Make input buffer configurable and find optimal size

Author: rhpvorderman

src/isal/igzip.py

@@ -167,8 +167,10 @@ def __init__(self, filename=None, mode=None,
                                                   isal_zlib.DEF_MEM_LEVEL,
                                                   0)
         if self.mode == READ:
-            raw = _IGzipReader(self.fileobj)
-            self._buffer = io.BufferedReader(raw, 64 * 1024)
+            # Having a large input buffer seems to work well for both normal
+            # gzip and BGZIP files.
+            raw = _IGzipReader(self.fileobj, 512 * 1024)
+            self._buffer = io.BufferedReader(raw)
 
     def __repr__(self):
         s = repr(self.fileobj)

src/isal/isal_zlib.pyi

@@ -66,4 +66,4 @@ def compressobj(level: int = ISAL_DEFAULT_COMPRESSION,
 def decompressobj(wbits: int = MAX_WBITS, zdict = None) -> Decompress: ...
 
 class _IGzipReader(io.RawIOBase):
-    def __init__(self, fp: typing.BinaryIO): ...
+    def __init__(self, fp: typing.BinaryIO, buffersize: int = 32 * 1024): ...

src/isal/isal_zlibmodule.c

@@ -1273,10 +1273,11 @@ static PyObject *
 IGzipReader__new__(PyTypeObject *type, PyObject *args, PyObject *kwargs)
 {
     PyObject *fp;
-    static char *keywords[] = {"fp", NULL};
-    static char *format = "O:IGzipReader";
+    Py_ssize_t buffer_size = 32 * 1024;
+    static char *keywords[] = {"fp", "buffersize", NULL};
+    static char *format = "O|n:IGzipReader";
     if (!PyArg_ParseTupleAndKeywords(
-            args, kwargs, format, keywords, &fp)) {
+            args, kwargs, format, keywords, &fp, &buffer_size)) {
         return NULL;
     }
     IGzipReader *self = PyObject_New(IGzipReader, type);
@@ -1296,7 +1297,7 @@ IGzipReader__new__(PyTypeObject *type, PyObject *args, PyObject *kwargs)
         PyErr_SetString(PyExc_MemoryError, "Unable to allocate lock");
         return NULL;
     }
-    self->buffer_size = 32 * 1024;
+    self->buffer_size = buffer_size;
     self->input_buffer = PyMem_Malloc(self->buffer_size);
     if (self->input_buffer == NULL) {
         Py_DECREF(self);