Edit

[EDIT] FILE: test_RefPat.py

from guppy.heapy.test import support
import io
import sys
import types
import unittest

class TestCase(support.TestCase):
    def setUp(self):
        support.TestCase.setUp(self)
        self.RefPat = self.heapy.RefPat
        self.iso = self.Use.iso

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 rp(self, dst, src, **kwds):
        iso = self.iso
        if src is not None:
            src = iso(src)
        rp = iso(dst).get_rp(src=src, **kwds)
        rp.mod.UniSet.summary_str.str_address = lambda a: '<address>'
        return rp

class RefPatCase(TestCase):
    def test_basic_methods(self):
        # Test basic methods: iteration, indexing, length, tree addressing via attribute access

# Test iteration

dst = src = []
        lists = [dst]
        for i in range(5):
            src = [src]
            lists.append(src)

rp = self.rp(dst, src, depth=10)
        for i, x in enumerate(rp):
            if i < len(lists):
                self.asis(lists[i], x.theone)

# Test indexing

# First case, when already iterated over

self.asis(rp[0].theone, lists[0])
        self.asis(rp[-2].theone, lists[-1])

# Second case, when not iterated over before

rp = self.rp(dst, src, depth=10)

self.asis(rp[0].theone, lists[0])
        self.asis(rp[-2].theone, lists[-1])

# Test length

self.aseq(len(rp), len(lists) + 1)
        rp = self.rp(dst, src, depth=10)
        self.aseq(len(rp), len(lists) + 1)

# Test attribute access

self.asis(rp._.theone, lists[0])
        self.asis(rp.a.theone, lists[1])

# Test attribute access, when not iterated over before

rp = self.rp(dst, src, depth=10)
        self.asis(rp.a2.theone, lists[2])
        self.asis(rp.a.theone, lists[1])

# Make sure attribute access is cached:
        # so it doesn't change when struct is changed

lists[2].append(lists[0])
        rp.View.clear_retainers()
        rp.View.update_referrers(self.iso(lists[0]))
        self.asis(rp.a.theone, lists[1])

# Test with recursive structure

dst = []
        dst.append(dst)
        src = [dst]
        rp = self.rp(dst, src)
        self.asis(rp._.theone, dst)
        self.aseq(rp.a, self.iso(dst, src))
        self.aseq(rp.a, rp.a2)
        self.aseq(rp.a, rp[1])

def test_presentation(self):

output = io.StringIO()

src = []

def write(x):
            print(x, file=output)

R = self.RefPat
        def test_pp(dst, src, result=None, **kwds):

rp = self.rp(dst, src, **kwds)
            write(repr(rp))
            return rp

dst = []
        src.append(dst)

test_pp(dst, src)

for i in range(5):
            x = dst
            dst = []
            x.append(dst)
        test_pp(dst, src)

src, dst = self.makegraph(5, 7)

test_pp(dst, src, depth=10)

# Test that pp() prints limited number of lines

src, dst = self.makegraph(5, 17)

rp = test_pp(dst, src, depth=17)

write(repr(rp.more))

# Test more of more

src, dst = self.makegraph(1, 30)

rp = test_pp(dst, src, depth=35)

m = rp.more

write(repr(m))
        write(repr(m.more))
        m1 = m.more
        write(repr(m1))
        m2 = m.more
        write(repr(m2.more))
        write(str(m1.more))  # Test also that str() is the same as repr()

# Test that we get back to start by .top

write(m1.top)

# Test that we get back to previous by .prev

write(m1.prev)

# Test that they won't say '...more lines...' if the # of lines is what is printed

src, dst = self.makegraph(1, 30)

rp = test_pp(dst, src, depth=10)

# Test how no more lines is printed

write(rp.more)
        write(rp.more.more)

# Test that one more line is printed rather than '1 more line'

src, dst = self.makegraph(1, 30)

rp = test_pp(dst, src, depth=21)
        write(rp.more)

# Test that we can do more without first printing

rp = self.rp(dst, src, depth=20)

write(rp.more)

self.aseq(output.getvalue(), """\
Reference Pattern by <[dict of] class>.
 0: _ --- [-] 1 list: <address>*0
 1: a      [-] 1 list: <address>*1
 2: aa ---- [R] 1 tuple: <address>*1
Reference Pattern by <[dict of] class>.
 0: _ --- [-] 1 list: <address>*0
 1: a      [-] 1 list: <address>*1
 2: aa ---- [-] 1 list: <address>*1
 3: a3       [-] 1 list: <address>*1
 4: a4 ------ [-] 1 list: <address>*1
 5: a5         [-] 1 list: <address>*1
 6: a6 -------- [-] 1 list: <address>*1
 7: a7           [R] 1 tuple: <address>*1
Reference Pattern by <[dict of] class>.
 0: _ --- [-] 1 list: <address>*0
 1: a      [-] 5 list: <address>*5, <address>*5, <address>*5, <address>*5...
 2: aa ---- [-] 5 list: <address>*5, <address>*5, <address>*5, <address>*5...
 3: a3       [-] 5 list: <address>*5, <address>*5, <address>*5, <address>*5...
 4: a4 ------ [-] 5 list: <address>*5, <address>*5, <address>*5, <address>*5...
 5: a5         [-] 5 list: <address>*5, <address>*5, <address>*5, <address>*5...
 6: a6 -------- [-] 5 list: <address>*5, <address>*5, <address>*5...
 7: a7           [-] 1 list: <address>*5
 8: a8 ---------- [R] 1 tuple: <address>*1
Reference Pattern by <[dict of] class>.
 0: _ --- [-] 1 list: <address>*0
 1: a      [-] 5 list: <address>*5, <address>*5, <address>*5, <address>*5...
 2: aa ---- [-] 5 list: <address>*5, <address>*5, <address>*5, <address>*5...
 3: a3       [-] 5 list: <address>*5, <address>*5, <address>*5, <address>*5...
 4: a4 ------ [-] 5 list: <address>*5, <address>*5, <address>*5, <address>*5...
 5: a5         [-] 5 list: <address>*5, <address>*5, <address>*5, <address>*5...
 6: a6 -------- [-] 5 list: <address>*5, <address>*5, <address>*5...
 7: a7           [-] 5 list: <address>*5, <address>*5, <address>*5...
 8: a8 ---------- [-] 5 list: <address>*5, <address>*5, <address>*5...
 9: a9             [-] 5 list: <address>*5, <address>*5, <address>*5...
<Type e.g. '_.more' for more.>
10: a10 ----------- [-] 5 list: <address>*5, <address>*5, <address>*5...
11: a11              [-] 5 list: <address>*5, <address>*5, <address>*5...
12: a12 ------------- [-] 5 list: <address>*5, <address>*5, <address>*5...
13: a13                [-] 5 list: <address>*5, <address>*5, <address>*5...
14: a14 --------------- [-] 5 list: <address>*5, <address>*5, <address>*5...
15: a15                  [-] 5 list: <address>*5, <address>*5, <address>*5...
16: a16 ----------------- [-] 5 list: <address>*5, <address>*5, <address>*5...
17: a17                    [+] 1 list: <address>*5
Reference Pattern by <[dict of] class>.
 0: _ --- [-] 1 list: <address>*0
 1: a      [-] 1 list: <address>*1
 2: aa ---- [-] 1 list: <address>*1
 3: a3       [-] 1 list: <address>*1
 4: a4 ------ [-] 1 list: <address>*1
 5: a5         [-] 1 list: <address>*1
 6: a6 -------- [-] 1 list: <address>*1
 7: a7           [-] 1 list: <address>*1
 8: a8 ---------- [-] 1 list: <address>*1
 9: a9             [-] 1 list: <address>*1
<Type e.g. '_.more' for more.>
10: a10 ----------- [-] 1 list: <address>*1
11: a11              [-] 1 list: <address>*1
12: a12 ------------- [-] 1 list: <address>*1
13: a13                [-] 1 list: <address>*1
14: a14 --------------- [-] 1 list: <address>*1
15: a15                  [-] 1 list: <address>*1
16: a16 ----------------- [-] 1 list: <address>*1
17: a17                    [-] 1 list: <address>*1
18: a18 ------------------- [-] 1 list: <address>*1
19: a19                      [-] 1 list: <address>*1
<Type e.g. '_.more' for more.>
20: a20 --------------------- [-] 1 list: <address>*1
21: a21                        [-] 1 list: <address>*1
22: a22 ----------------------- [-] 1 list: <address>*1
23: a23                          [-] 1 list: <address>*1
24: a24 ------------------------- [-] 1 list: <address>*1
25: a25                            [-] 1 list: <address>*1
26: a26 --------------------------- [-] 1 list: <address>*1
27: a27                              [-] 1 list: <address>*1
28: a28 ----------------------------- [-] 1 list: <address>*1
29: a29                                [-] 1 list: <address>*1
<Type e.g. '_.more' for more.>
20: a20 --------------------- [-] 1 list: <address>*1
21: a21                        [-] 1 list: <address>*1
22: a22 ----------------------- [-] 1 list: <address>*1
23: a23                          [-] 1 list: <address>*1
24: a24 ------------------------- [-] 1 list: <address>*1
25: a25                            [-] 1 list: <address>*1
26: a26 --------------------------- [-] 1 list: <address>*1
27: a27                              [-] 1 list: <address>*1
28: a28 ----------------------------- [-] 1 list: <address>*1
29: a29                                [-] 1 list: <address>*1
<Type e.g. '_.more' for more.>
30: a30 ------------------------------- [-] 1 list: <address>*1
31: a31                                  [R] 1 tuple: <address>*1
30: a30 ------------------------------- [-] 1 list: <address>*1
31: a31                                  [R] 1 tuple: <address>*1
Reference Pattern by <[dict of] class>.
 0: _ --- [-] 1 list: <address>*0
 1: a      [-] 1 list: <address>*1
 2: aa ---- [-] 1 list: <address>*1
 3: a3       [-] 1 list: <address>*1
 4: a4 ------ [-] 1 list: <address>*1
 5: a5         [-] 1 list: <address>*1
 6: a6 -------- [-] 1 list: <address>*1
 7: a7           [-] 1 list: <address>*1
 8: a8 ---------- [-] 1 list: <address>*1
 9: a9             [-] 1 list: <address>*1
<22 more lines. Type e.g. '_.more' for more.>
10: a10 ----------- [-] 1 list: <address>*1
11: a11              [-] 1 list: <address>*1
12: a12 ------------- [-] 1 list: <address>*1
13: a13                [-] 1 list: <address>*1
14: a14 --------------- [-] 1 list: <address>*1
15: a15                  [-] 1 list: <address>*1
16: a16 ----------------- [-] 1 list: <address>*1
17: a17                    [-] 1 list: <address>*1
18: a18 ------------------- [-] 1 list: <address>*1
19: a19                      [-] 1 list: <address>*1
<12 more lines. Type e.g. '_.more' for more.>
Reference Pattern by <[dict of] class>.
 0: _ --- [-] 1 list: <address>*0
 1: a      [-] 1 list: <address>*1
 2: aa ---- [-] 1 list: <address>*1
 3: a3       [-] 1 list: <address>*1
 4: a4 ------ [-] 1 list: <address>*1
 5: a5         [-] 1 list: <address>*1
 6: a6 -------- [-] 1 list: <address>*1
 7: a7           [-] 1 list: <address>*1
 8: a8 ---------- [-] 1 list: <address>*1
 9: a9             [-] 1 list: <address>*1
10: a10 ----------- [+] 1 list: <address>*1

Reference Pattern by <[dict of] class>.
 0: _ --- [-] 1 list: <address>*0
 1: a      [-] 1 list: <address>*1
 2: aa ---- [-] 1 list: <address>*1
 3: a3       [-] 1 list: <address>*1
 4: a4 ------ [-] 1 list: <address>*1
 5: a5         [-] 1 list: <address>*1
 6: a6 -------- [-] 1 list: <address>*1
 7: a7           [-] 1 list: <address>*1
 8: a8 ---------- [-] 1 list: <address>*1
 9: a9             [-] 1 list: <address>*1
<Type e.g. '_.more' for more.>
10: a10 ----------- [-] 1 list: <address>*1
11: a11              [-] 1 list: <address>*1
12: a12 ------------- [-] 1 list: <address>*1
13: a13                [-] 1 list: <address>*1
14: a14 --------------- [-] 1 list: <address>*1
15: a15                  [-] 1 list: <address>*1
16: a16 ----------------- [-] 1 list: <address>*1
17: a17                    [-] 1 list: <address>*1
18: a18 ------------------- [-] 1 list: <address>*1
19: a19                      [-] 1 list: <address>*1
<Type e.g. '_.more' for more.>
10: a10 ----------- [-] 1 list: <address>*1
11: a11              [-] 1 list: <address>*1
12: a12 ------------- [-] 1 list: <address>*1
13: a13                [-] 1 list: <address>*1
14: a14 --------------- [-] 1 list: <address>*1
15: a15                  [-] 1 list: <address>*1
16: a16 ----------------- [-] 1 list: <address>*1
17: a17                    [-] 1 list: <address>*1
18: a18 ------------------- [-] 1 list: <address>*1
19: a19                      [-] 1 list: <address>*1
20: a20 --------------------- [+] 1 list: <address>*1
""")

def test_referrer_registration(self):
        import gc

# The reference pattern should register itself as referrer target
        # so that after a gc, the rp target will still be included in the referrer target
        # Since the target is passed to referrers and update, it will still find the
        # referrers. It is an optimization issue: it should cover the referrers.
        # We test this by having two different-typed referrers
        # Accessing a referrer of the first one, then gc collecting, then checking that
        # the second one can be accessed without update: it was created automatically.
        # The test failed when not registering, but succeeded when registering was added.
        # It succeeds any case if no GC collection is made.

dst = []
        a = [dst]
        aa = [a]
        b = (dst,)
        ba = [b]
        src = [aa, ba]
        rp = self.rp(dst, src)

self.asis(rp._.theone, dst)
        gc.collect()
        self.asis(rp.aa.theone, aa)
        self.asis(rp.View.rg[b][0], ba)

def test_some_more_advanced_usages(self):
        import gc

# Test immediate dominators

dst = []
        src = [dst]
        src.append([dst])
        rp = self.rp(dst, src, depth=10, imdom=1)
        self.asis(rp._.theone, dst)
        self.asis(rp.a.theone, src)

# Test with mixed types

# In particular, dict owned by an instance

dst = []

class A:
            pass
        a = A()
        a.dst = dst
        b = {'dst': dst}

src = (a, b)

gc.collect()
        rp = self.rp(dst, src, depth=10)
        rp.mod.View._is_clear_drg_enabled = 0  # Note Apr 19 2005

if sys.getsizeof(b) > sys.getsizeof(a.__dict__):
            self.asis(rp.a.theone, b)
            self.asis(rp.b.theone, a.__dict__)
        else:
            self.asis(rp.a.theone, a.__dict__)
            self.asis(rp.b.theone, b)

# Test that the dict is eventually automatically removed from dictowners -
        # First test that dictowners is nonzero

ln = len(rp.mod.View.dict_ownership)
        self.assertTrue(ln > 0)

del src
        del a
        mod = rp.mod
        rp.mod.View._is_clear_drg_enabled = 1
        del rp
        # It is cleared after GC

gc.collect()
        lnnow = len(mod.View.dict_ownership)
        self.assertTrue(lnnow == 0)

class NewCase(TestCase):
    # Some new tests as they come up

def test_reset(self):
        # Test the .reset() method

dst = []
        a = [dst]
        b = [dst]
        src = [a, b]
        rp = self.rp(dst, src)
        self.aseq(rp.a, self.iso(a, b))

b.pop()
        rp.reset()
        self.aseq(rp.a, self.iso(a))

def test_paths(self):
        # Test the .paths() method

dst = []
        a = [dst]+[None]*40     # Make order well-defined. Note May 2 2005.
        b = [dst]
        src = [a, b]
        rp = self.rp(dst, src)

expected = """\
Paths from source 'a3' to target '_'.
 0: a3 [0] @ [0]
 1: aa [0]  @ [0]
 2: a  [0]   @ [0]
 3: _  [0]    = <1 list: <address>*0>
 4: aa [0]  @ [1]
 5: a  [1]   @ [0] -> #3"""

self.aseq(str(rp.paths('a3')), expected)

expected = expected[:expected.index('\n 4:')]

# Test the andsets argument, given as a dict

self.aseq(str(rp.paths('a3', andsets={'a': self.iso(a)})), expected)

# Test the andsets argument, given as a list

self.aseq(
            str(rp.paths('a3', andsets=[None, None, self.iso(a)])), expected)

def test_main(debug=0):
    support.run_unittest(RefPatCase, debug)
    support.run_unittest(NewCase, debug)

def test_leak():
    # Runs the tests in a loop and prints memory statistics,
    # to see if there are underlying low-level memory problems.
    # Requires Python to be compiled with debug support.
    from guppy.heapy.heapyc import xmemstats
    import gc
    import sys
    import time
    i = 0
    xmemstats()
    while 1:
        print('[%d]' % i, time.asctime())
        i += 1
        test_main()
        gc.collect()
        xmemstats()

if __name__ == "__main__":
    test_main()

CANCEL