From 77856c4fcd8510581df36f6db8ccbdf409e29272 Mon Sep 17 00:00:00 2001
From: Matthew Honnibal <honnibal@gmail.com>
Date: Thu, 17 Sep 2015 11:50:11 +1000
Subject: [PATCH] * Try giving Doc and Span objects vector and vector_norm
 attributes, and .similarity functions. Turns out to be bad idea.

---
 spacy/tokens/doc.pxd   |  3 +++
 spacy/tokens/doc.pyx   | 24 +++++++++++++++++++-----
 spacy/tokens/spans.pxd |  3 +++
 spacy/tokens/spans.pyx | 26 +++++++++++++++++++++++---
 4 files changed, 48 insertions(+), 8 deletions(-)

diff --git a/spacy/tokens/doc.pxd b/spacy/tokens/doc.pxd
index a13858175..ce1cfecc0 100644
--- a/spacy/tokens/doc.pxd
+++ b/spacy/tokens/doc.pxd
@@ -23,6 +23,9 @@ cdef class Doc:
     cdef readonly Pool mem
     cdef readonly Vocab vocab
 
+    cdef public object _vector
+    cdef public object _vector_norm
+
     cdef TokenC* data
 
     cdef public bint is_tagged
diff --git a/spacy/tokens/doc.pyx b/spacy/tokens/doc.pyx
index 0c3c0a2f7..5bdd5b22f 100644
--- a/spacy/tokens/doc.pyx
+++ b/spacy/tokens/doc.pyx
@@ -6,6 +6,7 @@ import numpy
 import numpy.linalg
 import struct
 cimport numpy as np
+import math
 
 from ..lexeme cimport Lexeme
 from ..lexeme cimport EMPTY_LEXEME
@@ -77,6 +78,7 @@ cdef class Doc:
         self.is_tagged = False
         self.is_parsed = False
         self._py_tokens = []
+        self._vector = None
 
     def __getitem__(self, object i):
         """Get a token.
@@ -133,12 +135,25 @@ cdef class Doc:
 
     property vector:
         def __get__(self):
-            return sum(t.vector for t in self if not t.is_stop) / len(self)
+            if self._vector is None:
+                self._vector = sum(t.vector for t in self) / len(self)
+            return self._vector
 
+        def __set__(self, value):
+            self._vector = value
 
     property vector_norm:
         def __get__(self):
-            return numpy.linalg.norm(self.vector)
+            cdef float value
+            if self._vector_norm is None:
+                self._vector_norm = 1e-20
+                for value in self.vector:
+                    self._vector_norm += value * value
+                self._vector_norm = math.sqrt(self._vector_norm)
+            return self._vector_norm
+        
+        def __set__(self, value):
+            self._vector_norm = value 
 
     @property
     def string(self):
@@ -304,15 +319,14 @@ cdef class Doc:
         cdef size_t count
         
         if counts is None:
-            counts = PreshCounter(self.length)
+            counts = PreshCounter()
             output_dict = True
         else:
             output_dict = False
         # Take this check out of the loop, for a bit of extra speed
         if exclude is None:
             for i in range(self.length):
-                attr = get_token_attr(&self.data[i], attr_id)
-                counts.inc(attr, 1)
+                counts.inc(get_token_attr(&self.data[i], attr_id), 1)
         else:
             for i in range(self.length):
                 if not exclude(self[i]):
diff --git a/spacy/tokens/spans.pxd b/spacy/tokens/spans.pxd
index d9704ad1f..8e9ef69ef 100644
--- a/spacy/tokens/spans.pxd
+++ b/spacy/tokens/spans.pxd
@@ -7,3 +7,6 @@ cdef class Span:
     cdef public int start
     cdef public int end
     cdef readonly int label
+
+    cdef public _vector
+    cdef public _vector_norm
diff --git a/spacy/tokens/spans.pyx b/spacy/tokens/spans.pyx
index 12ad6e425..88a72982f 100644
--- a/spacy/tokens/spans.pyx
+++ b/spacy/tokens/spans.pyx
@@ -3,6 +3,7 @@ from collections import defaultdict
 import numpy
 import numpy.linalg
 cimport numpy as np
+import math
 
 from ..structs cimport TokenC, LexemeC
 from ..typedefs cimport flags_t, attr_t
@@ -21,6 +22,8 @@ cdef class Span:
         self.start = start
         self.end = end
         self.label = label
+        self._vector = None
+        self._vector_norm = None
 
     def __richcmp__(self, Span other, int op):
         # Eq
@@ -60,15 +63,32 @@ cdef class Span:
 
     property vector:
         def __get__(self):
-            return sum(t.vector for t in self if not t.is_stop) / len(self)
+            if self._vector is None:
+                self._vector = sum(t.vector for t in self) / len(self)
+            return self._vector
+
+        def __set__(self, value):
+            self._vector = value
 
     property vector_norm:
         def __get__(self):
-            return numpy.linalg.norm(self.vector)
+            cdef float value
+            if self._vector_norm is None:
+                self._vector_norm = 1e-20
+                for value in self.vector:
+                    self._vector_norm += value * value
+                self._vector_norm = math.sqrt(self._vector_norm)
+            return self._vector_norm
+    
+        def __set__(self, value):
+            self._vector_norm = value
 
     property text:
         def __get__(self):
-            return u' '.join([t.text for t in self])
+            text = self.text_with_ws
+            if self[-1].whitespace_:
+                text = text[:-1]
+            return text
 
     property text_with_ws:
         def __get__(self):