From 4a60b68a2461932c60b066b9cf713f0f73c60dc3 Mon Sep 17 00:00:00 2001
From: Matthew Honnibal <honnibal@gmail.com>
Date: Fri, 3 Jul 2015 05:45:16 +0200
Subject: [PATCH] * Add encode_sense_strs function

---
 spacy/senses.pxd |  6 ++++++
 spacy/senses.pyx | 14 ++++++++++++++
 2 files changed, 20 insertions(+)

diff --git a/spacy/senses.pxd b/spacy/senses.pxd
index 282f5f590..392a92e05 100644
--- a/spacy/senses.pxd
+++ b/spacy/senses.pxd
@@ -59,5 +59,11 @@ cpdef enum:
     N_SENSES
 
 
+
+
 cdef flags_t[<int>parts_of_speech.N_UNIV_TAGS] POS_SENSES
 
+
+# Can we not define this compile time? Have to hard code? :(
+# DEF INVALID_SENSE = 1 << 54
+cdef flags_t encode_sense_strs(sense_names) except 18014398509481984
diff --git a/spacy/senses.pyx b/spacy/senses.pyx
index 99d193774..de6e19300 100644
--- a/spacy/senses.pyx
+++ b/spacy/senses.pyx
@@ -86,3 +86,17 @@ STRINGS = (
     'V_stative',
     'V_weather'
 )
+
+IDS = dict((sense_str, i) for i, sense_str in enumerate(STRINGS))
+
+
+cdef flags_t encode_sense_strs(sense_names) except 18014398509481984:
+    cdef flags_t sense_bits = 0
+    cdef flags_t sense_id = 0
+    for sense_str in sense_names:
+        if '.' in sense_str:
+            sense_str = sense_str[0].upper() + '_' + sense_str.split('.')[1]
+        if sense_str != 'N_Tops':
+            sense_id = IDS[sense_str]
+            sense_bits |= (1 << sense_id)
+    return sense_bits