Support for sparse/dense vectors in the python extension

- SparseVector, DenseVector
- improved Lattice
- Lattice translation
- Hypergraph reweighting, pruning

1 files changed, 56 insertions, 0 deletions

diff --git a/python/src/lattice.pxi b/python/src/lattice.pxi
new file mode 100644
index 00000000..493c6dcd
--- /dev/null
+++ b/python/src/lattice.pxi
@@ -0,0 +1,56 @@
+cimport lattice
+
+cdef class Lattice:
+    cdef lattice.Lattice* lattice
+
+    def __init__(self, inp):
+        if isinstance(inp, tuple):
+            self.lattice = new lattice.Lattice(len(inp))
+            for i, arcs in enumerate(inp):
+                self[i] = arcs
+        else:
+            if isinstance(inp, unicode):
+                inp = inp.encode('utf8')
+            if not isinstance(inp, str):
+                raise TypeError('Cannot create lattice from %s' % type(inp))
+            self.lattice = new lattice.Lattice()
+            lattice.ConvertTextToLattice(string(<char *>inp), self.lattice)
+
+    def __getitem__(self, int index):
+        if not 0 <= index < len(self):
+            raise IndexError('lattice index out of range')
+        arcs = []
+        cdef vector[lattice.LatticeArc] arc_vector = self.lattice[0][index]
+        cdef lattice.LatticeArc* arc
+        cdef str label
+        cdef unsigned i
+        for i in range(arc_vector.size()):
+            arc = &arc_vector[i]
+            label = TDConvert(arc.label)
+            arcs.append((label.decode('utf8'), arc.cost, arc.dist2next))
+        return tuple(arcs)
+
+    def __setitem__(self, int index, tuple arcs):
+        if not 0 <= index < len(self):
+            raise IndexError('lattice index out of range')
+        cdef lattice.LatticeArc* arc
+        for (label, cost, dist2next) in arcs:
+            if isinstance(label, unicode):
+                label = label.encode('utf8')
+            arc = new lattice.LatticeArc(TDConvert(<char *>label), cost, dist2next)
+            self.lattice[0][index].push_back(arc[0])
+            del arc
+
+    def __len__(self):
+        return self.lattice.size()
+
+    def __str__(self):
+        return hypergraph.AsPLF(self.lattice[0], True).c_str()
+
+    def __iter__(self):
+        cdef unsigned i
+        for i in range(len(self)):
+            yield self[i]
+
+    def __dealloc__(self):
+        del self.lattice