1
# Copyright (C) 2008, 2009 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
18
"""Pyrex extensions to btree node parsing."""
21
cdef extern from "python-compat.h":
24
cdef extern from "stdlib.h":
25
ctypedef unsigned size_t
27
cdef extern from "Python.h":
28
ctypedef int Py_ssize_t # Required for older pyrex versions
29
ctypedef struct PyObject:
31
int PyList_Append(object lst, object item) except -1
33
char *PyString_AsString(object p) except NULL
34
object PyString_FromStringAndSize(char *, Py_ssize_t)
35
PyObject *PyString_FromStringAndSize_ptr "PyString_FromStringAndSize" (char *, Py_ssize_t)
36
int PyString_CheckExact(object s)
37
int PyString_CheckExact_ptr "PyString_CheckExact" (PyObject *)
38
Py_ssize_t PyString_Size(object p)
39
Py_ssize_t PyString_GET_SIZE_ptr "PyString_GET_SIZE" (PyObject *)
40
char * PyString_AS_STRING_ptr "PyString_AS_STRING" (PyObject *)
41
int PyString_AsStringAndSize_ptr(PyObject *, char **buf, Py_ssize_t *len)
42
void PyString_InternInPlace(PyObject **)
43
int PyTuple_CheckExact(object t)
44
object PyTuple_New(Py_ssize_t n_entries)
45
void PyTuple_SET_ITEM(object, Py_ssize_t offset, object) # steals the ref
46
Py_ssize_t PyTuple_GET_SIZE(object t)
47
PyObject *PyTuple_GET_ITEM_ptr_object "PyTuple_GET_ITEM" (object tpl, int index)
48
void Py_INCREF(object)
49
void Py_DECREF_ptr "Py_DECREF" (PyObject *)
51
cdef extern from "string.h":
52
void *memcpy(void *dest, void *src, size_t n)
53
void *memchr(void *s, int c, size_t n)
55
# void *memrchr(void *s, int c, size_t n)
56
int strncmp(char *s1, char *s2, size_t n)
59
# TODO: Find some way to import this from _dirstate_helpers
60
cdef void* _my_memrchr(void *s, int c, size_t n):
61
# memrchr seems to be a GNU extension, so we have to implement it ourselves
62
# It is not present in any win32 standard library
74
# TODO: Import this from _dirstate_helpers when it is merged
75
cdef object safe_string_from_size(char *s, Py_ssize_t size):
78
'tried to create a string with an invalid size: %d @0x%x'
80
return PyString_FromStringAndSize(s, size)
83
cdef object safe_interned_string_from_size(char *s, Py_ssize_t size):
87
'tried to create a string with an invalid size: %d @0x%x'
89
py_str = PyString_FromStringAndSize_ptr(s, size)
90
PyString_InternInPlace(&py_str)
91
result = <object>py_str
92
# Casting a PyObject* to an <object> triggers an INCREF from Pyrex, so we
93
# DECREF it to avoid geting immortal strings
98
cdef class BTreeLeafParser:
99
"""Parse the leaf nodes of a BTree index.
101
:ivar bytes: The PyString object containing the uncompressed text for the
103
:ivar key_length: An integer describing how many pieces the keys have for
105
:ivar ref_list_length: An integer describing how many references this index
107
:ivar keys: A PyList of keys found in this node.
109
:ivar _cur_str: A pointer to the start of the next line to parse
110
:ivar _end_str: A pointer to the end of bytes
111
:ivar _start: Pointer to the location within the current line while
113
:ivar _header_found: True when we have parsed the header for this node
118
cdef int ref_list_length
123
# The current start point for parsing
126
cdef int _header_found
128
def __init__(self, bytes, key_length, ref_list_length):
130
self.key_length = key_length
131
self.ref_list_length = ref_list_length
135
self._header_found = 0
137
cdef extract_key(self, char * last):
140
:param last: points at the byte after the last byte permitted for the
144
cdef int loop_counter
146
key = PyTuple_New(self.key_length)
147
for loop_counter from 0 <= loop_counter < self.key_length:
149
temp_ptr = <char*>memchr(self._start, c'\0', last - self._start)
151
if loop_counter + 1 == self.key_length:
156
failure_string = ("invalid key, wanted segment from " +
157
repr(safe_string_from_size(self._start,
158
last - self._start)))
159
raise AssertionError(failure_string)
160
# capture the key string
161
# TODO: Consider using PyIntern_FromString, the only caveat is that
162
# it assumes a NULL-terminated string, so we have to check if
163
# temp_ptr[0] == c'\0' or some other char.
164
key_element = safe_interned_string_from_size(self._start,
165
temp_ptr - self._start)
166
# advance our pointer
167
self._start = temp_ptr + 1
168
Py_INCREF(key_element)
169
PyTuple_SET_ITEM(key, loop_counter, key_element)
172
cdef int process_line(self) except -1:
173
"""Process a line in the bytes."""
177
cdef char *next_start
178
cdef int loop_counter
180
self._start = self._cur_str
181
# Find the next newline
182
last = <char*>memchr(self._start, c'\n', self._end_str - self._start)
184
# Process until the end of the file
186
self._cur_str = self._end_str
188
# And the next string is right after it
189
self._cur_str = last + 1
190
# The last character is right before the '\n'
193
if last == self._start:
196
if last < self._start:
197
# Unexpected error condition - fail
199
if 0 == self._header_found:
200
# The first line in a leaf node is the header "type=leaf\n"
201
if strncmp("type=leaf", self._start, last - self._start) == 0:
202
self._header_found = 1
205
raise AssertionError('Node did not start with "type=leaf": %r'
206
% (safe_string_from_size(self._start, last - self._start)))
209
key = self.extract_key(last)
210
# find the value area
211
temp_ptr = <char*>_my_memrchr(self._start, c'\0', last - self._start)
216
# capture the value string
217
value = safe_string_from_size(temp_ptr + 1, last - temp_ptr - 1)
218
# shrink the references end point
220
if self.ref_list_length:
223
while loop_counter < self.ref_list_length:
225
# extract a reference list
226
loop_counter = loop_counter + 1
227
if last < self._start:
229
# find the next reference list end point:
230
temp_ptr = <char*>memchr(self._start, c'\t', last - self._start)
232
# Only valid for the last list
233
if loop_counter != self.ref_list_length:
236
raise AssertionError("invalid key")
238
# scan to the end of the ref list area
242
# scan to the end of this ref list
244
next_start = temp_ptr + 1
245
# Now, there may be multiple keys in the ref list.
246
while self._start < ref_ptr:
247
# loop finding keys and extracting them
248
temp_ptr = <char*>memchr(self._start, c'\r',
249
ref_ptr - self._start)
251
# key runs to the end
253
PyList_Append(ref_list, self.extract_key(temp_ptr))
254
PyList_Append(ref_lists, tuple(ref_list))
255
# prepare for the next reference list
256
self._start = next_start
257
ref_lists = tuple(ref_lists)
258
node_value = (value, ref_lists)
260
if last != self._start:
261
# unexpected reference data present
263
node_value = (value, ())
264
PyList_Append(self.keys, (key, node_value))
268
cdef Py_ssize_t byte_count
269
if not PyString_CheckExact(self.bytes):
270
raise AssertionError('self.bytes is not a string.')
271
byte_count = PyString_Size(self.bytes)
272
self._cur_str = PyString_AsString(self.bytes)
273
# This points to the last character in the string
274
self._end_str = self._cur_str + byte_count
275
while self._cur_str < self._end_str:
280
def _parse_leaf_lines(bytes, key_length, ref_list_length):
281
parser = BTreeLeafParser(bytes, key_length, ref_list_length)
282
return parser.parse()
285
def _flatten_node(node, reference_lists):
286
"""Convert a node into the serialized form.
288
:param node: A tuple representing a node:
289
(index, key_tuple, value, references)
290
:param reference_lists: Does this index have reference lists?
291
:return: (string_key, flattened)
292
string_key The serialized key for referencing this node
293
flattened A string with the serialized form for the contents
295
cdef int have_reference_lists
296
cdef Py_ssize_t flat_len
297
cdef Py_ssize_t key_len
298
cdef Py_ssize_t node_len
301
cdef Py_ssize_t value_len
303
cdef Py_ssize_t refs_len
304
cdef Py_ssize_t next_len
305
cdef int first_ref_list
306
cdef int first_reference
308
cdef PyObject *ref_bit
309
cdef Py_ssize_t ref_bit_len
311
if not PyTuple_CheckExact(node):
312
raise TypeError('We expected a tuple() for node not: %s'
314
node_len = PyTuple_GET_SIZE(node)
315
have_reference_lists = reference_lists
316
if have_reference_lists:
318
raise ValueError('With ref_lists, we expected 4 entries not: %s'
321
raise ValueError('Without ref_lists, we need at least 3 entries not: %s'
323
# I don't expect that we can do faster than string.join()
324
string_key = '\0'.join(<object>PyTuple_GET_ITEM_ptr_object(node, 1))
326
# TODO: instead of using string joins, precompute the final string length,
327
# and then malloc a single string and copy everything in.
329
# TODO: We probably want to use PySequenceFast, because we have lists and
330
# tuples, but we aren't sure which we will get.
332
# line := string_key NULL flat_refs NULL value LF
333
# string_key := BYTES (NULL BYTES)*
334
# flat_refs := ref_list (TAB ref_list)*
335
# ref_list := ref (CR ref)*
336
# ref := BYTES (NULL BYTES)*
339
if have_reference_lists:
340
# Figure out how many bytes it will take to store the references
341
ref_lists = <object>PyTuple_GET_ITEM_ptr_object(node, 3)
342
next_len = len(ref_lists) # TODO: use a Py function
344
# If there are no nodes, we don't need to do any work
345
# Otherwise we will need (len - 1) '\t' characters to separate
346
# the reference lists
347
refs_len = refs_len + (next_len - 1)
348
for ref_list in ref_lists:
349
next_len = len(ref_list)
351
# We will need (len - 1) '\r' characters to separate the
353
refs_len = refs_len + (next_len - 1)
354
for reference in ref_list:
355
if not PyTuple_CheckExact(reference):
357
'We expect references to be tuples not: %s'
359
next_len = PyTuple_GET_SIZE(reference)
361
# We will need (len - 1) '\x00' characters to
362
# separate the reference key
363
refs_len = refs_len + (next_len - 1)
364
for i from 0 <= i < next_len:
365
ref_bit = PyTuple_GET_ITEM_ptr_object(reference, i)
366
if not PyString_CheckExact_ptr(ref_bit):
367
raise TypeError('We expect reference bits'
368
' to be strings not: %s'
369
% type(<object>ref_bit))
370
refs_len = refs_len + PyString_GET_SIZE_ptr(ref_bit)
372
# So we have the (key NULL refs NULL value LF)
373
key_len = PyString_Size(string_key)
374
val = PyTuple_GET_ITEM_ptr_object(node, 2)
375
if not PyString_CheckExact_ptr(val):
376
raise TypeError('Expected a plain str for value not: %s'
378
value = PyString_AS_STRING_ptr(val)
379
value_len = PyString_GET_SIZE_ptr(val)
380
flat_len = (key_len + 1 + refs_len + 1 + value_len + 1)
381
line = PyString_FromStringAndSize(NULL, flat_len)
382
# Get a pointer to the new buffer
383
out = PyString_AsString(line)
384
memcpy(out, PyString_AsString(string_key), key_len)
390
for ref_list in ref_lists:
391
if first_ref_list == 0:
396
for reference in ref_list:
397
if first_reference == 0:
401
next_len = PyTuple_GET_SIZE(reference)
402
for i from 0 <= i < next_len:
406
ref_bit = PyTuple_GET_ITEM_ptr_object(reference, i)
407
ref_bit_len = PyString_GET_SIZE_ptr(ref_bit)
408
memcpy(out, PyString_AS_STRING_ptr(ref_bit), ref_bit_len)
409
out = out + ref_bit_len
412
memcpy(out, value, value_len)
413
out = out + value_len
415
return string_key, line