Rev 2483: Revert the pyrex implementation to its most basic in http://bzr.arbash-meinel.com/branches/bzr/0.17-dev/dirstate_pyrex

[John Arbash Meinel]

At http://bzr.arbash-meinel.com/branches/bzr/0.17-dev/dirstate_pyrex

------------------------------------------------------------
revno: 2483
revision-id: john at arbash-meinel.com-20070504041902-r5vxd4xpkduhbd0b
parent: john at arbash-meinel.com-20070504041242-lnhinwkv7wvsejg0
committer: John Arbash Meinel <john at arbash-meinel.com>
branch nick: dirstate_pyrex
timestamp: Thu 2007-05-03 23:19:02 -0500
message:
  Revert the pyrex implementation to its most basic
  The fancier ones were causing segfaults.
modified:
  bzrlib/compiled/dirstate_helpers.pyx dirstate_helpers.pyx-20070503201057-u425eni465q4idwn-3
-------------- next part --------------
=== modified file 'bzrlib/compiled/dirstate_helpers.pyx'
--- a/bzrlib/compiled/dirstate_helpers.pyx	2007-05-04 03:58:29 +0000
+++ b/bzrlib/compiled/dirstate_helpers.pyx	2007-05-04 04:19:02 +0000
@@ -22,9 +22,6 @@
 from bzrlib.dirstate import DirState
 
 
-cdef extern from *:
-    ctypedef int size_t
-
 cdef extern from "Python.h":
     # GetItem returns a borrowed reference
     void *PyDict_GetItem(object p, object key)
@@ -32,15 +29,6 @@
     object PyList_GetItem(object lst, int index)
     object PyTuple_GetItem(object tpl, int index)
 
-    char* PyString_AsString(object p)
-    int PyString_Size(object p)
-
-
-cdef extern from "string.h":
-    int strncmp(char *s1, char *s2, size_t len)
-    int strcmp(char *s1, char *s2)
-    char *strchr(char *s1, char c)
-
 
 cdef object _split_from_path(object cache, object path):
     """get the dirblock tuple for a given path.
@@ -57,108 +45,14 @@
     value_ptr = PyDict_GetItem(cache, path)
     if value_ptr == NULL:
         value = path.split('/')
-        PyDict_SetItem(cache, path, value)
+        cache[path] = value
     else:
         value = <object>value_ptr
 
     return value
 
 
-cdef int _bisect_dirblock_nocache(object dirblocks, object dirname, int _lo, int _hi):
-    cdef int _mid
-    cdef object cur
-    cdef object cur_split
-    cdef object dirname_split
-
-    dirname_split = dirname.split('/')
-
-    while _lo < _hi:
-        _mid = (_lo+_hi)/2
-        # Grab the dirname for the current dirblock
-        cur = PyTuple_GetItem(PyList_GetItem(dirblocks, _mid), 0)
-        cur_split = cur.split('/')
-        if cur_split < dirname_split: _lo = _mid+1
-        else: _hi = _mid
-    return _lo
-
-
-cdef int _cmp_dirblock_strings(char *path1, int size1, char *path2, int size2):
-    """This compares 2 strings separating on path sections.
-
-    This is equivalent to "cmp(path1.split('/'), path2.split('/'))"
-    However, we don't want to create an extra object for doing the split.
-
-    :param path1: The first path to compare
-    :param size1: The length of the first path
-    :param path2: The second path
-    :param size1: The length of the second path
-    :return: 0 if they are equal, -1 if path1 comes first, 1 if path2 comes
-        first
-    """
-    cdef char *base1
-    cdef char *base2
-    cdef char *tip1
-    cdef char *tip2
-    cdef char *end1
-    cdef char *end2
-    cdef int cur_len1
-    cdef int cur_len2
-    cdef int cmp_len
-    cdef int diff
-
-    base1 = path1
-    base2 = path2
-    end1 = base1 + size1
-    end2 = base2 + size2
-
-    # Ensure that we are pointing to the final NULL terminator on both ends
-    assert end1[0] == c'\x00'
-    assert end2[0] == c'\x00'
-
-    while base1 < end1 and base2 < end2:
-        # Find the next path separator
-        # (This is where you would like strchrnul)
-        tip1 = strchr(base1, c'/')
-        tip2 = strchr(base2, c'/')
-
-        if tip1 == NULL:
-            tip1 = end1
-        if tip2 == NULL:
-            tip2 = end2
-
-        cur_len1 = tip1 - base1
-        cur_len2 = tip2 - base2
-        cmp_len = cur_len1
-        if cur_len2 < cur_len1:
-            cmp_len = cur_len2
-
-        diff = strncmp(base1, base2, cmp_len)
-        # print 'comparing "%s", "%s", %d = %d' % (base1, base2, cmp_len, diff)
-        if diff != 0:
-            return diff
-        if cur_len1 < cur_len2:
-            return -1
-        elif cur_len1 > cur_len2:
-            return 1
-        base1 = tip1+1
-        base2 = tip2+1
-    # Do we still have uncompared characters?
-    if base1 < end1:
-        return 1
-    if base2 < end2:
-        return -1
-    return 0
-
-
-def cmp_dirblock_strings(path1, path2):
-    """Compare to python strings in dirblock fashion."""
-    return _cmp_dirblock_strings(PyString_AsString(path1),
-                                 PyString_Size(path1),
-                                 PyString_AsString(path2),
-                                 PyString_Size(path2))
-
-
-def bisect_dirblock(dirblocks, dirname, lo=0, hi=None, cache=None):
+def bisect_dirblock(dirblocks, dirname, lo=0, hi=None, cache={}):
     """Return the index where to insert dirname into the dirblocks.
 
     The return value idx is such that all directories blocks in dirblock[:idx]
@@ -180,15 +74,13 @@
         _hi = hi
 
     _lo = lo
-    if cache is None:
-        return _bisect_dirblock_nocache(dirblocks, dirname, _lo, _hi)
-
-    dirname_split = _split_from_path(cache, dirname)
+    dirname_split = dirname.split('/')
     while _lo < _hi:
         _mid = (_lo+_hi)/2
         # Grab the dirname for the current dirblock
-        cur = PyTuple_GetItem(PyList_GetItem(dirblocks, _mid), 0)
-        cur_split = _split_from_path(cache, cur)
+        #cur = PyTuple_GetItem(PyList_GetItem(dirblocks, _mid), 0)
+        cur = dirblocks[_mid][0]
+        cur_split = cur.split('/')
         if cur_split < dirname_split: _lo = _mid+1
         else: _hi = _mid
     return _lo