[cdec.sa] Explicit feature names in grammar extractor output

+ sparse features in extractor
+ hg.intersect(string)
+ basestring = str|unicode

1 files changed, 31 insertions, 3 deletions

diff --git a/python/src/sa/_sa.pxd b/python/src/sa/_sa.pxd
index d390bfc5..f1cd8e29 100644
--- a/python/src/sa/_sa.pxd
+++ b/python/src/sa/_sa.pxd
@@ -1,3 +1,31 @@
+from libc.stdio cimport FILE
+
+cdef class FloatList:
+    cdef int size
+    cdef int increment
+    cdef int len
+    cdef float* arr
+    cdef void set(self, int i, float v)
+    cdef void write_handle(self, FILE* f)
+    cdef void read_handle(self, FILE* f)
+
+cdef class IntList:
+    cdef int size
+    cdef int increment
+    cdef int len
+    cdef int* arr
+    cdef void set(self, int i, int val)
+    cdef void _append(self, int val)
+    cdef void _extend(self, IntList other)
+    cdef void _extend_arr(self, int* other, int other_len)
+    cdef void _clear(self)
+    cdef void write_handle(self, FILE* f)
+    cdef void read_handle(self, FILE* f)
+
+cdef class FeatureVector:
+    cdef IntList names
+    cdef FloatList values
+
 cdef class Phrase:
     cdef int *syms
     cdef int n, *varpos, n_vars
@@ -5,11 +33,11 @@ cdef class Phrase:
     cdef public int chunklen(self, int k)
 
 cdef class Rule:
-    cdef public int lhs
+    cdef int lhs
     cdef readonly Phrase f, e
-    cdef float *cscores
+    cdef FeatureVector scores
     cdef int n_scores
-    cdef public word_alignments
+    cdef word_alignments
 
 cdef char* sym_tostring(int sym)
 cdef char* sym_tocat(int sym)