+
+    def report_success(self, out, test, example, got):
+        """
+        Report that the given example ran successfully.  (Only
+        displays a message if verbose=True)
+        """
+        if self._verbose:
+            out("ok\n")
+
+    def report_failure(self, out, test, example, got):
+        """
+        Report that the given example failed.
+        """
+        out(self._failure_header(test, example) +
+            self._checker.output_difference(example, got, self.optionflags))
+
+    def report_unexpected_exception(self, out, test, example, exc_info):
+        """
+        Report that the given example raised an unexpected exception.
+        """
+        out(self._failure_header(test, example) +
+            'Exception raised:\n' + _indent(_exception_traceback(exc_info)))
+
+    def _failure_header(self, test, example):
+        out = [self.DIVIDER]
+        if test.filename:
+            if test.lineno is not None and example.lineno is not None:
+                lineno = test.lineno + example.lineno + 1
+            else:
+                lineno = '?'
+            out.append('File "%s", line %s, in %s' %
+                       (test.filename, lineno, test.name))
+        else:
+            out.append('Line %s, in %s' % (example.lineno+1, test.name))
+        out.append('Failed example:')
+        source = example.source
+        out.append(_indent(source))
+        return '\n'.join(out)
+
+    #/////////////////////////////////////////////////////////////////
+    # DocTest Running
+    #/////////////////////////////////////////////////////////////////
+
+    def __run(self, test, compileflags, out):
+        """
+        Run the examples in `test`.  Write the outcome of each example
+        with one of the `DocTestRunner.report_*` methods, using the
+        writer function `out`.  `compileflags` is the set of compiler
+        flags that should be used to execute examples.  Return a tuple
+        `(f, t)`, where `t` is the number of examples tried, and `f`
+        is the number of examples that failed.  The examples are run
+        in the namespace `test.globs`.
+        """
+        # Keep track of the number of failures and tries.
+        failures = tries = 0
+
+        # Save the option flags (since option directives can be used
+        # to modify them).
+        original_optionflags = self.optionflags
+
+        SUCCESS, FAILURE, BOOM = range(3) # `outcome` state
+
+        check = self._checker.check_output
+
+        # Process each example.
+        for examplenum, example in enumerate(test.examples):
+
+            # If REPORT_ONLY_FIRST_FAILURE is set, then supress
+            # reporting after the first failure.
+            quiet = (self.optionflags & REPORT_ONLY_FIRST_FAILURE and
+                     failures > 0)
+
+            # Merge in the example's options.
+            self.optionflags = original_optionflags
+            if example.options:
+                for (optionflag, val) in example.options.items():
+                    if val:
+                        self.optionflags |= optionflag
+                    else:
+                        self.optionflags &= ~optionflag
+
+            # Record that we started this example.
+            tries += 1
+            if not quiet:
+                self.report_start(out, test, example)
+
+            # Use a special filename for compile(), so we can retrieve
+            # the source code during interactive debugging (see
+            # __patched_linecache_getlines).
+            filename = '<doctest %s[%d]>' % (test.name, examplenum)
+
+            # Run the example in the given context (globs), and record
+            # any exception that gets raised.  (But don't intercept
+            # keyboard interrupts.)
+            try:
+                # Don't blink!  This is where the user's code gets run.
+                exec compile(example.source, filename, "single",
+                             compileflags, 1) in test.globs
+                self.debugger.set_continue() # ==== Example Finished ====
+                exception = None
+            except KeyboardInterrupt:
+                raise
+            except:
+                exception = sys.exc_info()
+                self.debugger.set_continue() # ==== Example Finished ====
+
+            got = self._fakeout.getvalue()  # the actual output
+            self._fakeout.truncate(0)
+            outcome = FAILURE   # guilty until proved innocent or insane
+
+            # If the example executed without raising any exceptions,
+            # verify its output.
+            if exception is None:
+                if check(example.want, got, self.optionflags):
+                    outcome = SUCCESS
+
+            # The example raised an exception:  check if it was expected.
+            else:
+                exc_info = sys.exc_info()
+                exc_msg = traceback.format_exception_only(*exc_info[:2])[-1]
+                if not quiet:
+                    got += _exception_traceback(exc_info)
+
+                # If `example.exc_msg` is None, then we weren't expecting
+                # an exception.
+                if example.exc_msg is None:
+                    outcome = BOOM
+
+                # We expected an exception:  see whether it matches.
+                elif check(example.exc_msg, exc_msg, self.optionflags):
+                    outcome = SUCCESS
+
+                # Another chance if they didn't care about the detail.
+                elif self.optionflags & IGNORE_EXCEPTION_DETAIL:
+                    m1 = re.match(r'[^:]*:', example.exc_msg)
+                    m2 = re.match(r'[^:]*:', exc_msg)
+                    if m1 and m2 and check(m1.group(0), m2.group(0),
+                                           self.optionflags):
+                        outcome = SUCCESS
+
+            # Report the outcome.
+            if outcome is SUCCESS:
+                if not quiet:
+                    self.report_success(out, test, example, got)
+            elif outcome is FAILURE:
+                if not quiet:
+                    self.report_failure(out, test, example, got)
+                failures += 1
+            elif outcome is BOOM:
+                if not quiet:
+                    self.report_unexpected_exception(out, test, example,
+                                                     exc_info)
+                failures += 1
+            else:
+                assert False, ("unknown outcome", outcome)
+
+        # Restore the option flags (in case they were modified)
+        self.optionflags = original_optionflags
+
+        # Record and return the number of failures and tries.
+        self.__record_outcome(test, failures, tries)
+        return failures, tries
+
+    def __record_outcome(self, test, f, t):
+        """
+        Record the fact that the given DocTest (`test`) generated `f`
+        failures out of `t` tried examples.
+        """
+        f2, t2 = self._name2ft.get(test.name, (0,0))
+        self._name2ft[test.name] = (f+f2, t+t2)
+        self.failures += f
+        self.tries += t
+
+    __LINECACHE_FILENAME_RE = re.compile(r'<doctest '
+                                         r'(?P<name>[\w\.]+)'
+                                         r'\[(?P<examplenum>\d+)\]>$')
+    def __patched_linecache_getlines(self, filename):
+        m = self.__LINECACHE_FILENAME_RE.match(filename)
+        if m and m.group('name') == self.test.name:
+            example = self.test.examples[int(m.group('examplenum'))]
+            return example.source.splitlines(True)
+        else:
+            return self.save_linecache_getlines(filename)
+
+    def run(self, test, compileflags=None, out=None, clear_globs=True):
+        """
+        Run the examples in `test`, and display the results using the
+        writer function `out`.
+
+        The examples are run in the namespace `test.globs`.  If
+        `clear_globs` is true (the default), then this namespace will
+        be cleared after the test runs, to help with garbage
+        collection.  If you would like to examine the namespace after
+        the test completes, then use `clear_globs=False`.
+
+        `compileflags` gives the set of flags that should be used by
+        the Python compiler when running the examples.  If not
+        specified, then it will default to the set of future-import
+        flags that apply to `globs`.
+
+        The output of each example is checked using
+        `DocTestRunner.check_output`, and the results are formatted by
+        the `DocTestRunner.report_*` methods.
+        """
+        self.test = test
+
+        if compileflags is None:
+            compileflags = _extract_future_flags(test.globs)
+
+        save_stdout = sys.stdout
+        if out is None:
+            out = save_stdout.write
+        sys.stdout = self._fakeout
+
+        # Patch pdb.set_trace to restore sys.stdout during interactive
+        # debugging (so it's not still redirected to self._fakeout).
+        # Note that the interactive output will go to *our*
+        # save_stdout, even if that's not the real sys.stdout; this
+        # allows us to write test cases for the set_trace behavior.
+        save_set_trace = pdb.set_trace
+        self.debugger = _OutputRedirectingPdb(save_stdout)
+        self.debugger.reset()
+        pdb.set_trace = self.debugger.set_trace
+
+        # Patch linecache.getlines, so we can see the example's source
+        # when we're inside the debugger.
+        self.save_linecache_getlines = linecache.getlines
+        linecache.getlines = self.__patched_linecache_getlines
+
+        try:
+            return self.__run(test, compileflags, out)
+        finally:
+            sys.stdout = save_stdout
+            pdb.set_trace = save_set_trace
+            linecache.getlines = self.save_linecache_getlines
+            if clear_globs:
+                test.globs.clear()
+
+    #/////////////////////////////////////////////////////////////////
+    # Summarization
+    #/////////////////////////////////////////////////////////////////
+    def summarize(self, verbose=None):
+        """
+        Print a summary of all the test cases that have been run by
+        this DocTestRunner, and return a tuple `(f, t)`, where `f` is
+        the total number of failed examples, and `t` is the total
+        number of tried examples.
+
+        The optional `verbose` argument controls how detailed the
+        summary is.  If the verbosity is not specified, then the
+        DocTestRunner's verbosity is used.
+        """
+        if verbose is None:
+            verbose = self._verbose
+        notests = []
+        passed = []
+        failed = []
+        totalt = totalf = 0
+        for x in self._name2ft.items():
+            name, (f, t) = x
+            assert f <= t
+            totalt += t
+            totalf += f
+            if t == 0:
+                notests.append(name)
+            elif f == 0:
+                passed.append( (name, t) )
+            else:
+                failed.append(x)
+        if verbose:
+            if notests:
+                print len(notests), "items had no tests:"
+                notests.sort()
+                for thing in notests:
+                    print "   ", thing
+            if passed:
+                print len(passed), "items passed all tests:"
+                passed.sort()
+                for thing, count in passed:
+                    print " %3d tests in %s" % (count, thing)
+        if failed:
+            print self.DIVIDER
+            print len(failed), "items had failures:"
+            failed.sort()
+            for thing, (f, t) in failed:
+                print " %3d of %3d in %s" % (f, t, thing)
+        if verbose:
+            print totalt, "tests in", len(self._name2ft), "items."
+            print totalt - totalf, "passed and", totalf, "failed."
+        if totalf:
+            print "***Test Failed***", totalf, "failures."
+        elif verbose:
+            print "Test passed."
+        return totalf, totalt
+
+    #/////////////////////////////////////////////////////////////////
+    # Backward compatibility cruft to maintain doctest.master.
+    #/////////////////////////////////////////////////////////////////
+    def merge(self, other):
+        d = self._name2ft
+        for name, (f, t) in other._name2ft.items():
+            if name in d:
+                print "*** DocTestRunner.merge: '" + name + "' in both" \
+                    " testers; summing outcomes."
+                f2, t2 = d[name]
+                f = f + f2
+                t = t + t2
+            d[name] = f, t
+
+class OutputChecker:
+    """
+    A class used to check the whether the actual output from a doctest
+    example matches the expected output.  `OutputChecker` defines two
+    methods: `check_output`, which compares a given pair of outputs,
+    and returns true if they match; and `output_difference`, which
+    returns a string describing the differences between two outputs.
+    """
+    def check_output(self, want, got, optionflags):
+        """
+        Return True iff the actual output from an example (`got`)
+        matches the expected output (`want`).  These strings are
+        always considered to match if they are identical; but
+        depending on what option flags the test runner is using,
+        several non-exact match types are also possible.  See the
+        documentation for `TestRunner` for more information about
+        option flags.
+        """
+        # Handle the common case first, for efficiency:
+        # if they're string-identical, always return true.
+        if got == want:
+            return True
+
+        # The values True and False replaced 1 and 0 as the return
+        # value for boolean comparisons in Python 2.3.
+        if not (optionflags & DONT_ACCEPT_TRUE_FOR_1):
+            if (got,want) == ("True\n", "1\n"):
+                return True
+            if (got,want) == ("False\n", "0\n"):
+                return True
+
+        # <BLANKLINE> can be used as a special sequence to signify a
+        # blank line, unless the DONT_ACCEPT_BLANKLINE flag is used.
+        if not (optionflags & DONT_ACCEPT_BLANKLINE):
+            # Replace <BLANKLINE> in want with a blank line.
+            want = re.sub('(?m)^%s\s*?$' % re.escape(BLANKLINE_MARKER),
+                          '', want)
+            # If a line in got contains only spaces, then remove the
+            # spaces.
+            got = re.sub('(?m)^\s*?$', '', got)
+            if got == want:
+                return True
+
+        # This flag causes doctest to ignore any differences in the
+        # contents of whitespace strings.  Note that this can be used
+        # in conjunction with the ELLIPSIS flag.
+        if optionflags & NORMALIZE_WHITESPACE:
+            got = ' '.join(got.split())
+            want = ' '.join(want.split())
+            if got == want:
+                return True
+
+        # The ELLIPSIS flag says to let the sequence "..." in `want`
+        # match any substring in `got`.
+        if optionflags & ELLIPSIS:
+            if _ellipsis_match(want, got):
+                return True
+
+        # We didn't find any match; return false.
+        return False
+
+    # Should we do a fancy diff?
+    def _do_a_fancy_diff(self, want, got, optionflags):
+        # Not unless they asked for a fancy diff.
+        if not optionflags & (REPORT_UDIFF |
+                              REPORT_CDIFF |
+                              REPORT_NDIFF):
+            return False
+
+        # If expected output uses ellipsis, a meaningful fancy diff is
+        # too hard ... or maybe not.  In two real-life failures Tim saw,
+        # a diff was a major help anyway, so this is commented out.
+        # [todo] _ellipsis_match() knows which pieces do and don't match,
+        # and could be the basis for a kick-ass diff in this case.
+        ##if optionflags & ELLIPSIS and ELLIPSIS_MARKER in want:
+        ##    return False
+
+        # ndiff does intraline difference marking, so can be useful even
+        # for 1-line differences.
+        if optionflags & REPORT_NDIFF:
+            return True
+
+        # The other diff types need at least a few lines to be helpful.
+        return want.count('\n') > 2 and got.count('\n') > 2
+
+    def output_difference(self, example, got, optionflags):
+        """
+        Return a string describing the differences between the
+        expected output for a given example (`example`) and the actual
+        output (`got`).  `optionflags` is the set of option flags used
+        to compare `want` and `got`.
+        """
+        want = example.want
+        # If <BLANKLINE>s are being used, then replace blank lines
+        # with <BLANKLINE> in the actual output string.
+        if not (optionflags & DONT_ACCEPT_BLANKLINE):
+            got = re.sub('(?m)^[ ]*(?=\n)', BLANKLINE_MARKER, got)
+
+        # Check if we should use diff.
+        if self._do_a_fancy_diff(want, got, optionflags):
+            # Split want & got into lines.
+            want_lines = want.splitlines(True)  # True == keep line ends
+            got_lines = got.splitlines(True)
+            # Use difflib to find their differences.
+            if optionflags & REPORT_UDIFF:
+                diff = difflib.unified_diff(want_lines, got_lines, n=2)
+                diff = list(diff)[2:] # strip the diff header
+                kind = 'unified diff with -expected +actual'
+            elif optionflags & REPORT_CDIFF:
+                diff = difflib.context_diff(want_lines, got_lines, n=2)
+                diff = list(diff)[2:] # strip the diff header
+                kind = 'context diff with expected followed by actual'
+            elif optionflags & REPORT_NDIFF:
+                engine = difflib.Differ(charjunk=difflib.IS_CHARACTER_JUNK)
+                diff = list(engine.compare(want_lines, got_lines))
+                kind = 'ndiff with -expected +actual'
+            else:
+                assert 0, 'Bad diff option'
+            # Remove trailing whitespace on diff output.
+            diff = [line.rstrip() + '\n' for line in diff]
+            return 'Differences (%s):\n' % kind + _indent(''.join(diff))
+
+        # If we're not using diff, then simply list the expected
+        # output followed by the actual output.
+        if want and got:
+            return 'Expected:\n%sGot:\n%s' % (_indent(want), _indent(got))
+        elif want:
+            return 'Expected:\n%sGot nothing\n' % _indent(want)
+        elif got:
+            return 'Expected nothing\nGot:\n%s' % _indent(got)
+        else:
+            return 'Expected nothing\nGot nothing\n'
+
+class DocTestFailure(Exception):
+    """A DocTest example has failed in debugging mode.
+
+    The exception instance has variables:
+
+    - test: the DocTest object being run
+
+    - excample: the Example object that failed
+
+    - got: the actual output
+    """
+    def __init__(self, test, example, got):
+        self.test = test
+        self.example = example
+        self.got = got
+
+    def __str__(self):
+        return str(self.test)
+
+class UnexpectedException(Exception):
+    """A DocTest example has encountered an unexpected exception
+
+    The exception instance has variables:
+
+    - test: the DocTest object being run
+
+    - excample: the Example object that failed
+
+    - exc_info: the exception info
+    """
+    def __init__(self, test, example, exc_info):
+        self.test = test
+        self.example = example
+        self.exc_info = exc_info
+
+    def __str__(self):
+        return str(self.test)
+
+class DebugRunner(DocTestRunner):
+    r"""Run doc tests but raise an exception as soon as there is a failure.
+
+       If an unexpected exception occurs, an UnexpectedException is raised.
+       It contains the test, the example, and the original exception:
+
+         >>> runner = DebugRunner(verbose=False)
+         >>> test = DocTestParser().get_doctest('>>> raise KeyError\n42',
+         ...                                    {}, 'foo', 'foo.py', 0)
+         >>> try:
+         ...     runner.run(test)
+         ... except UnexpectedException, failure:
+         ...     pass
+
+         >>> failure.test is test
+         True
+
+         >>> failure.example.want
+         '42\n'
+
+         >>> exc_info = failure.exc_info
+         >>> raise exc_info[0], exc_info[1], exc_info[2]
+         Traceback (most recent call last):
+         ...
+         KeyError
+
+       We wrap the original exception to give the calling application
+       access to the test and example information.
+
+       If the output doesn't match, then a DocTestFailure is raised:
+
+         >>> test = DocTestParser().get_doctest('''
+         ...      >>> x = 1
+         ...      >>> x
+         ...      2
+         ...      ''', {}, 'foo', 'foo.py', 0)
+
+         >>> try:
+         ...    runner.run(test)
+         ... except DocTestFailure, failure:
+         ...    pass
+
+       DocTestFailure objects provide access to the test:
+
+         >>> failure.test is test
+         True
+
+       As well as to the example:
+
+         >>> failure.example.want
+         '2\n'
+
+       and the actual output:
+
+         >>> failure.got
+         '1\n'
+
+       If a failure or error occurs, the globals are left intact:
+
+         >>> del test.globs['__builtins__']
+         >>> test.globs
+         {'x': 1}
+
+         >>> test = DocTestParser().get_doctest('''
+         ...      >>> x = 2
+         ...      >>> raise KeyError
+         ...      ''', {}, 'foo', 'foo.py', 0)
+
+         >>> runner.run(test)
+         Traceback (most recent call last):
+         ...
+         UnexpectedException: <DocTest foo from foo.py:0 (2 examples)>
+
+         >>> del test.globs['__builtins__']
+         >>> test.globs
+         {'x': 2}
+
+       But the globals are cleared if there is no error:
+
+         >>> test = DocTestParser().get_doctest('''
+         ...      >>> x = 2
+         ...      ''', {}, 'foo', 'foo.py', 0)
+
+         >>> runner.run(test)
+         (0, 1)
+
+         >>> test.globs
+         {}
+
+       """
+
+    def run(self, test, compileflags=None, out=None, clear_globs=True):
+        r = DocTestRunner.run(self, test, compileflags, out, False)
+        if clear_globs:
+            test.globs.clear()
+        return r
+
+    def report_unexpected_exception(self, out, test, example, exc_info):
+        raise UnexpectedException(test, example, exc_info)
+
+    def report_failure(self, out, test, example, got):
+        raise DocTestFailure(test, example, got)
+
+######################################################################
+## 6. Test Functions
+######################################################################
+# These should be backwards compatible.
+
+# For backward compatibility, a global instance of a DocTestRunner
+# class, updated by testmod.
+master = None
+
+def testmod(m=None, name=None, globs=None, verbose=None, isprivate=None,
+            report=True, optionflags=0, extraglobs=None,
+            raise_on_error=False, exclude_empty=False):
+    """m=None, name=None, globs=None, verbose=None, isprivate=None,
+       report=True, optionflags=0, extraglobs=None, raise_on_error=False,
+       exclude_empty=False
+
+    Test examples in docstrings in functions and classes reachable
+    from module m (or the current module if m is not supplied), starting
+    with m.__doc__.  Unless isprivate is specified, private names
+    are not skipped.
+
+    Also test examples reachable from dict m.__test__ if it exists and is
+    not None.  m.__test__ maps names to functions, classes and strings;
+    function and class docstrings are tested even if the name is private;
+    strings are tested directly, as if they were docstrings.
+
+    Return (#failures, #tests).
+
+    See doctest.__doc__ for an overview.
+
+    Optional keyword arg "name" gives the name of the module; by default
+    use m.__name__.
+
+    Optional keyword arg "globs" gives a dict to be used as the globals
+    when executing examples; by default, use m.__dict__.  A copy of this
+    dict is actually used for each docstring, so that each docstring's
+    examples start with a clean slate.
+
+    Optional keyword arg "extraglobs" gives a dictionary that should be
+    merged into the globals that are used to execute examples.  By
+    default, no extra globals are used.  This is new in 2.4.
+
+    Optional keyword arg "verbose" prints lots of stuff if true, prints
+    only failures if false; by default, it's true iff "-v" is in sys.argv.
+
+    Optional keyword arg "report" prints a summary at the end when true,
+    else prints nothing at the end.  In verbose mode, the summary is
+    detailed, else very brief (in fact, empty if all tests passed).
+
+    Optional keyword arg "optionflags" or's together module constants,
+    and defaults to 0.  This is new in 2.3.  Possible values (see the
+    docs for details):
+
+        DONT_ACCEPT_TRUE_FOR_1
+        DONT_ACCEPT_BLANKLINE
+        NORMALIZE_WHITESPACE
+        ELLIPSIS
+        IGNORE_EXCEPTION_DETAIL
+        REPORT_UDIFF
+        REPORT_CDIFF
+        REPORT_NDIFF
+        REPORT_ONLY_FIRST_FAILURE
+
+    Optional keyword arg "raise_on_error" raises an exception on the
+    first unexpected exception or failure. This allows failures to be
+    post-mortem debugged.
+
+    Deprecated in Python 2.4:
+    Optional keyword arg "isprivate" specifies a function used to
+    determine whether a name is private.  The default function is
+    treat all functions as public.  Optionally, "isprivate" can be
+    set to doctest.is_private to skip over functions marked as private
+    using the underscore naming convention; see its docs for details.
+
+    Advanced tomfoolery:  testmod runs methods of a local instance of
+    class doctest.Tester, then merges the results into (or creates)
+    global Tester instance doctest.master.  Methods of doctest.master
+    can be called directly too, if you want to do something unusual.
+    Passing report=0 to testmod is especially useful then, to delay
+    displaying a summary.  Invoke doctest.master.summarize(verbose)
+    when you're done fiddling.
+    """
+    global master
+
+    if isprivate is not None:
+        warnings.warn("the isprivate argument is deprecated; "
+                      "examine DocTestFinder.find() lists instead",
+                      DeprecationWarning)
+
+    # If no module was given, then use __main__.
+    if m is None:
+        # DWA - m will still be None if this wasn't invoked from the command
+        # line, in which case the following TypeError is about as good an error
+        # as we should expect
+        m = sys.modules.get('__main__')
+
+    # Check that we were actually given a module.
+    if not inspect.ismodule(m):
+        raise TypeError("testmod: module required; %r" % (m,))
+
+    # If no name was given, then use the module's name.
+    if name is None:
+        name = m.__name__
+
+    # Find, parse, and run all tests in the given module.
+    finder = DocTestFinder(_namefilter=isprivate, exclude_empty=exclude_empty)
+
+    if raise_on_error:
+        runner = DebugRunner(verbose=verbose, optionflags=optionflags)
+    else:
+        runner = DocTestRunner(verbose=verbose, optionflags=optionflags)
+
+    for test in finder.find(m, name, globs=globs, extraglobs=extraglobs):
+        runner.run(test)
+
+    if report:
+        runner.summarize()
+
+    if master is None:
+        master = runner
+    else:
+        master.merge(runner)
+
+    return runner.failures, runner.tries
+
+def testfile(filename, module_relative=True, name=None, package=None,
+             globs=None, verbose=None, report=True, optionflags=0,
+             extraglobs=None, raise_on_error=False, parser=DocTestParser()):
+    """
+    Test examples in the given file.  Return (#failures, #tests).
+
+    Optional keyword arg "module_relative" specifies how filenames
+    should be interpreted:
+
+      - If "module_relative" is True (the default), then "filename"
+         specifies a module-relative path.  By default, this path is
+         relative to the calling module's directory; but if the
+         "package" argument is specified, then it is relative to that
+         package.  To ensure os-independence, "filename" should use
+         "/" characters to separate path segments, and should not
+         be an absolute path (i.e., it may not begin with "/").
+
+      - If "module_relative" is False, then "filename" specifies an
+        os-specific path.  The path may be absolute or relative (to
+        the current working directory).
+
+    Optional keyword arg "name" gives the name of the test; by default
+    use the file's basename.
+
+    Optional keyword argument "package" is a Python package or the
+    name of a Python package whose directory should be used as the
+    base directory for a module relative filename.  If no package is
+    specified, then the calling module's directory is used as the base
+    directory for module relative filenames.  It is an error to
+    specify "package" if "module_relative" is False.
+
+    Optional keyword arg "globs" gives a dict to be used as the globals
+    when executing examples; by default, use {}.  A copy of this dict
+    is actually used for each docstring, so that each docstring's
+    examples start with a clean slate.
+
+    Optional keyword arg "extraglobs" gives a dictionary that should be
+    merged into the globals that are used to execute examples.  By
+    default, no extra globals are used.
+
+    Optional keyword arg "verbose" prints lots of stuff if true, prints
+    only failures if false; by default, it's true iff "-v" is in sys.argv.
+
+    Optional keyword arg "report" prints a summary at the end when true,
+    else prints nothing at the end.  In verbose mode, the summary is
+    detailed, else very brief (in fact, empty if all tests passed).
+
+    Optional keyword arg "optionflags" or's together module constants,
+    and defaults to 0.  Possible values (see the docs for details):
+
+        DONT_ACCEPT_TRUE_FOR_1
+        DONT_ACCEPT_BLANKLINE
+        NORMALIZE_WHITESPACE
+        ELLIPSIS
+        IGNORE_EXCEPTION_DETAIL
+        REPORT_UDIFF
+        REPORT_CDIFF
+        REPORT_NDIFF
+        REPORT_ONLY_FIRST_FAILURE
+
+    Optional keyword arg "raise_on_error" raises an exception on the
+    first unexpected exception or failure. This allows failures to be
+    post-mortem debugged.
+
+    Optional keyword arg "parser" specifies a DocTestParser (or
+    subclass) that should be used to extract tests from the files.
+
+    Advanced tomfoolery:  testmod runs methods of a local instance of
+    class doctest.Tester, then merges the results into (or creates)
+    global Tester instance doctest.master.  Methods of doctest.master
+    can be called directly too, if you want to do something unusual.
+    Passing report=0 to testmod is especially useful then, to delay
+    displaying a summary.  Invoke doctest.master.summarize(verbose)
+    when you're done fiddling.
+    """
+    global master
+
+    if package and not module_relative:
+        raise ValueError("Package may only be specified for module-"
+                         "relative paths.")
+
+    # Relativize the path
+    if module_relative:
+        package = _normalize_module(package)
+        filename = _module_relative_path(package, filename)
+
+    # If no name was given, then use the file's name.
+    if name is None:
+        name = os.path.basename(filename)
+
+    # Assemble the globals.
+    if globs is None:
+        globs = {}
+    else:
+        globs = globs.copy()
+    if extraglobs is not None:
+        globs.update(extraglobs)
+
+    if raise_on_error:
+        runner = DebugRunner(verbose=verbose, optionflags=optionflags)
+    else:
+        runner = DocTestRunner(verbose=verbose, optionflags=optionflags)
+
+    # Read the file, convert it to a test, and run it.
+    s = open(filename).read()
+    test = parser.get_doctest(s, globs, name, filename, 0)
+    runner.run(test)
+
+    if report:
+        runner.summarize()
+
+    if master is None:
+        master = runner
+    else:
+        master.merge(runner)
+
+    return runner.failures, runner.tries
+
+def run_docstring_examples(f, globs, verbose=False, name="NoName",
+                           compileflags=None, optionflags=0):
+    """
+    Test examples in the given object's docstring (`f`), using `globs`
+    as globals.  Optional argument `name` is used in failure messages.
+    If the optional argument `verbose` is true, then generate output
+    even if there are no failures.
+
+    `compileflags` gives the set of flags that should be used by the
+    Python compiler when running the examples.  If not specified, then
+    it will default to the set of future-import flags that apply to
+    `globs`.
+
+    Optional keyword arg `optionflags` specifies options for the
+    testing and output.  See the documentation for `testmod` for more
+    information.
+    """
+    # Find, parse, and run all tests in the given module.
+    finder = DocTestFinder(verbose=verbose, recurse=False)
+    runner = DocTestRunner(verbose=verbose, optionflags=optionflags)
+    for test in finder.find(f, name, globs=globs):
+        runner.run(test, compileflags=compileflags)
+
+######################################################################
+## 7. Tester
+######################################################################
+# This is provided only for backwards compatibility.  It's not
+# actually used in any way.
+
+class Tester:
+    def __init__(self, mod=None, globs=None, verbose=None,
+                 isprivate=None, optionflags=0):
+
+        warnings.warn("class Tester is deprecated; "
+                      "use class doctest.DocTestRunner instead",
+                      DeprecationWarning, stacklevel=2)
+        if mod is None and globs is None:
+            raise TypeError("Tester.__init__: must specify mod or globs")
+        if mod is not None and not inspect.ismodule(mod):
+            raise TypeError("Tester.__init__: mod must be a module; %r" %
+                            (mod,))
+        if globs is None:
+            globs = mod.__dict__
+        self.globs = globs
+
+        self.verbose = verbose
+        self.isprivate = isprivate
+        self.optionflags = optionflags
+        self.testfinder = DocTestFinder(_namefilter=isprivate)
+        self.testrunner = DocTestRunner(verbose=verbose,
+                                        optionflags=optionflags)
+
+    def runstring(self, s, name):
+        test = DocTestParser().get_doctest(s, self.globs, name, None, None)
+        if self.verbose:
+            print "Running string", name
+        (f,t) = self.testrunner.run(test)
+        if self.verbose:
+            print f, "of", t, "examples failed in string", name
+        return (f,t)
+
+    def rundoc(self, object, name=None, module=None):
+        f = t = 0
+        tests = self.testfinder.find(object, name, module=module,
+                                     globs=self.globs)
+        for test in tests:
+            (f2, t2) = self.testrunner.run(test)
+            (f,t) = (f+f2, t+t2)
+        return (f,t)
+
+    def rundict(self, d, name, module=None):
+        import new
+        m = new.module(name)
+        m.__dict__.update(d)
+        if module is None:
+            module = False
+        return self.rundoc(m, name, module)
+
+    def run__test__(self, d, name):
+        import new
+        m = new.module(name)
+        m.__test__ = d
+        return self.rundoc(m, name)
+
+    def summarize(self, verbose=None):
+        return self.testrunner.summarize(verbose)
+
+    def merge(self, other):
+        self.testrunner.merge(other.testrunner)
+
+######################################################################
+## 8. Unittest Support
+######################################################################
+
+_unittest_reportflags = 0
+
+def set_unittest_reportflags(flags):
+    """Sets the unittest option flags.
+
+    The old flag is returned so that a runner could restore the old
+    value if it wished to:
+
+      >>> old = _unittest_reportflags
+      >>> set_unittest_reportflags(REPORT_NDIFF |
+      ...                          REPORT_ONLY_FIRST_FAILURE) == old
+      True
+
+      >>> import doctest
+      >>> doctest._unittest_reportflags == (REPORT_NDIFF |
+      ...                                   REPORT_ONLY_FIRST_FAILURE)
+      True
+
+    Only reporting flags can be set:
+
+      >>> set_unittest_reportflags(ELLIPSIS)
+      Traceback (most recent call last):
+      ...
+      ValueError: ('Only reporting flags allowed', 8)
+
+      >>> set_unittest_reportflags(old) == (REPORT_NDIFF |
+      ...                                   REPORT_ONLY_FIRST_FAILURE)
+      True
+    """
+    global _unittest_reportflags
+
+    if (flags & REPORTING_FLAGS) != flags:
+        raise ValueError("Only reporting flags allowed", flags)
+    old = _unittest_reportflags
+    _unittest_reportflags = flags
+    return old
+
+
+class DocTestCase(unittest.TestCase):
+
+    def __init__(self, test, optionflags=0, setUp=None, tearDown=None,
+                 checker=None):
+
+        unittest.TestCase.__init__(self)
+        self._dt_optionflags = optionflags
+        self._dt_checker = checker
+        self._dt_test = test
+        self._dt_setUp = setUp
+        self._dt_tearDown = tearDown
+
+    def setUp(self):
+        test = self._dt_test
+
+        if self._dt_setUp is not None:
+            self._dt_setUp(test)
+
+    def tearDown(self):
+        test = self._dt_test
+
+        if self._dt_tearDown is not None:
+            self._dt_tearDown(test)
+
+        test.globs.clear()
+
+    def runTest(self):
+        test = self._dt_test
+        old = sys.stdout
+        new = StringIO()
+        optionflags = self._dt_optionflags
+
+        if not (optionflags & REPORTING_FLAGS):
+            # The option flags don't include any reporting flags,
+            # so add the default reporting flags
+            optionflags |= _unittest_reportflags
+
+        runner = DocTestRunner(optionflags=optionflags,
+                               checker=self._dt_checker, verbose=False)
+
+        try:
+            runner.DIVIDER = "-"*70
+            failures, tries = runner.run(
+                test, out=new.write, clear_globs=False)
+        finally:
+            sys.stdout = old
+
+        if failures:
+            raise self.failureException(self.format_failure(new.getvalue()))
+
+    def format_failure(self, err):
+        test = self._dt_test
+        if test.lineno is None:
+            lineno = 'unknown line number'
+        else:
+            lineno = '%s' % test.lineno
+        lname = '.'.join(test.name.split('.')[-1:])
+        return ('Failed doctest test for %s\n'
+                '  File "%s", line %s, in %s\n\n%s'
+                % (test.name, test.filename, lineno, lname, err)
+                )
+
+    def debug(self):
+        r"""Run the test case without results and without catching exceptions
+
+           The unit test framework includes a debug method on test cases
+           and test suites to support post-mortem debugging.  The test code
+           is run in such a way that errors are not caught.  This way a
+           caller can catch the errors and initiate post-mortem debugging.
+
+           The DocTestCase provides a debug method that raises
+           UnexpectedException errors if there is an unexepcted
+           exception:
+
+             >>> test = DocTestParser().get_doctest('>>> raise KeyError\n42',
+             ...                {}, 'foo', 'foo.py', 0)
+             >>> case = DocTestCase(test)
+             >>> try:
+             ...     case.debug()
+             ... except UnexpectedException, failure:
+             ...     pass
+
+           The UnexpectedException contains the test, the example, and
+           the original exception:
+
+             >>> failure.test is test
+             True
+
+             >>> failure.example.want
+             '42\n'
+
+             >>> exc_info = failure.exc_info
+             >>> raise exc_info[0], exc_info[1], exc_info[2]
+             Traceback (most recent call last):
+             ...
+             KeyError
+
+           If the output doesn't match, then a DocTestFailure is raised:
+
+             >>> test = DocTestParser().get_doctest('''
+             ...      >>> x = 1
+             ...      >>> x
+             ...      2
+             ...      ''', {}, 'foo', 'foo.py', 0)
+             >>> case = DocTestCase(test)
+
+             >>> try:
+             ...    case.debug()
+             ... except DocTestFailure, failure:
+             ...    pass
+
+           DocTestFailure objects provide access to the test:
+
+             >>> failure.test is test
+             True
+
+           As well as to the example:
+
+             >>> failure.example.want
+             '2\n'
+
+           and the actual output:
+
+             >>> failure.got
+             '1\n'
+
+           """
+
+        self.setUp()
+        runner = DebugRunner(optionflags=self._dt_optionflags,
+                             checker=self._dt_checker, verbose=False)
+        runner.run(self._dt_test)
+        self.tearDown()
+
+    def id(self):
+        return self._dt_test.name
+
+    def __repr__(self):
+        name = self._dt_test.name.split('.')
+        return "%s (%s)" % (name[-1], '.'.join(name[:-1]))
+
+    __str__ = __repr__
+
+    def shortDescription(self):
+        return "Doctest: " + self._dt_test.name
+
+def DocTestSuite(module=None, globs=None, extraglobs=None, test_finder=None,
+                 **options):
+    """
+    Convert doctest tests for a module to a unittest test suite.
+
+    This converts each documentation string in a module that
+    contains doctest tests to a unittest test case.  If any of the
+    tests in a doc string fail, then the test case fails.  An exception
+    is raised showing the name of the file containing the test and a
+    (sometimes approximate) line number.
+
+    The `module` argument provides the module to be tested.  The argument
+    can be either a module or a module name.
+
+    If no argument is given, the calling module is used.
+
+    A number of options may be provided as keyword arguments:
+
+    setUp
+      A set-up function.  This is called before running the
+      tests in each file. The setUp function will be passed a DocTest
+      object.  The setUp function can access the test globals as the
+      globs attribute of the test passed.
+
+    tearDown
+      A tear-down function.  This is called after running the
+      tests in each file.  The tearDown function will be passed a DocTest
+      object.  The tearDown function can access the test globals as the
+      globs attribute of the test passed.
+
+    globs
+      A dictionary containing initial global variables for the tests.
+
+    optionflags
+       A set of doctest option flags expressed as an integer.
+    """
+
+    if test_finder is None:
+        test_finder = DocTestFinder()
+
+    module = _normalize_module(module)
+    tests = test_finder.find(module, globs=globs, extraglobs=extraglobs)
+    if globs is None:
+        globs = module.__dict__
+    if not tests:
+        # Why do we want to do this? Because it reveals a bug that might
+        # otherwise be hidden.
+        raise ValueError(module, "has no tests")
+
+    tests.sort()
+    suite = unittest.TestSuite()
+    for test in tests:
+        if len(test.examples) == 0:
+            continue
+        if not test.filename:
+            filename = module.__file__
+            if filename[-4:] in (".pyc", ".pyo"):
+                filename = filename[:-1]
+            test.filename = filename
+        suite.addTest(DocTestCase(test, **options))
+
+    return suite
+
+class DocFileCase(DocTestCase):
+
+    def id(self):
+        return '_'.join(self._dt_test.name.split('.'))
+
+    def __repr__(self):
+        return self._dt_test.filename
+    __str__ = __repr__
+
+    def format_failure(self, err):
+        return ('Failed doctest test for %s\n  File "%s", line 0\n\n%s'
+                % (self._dt_test.name, self._dt_test.filename, err)
+                )
+
+def DocFileTest(path, module_relative=True, package=None,
+                globs=None, parser=DocTestParser(), **options):
+    if globs is None:
+        globs = {}
+
+    if package and not module_relative:
+        raise ValueError("Package may only be specified for module-"
+                         "relative paths.")
+
+    # Relativize the path.
+    if module_relative:
+        package = _normalize_module(package)
+        path = _module_relative_path(package, path)
+
+    # Find the file and read it.
+    name = os.path.basename(path)
+    doc = open(path).read()
+
+    # Convert it to a test, and wrap it in a DocFileCase.
+    test = parser.get_doctest(doc, globs, name, path, 0)
+    return DocFileCase(test, **options)
+
+def DocFileSuite(*paths, **kw):
+    """A unittest suite for one or more doctest files.
+
+    The path to each doctest file is given as a string; the
+    interpretation of that string depends on the keyword argument
+    "module_relative".
+
+    A number of options may be provided as keyword arguments:
+
+    module_relative
+      If "module_relative" is True, then the given file paths are
+      interpreted as os-independent module-relative paths.  By
+      default, these paths are relative to the calling module's
+      directory; but if the "package" argument is specified, then
+      they are relative to that package.  To ensure os-independence,
+      "filename" should use "/" characters to separate path
+      segments, and may not be an absolute path (i.e., it may not
+      begin with "/").
+
+      If "module_relative" is False, then the given file paths are
+      interpreted as os-specific paths.  These paths may be absolute
+      or relative (to the current working directory).
+
+    package
+      A Python package or the name of a Python package whose directory
+      should be used as the base directory for module relative paths.
+      If "package" is not specified, then the calling module's
+      directory is used as the base directory for module relative
+      filenames.  It is an error to specify "package" if
+      "module_relative" is False.
+
+    setUp
+      A set-up function.  This is called before running the
+      tests in each file. The setUp function will be passed a DocTest
+      object.  The setUp function can access the test globals as the
+      globs attribute of the test passed.
+
+    tearDown
+      A tear-down function.  This is called after running the
+      tests in each file.  The tearDown function will be passed a DocTest
+      object.  The tearDown function can access the test globals as the
+      globs attribute of the test passed.
+
+    globs
+      A dictionary containing initial global variables for the tests.
+
+    optionflags
+      A set of doctest option flags expressed as an integer.
+
+    parser
+      A DocTestParser (or subclass) that should be used to extract
+      tests from the files.
+    """
+    suite = unittest.TestSuite()
+
+    # We do this here so that _normalize_module is called at the right
+    # level.  If it were called in DocFileTest, then this function
+    # would be the caller and we might guess the package incorrectly.
+    if kw.get('module_relative', True):
+        kw['package'] = _normalize_module(kw.get('package'))
+
+    for path in paths:
+        suite.addTest(DocFileTest(path, **kw))
+
+    return suite
+
+######################################################################
+## 9. Debugging Support
+######################################################################
+
+def script_from_examples(s):
+    r"""Extract script from text with examples.
+
+       Converts text with examples to a Python script.  Example input is
+       converted to regular code.  Example output and all other words
+       are converted to comments:
+
+       >>> text = '''
+       ...       Here are examples of simple math.
+       ...
+       ...           Python has super accurate integer addition
+       ...
+       ...           >>> 2 + 2
+       ...           5
+       ...
+       ...           And very friendly error messages:
+       ...
+       ...           >>> 1/0
+       ...           To Infinity
+       ...           And
+       ...           Beyond
+       ...
+       ...           You can use logic if you want:
+       ...
+       ...           >>> if 0:
+       ...           ...    blah
+       ...           ...    blah
+       ...           ...
+       ...
+       ...           Ho hum
+       ...           '''
+
+       >>> print script_from_examples(text)
+       # Here are examples of simple math.
+       #
+       #     Python has super accurate integer addition
+       #
+       2 + 2
+       # Expected:
+       ## 5
+       #
+       #     And very friendly error messages:
+       #
+       1/0
+       # Expected:
+       ## To Infinity
+       ## And
+       ## Beyond
+       #
+       #     You can use logic if you want:
+       #
+       if 0:
+          blah
+          blah
+       #
+       #     Ho hum
+       """
+    output = []
+    for piece in DocTestParser().parse(s):
+        if isinstance(piece, Example):
+            # Add the example's source code (strip trailing NL)
+            output.append(piece.source[:-1])
+            # Add the expected output:
+            want = piece.want
+            if want:
+                output.append('# Expected:')
+                output += ['## '+l for l in want.split('\n')[:-1]]
+        else:
+            # Add non-example text.
+            output += [_comment_line(l)
+                       for l in piece.split('\n')[:-1]]
+
+    # Trim junk on both ends.
+    while output and output[-1] == '#':
+        output.pop()
+    while output and output[0] == '#':
+        output.pop(0)
+    # Combine the output, and return it.
+    return '\n'.join(output)
+
+def testsource(module, name):
+    """Extract the test sources from a doctest docstring as a script.
+
+    Provide the module (or dotted name of the module) containing the
+    test to be debugged and the name (within the module) of the object
+    with the doc string with tests to be debugged.
+    """
+    module = _normalize_module(module)
+    tests = DocTestFinder().find(module)
+    test = [t for t in tests if t.name == name]
+    if not test:
+        raise ValueError(name, "not found in tests")
+    test = test[0]
+    testsrc = script_from_examples(test.docstring)
+    return testsrc
+
+def debug_src(src, pm=False, globs=None):
+    """Debug a single doctest docstring, in argument `src`'"""
+    testsrc = script_from_examples(src)
+    debug_script(testsrc, pm, globs)
+
+def debug_script(src, pm=False, globs=None):
+    "Debug a test script.  `src` is the script, as a string."
+    import pdb
+
+    # Note that tempfile.NameTemporaryFile() cannot be used.  As the
+    # docs say, a file so created cannot be opened by name a second time
+    # on modern Windows boxes, and execfile() needs to open it.
+    srcfilename = tempfile.mktemp(".py", "doctestdebug")
+    f = open(srcfilename, 'w')
+    f.write(src)
+    f.close()
+
+    try:
+        if globs:
+            globs = globs.copy()
+        else:
+            globs = {}
+
+        if pm:
+            try:
+                execfile(srcfilename, globs, globs)
+            except:
+                print sys.exc_info()[1]
+                pdb.post_mortem(sys.exc_info()[2])
+        else:
+            # Note that %r is vital here.  '%s' instead can, e.g., cause
+            # backslashes to get treated as metacharacters on Windows.
+            pdb.run("execfile(%r)" % srcfilename, globs, globs)
+
+    finally:
+        os.remove(srcfilename)
+
+def debug(module, name, pm=False):
+    """Debug a single doctest docstring.
+
+    Provide the module (or dotted name of the module) containing the
+    test to be debugged and the name (within the module) of the object
+    with the docstring with tests to be debugged.
+    """
+    module = _normalize_module(module)
+    testsrc = testsource(module, name)
+    debug_script(testsrc, pm, module.__dict__)
+
+######################################################################
+## 10. Example Usage
+######################################################################
+class _TestClass:
+    """
+    A pointless class, for sanity-checking of docstring testing.
+
+    Methods:
+        square()
+        get()
+
+    >>> _TestClass(13).get() + _TestClass(-12).get()
+    1
+    >>> hex(_TestClass(13).square().get())
+    '0xa9'
+    """
+
+    def __init__(self, val):
+        """val -> _TestClass object with associated value val.
+
+        >>> t = _TestClass(123)
+        >>> print t.get()
+        123
+        """
+
+        self.val = val
+
+    def square(self):
+        """square() -> square TestClass's associated value
+
+        >>> _TestClass(13).square().get()
+        169
+        """
+
+        self.val = self.val ** 2
+        return self
+
+    def get(self):
+        """get() -> return TestClass's associated value.
+
+        >>> x = _TestClass(-42)
+        >>> print x.get()
+        -42
+        """
+
+        return self.val
+
+__test__ = {"_TestClass": _TestClass,
+            "string": r"""
+                      Example of a string object, searched as-is.
+                      >>> x = 1; y = 2
+                      >>> x + y, x * y
+                      (3, 2)
+                      """,
+
+            "bool-int equivalence": r"""
+                                    In 2.2, boolean expressions displayed
+                                    0 or 1.  By default, we still accept
+                                    them.  This can be disabled by passing
+                                    DONT_ACCEPT_TRUE_FOR_1 to the new
+                                    optionflags argument.
+                                    >>> 4 == 4
+                                    1
+                                    >>> 4 == 4
+                                    True
+                                    >>> 4 > 4
+                                    0
+                                    >>> 4 > 4
+                                    False
+                                    """,
+
+            "blank lines": r"""
+                Blank lines can be marked with <BLANKLINE>:
+                    >>> print 'foo\n\nbar\n'
+                    foo
+                    <BLANKLINE>
+                    bar
+                    <BLANKLINE>
+            """,
+
+            "ellipsis": r"""
+                If the ellipsis flag is used, then '...' can be used to
+                elide substrings in the desired output:
+                    >>> print range(1000) #doctest: +ELLIPSIS
+                    [0, 1, 2, ..., 999]
+            """,
+
+            "whitespace normalization": r"""
+                If the whitespace normalization flag is used, then
+                differences in whitespace are ignored.
+                    >>> print range(30) #doctest: +NORMALIZE_WHITESPACE
+                    [0, 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]
+            """,
+           }
+
+def _test():
+    r = unittest.TextTestRunner()
+    r.run(DocTestSuite())
+
+if __name__ == "__main__":
+    _test()
+
diff --git a/setuptools/tests/test_resources.py b/setuptools/tests/test_resources.py
index 0a4cb183..1cb841da 100644
--- a/setuptools/tests/test_resources.py
+++ b/setuptools/tests/test_resources.py
@@ -54,23 +54,23 @@ class DistroTests(TestCase):
             [dist.version for dist in ad['FooPkg']], ['1.9','1.4','1.2']
         )
 
-        path = []
+        ws = WorkingSet([])
+        foo12 = Distribution.from_filename("FooPkg-1.2-py2.4.egg")
+        foo14 = Distribution.from_filename("FooPkg-1.4-py2.4-win32.egg")
         req, = parse_requirements("FooPkg>=1.3")
 
         # Nominal case: no distros on path, should yield all applicable
-        self.assertEqual(ad.best_match(req,path).version, '1.9')
-
+        self.assertEqual(ad.best_match(req,ws).version, '1.9')
         # If a matching distro is already installed, should return only that
-        path.append("FooPkg-1.4-py2.4-win32.egg")
-        self.assertEqual(ad.best_match(req,path).version, '1.4')
+        ws.add(foo14); self.assertEqual(ad.best_match(req,ws).version, '1.4')
 
         # If the first matching distro is unsuitable, it's a version conflict
-        path.insert(0,"FooPkg-1.2-py2.4.egg")
-        self.assertRaises(VersionConflict, ad.best_match, req, path)
+        ws = WorkingSet([]); ws.add(foo12); ws.add(foo14)
+        self.assertRaises(VersionConflict, ad.best_match, req, ws)
 
         # If more than one match on the path, the first one takes precedence
-        path.insert(0,"FooPkg-1.4-py2.4-win32.egg")
-        self.assertEqual(ad.best_match(req,path).version, '1.4')
+        ws = WorkingSet([]); ws.add(foo14); ws.add(foo12); ws.add(foo14);
+        self.assertEqual(ad.best_match(req,ws).version, '1.4')
 
     def checkFooPkg(self,d):
         self.assertEqual(d.project_name, "FooPkg")
@@ -86,8 +86,6 @@ class DistroTests(TestCase):
             project_name="FooPkg",version="1.3-1",py_version="2.4",platform="win32"
         )
         self.checkFooPkg(d)
-        self.failUnless(d.installed_on(["/some/path"]))
-        self.failIf(d.installed_on([]))
 
         d = Distribution("/some/path")
         self.assertEqual(d.py_version, sys.version[:3])
@@ -121,15 +119,17 @@ class DistroTests(TestCase):
             self.checkDepends(self.distDepends(v), v)
 
 
+
+
     def testResolve(self):
-        ad = AvailableDistributions([])
+        ad = AvailableDistributions([]); ws = WorkingSet([])
 
         # Resolving no requirements -> nothing to install
-        self.assertEqual( list(ad.resolve([],[])), [] )
+        self.assertEqual( list(ws.resolve([],ad)), [] )
 
         # Request something not in the collection -> DistributionNotFound
         self.assertRaises(
-            DistributionNotFound, ad.resolve, parse_requirements("Foo"), []
+            DistributionNotFound, ws.resolve, parse_requirements("Foo"), ad
         )
 
         Foo = Distribution.from_filename(
@@ -138,28 +138,28 @@ class DistroTests(TestCase):
         )
         ad.add(Foo)
 
-        # Request thing(s) that are available -> list to install
+        # Request thing(s) that are available -> list to activate
         self.assertEqual(
-            list(ad.resolve(parse_requirements("Foo"),[])), [Foo]
+            list(ws.resolve(parse_requirements("Foo"), ad)), [Foo]
         )
 
-        # Request an option that causes an unresolved dependency for "Baz"
+        # Request an extra that causes an unresolved dependency for "Baz"
         self.assertRaises(
-            DistributionNotFound, ad.resolve,parse_requirements("Foo[bar]"),[]
+            DistributionNotFound, ws.resolve,parse_requirements("Foo[bar]"), ad
         )      
         Baz = Distribution.from_filename(
             "/foo_dir/Baz-2.1.egg", metadata=Metadata(('depends.txt', "Foo"))
         )
         ad.add(Baz)
 
-        # Install list now includes resolved dependency
+        # Activation list now includes resolved dependency
         self.assertEqual(
-            list(ad.resolve(parse_requirements("Foo[bar]"),[])), [Foo,Baz]
+            list(ws.resolve(parse_requirements("Foo[bar]"), ad)), [Foo,Baz]
         )
         # Requests for conflicting versions produce VersionConflict
         self.assertRaises(
             VersionConflict,
-            ad.resolve, parse_requirements("Foo==1.2\nFoo!=1.2"), []
+            ws.resolve, parse_requirements("Foo==1.2\nFoo!=1.2"), ad
         )
 
     def testDistroDependsOptions(self):
@@ -208,7 +208,7 @@ class RequirementsTests(TestCase):
     def testBasics(self):
         r = Requirement.parse("Twisted>=1.2")
         self.assertEqual(str(r),"Twisted>=1.2")
-        self.assertEqual(repr(r),"Requirement('Twisted', [('>=', '1.2')], ())")
+        self.assertEqual(repr(r),"Requirement.parse('Twisted>=1.2')")
         self.assertEqual(r, Requirement("Twisted", [('>=','1.2')]))
         self.assertEqual(r, Requirement("twisTed", [('>=','1.2')]))
         self.assertNotEqual(r, Requirement("Twisted", [('>=','2.0')]))
@@ -285,13 +285,6 @@ class RequirementsTests(TestCase):
 
 
 
-
-
-
-
-
-
-
 class ParseTests(TestCase):
 
     def testEmptyParse(self):