Python 3 support

.gitignore

@@ -34,3 +34,4 @@ nosetests.xml
 .mr.developer.cfg
 .project
 .pydevproject
+.idea/

MANIFEST.in

@@ -1,2 +1,2 @@
 include LICENSE
-include README.rst
+include README.md

README.md

@@ -0,0 +1,16 @@
+# chembl_ikey
+
+This is pure python implementation of InChiKey generation algorithm based on the original C code.
+
+This is a improved version of the chembl_ikey package developed at Chembl group, EMBL-EBI, Cambridge, UK.
+
+The original version is only compatible with python2, and this version could run with python2 or python3. Some bugs are also fixed.
+
+## How to use
+
+1. Download and unzip (or clone) the package
+2. Install the package by using `pip install setup.py`
+3. Import the interface by using `from chembl_ikey import inchi_to_inchikey`
+4. Convert inchi to inchikey by calling `inchi_to_inchikey(inchi)`
+
+A sample of how to use the code could be found in chembl_ikey/test.py 

chembl_ikey/__init__.py

@@ -1 +1,3 @@
-__author__ = 'mnowotka'
+__author__ = 'mnowotka, liwt31'
+
+from .ikey import inchi_to_inchikey

chembl_ikey/ikey.py

@@ -1,29 +1,22 @@
-__author__ = 'mnowotka'
-
 import hashlib
-from chembl_ikey import ikey_base26
-from ikey_base26 import base26_triplet_1
-from ikey_base26 import base26_triplet_2
-from ikey_base26 import base26_triplet_3
-from ikey_base26 import base26_triplet_4
-from ikey_base26 import base26_dublet_for_bits_56_to_64
-from ikey_base26 import base26_dublet_for_bits_28_to_36
+from .ikey_base26 import base26_triplet_1, base26_triplet_2, base26_triplet_3, base26_triplet_4, \
+                         base26_dublet_for_bits_56_to_64, base26_dublet_for_bits_28_to_36
 
-#-----------------------------------------------------------------------------------------------------------------------
+__author__ = 'mnowotka, liwt31'
 
 INCHI_STRING_PREFIX = "InChI="
 LEN_INCHI_STRING_PREFIX = len(INCHI_STRING_PREFIX)
 
-#-----------------------------------------------------------------------------------------------------------------------
 
 def get_sha256(text):
     hash = hashlib.sha256()
-    hash.update(text)
-    return hash.digest()
+    hash.update(text.encode('ascii'))
+    digest = hash.digest()
+    digest_bytes_list = [ord(digest_byte) for digest_byte in digest] if isinstance(digest, str) else list(digest)
+    return digest_bytes_list
 
-#-----------------------------------------------------------------------------------------------------------------------
 
-def inchiToInchiKey(szINCHISource):
+def inchi_to_inchikey(szINCHISource):
 
     flagstd = 'S'
     flagnonstd = 'N'
@@ -79,9 +72,11 @@ def inchiToInchiKey(szINCHISource):
                 return None
             end = idx
             break
+    else:
+        end = slen
 
-    if end == (slen - 1):
-        end += 1
+    #if end == (slen - 1):
+    #    end += 1
 
     if not proto:
         smajor = aux[:end]
@@ -118,9 +113,4 @@ def inchiToInchiKey(szINCHISource):
                                 base26_triplet_4(digest_major) + base26_dublet_for_bits_56_to_64(digest_major)
     minor = base26_triplet_1(digest_minor) + base26_triplet_2(digest_minor) + \
                                 base26_dublet_for_bits_28_to_36(digest_minor)
-    return "%s-%s%s%s-%s" % (major, minor, flag, flagver, flagproto)
-
-
-#-----------------------------------------------------------------------------------------------------------------------
-
-
+    return "%s-%s%s%s-%s" % (major, minor, flag, flagver, flagproto)

chembl_ikey/ikey_base26.py

@@ -1,64 +1,53 @@
-__author__ = 'mnowotka'
-
 from itertools import product
 from string import ascii_uppercase
 
-#-----------------------------------------------------------------------------------------------------------------------
+__author__ = 'mnowotka, liwt31'
 
-t26 = [ t for t in map(''.join, product(ascii_uppercase, repeat=3)) if t[0]!='E' and (t < 'TAA' or t > 'TTV')]
-d26 = map(''.join, product(ascii_uppercase, repeat=2))
+t26 = [t for t in map(''.join, product(ascii_uppercase, repeat=3)) if t[0] != 'E' and (t < 'TAA' or t > 'TTV')]
+d26 = list(map(''.join, product(ascii_uppercase, repeat=2)))
 
-#-----------------------------------------------------------------------------------------------------------------------
 
 def base26_triplet_1(a):
-    b0 = ord(a[0])
-    b1 = ord(a[1]) & 0x3f
+    b0 = a[0]
+    b1 = a[1] & 0x3f
     h = b0 | b1 << 8
     return t26[h]
 
-#-----------------------------------------------------------------------------------------------------------------------
 
 def base26_triplet_2(a):
-    b0 = ord(a[1]) & 0xc0
-    b1 = ord(a[2])
-    b2 = ord(a[3]) & 0x0f
-    h =  (b0 | b1 << 8 | b2 << 16) >> 6
+    b0 = a[1] & 0xc0
+    b1 = a[2]
+    b2 = a[3] & 0x0f
+    h = (b0 | b1 << 8 | b2 << 16) >> 6
     return t26[h]
 
-#-----------------------------------------------------------------------------------------------------------------------
 
 def base26_triplet_3(a):
-    b0 = ord(a[3]) & 0xf0
-    b1 = ord(a[4])
-    b2 = ord(a[5]) & 0x03
-    h =  (b0 | b1 << 8  | b2 << 16) >> 4
+    b0 = a[3] & 0xf0
+    b1 = a[4]
+    b2 = a[5] & 0x03
+    h = (b0 | b1 << 8 | b2 << 16) >> 4
     return t26[h]
 
-#-----------------------------------------------------------------------------------------------------------------------
 
 def base26_triplet_4(a):
-    b0 = ord(a[5]) & 0xfc
-    b1 = ord(a[6])
-    h =  (b0 | b1 << 8) >> 2
+    b0 = a[5] & 0xfc
+    b1 = a[6]
+    h = (b0 | b1 << 8) >> 2
     return t26[h]
 
-#-----------------------------------------------------------------------------------------------------------------------
 
 def base26_dublet_for_bits_28_to_36(a):
-    b0 = ord(a[3]) & 0xf0
-    b1 = ord(a[4]) & 0x1f
-    h =  (b0 | b1 << 8) >> 4
+    b0 = a[3] & 0xf0
+    b1 = a[4] & 0x1f
+    h = (b0 | b1 << 8) >> 4
     return d26[h]
 
-#-----------------------------------------------------------------------------------------------------------------------
 
 def base26_dublet_for_bits_56_to_64(a):
-    b0 = ord(a[7])
-    b1 = ord(a[8]) & 0x01
-    h =  b0 | b1 << 8
+    b0 = a[7]
+    b1 = a[8] & 0x01
+    h = b0 | b1 << 8
     return d26[h]
 
-#-----------------------------------------------------------------------------------------------------------------------
-
-
 

chembl_ikey/tests.py

@@ -1,11 +1,11 @@
-__author__ = 'mnowotka'
-
 import unittest
-from chembl_ikey.ikey import inchiToInchiKey
+from chembl_ikey import inchi_to_inchikey
+
+__author__ = 'mnowotka, weitangli'
 
 class TestIKey(unittest.TestCase):
     def test_morphineInChIKey(self):
-        key = inchiToInchiKey("InChI=1S/C17H19NO3/c1-18-7-6-17-10-3-5-13(20)16(17)21-15-12(19)4-2-9(14(15)17)8-11(10)18/h2-5,10-11,13,16,19-20H,6-8H2,1H3/t10-,11+,13-,16-,17-/m0/s1")
+        key = inchi_to_inchikey("InChI=1S/C17H19NO3/c1-18-7-6-17-10-3-5-13(20)16(17)21-15-12(19)4-2-9(14(15)17)8-11(10)18/h2-5,10-11,13,16,19-20H,6-8H2,1H3/t10-,11+,13-,16-,17-/m0/s1")
         self.assertEqual(key,'BQJCRHHNABKAKU-KBQPJGBKSA-N')
 
 

setup.py

@@ -12,14 +12,14 @@
 
 setup(
     name='chembl_ikey',
-    version='0.0.1',
-    author='Michal Nowotka',
-    author_email='[email protected]',
+    version='0.0.2',
+    author='Michal Nowotka, Weitang Li',
+    author_email='[email protected], [email protected]',
     description='Pure python implementation of InChiKey generation algorithm based on the original C code',
     url='https://www.ebi.ac.uk/chembl/',
     license='CC BY-SA 3.0',
     packages=['chembl_ikey'],
-    long_description=open('README.rst').read(),
+    long_description=open('README.md').read(),
     include_package_data=False,
     classifiers=['Development Status :: 2 - Pre-Alpha',
                  'Intended Audience :: Developers',