Show More
@@ -1,312 +1,339 b'' | |||
|
1 | 1 | # nodemap.py - nodemap related code and utilities |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2019 Pierre-Yves David <pierre-yves.david@octobus.net> |
|
4 | 4 | # Copyright 2019 George Racinet <georges.racinet@octobus.net> |
|
5 | 5 | # |
|
6 | 6 | # This software may be used and distributed according to the terms of the |
|
7 | 7 | # GNU General Public License version 2 or any later version. |
|
8 | 8 | |
|
9 | 9 | from __future__ import absolute_import |
|
10 | 10 | |
|
11 | 11 | import os |
|
12 | 12 | import re |
|
13 | 13 | import struct |
|
14 | 14 | |
|
15 | 15 | from .. import ( |
|
16 | 16 | error, |
|
17 | 17 | node as nodemod, |
|
18 | 18 | util, |
|
19 | 19 | ) |
|
20 | 20 | |
|
21 | 21 | |
|
22 | 22 | class NodeMap(dict): |
|
23 | 23 | def __missing__(self, x): |
|
24 | 24 | raise error.RevlogError(b'unknown node: %s' % x) |
|
25 | 25 | |
|
26 | 26 | |
|
27 | 27 | def persisted_data(revlog): |
|
28 | 28 | """read the nodemap for a revlog from disk""" |
|
29 | 29 | if revlog.nodemap_file is None: |
|
30 | 30 | return None |
|
31 | 31 | pdata = revlog.opener.tryread(revlog.nodemap_file) |
|
32 | 32 | if not pdata: |
|
33 | 33 | return None |
|
34 | 34 | offset = 0 |
|
35 | 35 | (version,) = S_VERSION.unpack(pdata[offset : offset + S_VERSION.size]) |
|
36 | 36 | if version != ONDISK_VERSION: |
|
37 | 37 | return None |
|
38 | 38 | offset += S_VERSION.size |
|
39 | 39 | (uuid_size,) = S_HEADER.unpack(pdata[offset : offset + S_HEADER.size]) |
|
40 | 40 | offset += S_HEADER.size |
|
41 | 41 | uid = pdata[offset : offset + uuid_size] |
|
42 | 42 | |
|
43 | 43 | filename = _rawdata_filepath(revlog, uid) |
|
44 | 44 | return revlog.opener.tryread(filename) |
|
45 | 45 | |
|
46 | 46 | |
|
47 | 47 | def setup_persistent_nodemap(tr, revlog): |
|
48 | 48 | """Install whatever is needed transaction side to persist a nodemap on disk |
|
49 | 49 | |
|
50 | 50 | (only actually persist the nodemap if this is relevant for this revlog) |
|
51 | 51 | """ |
|
52 | 52 | if revlog._inline: |
|
53 | 53 | return # inlined revlog are too small for this to be relevant |
|
54 | 54 | if revlog.nodemap_file is None: |
|
55 | 55 | return # we do not use persistent_nodemap on this revlog |
|
56 | 56 | callback_id = b"revlog-persistent-nodemap-%s" % revlog.nodemap_file |
|
57 | 57 | if tr.hasfinalize(callback_id): |
|
58 | 58 | return # no need to register again |
|
59 | 59 | tr.addfinalize(callback_id, lambda tr: _persist_nodemap(tr, revlog)) |
|
60 | 60 | |
|
61 | 61 | |
|
62 | 62 | def _persist_nodemap(tr, revlog): |
|
63 | 63 | """Write nodemap data on disk for a given revlog |
|
64 | 64 | """ |
|
65 | 65 | if getattr(revlog, 'filteredrevs', ()): |
|
66 | 66 | raise error.ProgrammingError( |
|
67 | 67 | "cannot persist nodemap of a filtered changelog" |
|
68 | 68 | ) |
|
69 | 69 | if revlog.nodemap_file is None: |
|
70 | 70 | msg = "calling persist nodemap on a revlog without the feature enableb" |
|
71 | 71 | raise error.ProgrammingError(msg) |
|
72 | 72 | if util.safehasattr(revlog.index, "nodemap_data_all"): |
|
73 | 73 | data = revlog.index.nodemap_data_all() |
|
74 | 74 | else: |
|
75 | 75 | data = persistent_data(revlog.index) |
|
76 | 76 | uid = _make_uid() |
|
77 | 77 | datafile = _rawdata_filepath(revlog, uid) |
|
78 | 78 | olds = _other_rawdata_filepath(revlog, uid) |
|
79 | 79 | if olds: |
|
80 | 80 | realvfs = getattr(revlog, '_realopener', revlog.opener) |
|
81 | 81 | |
|
82 | 82 | def cleanup(tr): |
|
83 | 83 | for oldfile in olds: |
|
84 | 84 | realvfs.tryunlink(oldfile) |
|
85 | 85 | |
|
86 | 86 | callback_id = b"revlog-cleanup-nodemap-%s" % revlog.nodemap_file |
|
87 | 87 | tr.addpostclose(callback_id, cleanup) |
|
88 | 88 | # EXP-TODO: if this is a cache, this should use a cache vfs, not a |
|
89 | 89 | # store vfs |
|
90 | 90 | with revlog.opener(datafile, b'w') as fd: |
|
91 | 91 | fd.write(data) |
|
92 | 92 | # EXP-TODO: if this is a cache, this should use a cache vfs, not a |
|
93 | 93 | # store vfs |
|
94 | 94 | with revlog.opener(revlog.nodemap_file, b'w', atomictemp=True) as fp: |
|
95 | 95 | fp.write(_serialize_docket(uid)) |
|
96 | 96 | # EXP-TODO: if the transaction abort, we should remove the new data and |
|
97 | 97 | # reinstall the old one. |
|
98 | 98 | |
|
99 | 99 | |
|
100 | 100 | ### Nodemap docket file |
|
101 | 101 | # |
|
102 | 102 | # The nodemap data are stored on disk using 2 files: |
|
103 | 103 | # |
|
104 | 104 | # * a raw data files containing a persistent nodemap |
|
105 | 105 | # (see `Nodemap Trie` section) |
|
106 | 106 | # |
|
107 | 107 | # * a small "docket" file containing medatadata |
|
108 | 108 | # |
|
109 | 109 | # While the nodemap data can be multiple tens of megabytes, the "docket" is |
|
110 | 110 | # small, it is easy to update it automatically or to duplicated its content |
|
111 | 111 | # during a transaction. |
|
112 | 112 | # |
|
113 | 113 | # Multiple raw data can exist at the same time (The currently valid one and a |
|
114 | 114 | # new one beind used by an in progress transaction). To accomodate this, the |
|
115 | 115 | # filename hosting the raw data has a variable parts. The exact filename is |
|
116 | 116 | # specified inside the "docket" file. |
|
117 | 117 | # |
|
118 | 118 | # The docket file contains information to find, qualify and validate the raw |
|
119 | 119 | # data. Its content is currently very light, but it will expand as the on disk |
|
120 | 120 | # nodemap gains the necessary features to be used in production. |
|
121 | 121 | |
|
122 | 122 | # version 0 is experimental, no BC garantee, do no use outside of tests. |
|
123 | 123 | ONDISK_VERSION = 0 |
|
124 | 124 | |
|
125 | 125 | S_VERSION = struct.Struct(">B") |
|
126 | 126 | S_HEADER = struct.Struct(">B") |
|
127 | 127 | |
|
128 | 128 | ID_SIZE = 8 |
|
129 | 129 | |
|
130 | 130 | |
|
131 | 131 | def _make_uid(): |
|
132 | 132 | """return a new unique identifier. |
|
133 | 133 | |
|
134 | 134 | The identifier is random and composed of ascii characters.""" |
|
135 | 135 | return nodemod.hex(os.urandom(ID_SIZE)) |
|
136 | 136 | |
|
137 | 137 | |
|
138 | 138 | def _serialize_docket(uid): |
|
139 | 139 | """return serialized bytes for a docket using the passed uid""" |
|
140 | 140 | data = [] |
|
141 | 141 | data.append(S_VERSION.pack(ONDISK_VERSION)) |
|
142 | 142 | data.append(S_HEADER.pack(len(uid))) |
|
143 | 143 | data.append(uid) |
|
144 | 144 | return b''.join(data) |
|
145 | 145 | |
|
146 | 146 | |
|
147 | 147 | def _rawdata_filepath(revlog, uid): |
|
148 | 148 | """The (vfs relative) nodemap's rawdata file for a given uid""" |
|
149 | 149 | prefix = revlog.nodemap_file[:-2] |
|
150 | 150 | return b"%s-%s.nd" % (prefix, uid) |
|
151 | 151 | |
|
152 | 152 | |
|
153 | 153 | def _other_rawdata_filepath(revlog, uid): |
|
154 | 154 | prefix = revlog.nodemap_file[:-2] |
|
155 | 155 | pattern = re.compile(b"(^|/)%s-[0-9a-f]+\.nd$" % prefix) |
|
156 | 156 | new_file_path = _rawdata_filepath(revlog, uid) |
|
157 | 157 | new_file_name = revlog.opener.basename(new_file_path) |
|
158 | 158 | dirpath = revlog.opener.dirname(new_file_path) |
|
159 | 159 | others = [] |
|
160 | 160 | for f in revlog.opener.listdir(dirpath): |
|
161 | 161 | if pattern.match(f) and f != new_file_name: |
|
162 | 162 | others.append(f) |
|
163 | 163 | return others |
|
164 | 164 | |
|
165 | 165 | |
|
166 | 166 | ### Nodemap Trie |
|
167 | 167 | # |
|
168 | 168 | # This is a simple reference implementation to compute and persist a nodemap |
|
169 | 169 | # trie. This reference implementation is write only. The python version of this |
|
170 | 170 | # is not expected to be actually used, since it wont provide performance |
|
171 | 171 | # improvement over existing non-persistent C implementation. |
|
172 | 172 | # |
|
173 | 173 | # The nodemap is persisted as Trie using 4bits-address/16-entries block. each |
|
174 | 174 | # revision can be adressed using its node shortest prefix. |
|
175 | 175 | # |
|
176 | 176 | # The trie is stored as a sequence of block. Each block contains 16 entries |
|
177 | 177 | # (signed 64bit integer, big endian). Each entry can be one of the following: |
|
178 | 178 | # |
|
179 | 179 | # * value >= 0 -> index of sub-block |
|
180 | 180 | # * value == -1 -> no value |
|
181 | 181 | # * value < -1 -> a revision value: rev = -(value+10) |
|
182 | 182 | # |
|
183 | 183 | # The implementation focus on simplicity, not on performance. A Rust |
|
184 | 184 | # implementation should provide a efficient version of the same binary |
|
185 | 185 | # persistence. This reference python implementation is never meant to be |
|
186 | 186 | # extensively use in production. |
|
187 | 187 | |
|
188 | 188 | |
|
189 | 189 | def persistent_data(index): |
|
190 | 190 | """return the persistent binary form for a nodemap for a given index |
|
191 | 191 | """ |
|
192 | 192 | trie = _build_trie(index) |
|
193 | 193 | return _persist_trie(trie) |
|
194 | 194 | |
|
195 | 195 | |
|
196 | 196 | S_BLOCK = struct.Struct(">" + ("l" * 16)) |
|
197 | 197 | |
|
198 | 198 | NO_ENTRY = -1 |
|
199 | 199 | # rev 0 need to be -2 because 0 is used by block, -1 is a special value. |
|
200 | 200 | REV_OFFSET = 2 |
|
201 | 201 | |
|
202 | 202 | |
|
203 | 203 | def _transform_rev(rev): |
|
204 | 204 | """Return the number used to represent the rev in the tree. |
|
205 | 205 | |
|
206 | 206 | (or retrieve a rev number from such representation) |
|
207 | 207 | |
|
208 | 208 | Note that this is an involution, a function equal to its inverse (i.e. |
|
209 | 209 | which gives the identity when applied to itself). |
|
210 | 210 | """ |
|
211 | 211 | return -(rev + REV_OFFSET) |
|
212 | 212 | |
|
213 | 213 | |
|
214 | 214 | def _to_int(hex_digit): |
|
215 | 215 | """turn an hexadecimal digit into a proper integer""" |
|
216 | 216 | return int(hex_digit, 16) |
|
217 | 217 | |
|
218 | 218 | |
|
219 | 219 | class Block(dict): |
|
220 | 220 | """represent a block of the Trie |
|
221 | 221 | |
|
222 | 222 | contains up to 16 entry indexed from 0 to 15""" |
|
223 | 223 | |
|
224 | 224 | def __iter__(self): |
|
225 | 225 | return iter(self.get(i) for i in range(16)) |
|
226 | 226 | |
|
227 | 227 | |
|
228 | 228 | def _build_trie(index): |
|
229 | 229 | """build a nodemap trie |
|
230 | 230 | |
|
231 | 231 | The nodemap stores revision number for each unique prefix. |
|
232 | 232 | |
|
233 | 233 | Each block is a dictionary with keys in `[0, 15]`. Values are either |
|
234 | 234 | another block or a revision number. |
|
235 | 235 | """ |
|
236 | 236 | root = Block() |
|
237 | 237 | for rev in range(len(index)): |
|
238 | 238 | hex = nodemod.hex(index[rev][7]) |
|
239 | 239 | _insert_into_block(index, 0, root, rev, hex) |
|
240 | 240 | return root |
|
241 | 241 | |
|
242 | 242 | |
|
243 | 243 | def _insert_into_block(index, level, block, current_rev, current_hex): |
|
244 | 244 | """insert a new revision in a block |
|
245 | 245 | |
|
246 | 246 | index: the index we are adding revision for |
|
247 | 247 | level: the depth of the current block in the trie |
|
248 | 248 | block: the block currently being considered |
|
249 | 249 | current_rev: the revision number we are adding |
|
250 | 250 | current_hex: the hexadecimal representation of the of that revision |
|
251 | 251 | """ |
|
252 | 252 | hex_digit = _to_int(current_hex[level : level + 1]) |
|
253 | 253 | entry = block.get(hex_digit) |
|
254 | 254 | if entry is None: |
|
255 | 255 | # no entry, simply store the revision number |
|
256 | 256 | block[hex_digit] = current_rev |
|
257 | 257 | elif isinstance(entry, dict): |
|
258 | 258 | # need to recurse to an underlying block |
|
259 | 259 | _insert_into_block(index, level + 1, entry, current_rev, current_hex) |
|
260 | 260 | else: |
|
261 | 261 | # collision with a previously unique prefix, inserting new |
|
262 | 262 | # vertices to fit both entry. |
|
263 | 263 | other_hex = nodemod.hex(index[entry][7]) |
|
264 | 264 | other_rev = entry |
|
265 | 265 | new = Block() |
|
266 | 266 | block[hex_digit] = new |
|
267 | 267 | _insert_into_block(index, level + 1, new, other_rev, other_hex) |
|
268 | 268 | _insert_into_block(index, level + 1, new, current_rev, current_hex) |
|
269 | 269 | |
|
270 | 270 | |
|
271 | 271 | def _persist_trie(root): |
|
272 | 272 | """turn a nodemap trie into persistent binary data |
|
273 | 273 | |
|
274 | 274 | See `_build_trie` for nodemap trie structure""" |
|
275 | 275 | block_map = {} |
|
276 | 276 | chunks = [] |
|
277 | 277 | for tn in _walk_trie(root): |
|
278 | 278 | block_map[id(tn)] = len(chunks) |
|
279 | 279 | chunks.append(_persist_block(tn, block_map)) |
|
280 | 280 | return b''.join(chunks) |
|
281 | 281 | |
|
282 | 282 | |
|
283 | 283 | def _walk_trie(block): |
|
284 | 284 | """yield all the block in a trie |
|
285 | 285 | |
|
286 | 286 | Children blocks are always yield before their parent block. |
|
287 | 287 | """ |
|
288 | 288 | for (_, item) in sorted(block.items()): |
|
289 | 289 | if isinstance(item, dict): |
|
290 | 290 | for sub_block in _walk_trie(item): |
|
291 | 291 | yield sub_block |
|
292 | 292 | yield block |
|
293 | 293 | |
|
294 | 294 | |
|
295 | 295 | def _persist_block(block_node, block_map): |
|
296 | 296 | """produce persistent binary data for a single block |
|
297 | 297 | |
|
298 | 298 | Children block are assumed to be already persisted and present in |
|
299 | 299 | block_map. |
|
300 | 300 | """ |
|
301 | 301 | data = tuple(_to_value(v, block_map) for v in block_node) |
|
302 | 302 | return S_BLOCK.pack(*data) |
|
303 | 303 | |
|
304 | 304 | |
|
305 | 305 | def _to_value(item, block_map): |
|
306 | 306 | """persist any value as an integer""" |
|
307 | 307 | if item is None: |
|
308 | 308 | return NO_ENTRY |
|
309 | 309 | elif isinstance(item, dict): |
|
310 | 310 | return block_map[id(item)] |
|
311 | 311 | else: |
|
312 | 312 | return _transform_rev(item) |
|
313 | ||
|
314 | ||
|
315 | def parse_data(data): | |
|
316 | """parse parse nodemap data into a nodemap Trie""" | |
|
317 | if (len(data) % S_BLOCK.size) != 0: | |
|
318 | msg = "nodemap data size is not a multiple of block size (%d): %d" | |
|
319 | raise error.Abort(msg % (S_BLOCK.size, len(data))) | |
|
320 | if not data: | |
|
321 | return Block() | |
|
322 | block_map = {} | |
|
323 | new_blocks = [] | |
|
324 | for i in range(0, len(data), S_BLOCK.size): | |
|
325 | block = Block() | |
|
326 | ondisk_id = len(block_map) | |
|
327 | block_map[ondisk_id] = block | |
|
328 | block_data = data[i : i + S_BLOCK.size] | |
|
329 | values = S_BLOCK.unpack(block_data) | |
|
330 | new_blocks.append((block, values)) | |
|
331 | for b, values in new_blocks: | |
|
332 | for idx, v in enumerate(values): | |
|
333 | if v == NO_ENTRY: | |
|
334 | continue | |
|
335 | elif v >= 0: | |
|
336 | b[idx] = block_map[v] | |
|
337 | else: | |
|
338 | b[idx] = _transform_rev(v) | |
|
339 | return block |
General Comments 0
You need to be logged in to leave comments.
Login now