Edit

[EDIT] FILE: test_Path.py

from guppy.heapy.test import support
import inspect
import unittest

class TestCase(support.TestCase):
    def setUp(self):
        support.TestCase.setUp(self)
        self.Path = self.heapy.Path

def chkrel(self, src, dst, relstr=None, clas=None):
        rel = self.relation(src, dst)
        if clas is not None:
            self.assertTrue(isinstance(rel, clas))
        if relstr is None:
            print(rel)
        else:
            sr = str(rel)
            if sr.startswith('<') and not relstr.startswith('<'):
                self.assertTrue(sr.endswith('>'))
                sr = sr[1:-1].split(',')
                self.assertTrue(relstr in sr)
            else:
                self.aseq(sr, relstr)

def chkrelattr(self, src, *attrs):
        for attr in attrs:
            self.chkrel(src, getattr(src, attr), '%s.'+attr)

def chkpath(self, src, dst, expect=None):
        rel = self.shpaths(dst, src)
        if expect is None:
            print(rel)
        else:
            li = rel.aslist()
            if len(li) == 1:
                li = li[0]
            self.aseq(str(li), str(expect), -1)

def relation(self, src, dst):
        return self.Path.relation(src, dst)

def shpaths(self, dst, src=None, *args, **kwds):
        # return self.Path.shpaths(dst, src, *args, **kwds)

dst = self.iso(dst)
        if src is not None:
            src = self.iso(src)
        return dst.get_shpaths(src, *args, **kwds)

class RelationTestCase(TestCase):
    # Test relations from standard types and some simple paths

def test_cell_relation(self):
        cellvalue = []

def f():
            return cellvalue
        self.chkrel(f.__closure__[0], cellvalue, '%s.cell_contents')

def test_code_relation(self):
        def f():
            a = 3
            return self, a
        co = f.__code__
        # xxx brittle test but catches a bug
        self.chkpath(co, 3, '%s.co_consts[1]')
        # commented in notes Sep 27 2004
        relAttr = ('co_code', 'co_consts', 'co_names',
                   'co_filename', 'co_name')
        if self.version_info < (3, 11):
            relAttr += ('co_varnames', 'co_freevars', 'co_cellvars')
        else:
            relAttr += ('co_exceptiontable', 'co_qualname')
        if self.version_info >= (3, 10):
            relAttr += ('co_linetable',)
        else:
            relAttr += ('co_lnotab',)
        self.chkrelattr(co, *relAttr)

def test_dict_relation(self):
        k1 = 'k1'
        k2 = 'k2'
        v1 = 'v1'
        v2 = 'v2'
        k3 = tuple(range(100))
        v3 = tuple(range(100, 200))
        x = {k1: v1, k2: v2, k3: v3}
        self.chkrel(x, v1, "%s['k1']")
        self.chkrel(x, v2, "%s['k2']")
        self.chkrel(x, v3, "%s[(0, 1, 2, 3, 4, 5, ...)]")
        ks = [str(self.relation(x, k1)),
              str(self.relation(x, k2)),
              str(self.relation(x, k3))]
        ks.sort()
        self.aseq(ks, ['%s.keys()[0]', '%s.keys()[1]', '%s.keys()[2]'])

def test_dictproxy_relation(self):
        v1 = 'v1'

class T(object):
            k1 = v1
        x = T.__dict__
        self.chkpath(x, v1, "%s->mapping['k1']")
        self.chkrel(x, v1, "%s['k1']")

def test_frame_relation(self):
        f = inspect.currentframe()
        f.f_trace = lambda: None
        self.chkrelattr(f, 'f_back', 'f_builtins', 'f_code', 'f_globals',
                        'f_locals', 'f_trace')

a = []
        # The representation of local variables is how they may be accessed
        # - not how they are really stored.
        # xxx this may be confusing/lack information?
        # The information is available in the relation object class,
        # it is just not represented with str()...
        self.chkrel(f, a, "%s.f_locals['a']", clas=self.Path.R_LOCAL_VAR)

x = []
        z = []

def func(x, y=3):
            frame = inspect.currentframe()
            return self, frame, z
        _, frame, __ = func(0)
        del _, __
        self.chkrel(frame, self, "%s.f_locals ['self']", clas=self.Path.R_CELL)

self.chkrel(f, x, "%s.f_locals['x']", clas=self.Path.R_LOCAL_VAR)
        self.chkrel(f, z, "%s.f_locals ['z']", clas=self.Path.R_CELL)
        # self becomes both a local var and a cell var, since it is an argument.
        # FIXME: But not in py3... it seems
        # self.chkrel(f, self, "<%s.f_locals['self'],%s.f_locals ['self']>")

# Stack variables doesn't work (Because ceval.c doesn't update
        # the f_stacktop index.) so the corresponding part of frame_relate is not tested.

def test_function_relation(self):
        def f(x, y=3):
            return self
        f.a = []

self.chkrelattr(f, '__code__', '__globals__', '__defaults__',
                        '__closure__', '__doc__', '__name__', '__dict__',
                        'a')

def test_instancemethod_relation(self):
        class T:
            def f(x):
                pass
        t = T()
        self.chkrelattr(t.f, '__func__', '__self__')

def test_list_relation(self):
        v1 = 'v1'
        v2 = 'v2'
        v3 = list(range(100, 200))
        x = [v1, v2, v3]
        self.chkrel(x, v1, '%s[0]')
        self.chkrel(x, v2, '%s[1]')
        self.chkrel(x, v3, '%s[2]')

def test_meth_relation(self):
        x = []
        self.chkrel(x.append, x, '%s.__self__')

def test_module_relation(self):
        self.chkrelattr(unittest, '__dict__', 'TestCase')

def test_nodegraph_relation(self):
        a = 0
        b = 1
        rl = [a, b]
        rg = self.heapy.heapyc.NodeGraph([(a, rl), (b, rl)])
        self.chkrel(rg, a, '%s->edges[0].src')
        self.chkrel(rg, b, '%s->edges[1].src')
        self.chkrel(rg, rl, '<%s->edges[0].tgt,%s->edges[1].tgt>')
        self.chkpath(rg, a, '%s->edges[0].src')
        self.chkpath(rg, rl, ['%s->edges[0].tgt', '%s->edges[1].tgt'])

def test_nodeset_relation(self):
        from guppy.sets import immnodeset, mutnodeset
        x = ['a']
        for s in (immnodeset(x), mutnodeset(x)):
            for i in range(len(x)):
                self.chkrel(s, x[i], 'list(%%s)[%s]' % i)

def test_object_relation(self):
        class T(object):
            __slots__ = 'a', 'b'
        t = T()
        a = []
        t.a = a
        b = []
        t.b = b
        # self.chkrel(t, T, 'type(%s)')
        self.chkrel(t, T, '%s->ob_type')
        self.chkrelattr(t, 'a', 'b')
        # We shouldn't have a __dict__ here - just make sure this is the case
        self.assertRaises(AttributeError, lambda: t.__dict__)

class U(T):
            pass
        u = U()
        u.a = a
        self.chkpath(u, T, "%s->ob_type.__base__")
        self.chkrel(u, a, '%s.a')

c = []
        u.c = c
        self.chkrel(u, c, '%s.c')
        self.chkrel(u, u.__dict__, '%s.__dict__')

class V(U):
            pass
        v = V()
        v.c = c
        self.chkrelattr(v, '__dict__')

class W(V):
            __slots__ = 'c', 'd', 'b'
            pass
        w = W()
        w.a = a
        w.b = b
        w.c = c
        w.d = []
        w.e = []
        self.chkrelattr(w, '__dict__', 'a', 'b', 'c', 'd', 'e')
        self.chkpath(w, w.a, '%s.a')
        self.chkpath(w, w.b, '%s.b')
        self.chkpath(w, w.c, '%s.c')
        self.chkpath(w, w.d, '%s.d')
        self.chkpath(w, w.e, "%s.__dict__['e']")

class R(object):
            rvar = []

class S(R, T):
            svar = []

s = S()
        s.a = a
        s.b = b
        s.c = c
        self.chkrelattr(s, '__dict__', 'a', 'b', 'c')
        self.chkpath(s, s.a, '%s.a')
        self.chkpath(s, s.b, '%s.b')
        self.chkpath(s, s.c, "%s.__dict__['c']")

# Class variables are not directly related- should they be that?
        # Possibly, but the compression could as well be done in Python.
        # We just check that we can get the path.
        self.chkpath(s, s.svar, "%s->ob_type.__dict__['svar']")
        self.chkpath(s, s.rvar, ["%s->ob_type.__bases__[0].__dict__['rvar']",
                                 "%s->ob_type.__mro__[1].__dict__['rvar']"])
        self.chkpath(s, s.__slots__,
                     "%s->ob_type.__base__.__dict__['__slots__']")

def test_traceback_relation(self):
        import sys
        try:
            def g():
                1/0
            g()
        except ZeroDivisionError:
            type, value, traceback = sys.exc_info()
        self.chkrelattr(traceback, 'tb_next', 'tb_frame')

def test_tuple_relation(self):
        v1 = 'v1'
        v2 = 'v2'
        v3 = list(range(100, 200))
        x = (v1, v2, v3)
        self.chkrel(x, v1, '%s[0]')
        self.chkrel(x, v2, '%s[1]')
        self.chkrel(x, v3, '%s[2]')

def test_type_relation(self):
        name = 'T'
        base = object
        bases = (base,)
        dict = {'__slots__': ('a', 'b')}
        T = type(name, bases, dict)
        # tp_dict can't be directly tested since .__dict__ returns a proxy
        # and the dict passed is not used directly.
        # We test it indirectly by getting a path through it.
        self.chkpath(T, T.a, "%s.__dict__['a']")
        # The C-struct __slots__ field can't be tested directly
        # This just tests the ordinary attribute
        self.chkpath(T, T.__slots__, "%s.__dict__['__slots__']")
        self.chkrelattr(T, '__mro__', '__base__', '__bases__')
        # tp_cache and tp_subclasses can also not be tested directly

# Inheritance is tested via test_object_relation()

class RootTestCase(TestCase):
    def test_1(self):
        import sys
        import builtins
        root = self.View.root
        # Interpreter attributes

rel = str(self.relation(root, sys.modules))

self.assertTrue(eval(rel % 'root') is sys.modules)
        self.aseq(rel, '%s.i0_modules')

rel = str(self.relation(root, sys.__dict__))
        self.assertTrue(eval(rel % 'root') is sys.__dict__)
        self.aseq(rel, '%s.i0_sysdict')

rel = str(self.relation(root, builtins.__dict__))
        self.assertTrue(eval(rel % 'root') is builtins.__dict__)
        self.aseq(rel, '%s.i0_builtins')

for name in "codec_search_path", "codec_search_cache", "codec_error_registry":
            attr = "i0_%s" % name
            rel = str(self.relation(root, getattr(root, attr)))
            self.aseq(rel, '%%s.%s' % attr)

# Thread attributes

try:
            1/0
        except ZeroDivisionError:
            exc_type, exc_value, exc_traceback = sys.exc_info()
            if sys.version_info >= (3, 11):
                rel = str(self.relation(root, exc_value))
                self.asis(eval(rel % 'root'), exc_value)
            else:
                for name in 'exc_type', 'exc_value', 'exc_traceback':
                    rel = str(self.relation(root, eval(name)))
                    self.asis(eval(rel % 'root'), eval(name))

# There are more, untested, attributes, but the code is farily regular...
            # More complication is to do with frames which I concentrate on for now.

# We need to find out what level we are at - count to lowest frame
            level = 0
            frame = exc_traceback.tb_frame

while frame.f_back:
                frame = frame.f_back
                level += 1
            rel = str(self.relation(root, frame))
            self.assertTrue(rel.endswith('_f0'))
            rel = str(self.relation(root, exc_traceback.tb_frame))
            self.asis(eval(rel % 'root'), exc_traceback.tb_frame)
            self.assertTrue(rel.endswith('_f%d' % level))

def test_thread(self):
        try:
            import _thread
        except ImportError:
            print('threading not enabled - skipping test')
            return

root = self.View.root

def task(self):
            import sys
            try:
                1/0
            except ZeroDivisionError:
                exc_type, exc_value, exc_traceback = sys.exc_info()
                self.exc_value = exc_value
                self.sync = 1
                while self.sync:
                    pass
                self.sync = 1

self.sync = 0
        _thread.start_new_thread(task, (self,))
        while not self.sync:
            pass
        exc_value = self.exc_value
        rel = str(self.relation(root, exc_value))
        self.asis(eval(rel % 'root'), exc_value)
        self.sync = 0
        while not self.sync:
            pass

def task(self):
            self.test_1()
            self.sync = 1
        self.sync = 0
        _thread.start_new_thread(task, (self,))
        while not self.sync:
            pass

def test_secondary_interpreter(self):
        import sys
        if sys.version_info >= (3, 9):
            print('multi-interpreter not supported past Python 3.9')
            return

try:
            import _thread
        except ImportError:
            print('threading not enabled - skipping test')
            return

import_remote = """\
import sys
import _thread
import time

def task():
    time.sleep(1)

self.sysdict = sys.__dict__
    self.sync = 1

while self.sync:
        pass

_thread.start_new_thread(task, ())
"""

self.sync = 0
        thid = self.heapy.heapyc.interpreter(import_remote, {'self': self})

root = self.View.root

import sys
        sysdict = sys.__dict__
        rel = str(self.relation(root, sysdict))
        self.aseq(rel, '%s.i0_sysdict')

while not self.sync:
            pass

rel = str(self.relation(root, self.sysdict))
        self.aseq(rel, '%s.i1_sysdict')

self.sync = 0

class PathTestCase(TestCase):
    def makegraph(self, width, length):
        # Generate a structure which will yield a high number
        # of shortest paths.
        # Returns a pair src, dst which are connected via a noncyclic graph
        # with many edges.
        # The length of each path (all shortest), number of edges will be length
        # The number of nodes will be 2 + width * (length - 1)
        # The number of paths will be
        #       width ** length, if width >= 1 and length >= 1

dst = []
        ls = []
        for i in range(width):
            ls.append([dst])
        ls = [dst] * width
        for i in range(length-1):
            xs = []
            for j in range(width):
                ys = []
                xs.append(ys)
                for k in range(width):
                    ys.append(ls[k])
            ls = xs
        src = ls
        return src, dst

def chkgraph(self, width, length, expect=None):
        src, dst = self.makegraph(width, length)
        self.chkpath(src, dst, expect)

def test_path(self):
        dst = 'dst'
        self.chkpath([dst], dst, '%s[0]')
        self.chkpath([[], dst], dst, '%s[1]')
        self.chkpath([dst, dst], dst, "['%s[0]', '%s[1]']")
        self.chkpath([[dst, 0], dst, [dst, 2]], dst, "%s[1]")
        self.chkpath([[dst, 0], [dst, 2]], dst, "['%s[0][0]', '%s[1][0]']")

src, dst = self.makegraph(1, 1)

self.chkgraph(1, 1, '%s[0]')
        self.chkgraph(1, 2, '%s[0][0]')
        self.chkgraph(2, 1, ['%s[0]', '%s[1]'])
        self.chkgraph(3, 2, ['%s[0][0]', '%s[0][1]',     '%s[0][2]',
                             '%s[1][0]', '%s[1][1]',     '%s[1][2]',
                             '%s[2][0]', '%s[2][1]',     '%s[2][2]'])

def test_numpaths(self):
        for (width, length) in [(2, 1), (7, 3), (3, 7), (10, 20)]:
            src, dst = self.makegraph(width, length)
            p = self.shpaths(dst, src)
            self.aseq(p.numpaths, width**length)

def test_iter(self):
        src, dst = self.makegraph(2, 2)
        p = self.shpaths(dst, src)
        it = iter(p)
        ss = []
        for i in it:
            ss.append(str(i))
        ss.sort()
        self.aseq(ss, ['%s[0][0]', '%s[0][1]', '%s[1][0]', '%s[1][1]'])

# Check that we can get some of the first values from the iterator
        # of a graph with an astronomical number of paths.

width = 11
        length = 13
        numpaths = 20
        src, dst = self.makegraph(width, length)
        p = self.shpaths(dst, src)
        it = iter(p)
        for i in range(numpaths):
            path = next(it)
            sp = str(path)
            div, mod = divmod(i, width)
            self.aseq(sp, '%s'+'[0]'*(length-2)+'[%d][%d]' % (div, mod))

# Check that the iterator works even if the graph initially
        # would yield astronomical numbers of dead ends.
        # (The initial algorithm took astronomically long time.)

osrc = src

src, dst = self.makegraph(width, length)
        src[0] = osrc
        p = self.shpaths(dst, src)
        it = iter(p)
        for i in range(numpaths):
            path = next(it)
            sp = str(path)
            div, mod = divmod(i, width)
            self.aseq(sp, '%s[1]'+'[0]'*(length-3)+'[%d][%d]' % (div, mod))

# Test iterating with a negative start and a large positive start

numfromend = width / 2
        for it in [p.iter(-numfromend), p.iter(p.numpaths-numfromend)]:
            for i, path in enumerate(it):
                sp = str(path)
                self.aseq(sp, '%s'+('[%d]' % (width-1)) *
                          (length-1)+'[%d]' % (width-numfromend+i))

# Test iterating with start and stop

start = 5
        stop = 25
        i = start
        for path in p.iter(start, stop):
            sp = str(path)
            div, mod = divmod(i, width)
            self.aseq(sp, '%s[1]'+'[0]'*(length-3)+'[%d][%d]' % (div, mod))
            self.aseq(path.index, i)
            i += 1
        self.aseq(i, stop)

def test_str(self):
        # Make sure large number of paths will yield reasonable representations
        width = 11
        length = 4
        src, dst = self.makegraph(width, length)
        p = self.shpaths(dst, src)
        p.maxpaths = 1
        self.aseq(str(p), " 0: Src[0][0][0][0]\n<... 14640 more paths ...>")
        p.maxpaths = 2
        self.aseq(
            str(p), " 0: Src[0][0][0][0]\n 1: Src[0][0][0][1]\n<... 14639 more paths ...>")

def test_printing(self):
        # Test the pretty-printing and moreing methods
        from io import StringIO
        output = StringIO()
        self.Path.output = output
        width = 11
        length = 4
        src, dst = self.makegraph(width, length)
        p = self.shpaths(dst, src)
        p.maxpaths = 2
        self.aseq(str(p), """\
 0: Src[0][0][0][0]
 1: Src[0][0][0][1]
<... 14639 more paths ...>""")
        self.aseq(str(p.more), """\
 2: Src[0][0][0][2]
 3: Src[0][0][0][3]
<... 14637 more paths ...>""")

def test_subscript(self):
        # Test subscripting
        width = 3
        length = 40
        src, dst = self.makegraph(width, length)
        p = self.shpaths(dst, src)
        np = width**length
        self.aseq(np, p.numpaths)
        # p[0].pp(p.output)
        self.aseq(str(p[0]), '%s'+'[0]'*length)
        self.aseq(str(p[-np]), '%s'+'[0]'*length)
        self.aseq(str(p[width-1]), '%s'+'[0]'*(length-1) + '[%d]' % (width-1))
        self.aseq(str(p[width]), '%s'+'[0]'*(length-2) + '[1][0]')
        self.aseq(str(p[width+1]), '%s'+'[0]'*(length-2) + '[1][1]')
        self.aseq(str(p[np-1]), '%s'+('[%d]' % (width-1))*length)
        self.aseq(str(p[-1]), '%s'+('[%d]' % (width-1))*length)
        self.assertRaises(IndexError, lambda: p[np])
        self.assertRaises(IndexError, lambda: p[-np-1])

class MultiTestCase(TestCase):
    def test_pp(self):
        # Test printing of multi relations
        iso = self.iso
        dst = [[], []]
        src = iso(dst[:]*2)
        dst = [iso(x) for x in dst]
        p = self.Path.shpgraph(dst, src)
        self.aseq(str(p), """\
--- Dst[0] ---
 0: Src[0]
 1: Src[2]
--- Dst[1] ---
 0: Src[1]
 1: Src[3]""")
        p = self.Path.shpgraph(dst, src, srcname='A', dstname='B')
        self.aseq(str(p), """\
--- B[0] ---
 0: A[0]
 1: A[2]
--- B[1] ---
 0: A[1]
 1: A[3]""")

class AvoidTestCase(TestCase):
    def test_1(self):
        # Test that we can find new paths by avoiding edges
        # selected from previously found paths.
        # First we generate a graph with paths of various lengths...

src = ['src']
        a = src
        for i in range(3):
            b = ['b%d' % i]
            c = ['c%d' % i, b]
            a.append(b)
            a.append(c)
            a = b
        dst = a
        p = self.shpaths(dst, src)

for avoid, result in [
            ([],        '%s[1][1][1]'),
            ([0],       '%s[2][1][1][1]'),
            ([1],       '%s[1][2][1][1]'),
            ([2],       '%s[1][1][2][1]'),
            ([0, 1],    '%s[2][1][2][1][1]'),
            ([1, 2],    '%s[1][2][1][2][1]'),
            #           ([1, -1],   '%s[1][2][1][2][1]'),
            ([0, 2],    '%s[2][1][1][2][1]'),
            ([0, 1, 2], '%s[2][1][2][1][2][1]'),
            ([2, 1, 0], '%s[2][1][2][1][2][1]'),

]:
            result = result % ' 0: Src'
            # Find new path by avoiding edges from the original path
            q = self.shpaths(dst, src, avoid_edges=p.edges_at(*avoid))
            self.aseq(str(q), result)
            # Find the same path but via a direct method
            q = p.copy_but_avoid_edges_at_levels(*avoid)
            self.aseq(str(q), result)
            # The same, but via a shorter method name
            q = p.avoided(*avoid)
            self.aseq(str(q), result)

# Test that the avoided set is carried on to copies

q = p.avoided(0).avoided(2)
        self.aseq(str(q), ' 0: Src[2][1][2][1][1]')

class NewTestCase(TestCase):
    def test_1(self):
        import sys
        o = self.python.io.StringIO()
        iso = self.iso
        x = iso(sys.__dict__)
        print(x.shpaths, file=o)
        # This used to include a path via parameter avoid_edges
        # which was confusing
        print(x.shpaths.avoided(0), file=o)

# repr() used to be quite useless. I have it now defined as .pp(),
        # but without trailin newline.

print(repr(x.shpaths), file=o)
        print(repr(x.shpaths), file=o)

# The shpaths object could sometimes disturb a shpath calculation
        # because dst was free in it.

x = []
        y = [[[x]]]

sp = iso(x).get_shpaths(iso(y))
        print(sp, file=o)

y.append(sp)
        print(iso(x).get_shpaths(iso(y)), file=o)

# Test that the shortest paths to a set of objects, is the shortest
        # paths to those that can be reached by the shortest paths, only

x = []
        y = [x]
        z = [y]

print(iso(x, y).get_shpaths(iso(z)), file=o)

# Test that we can relate objects that inherits from a class and object
        # (Used to segfault)

class C:
            pass

class O(C, object):
            __slots__ = 'x',

ob = O()
        ob.x = x

print(iso(x).get_shpaths(iso(ob)), file=o)

# Test that generalization to a set of sources makes some sense
        # The shortest paths are from the closest sources

# Hack to make a constant address rendering, for test comparison.
        # This doesn't change anything permanently.
        # XXX come up with an official way to do this.
        summary_str = self.heapy.UniSet.summary_str
        def str_address(x): return '<address>'
        str_address._idpart_header = getattr(
            summary_str.str_address, '_idpart_header', None)
        str_address._idpart_sortrender = getattr(
            summary_str.str_address, '_idpart_sortrender', None)
        summary_str.str_address = str_address

S = iso()
        shp = iso(x).get_shpaths(iso(y, z))
        print(shp, file=o)
        print(repr(shp), file=o)
        for p in shp:
            S = S ^ p.src
        self.aseq(S, iso(y))

shp = iso(x).get_shpaths(iso(ob, y, z))
        print(str(shp), file=o)
        print(repr(shp), file=o)
        S = iso()
        for i, p in enumerate(shp):
            S = S ^ p.src
            self.aseq(p.src, shp[i].src)
        self.aseq(S, iso(ob, y))

# Test that the iter can be restarted
        # even after multiple sources handling was added

it = iter(shp)
        a = list(it)
        it.isatend = 0
        b = list(it)
        self.aseq(str(a), str(b))

# The sort order is based on the source set's byid partition (see
        # Path.PathsIter.reset), which sorts based on the size, then render
        # (see Part.IdentityPartition.__init__)
        if sys.getsizeof(y) > sys.getsizeof(ob):
            self.aseq(o.getvalue(), """\
 0: hpy().Root.i0_sysdict
 0: Src.i0_modules['sys'].__dict__
 0: hpy().Root.i0_sysdict
 0: hpy().Root.i0_sysdict
 0: Src[0][0][0]
 0: Src[0][0][0]
 0: Src[0]
 0: Src.x
 0: <1 list: <address>*1>[0]
 0: <1 list: <address>*1>[0]
 0: <1 list: <address>*1>[0]
 1: <1 __main__.O: <address>>.x
 0: <1 list: <address>*1>[0]
 1: <1 __main__.O: <address>>.x
""".replace('__main__', self.__module__))
        else:
            self.aseq(o.getvalue(), """\
 0: hpy().Root.i0_sysdict
 0: Src.i0_modules['sys'].__dict__
 0: hpy().Root.i0_sysdict
 0: hpy().Root.i0_sysdict
 0: Src[0][0][0]
 0: Src[0][0][0]
 0: Src[0]
 0: Src.x
 0: <1 list: <address>*1>[0]
 0: <1 list: <address>*1>[0]
 0: <1 __main__.O: <address>>.x
 1: <1 list: <address>*1>[0]
 0: <1 __main__.O: <address>>.x
 1: <1 list: <address>*1>[0]
""".replace('__main__', self.__module__))

def test_2(self):
        # To assist interactivity,
        # the more attribute is defined to return an object which
        # the repr() of gives more lines; and has a similar more attribute.
        # Testing this functionality here.

o = self.python.io.StringIO()
        iso = self.iso
        dst = []
        src = [dst]*20
        print(repr(iso(dst).get_shpaths(iso(src))), file=o)
        print(repr(iso(dst).get_shpaths(iso(src)).more), file=o)
        p = iso(dst).get_shpaths(iso(src))
        print(repr(p.more), file=o)

self.aseq(o.getvalue(), """\
 0: Src[0]
 1: Src[1]
 2: Src[2]
 3: Src[3]
 4: Src[4]
 5: Src[5]
 6: Src[6]
 7: Src[7]
 8: Src[8]
 9: Src[9]
<... 10 more paths ...>
10: Src[10]
11: Src[11]
12: Src[12]
13: Src[13]
14: Src[14]
15: Src[15]
16: Src[16]
17: Src[17]
18: Src[18]
19: Src[19]
10: Src[10]
11: Src[11]
12: Src[12]
13: Src[13]
14: Src[14]
15: Src[15]
16: Src[16]
17: Src[17]
18: Src[18]
19: Src[19]
""")

def test_empty(self):
        # Test empty paths
        iso = self.iso
        dst = []
        self.assertTrue(len(list(iso(dst).get_shpaths(iso()))) == 0)

def test_3(self):
        # Test that Edges is not included in the shortest path

iso = self.iso
        dst = []
        shp = iso(dst).shpaths
        del dst
        self.assertTrue('Edges' not in str(shp.avoided(0)))

dst = []

src = [dst]

shp = iso(dst).get_shpaths(iso(src))

src[0] = shp

dst = iso(dst)
        src = iso(src)

self.assertTrue(dst.get_shpaths(src).numpaths == 0)

# Test the sets attribute

dst = []
        src = [dst]
        dst = iso(dst)
        src = iso(src)
        self.aseq(dst.get_shpaths(src).sets, (src, dst))

# Test that srs doesn't disturb the path calculation

class C:
            pass

c = C()
        cd = iso(c.__dict__)

p = cd.shpaths
        repr(p)

del c
        q = cd.shpaths
        self.aseq(repr(q).strip(), "")

del p, q

# Test that internals of ShortestPath are hidden in general
        # (via hiding_tag), to consistent result when used interactively,
        # as commented on in notes.txt per Nov 30 2004.

dst = []
        src = [[[[dst]]]]

d = iso(dst)
        s = iso(src)

p = d.get_shpaths(s)
        self.aseq(str(p), " 0: Src[0][0][0][0]")
        src.append(p)
        p._XX_ = dst    # A shorter path, but it should be hidden
        self.aseq(str(d.get_shpaths(s)), " 0: Src[0][0][0][0]")

# Test what .more prints finally

self.aseq(str(p.more), '<No more paths>')

# Test that .top is idempotent

self.asis(p.more.top.top, p)

def test_4(self):
        # Test that if one extra path will be printed if there are
        # exactly 11 paths

iso = self.iso
        o = self.python.io.StringIO()

dst = []
        src = [dst] * 11

shp = iso(dst).get_shpaths(iso(src))
        print(str(shp), file=o)
        self.aseq(o.getvalue(), """\
 0: Src[0]
 1: Src[1]
 2: Src[2]
 3: Src[3]
 4: Src[4]
 5: Src[5]
 6: Src[6]
 7: Src[7]
 8: Src[8]
 9: Src[9]
10: Src[10]
""")

def test_comparison(self):
        # Test that non-compariable keys won't crash
        # output order may be arbitrary not the output is not tested

iso = self.iso
        dst = []

shp = iso(dst).get_shpaths(iso({0: dst, '': dst}))
        str(shp)

shp = iso(dst).get_shpaths(iso({object(): dst, object(): dst}))
        str(shp)

def run_test(case, debug=0):
    support.run_unittest(case, debug)

def test_main(debug=0):
    run_test(NewTestCase, debug)
    run_test(RelationTestCase, debug)
    run_test(RootTestCase, debug)
    run_test(PathTestCase, debug)
    run_test(MultiTestCase, debug)
    run_test(AvoidTestCase, debug)

if __name__ == "__main__":
    test_main()

CANCEL