more tests and tweak _xdelta3.c

Author: samuelcolvin

.gitignore

@@ -1,5 +1,6 @@
 .idea/
 env/
+env35/
 *.py[cod]
 *.egg-info/
 build/

README.rst

@@ -8,7 +8,7 @@ Fast delta encoding in python using xdelta3.
 Requirements
 ------------
 
-* **Python 3.5 or 3.6** - it's 2017: you should be using python 3.6 by now anyway.
+* **Python 3.5 or 3.6** - it's 2017, you should be using python 3.6 by now anyway.
 * **linux** - compilation only tested on ubuntu, might work on other platform.
 
 Installation
@@ -38,8 +38,8 @@ How fast?
 
 *xdelta3-python* is a thin wrapper around `xdelta 3.1.1 <https://github.com/jmacd/xdelta/>`_
 which is a highly optimised c library for delta calculation and compression.
-It can encode a delta and decode it again for 5 small changes in a 5.4M character string
-(the complete works of shakespeare) in around 30ms. Boom.
+It can encode a delta and decode it again for 5 small changes in a 5.5 million character string
+(the complete works of shakespeare) in around 10ms (or 30ms with the highest compression level). Boom.
 See `performance.py <https://github.com/samuelcolvin/xdelta3-python/blob/master/performance.py>`_.
 
 .. |BuildStatus| image:: https://travis-ci.org/samuelcolvin/xdelta3-python.svg?branch=master

performance.py

@@ -17,9 +17,9 @@
 v2 = Path('shakespeare_changed.txt').read_bytes()
 
 times = []
-for i in range(10):
+for i in range(50):
     start = time()
-    delta = xdelta3.encode(v1, v2)
+    delta = xdelta3.encode(v1, v2, xdelta3.Flags.COMPLEVEL_1)
     v22 = xdelta3.decode(v1, delta)
     time_taken = (time() - start) * 1000
     times.append(time_taken)

tests/test_main.py

@@ -1,20 +1,35 @@
+import base64
+import os
+import string
+
 import pytest
 
 import xdelta3
 
 value_one = b'this is a short string to test with. It is suitable for delta encoding.'
 value_two = b'this is a different short string to test with. It is suitable for delta encoding.'
+expected_delta = b'\xd6\xc3\xc4\x00\x00\x01G\x00\x14Q\x00\t\x04\x02different\x1a\n\x13>\x00\t'
 
 
 def test_encode_decode():
     delta = xdelta3.encode(value_one, value_two)
+    assert delta == expected_delta
     value_two2 = xdelta3.decode(value_one, delta)
     assert value_two == value_two2
 
 
-def test_encode():
-    result = xdelta3.encode(value_one, value_two)
-    assert result == b'\xd6\xc3\xc4\x00\x00\x01G\x00\x14Q\x00\t\x04\x02different\x1a\n\x13>\x00\t'
+def test_long_random():
+    v1 = base64.b32encode(os.urandom(1000))
+    v2 = b'x' + v1 + b'x'
+    delta = xdelta3.encode(v1, v2)
+    v22 = xdelta3.decode(v1, delta)
+    assert v2 == v22
+
+
+def test_decode_error():
+    with pytest.raises(xdelta3.XDeltaError) as exc_info:
+        xdelta3.decode(expected_delta, value_one)
+    assert exc_info.value.args[0] == 'Error occur executing xdelta3: XD3_INVALID_INPUT'
 
 
 def test_no_delta():
@@ -23,6 +38,20 @@ def test_no_delta():
     assert exc_info.value.args[0] == 'No delta found shorter than the input value'
 
 
+def test_different_compression():
+    all_ascii = (string.ascii_letters + string.digits).encode()
+    v1 = all_ascii * 1000
+    v2 = all_ascii * 900 + string.ascii_letters.encode() * 100 + all_ascii * 100
+    delta_a = xdelta3.encode(v1, v2, xdelta3.Flags.COMPLEVEL_9)
+    v2_a = xdelta3.decode(v1, delta_a)
+    assert v2 == v2_a
+
+    delta_b = xdelta3.encode(v1, v2, xdelta3.Flags.COMPLEVEL_1)
+    v2_b = xdelta3.decode(v1, delta_b)
+    assert v2 == v2_b
+    assert len(delta_a) < len(delta_b)
+
+
 def test_readme():
     value_one = b'wonderful string to demonstrate xdelta3, much of these two strings is the same.'
     value_two = b'different string to demonstrate xdelta3, much of these two strings is the same.'

xdelta3/_xdelta3.c

@@ -19,7 +19,6 @@ static PyObject * xdelta3_execute(PyObject *self, PyObject *args)
   source_size = (size_t)source_len;
   input_size = (size_t)input_len;
 
-
   if (action == 0) {
     // if the output would be longer than the input itself, there's no point using delta encoding
     output_alloc = input_size;
@@ -59,30 +58,30 @@ static PyObject * xdelta3_execute(PyObject *self, PyObject *args)
 
 static PyMethodDef xdelta3_methods[] = {
   {"execute",  xdelta3_execute, METH_VARARGS, "xdelta3 encode or decode"},
-  {NULL, NULL, 0, NULL}    /* Sentinel */
+  {NULL, NULL, 0, NULL}
 };
 
 static struct PyModuleDef xdelta3_module = {
-   PyModuleDef_HEAD_INIT,
-   "xdelta3",   /* name of module */
-   "xdelta3", /* module documentation, may be NULL */
-   0,     /* size of per-interpreter state of the module, or -1 if the module keeps state in global variables. */
-   xdelta3_methods
+  PyModuleDef_HEAD_INIT,
+  "_xdelta3",
+  NULL,
+  0,
+  xdelta3_methods
 };
 
 PyMODINIT_FUNC PyInit__xdelta3(void) {
-    PyObject *m;
+  PyObject *m;
 
-    m = PyModule_Create(&xdelta3_module);
-    if (m == NULL)
-        return NULL;
+  m = PyModule_Create(&xdelta3_module);
+  if (m == NULL)
+    return NULL;
 
-    XDeltaError = PyErr_NewException("xdelta3.XDeltaError", NULL, NULL);
-    Py_INCREF(XDeltaError);
-    PyModule_AddObject(m, "XDeltaError", XDeltaError);
+  XDeltaError = PyErr_NewException("xdelta3.XDeltaError", NULL, NULL);
+  Py_INCREF(XDeltaError);
+  PyModule_AddObject(m, "XDeltaError", XDeltaError);
 
-    NoDeltaFound = PyErr_NewException("xdelta3.NoDeltaFound", NULL, NULL);
-    Py_INCREF(NoDeltaFound);
-    PyModule_AddObject(m, "NoDeltaFound", NoDeltaFound);
-    return m;
+  NoDeltaFound = PyErr_NewException("xdelta3.NoDeltaFound", NULL, NULL);
+  Py_INCREF(NoDeltaFound);
+  PyModule_AddObject(m, "NoDeltaFound", NoDeltaFound);
+  return m;
 }

xdelta3/version.py

@@ -2,4 +2,4 @@
 
 __all__ = ['VERSION']
 
-VERSION = StrictVersion('0.0.4')
+VERSION = StrictVersion('0.0.5')