General Improvements: C extension, Py2 Py3 compatibility, Performance

MANIFEST.txt

@@ -0,0 +1 @@
+python/include *.txt

README.md

@@ -20,7 +20,15 @@ USAGE
 
 ### C ###
 
-    #include <base92/base92.h>
+To build as a shared library in $PREFIX.
+
+    mkdir -p $PREFIX/include/ $PREFIX/lib/
+    gcc -shared -Wl,-soname,libbase92 -o $PREFIX/lib/libbase92.so -fPIC -Ic/src c/src/base92.c
+    cp -a c/src/base92.h $PREFIX/include/
+
+Use:
+
+    #include <base92.h>
     ...
     strcmp(base92encode("hello world", 11), "Fc_$aOTdKnsM*k") == 0;
     base92decode("Fc_$aOTdKnsM*k", &length);
@@ -29,6 +37,12 @@ USAGE
 
 ### Python ###
 
+To build:
+
+    cd python
+    python setup.py build_ext --inplace
+    python setup.py install
+
 Fire up your favorite python:
 
     >>> import base92

c/src/base92.c

@@ -79,9 +79,9 @@ unsigned char* base92encode(unsigned char* str, int len) {
         int tmp;
         unsigned char c;
         unsigned char *res;
-        
+
         if (len == 0) {
-                return "~";
+                return (unsigned char*)"~";
         }
         // precalculate how much space we need to malloc
         size = (len * 8) % 13;
@@ -159,17 +159,13 @@ unsigned char* base92decode(unsigned char* str, int* len) {
         unsigned char* res;
         unsigned long workspace;
         unsigned short wssize;
-        size = strlen(str);
+        size = strlen((char*)str);
         // handle small cases first
-        if (strcmp(str, "~") == 0 || size == 0) {
+        if (strcmp((char*)str, "~") == 0 || size == 0) {
                 res = (unsigned char*)malloc(sizeof(char) * 1);
                 res[0] = 0;
                 return res;
         }
-        // this case does not fit the specs
-        if (size < 2) {
-                res = NULL;
-        }
         // calculate size
         *len = ((size/2 * 13) + (size%2 * 6)) / 8;
         res = (unsigned char *)malloc(sizeof(char) * (*len));

python/README.txt

@@ -6,6 +6,17 @@ A little library for encoding byte-strings into strings easily
 typeable on a standard US 101-key keyboard, with strictly better
 information density than base64 or base85 encodings.
 
+It is python3 compatible and has a C backend.
+
+--------
+BUILDING
+--------
+
+Compile the C extension and install.
+
+    python setup.py build_ext --inplace  # creates base92/base92_extension.so
+    python setup.py install
+
 -----
 USAGE
 -----
@@ -17,6 +28,26 @@ Fire up your favorite python::
     'hello world'
     >>> base92.encode('\x61\xf2\x05\x99\x42')
     'DJ8gER!'
+
+    >>> import base92.test
+    >>> base92.test.run()
+    testing and cross validating encoders and decoders from modules [<module 'base92.cbase92' from 'base92/cbase92.pyc'>, <module 'base92.base92' from 'base92/base92.py'>]
+    selected regression tests passed
+    generating 10000 random byte strings
+    10000 randomized X == decode(encode(X)) tests passed
+    performance of module <module 'base92.cbase92' from 'base92/cbase92.pyc'> on the 10000 random byte strings
+    - encoding: 0.00835490226746s
+    - decoding: 0.00846481323242s
+    performance of module <module 'base92.base92' from 'base92/base92.py'> on the 10000 random byte strings
+    - encoding: 1.75639009476s
+    - decoding: 1.28861784935s
+
+If the C backend is not available, the python backend will be used:
+
+    rm -f base92/base92_extension.so
+
+    >>> import base92
+    Falling back to base92 python backend due to: No module named base92_extension
 
 We use doctests, so running the tests is as easy as executing the
 base92.py library file with your python.
@@ -25,10 +56,7 @@ base92.py library file with your python.
 MISC
 ----
 
-This library is pure python: there may be a cbase92 forthcoming,
-backed by a C library.
-
-This library has not been tested with python3.
+This library has a C extension as a backend and falls back to python if the backend isn't available.
 
 There is more information available at
 <https://github.com/thenoviceoof/base92>

python/base92/__init__.py

@@ -1,5 +1,5 @@
-'''
-Import routines from base92.base92 for manipulating base92 encoded strings.
+"""
+Import routines from base92.cbase92 or base92.base92 for manipulating base92 encoded strings.
 
 Example:
 
@@ -9,6 +9,18 @@
 'Fc_$aOTdKnsM*k'
 >>> decode(x)
 'hello world'
-'''
+"""
 
-from base92 import encode, decode, b92encode, b92decode, __version__
+from . import base92
+
+try:
+    from . import cbase92
+    preferred_base92 = cbase92
+except (ImportError, OSError) as e:
+    print('Falling back to base92 python backend due to: {}'.format(e))
+    preferred_base92 = base92
+    cbase92 = None
+
+encode = b92encode = preferred_base92.encode
+decode = b92decode = preferred_base92.decode
+__version__ = base92.__version__