SF patch 580331 by Oren Tirosh: make file objects their own iterator.
For a file f, iter(f) now returns f (unless f is closed), and f.next()
is similar to f.readline() when EOF is not reached; however, f.next()
uses a readahead buffer that messes up the file position, so mixing
f.next() and f.readline() (or other methods) doesn't work right.
Calling f.seek() drops the readahead buffer, but other operations
don't.
The real purpose of this change is to reduce the confusion between
objects and their iterators. By making a file its own iterator, it's
made clearer that using the iterator modifies the file object's state
(in particular the current position).
A nice side effect is that this speeds up "for line in f:" by not
having to use the xreadlines module. The f.xreadlines() method is
still supported for backwards compatibility, though it is the same as
iter(f) now.
(I made some cosmetic changes to Oren's code, and added a test for
"file closed" to file_iternext() and file_iter().)
diff --git a/Objects/fileobject.c b/Objects/fileobject.c
index 0b9bab8..726621a 100644
--- a/Objects/fileobject.c
+++ b/Objects/fileobject.c
@@ -1,4 +1,3 @@
-
/* File object implementation */
#include "Python.h"
@@ -116,6 +115,7 @@
f->f_close = close;
f->f_softspace = 0;
f->f_binary = strchr(mode,'b') != NULL;
+ f->f_buf = NULL;
#ifdef WITH_UNIVERSAL_NEWLINES
f->f_univ_newline = (strchr(mode, 'U') != NULL);
f->f_newlinetypes = NEWLINE_UNKNOWN;
@@ -271,6 +271,8 @@
return NULL;
}
+void drop_readahead(PyFileObject *);
+
/* Methods */
static void
@@ -283,6 +285,7 @@
}
Py_XDECREF(f->f_name);
Py_XDECREF(f->f_mode);
+ drop_readahead(f);
f->ob_type->tp_free((PyObject *)f);
}
@@ -405,6 +408,7 @@
if (f->f_fp == NULL)
return err_closed();
+ drop_readahead(f);
whence = 0;
if (!PyArg_ParseTuple(args, "O|i:seek", &offobj, &whence))
return NULL;
@@ -1178,28 +1182,6 @@
}
static PyObject *
-file_xreadlines(PyFileObject *f)
-{
- static PyObject* xreadlines_function = NULL;
-
- if (f->f_fp == NULL)
- return err_closed();
- if (!xreadlines_function) {
- PyObject *xreadlines_module =
- PyImport_ImportModule("xreadlines");
- if(!xreadlines_module)
- return NULL;
-
- xreadlines_function = PyObject_GetAttrString(xreadlines_module,
- "xreadlines");
- Py_DECREF(xreadlines_module);
- if(!xreadlines_function)
- return NULL;
- }
- return PyObject_CallFunction(xreadlines_function, "(O)", f);
-}
-
-static PyObject *
file_readlines(PyFileObject *f, PyObject *args)
{
long sizehint = 0;
@@ -1462,6 +1444,15 @@
#undef CHUNKSIZE
}
+static PyObject *
+file_getiter(PyFileObject *f)
+{
+ if (f->f_fp == NULL)
+ return err_closed();
+ Py_INCREF(f);
+ return (PyObject *)f;
+}
+
PyDoc_STRVAR(readline_doc,
"readline([size]) -> next line from the file, as a string.\n"
"\n"
@@ -1517,10 +1508,10 @@
"total number of bytes in the lines returned.");
PyDoc_STRVAR(xreadlines_doc,
-"xreadlines() -> next line from the file, as a string.\n"
+"xreadlines() -> returns self.\n"
"\n"
-"Equivalent to xreadlines.xreadlines(file). This is like readline(), but\n"
-"often quicker, due to reading ahead internally.");
+"For backward compatibility. File objects now include the performance\n"
+"optimizations previously implemented in the xreadlines module.");
PyDoc_STRVAR(writelines_doc,
"writelines(sequence_of_strings) -> None. Write the strings to the file.\n"
@@ -1554,7 +1545,7 @@
{"tell", (PyCFunction)file_tell, METH_NOARGS, tell_doc},
{"readinto", (PyCFunction)file_readinto, METH_VARARGS, readinto_doc},
{"readlines", (PyCFunction)file_readlines, METH_VARARGS, readlines_doc},
- {"xreadlines", (PyCFunction)file_xreadlines, METH_NOARGS, xreadlines_doc},
+ {"xreadlines", (PyCFunction)file_getiter, METH_NOARGS, xreadlines_doc},
{"writelines", (PyCFunction)file_writelines, METH_O, writelines_doc},
{"flush", (PyCFunction)file_flush, METH_NOARGS, flush_doc},
{"close", (PyCFunction)file_close, METH_NOARGS, close_doc},
@@ -1617,12 +1608,120 @@
{0},
};
-static PyObject *
-file_getiter(PyObject *f)
+void
+drop_readahead(PyFileObject *f)
{
- return PyObject_CallMethod(f, "xreadlines", "");
+ if (f->f_buf != NULL) {
+ PyMem_Free(f->f_buf);
+ f->f_buf = NULL;
+ }
}
+/* Make sure that file has a readahead buffer with at least one byte
+ (unless at EOF) and no more than bufsize. Returns negative value on
+ error */
+int readahead(PyFileObject *f, int bufsize) {
+ int chunksize;
+
+ if (f->f_buf != NULL) {
+ if( (f->f_bufend - f->f_bufptr) >= 1)
+ return 0;
+ else
+ drop_readahead(f);
+ }
+ if ((f->f_buf = PyMem_Malloc(bufsize)) == NULL) {
+ return -1;
+ }
+ Py_BEGIN_ALLOW_THREADS
+ errno = 0;
+ chunksize = Py_UniversalNewlineFread(
+ f->f_buf, bufsize, f->f_fp, (PyObject *)f);
+ Py_END_ALLOW_THREADS
+ if (chunksize == 0) {
+ if (ferror(f->f_fp)) {
+ PyErr_SetFromErrno(PyExc_IOError);
+ clearerr(f->f_fp);
+ drop_readahead(f);
+ return -1;
+ }
+ }
+ f->f_bufptr = f->f_buf;
+ f->f_bufend = f->f_buf + chunksize;
+ return 0;
+}
+
+/* Used by file_iternext. The returned string will start with 'skip'
+ uninitialized bytes followed by the remainder of the line. Don't be
+ horrified by the recursive call: maximum recursion depth is limited by
+ logarithmic buffer growth to about 50 even when reading a 1gb line. */
+
+PyStringObject *
+readahead_get_line_skip(PyFileObject *f, int skip, int bufsize) {
+ PyStringObject* s;
+ char *bufptr;
+ char *buf;
+ int len;
+
+ if (f->f_buf == NULL)
+ if (readahead(f, bufsize) < 0)
+ return NULL;
+
+ len = f->f_bufend - f->f_bufptr;
+ if (len == 0)
+ return (PyStringObject *)
+ PyString_FromStringAndSize(NULL, skip);
+ bufptr = memchr(f->f_bufptr, '\n', len);
+ if (bufptr != NULL) {
+ bufptr++; /* Count the '\n' */
+ len = bufptr - f->f_bufptr;
+ s = (PyStringObject *)
+ PyString_FromStringAndSize(NULL, skip+len);
+ if (s == NULL)
+ return NULL;
+ memcpy(PyString_AS_STRING(s)+skip, f->f_bufptr, len);
+ f->f_bufptr = bufptr;
+ if (bufptr == f->f_bufend)
+ drop_readahead(f);
+ } else {
+ bufptr = f->f_bufptr;
+ buf = f->f_buf;
+ f->f_buf = NULL; /* Force new readahead buffer */
+ s = readahead_get_line_skip(
+ f, skip+len, bufsize + (bufsize>>2) );
+ if (s == NULL) {
+ PyMem_Free(buf);
+ return NULL;
+ }
+ memcpy(PyString_AS_STRING(s)+skip, bufptr, len);
+ PyMem_Free(buf);
+ }
+ return s;
+}
+
+/* A larger buffer size may actually decrease performance. */
+#define READAHEAD_BUFSIZE 8192
+
+static PyObject *
+file_iternext(PyFileObject *f)
+{
+ PyStringObject* l;
+
+ int i;
+
+ if (f->f_fp == NULL)
+ return err_closed();
+
+ i = f->f_softspace;
+
+ l = readahead_get_line_skip(f, 0, READAHEAD_BUFSIZE);
+ if (l == NULL || PyString_GET_SIZE(l) == 0) {
+ Py_XDECREF(l);
+ return NULL;
+ }
+ return (PyObject *)l;
+}
+
+
static PyObject *
file_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
{
@@ -1742,8 +1841,8 @@
0, /* tp_clear */
0, /* tp_richcompare */
0, /* tp_weaklistoffset */
- file_getiter, /* tp_iter */
- 0, /* tp_iternext */
+ (getiterfunc)file_getiter, /* tp_iter */
+ (iternextfunc)file_iternext, /* tp_iternext */
file_methods, /* tp_methods */
file_memberlist, /* tp_members */
file_getsetlist, /* tp_getset */