corrected bz2 algorithm & added to doctests

chrislit · Oct 22, 2018 · 5fcdb58 · 5fcdb58
1 parent 474bc8a
commit 5fcdb58
Show file tree

Hide file tree

Showing 2 changed files with 69 additions and 9 deletions.
diff --git a/README.rst b/README.rst
@@ -76,7 +76,7 @@ Abydos
      :target: https://app.fossa.io/projects/git%2Bgithub.com%2Fchrislit%2Fabydos?ref=badge_shield
      :alt: FOSSA Status
 
-.. |pylint| image:: https://img.shields.io/badge/Pylint-9.45/10-yellowgreen.svg
+.. |pylint| image:: https://img.shields.io/badge/Pylint-9.44/10-yellowgreen.svg
    :target: #
    :alt: Pylint Score
 

diff --git a/abydos/distance/compression.py b/abydos/distance/compression.py
@@ -53,8 +53,14 @@ def dist_ncd_arith(src, tar, probs=None):
     :returns: compression distance
     :rtype: float
 
+    >>> dist_ncd_arith('cat', 'hat')
+    0.5454545454545454
     >>> dist_ncd_arith('Niall', 'Neil')
     0.6875
+    >>> dist_ncd_arith('aluminum', 'Catalan')
+    0.8275862068965517
+    >>> dist_ncd_arith('ATCG', 'TAGC')
+    0.6923076923076923
     """
     if src == tar:
         return 0.0
@@ -83,8 +89,14 @@ def sim_ncd_arith(src, tar, probs=None):
     :returns: compression similarity
     :rtype: float
 
+    >>> sim_ncd_arith('cat', 'hat')
+    0.4545454545454546
     >>> sim_ncd_arith('Niall', 'Neil')
     0.3125
+    >>> sim_ncd_arith('aluminum', 'Catalan')
+    0.1724137931034483
+    >>> sim_ncd_arith('ATCG', 'TAGC')
+    0.3076923076923077
     """
     return 1-dist_ncd_arith(src, tar, probs)
 
@@ -101,8 +113,14 @@ def dist_ncd_rle(src, tar, use_bwt=False):
     :returns: compression distance
     :rtype: float
 
+    >>> dist_ncd_rle('cat', 'hat')
+    1.0
     >>> dist_ncd_rle('Niall', 'Neil')
     1.0
+    >>> dist_ncd_rle('aluminum', 'Catalan')
+    1.0
+    >>> dist_ncd_rle('ATCG', 'TAGC')
+    1.0
     """
     if src == tar:
         return 0.0
@@ -129,8 +147,14 @@ def sim_ncd_rle(src, tar, use_bwt=False):
     :returns: compression similarity
     :rtype: float
 
+    >>> sim_ncd_rle('cat', 'hat')
+    0.0
     >>> sim_ncd_rle('Niall', 'Neil')
     0.0
+    >>> sim_ncd_rle('aluminum', 'Catalan')
+    0.0
+    >>> sim_ncd_rle('ATCG', 'TAGC')
+    0.0
     """
     return 1 - dist_ncd_rle(src, tar, use_bwt)
 
@@ -145,8 +169,14 @@ def dist_ncd_bwtrle(src, tar):
     :returns: compression distance
     :rtype: float
 
+    >>> dist_ncd_bwtrle('cat', 'hat')
+    0.75
     >>> dist_ncd_bwtrle('Niall', 'Neil')
     0.8333333333333334
+    >>> dist_ncd_bwtrle('aluminum', 'Catalan')
+    1.0
+    >>> dist_ncd_bwtrle('ATCG', 'TAGC')
+    0.8
     """
     return dist_ncd_rle(src, tar, True)
 
@@ -161,8 +191,14 @@ def sim_ncd_bwtrle(src, tar):
     :returns: compression similarity
     :rtype: float
 
+    >>> sim_ncd_bwtrle('cat', 'hat')
+    0.25
     >>> sim_ncd_bwtrle('Niall', 'Neil')
     0.16666666666666663
+    >>> sim_ncd_bwtrle('aluminum', 'Catalan')
+    0.0
+    >>> sim_ncd_bwtrle('ATCG', 'TAGC')
+    0.19999999999999996
     """
     return 1 - dist_ncd_bwtrle(src, tar)
 
@@ -177,8 +213,14 @@ def dist_ncd_zlib(src, tar):
     :returns: compression distance
     :rtype: float
 
+    >>> dist_ncd_zlib('cat', 'hat')
+    0.3333333333333333
     >>> dist_ncd_zlib('Niall', 'Neil')
     0.45454545454545453
+    >>> dist_ncd_zlib('aluminum', 'Catalan')
+    0.5714285714285714
+    >>> dist_ncd_zlib('ATCG', 'TAGC')
+    0.4
     """
     if src == tar:
         return 0.0
@@ -206,8 +248,14 @@ def sim_ncd_zlib(src, tar):
     :returns: compression similarity
     :rtype: float
 
+    >>> sim_ncd_zlib('cat', 'hat')
+    0.6666666666666667
     >>> sim_ncd_zlib('Niall', 'Neil')
     0.5454545454545454
+    >>> sim_ncd_zlib('aluminum', 'Catalan')
+    0.4285714285714286
+    >>> sim_ncd_zlib('ATCG', 'TAGC')
+    0.6
     """
     return 1 - dist_ncd_zlib(src, tar)
 
@@ -224,23 +272,23 @@ def dist_ncd_bz2(src, tar):
 
     >>> dist_ncd_bz2('cat', 'hat')
     0.08
+    >>> dist_ncd_bz2('Niall', 'Neil')
+    0.037037037037037035
     >>> dist_ncd_bz2('aluminum', 'Catalan')
     0.20689655172413793
     >>> dist_ncd_bz2('ATCG', 'TAGC')
     0.037037037037037035
-    >>> dist_ncd_bz2('Niall', 'Neil')
-    0.037037037037037035
     """
     if src == tar:
         return 0.0
 
     src = src.encode('utf-8')
     tar = tar.encode('utf-8')
 
-    src_comp = encode(src, 'bz2_codec')[2:]
-    tar_comp = encode(tar, 'bz2_codec')[2:]
-    concat_comp = encode(src + tar, 'bz2_codec')[2:]
-    concat_comp2 = encode(tar + src, 'bz2_codec')[2:]
+    src_comp = encode(src, 'bz2_codec')[15:]
+    tar_comp = encode(tar, 'bz2_codec')[15:]
+    concat_comp = encode(src + tar, 'bz2_codec')[15:]
+    concat_comp2 = encode(tar + src, 'bz2_codec')[15:]
 
     return ((min(len(concat_comp), len(concat_comp2)) -
              min(len(src_comp), len(tar_comp))) /
@@ -259,12 +307,12 @@ def sim_ncd_bz2(src, tar):
 
     >>> sim_ncd_bz2('cat', 'hat')
     0.92
+    >>> sim_ncd_bz2('Niall', 'Neil')
+    0.962962962962963
     >>> sim_ncd_bz2('aluminum', 'Catalan')
     0.7931034482758621
     >>> sim_ncd_bz2('ATCG', 'TAGC')
     0.962962962962963
-    >>> sim_ncd_bz2('Niall', 'Neil')
-    0.962962962962963
     """
     return 1 - dist_ncd_bz2(src, tar)
 
@@ -279,8 +327,14 @@ def dist_ncd_lzma(src, tar):
     :returns: compression distance
     :rtype: float
 
+    >>> dist_ncd_lzma('cat', 'hat')
+    0.08695652173913043
     >>> dist_ncd_lzma('Niall', 'Neil')
     0.16
+    >>> dist_ncd_lzma('aluminum', 'Catalan')
+    0.16
+    >>> dist_ncd_lzma('ATCG', 'TAGC')
+    0.08695652173913043
     """
     if src == tar:
         return 0.0
@@ -312,8 +366,14 @@ def sim_ncd_lzma(src, tar):
     :returns: compression similarity
     :rtype: float
 
+    >>> sim_ncd_lzma('cat', 'hat')
+    0.9130434782608696
     >>> sim_ncd_lzma('Niall', 'Neil')
     0.84
+    >>> sim_ncd_lzma('aluminum', 'Catalan')
+    0.84
+    >>> sim_ncd_lzma('ATCG', 'TAGC')
+    0.9130434782608696
     """
     return 1 - dist_ncd_lzma(src, tar)