Edit

[EDIT] FILE: Path.py

import functools

from guppy.etc.Descriptor import property_nondata

class R_NORELATION:
    code = -1
    r = None

def stra(self, a, safe=True):
        return '%s.??' % a

class R_IDENTITY:
    code = 0

def stra(self, a, safe=True):
        return a

class R_ATTRIBUTE:
    code = 1
    strpat = '%s.%s'

class R_INDEXVAL:
    code = 2

def stra(self, a, safe=True):
        if safe:
            return '%s[%s]' % (a, self.saferepr(self.r))
        else:
            return '%s[%r]' % (a, self.r)

class R_INDEXKEY:
    code = 3
    strpat = '%s.keys()[%r]'

class R_INTERATTR:
    code = 4
    strpat = '%s->%s'

class R_HASATTR:
    code = 5
    strpat = '%s.__dict__.keys()[%r]'

class R_LOCAL_VAR:
    code = 6
    strpat = '%s.f_locals[%r]'

class R_CELL:
    code = 7
    strpat = '%s.f_locals [%r]'

class R_STACK:
    code = 8
    strpat = '%s->f_valuestack[%d]'

class R_INSET:
    code = 9
    strpat = 'list(%s)[%d]'

class R_RELSRC:
    code = 10

def stra(self, a, safe=True):
        return self.r % (a,)

class R_LIMIT:
    code = 11

@functools.total_ordering
class RelationBase(object):
    __slots__ = 'r', 'isinverted'

def __init__(self, r, isinverted=0):
        self.r = r
        self.isinverted = isinverted

def __lt__(self, other):
        if not isinstance(other, RelationBase):
            return id(type(self)) < id(type(other))
        if self.code != other.code:
            return self.code < other.code
        if type(self.r) is not type(other.r):
            return id(type(self.r)) < id(type(other.r))
        try:
            return self.r < other.r
        except TypeError:
            return id(self.r) < id(other.r)

def __eq__(self, other):
        if isinstance(other, RelationBase):
            if self.code != other.code:
                return False
            return self.r == other.r
        else:
            return False

def __str__(self):
        return self.stra('%s')

def inverted(self):
        return self.__class__(self.r, not self.isinverted)

def stra(self, a, safe=True):
        return self.strpat % (a, self.r)

class MultiRelation(RelationBase):
    def __init__(self, rels):
        self.rels = rels

def stra(self, a, safe=True):
        return '<'+','.join([x.stra(a, safe=safe) for x in self.rels])+'>'

@functools.total_ordering
class Path:
    def __init__(self, mod, path, index, srcname):
        self.mod = mod
        self.path = path[1:]
        self.index = index
        self.src = path[1]
        self.tgt = path[-1]
        self.strprefix = '%s'
        if srcname == '_str_of_src_':
            srcname = self.src.brief
        if callable(srcname):
            srcname = srcname(self)
        self.srcname = srcname

def __lt__(self, other):
        return str(self) < str(other)

def __eq__(self, other):
        return str(self) == str(other)

def __len__(self):
        return int((len(self.path) - 1) / 2)

def stra(self, safe=True):
        if self.path:
            s = self.strprefix
            for i in range(1, len(self.path), 2):
                r = self.path[i]
                s = r.stra(s, safe=safe)
        else:
            s = '<Empty Path>'
        return s

def __str__(self):
        return self.stra()

def __repr__(self):
        return repr(str(self))

def _get_line_iter(self):
        yield '%2d: %s' % (self.index, str(self) % self.srcname)

def types(self):
        return [type(x) for x in self.path]

class PathsIter:
    def __init__(self, paths, start=None, stop=None):
        self.paths = paths
        self.mod = paths.mod
        self.stop = stop
        self.reset(start)

def __iter__(self):
        return self

def reset(self, idx=None):
        if idx is None:
            idx = 0
        if idx != 0:  # Optimization: don't calculate numpaths in common case.
            ln = self.paths.numpaths
            if idx < 0:
                idx = ln + idx
            if not (0 <= idx < ln):
                self.isatend = 1
                return
        Src = self.paths.Src
        sr = [('%s', src.by(Src.er)) for src in Src.byid.parts]
        srs = []
        idxs = []
        np = 0
        while sr:
            if idx == 0:
                i, (rel, src) = 0, sr[0]
            else:
                for i, (rel, src) in enumerate(sr):
                    npnext = np + self.paths.numpaths_from(src)
                    if idx < npnext:
                        break
                    np = npnext
                else:
                    assert 0
            idxs.append(i)
            srs.append(sr)
            sr = self.mod.sortedrels(self.paths.IG, src)
        self.pos = idx
        self.idxs = idxs
        self.srs = srs
        self.isatend = not idxs

def __next__(self):
        paths = self.paths
        if (self.isatend or
                self.stop is not None and self.pos >= self.stop):
            raise StopIteration
        path = []
        for row, col in enumerate(self.idxs):
            sr = self.srs[row]
            if sr is None:
                sr = self.mod.sortedrels(paths.IG, path[-1])
                self.srs[row] = sr
            rel, dst = sr[col]
            path.append(rel)
            path.append(dst)
        rp = self.mod.Path(paths, path, self.pos, paths.srcname)
        self.pos += 1
        while row >= 0:
            self.idxs[row] += 1
            if self.idxs[row] < len(self.srs[row]):
                break
            if row > 0:
                self.srs[row] = None
            self.idxs[row] = 0
            row -= 1
        else:
            self.isatend = 1
            self.pos = 0
        return rp

class ShortestPaths:
    def __init__(self, sg, Dst):
        self.sg = sg
        self.Dst = Dst
        self.mod = mod = sg.mod
        self._hiding_tag_ = mod._hiding_tag_
        self.srcname = sg.srcname
        self.top = self

self.IG = IG = mod.nodegraph()
        Edges = []
        Y = Dst.nodes
        while Y:
            R = sg.G.domain_restricted(Y)
            R.invert()
            IG.update(R)
            Edges.append(R)
            Y = R.get_domain()
        if Edges:
            Edges.pop()
            Edges.reverse()
            self.Src = mod.idset(Edges[0].get_domain())
        else:
            self.Src = mod.iso()

self.edges = tuple(Edges)
        sets = []
        for i, e in enumerate(Edges):
            if i == 0:
                sets.append(mod.idset(e.get_domain()))
            sets.append(mod.idset(e.get_range()))
        self.sets = tuple(sets)

mod.OutputHandling.setup_printing(self)

self.maxpaths = 10

def __getitem__(self, idx):
        try:
            return next(self.iter(start=idx))
        except StopIteration:
            raise IndexError

def __iter__(self):
        return self.iter()

def iter(self, start=0, stop=None):
        return PathsIter(self, start, stop)

def aslist(self):
        return list(self)

def copy_but_avoid_edges_at_levels(self, *args):
        avoid = self.edges_at(*args).updated(self.sg.AvoidEdges)
        assert avoid._hiding_tag_ is self.mod._hiding_tag_
        return self.mod.shpaths(self.Dst, self.Src, avoid_edges=avoid)
        # return self.mod.shpaths(self.dst, self.src, avoid_edges=avoid)

avoided = copy_but_avoid_edges_at_levels

# The builtin __len__ doesn't always work due to builtin Python restriction to int result:
    # so we don't provide it at all to avoid unsuspected errors sometimes.
    # Use .numpaths attribute instead.
    #   def __len__(self):
    #      return self.numpaths

def depth(self):
        pass

def edges_at(self, *args):
        E = self.mod.nodegraph()
        for col in args:
            E.update(self.edges[col])
        assert E._hiding_tag_ == self.mod._parent.View._hiding_tag_
        return E

def numpaths_from(self, Src):
        try:
            NP = self.NP
        except AttributeError:
            NP = self.mod.nodegraph(is_mapping=True)
            NP.add_edges_n1(self.IG.get_domain(), None)
            for dst in self.Dst.nodes:
                NP.add_edge(dst, 1)
            self.NP = NP
        numedges = self.mod.hv.numedges
        IG = self.IG

def np(y):
            n = NP[y]
            if n is None:
                n = 0
                for z in IG[y]:
                    sn = NP[z]
                    if sn is None:
                        sn = np(z)
                    n += sn * numedges(y, z)
                NP[y] = n
            return n
        num = 0
        for src in Src.nodes:
            num += np(src)
        return num

def _get_numpaths(self):
        num = self.numpaths_from(self.Src)
        self.numpaths = num
        return num

numpaths = property_nondata(fget=_get_numpaths)

@property
    def maxpaths(self):
        return self.printer.max_more_lines

@maxpaths.setter
    def maxpaths(self, value):
        self.printer.max_more_lines = value

def _oh_get_num_lines(self):
        return self.numpaths

def _oh_get_line_iter(self):
        for el in self:
            yield from el._get_line_iter()

def _oh_get_more_msg(self, start_lineno, end_lineno):
        nummore = self.numpaths-(end_lineno+1)
        return '<... %d more paths ...>' % nummore

def _oh_get_empty_msg(self):
        if self.numpaths:
            return '<No more paths>'
        return None

class ShortestGraph:
    def __init__(self, mod, G, DstSets, Src, AvoidEdges,
                 srcname=None, dstname=None):
        self.mod = mod
        self.G = G
        self.Src = Src
        self.DstSets = DstSets
        self.AvoidEdges = AvoidEdges
        if srcname is None:
            if Src.count == 1:
                srcname = mod.srcname_1
            else:
                srcname = mod.srcname_n
        self.srcname = srcname
        if dstname is None:
            dstname = mod.dstname
        self.dstname = dstname

def __getitem__(self, idx):
        return self.mod.ShortestPaths(self, self.DstSets[idx])

def __len__(self):
        return len(self.DstSets)

def __repr__(self):
        lst = []
        for i, p in enumerate(self):
            lst.append('--- %s[%d] ---' % (self.dstname, i))
            lst.append(str(p))

return '\n'.join(lst)

class _GLUECLAMP_:
    _preload_ = ('_hiding_tag_',)
    _chgable_ = ('output', 'srcname_1', 'srcname_n')

srcname_1 = 'Src'
    srcname_n = '_str_of_src_'
    dstname = 'Dst'

_imports_ = (
        '_parent.ImpSet:mutnodeset',
        '_parent:OutputHandling',
        '_parent.Use:idset',
        '_parent.Use:iso',
        '_parent.Use:Nothing',
        '_parent.Use:reprefix',
        '_parent.UniSet:idset_adapt',
        '_parent.View:hv',
        '_parent.View:nodegraph',
        '_parent:View',  # NOT View.root, since it may change
    )

def _get_rel_table(self):
        table = {}
        for name in dir(self._module):
            if name.startswith('R_'):
                c = getattr(self, name)

class r(c, self.RelationBase):
                    repr = self.saferepr
                    saferepr = self.saferepr
                r.__qualname__ = r.__name__ = 'Based_'+name
                table[c.code] = r
        return table

def _get__hiding_tag_(self): return self._parent.View._hiding_tag_
    def _get_identity(self): return self.rel_table[R_IDENTITY.code]('')
    def _get_norelation(self): return self.rel_table[R_NORELATION.code]('')
    def _get_saferepr(self): return self._root.reprlib.repr
    def _get_shpathstep(self): return self.hv.shpathstep

def sortedrels(self, IG, Src):
        t = []
        iso = self.iso
        for src in Src.nodes:
            for dst in IG[src]:
                Dst = iso(dst)
                for rel in self.relations(src, dst):
                    t.append((rel, Dst))
        t.sort(key=lambda x: x[0])
        return t

def prunedinverted(self, G, Y):
        IG = self.nodegraph()
        while Y:
            R = G.domain_restricted(Y)
            R.invert()
            IG.update(R)
            Y = R.get_domain()
        return IG

def relation(self, src, dst):
        tab = self.relations(src, dst)
        if len(tab) > 1:
            r = MultiRelation(tab)
        elif not tab:
            r = self.norelation
        else:
            r = tab[0]
        return r

def relations(self, src, dst):
        tab = []
        if src is dst:
            tab.append(self.identity)
        rawrel = self.hv.relate(src, dst)
        for i, rs in enumerate(rawrel):
            for r in rs:
                tab.append(self.rel_table[i](r))
        if not tab:
            tab = [self.norelation]
        return tab

def shpaths(self, dst, src=None, avoid_nodes=None, avoid_edges=()):
        return self.shpgraph([dst], src, avoid_nodes, avoid_edges)[0]

def shpgraph(self, DstSets, src=None, avoid_nodes=None, avoid_edges=(),
                 srcname=None, dstname=None):
        if src is None:
            Src = self.iso(self.View.root)
            if srcname is None and self.View.root is self.View.heapyc.RootState:
                srcname = '%sRoot' % self.reprefix
        else:
            Src = self.idset_adapt(src)
        if avoid_nodes is None:
            AvoidNodes = self.Nothing
        else:
            AvoidNodes = self.idset_adapt(avoid_nodes)
        AvoidEdges = self.nodegraph(avoid_edges)
        G, DstSets = self.shpgraph_algorithm(
            DstSets, Src, AvoidNodes, AvoidEdges)

return self.ShortestGraph(self, G, DstSets, Src, AvoidEdges,
                                  srcname, dstname)

def shpgraph_algorithm(self, DstSets, Src, AvoidNodes, AvoidEdges):
        U = (Src - AvoidNodes).nodes
        S = self.mutnodeset(AvoidNodes.nodes)
        G = self.nodegraph()
        unseen = list(enumerate(DstSets))
        DstSets = [self.Nothing]*len(DstSets)
        while U and unseen:
            S |= U
            U = self.shpathstep(G, U, S, AvoidEdges)
            unseen_ = []
            for i, D in unseen:
                D_ = D & U
                if D_:
                    DstSets[i] = D_
                else:
                    unseen_.append((i, D))
            unseen = unseen_
        return G, [self.idset_adapt(D) for D in DstSets]

class _Specification_:
    class GlueTypeExpr:
        exec("""\
shpgraph        <in>    callable
""".replace('<in>', ' = lambda IN : '))

CANCEL