suwako / LittleBeast

master
ブランチ master
LittleBeast / src / Std / Lib / Git / ObjectDb.php
10.9 KB、290行 生 Blame 履歴
1<?php
2/*************************************************************
3# 076 License
4
5Copyright (c) テクニカル諏訪子
6
7Permission is hereby granted to any person obtaining a copy of the software
8Little Beast (the "Software") to use, modify, merge, copy, publish, distribute,
9sublicense, and/or sell copies of the Software, subject to the following conditions:
10
11 1. **Origin Attribution**:
12 - You must not misrepresent the origin of the Software; you must not claim
13 you created the original Software.
14 - If the Software is used in a product, you must either:
15 a. Provide clear attribution in the product's documentation, user interface,
16 or other visible areas, **OR**
17 b. Pay the original developers a fee they specify in writing.
18 2. **Usage Restriction**:
19 - The Software, or any derivative works, dependencies, or libraries
20 incorporating it, must not be used for censorship or to suppress freedom of
21 speech, expression, or creativity. Prohibited uses include, but are not
22 limited to:
23 - Censorship of so-called "hate speech", visuals, non-mainstream opinions,
24 ideas, or objective reality.
25 - Tools or systems designed to restrict access to information or
26 artistic works.
27 3. **Notice Preservation**:
28 - This license and the above copyright notice must remain intact in all copies
29 of the source code.
30
31THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED,
32INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A
33PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
34HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF
35CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE
36OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
37 */
38namespace Std\Lib\Git;
39
40/**
41 * オブジェクトの保存（objects/）：ルーズオブジェクトとパック、objects/info/alternates
42 *
43 * IDは16進数の文字列。読むと[種類, データ]（種類はcommit、tree、blob、tag）。
44 */
45final class ObjectDb {
46 /** @var string[] オブジェクトのディレクトリ（自分とalternates） */
47 private array $dirs;
48 /** @var Pack[] パス => パック */
49 private array $packs = [];
50 private int $scanned = 0;
51 /** 差分の元のキャッシュ（キー => [種類, データ]） */
52 private array $cache = [];
53 private int $cacheBytes = 0;
54 private const CACHE_LIMIT = 33554432;
55
56 public function __construct(private Repository $repo) {
57 $this->dirs = [$repo->path.'/objects'];
58 $alt = $repo->path.'/objects/info/alternates';
59 if (is_file($alt)) {
60 foreach (preg_split('/\r?\n/', (string)file_get_contents($alt)) as $line) {
61 $line = trim($line);
62 if ($line === '' || $line[0] === '#') continue;
63 $dir = preg_match('#^(/|[A-Za-z]:[/\\\\])#', $line) ? $line : $repo->path.'/objects/'.$line;
64 if (is_dir($dir)) $this->dirs[] = rtrim($dir, '/\\');
65 }
66 }
67 }
68
69 /** @return Pack[] */
70 public function packs(bool $rescan = false): array {
71 if ($rescan || $this->scanned === 0) {
72 foreach ($this->dirs as $dir) {
73 foreach (glob($dir.'/pack/pack-*.idx') ?: [] as $idx) {
74 $pack = substr($idx, 0, -4).'.pack';
75 if (!isset($this->packs[$pack]) && is_file($pack)) $this->packs[$pack] = new Pack($pack, $this->repo->hashLen);
76 }
77 }
78 $this->scanned++;
79 }
80 return $this->packs;
81 }
82
83 /**
84 * オブジェクトの場所
85 *
86 * @return array|null ['loose', パス] 又は ['pack', Pack, オフセット]
87 */
88 public function locate(string $id): ?array {
89 $id = strtolower($id);
90 if (!$this->repo->isId($id)) return null;
91 foreach ($this->dirs as $dir) {
92 $p = $dir.'/'.substr($id, 0, 2).'/'.substr($id, 2);
93 if (is_file($p)) return ['loose', $p];
94 }
95 $bin = hex2bin($id);
96 foreach ([false, true] as $rescan) {
97 foreach ($this->packs($rescan) as $pack) {
98 $off = $pack->find($bin);
99 if ($off !== null) return ['pack', $pack, $off];
100 }
101 }
102 return null;
103 }
104
105 public function exists(string $id): bool {
106 return $this->locate($id) !== null;
107 }
108
109 /**
110 * オブジェクトを読む
111 *
112 * @return array|null [種類, データ]
113 */
114 public function read(string $id): ?array {
115 $loc = $this->locate($id);
116 if ($loc === null) return null;
117 if ($loc[0] === 'loose') return $this->readLoose($loc[1], $id);
118 return $this->readPacked($loc[1], $loc[2]);
119 }
120
121 /** 必ず有るオブジェクト */
122 public function get(string $id, ?string $type = null): string {
123 $o = $this->read($id);
124 if ($o === null) throw new GitException("オブジェクトが有りません：{$id}");
125 if ($type !== null && $o[0] !== $type) throw new GitException("{$id}は{$type}ではなく{$o[0]}です");
126 return $o[1];
127 }
128
129 /**
130 * 種類と大きさだけ（データを全部は読まない）
131 *
132 * @return array|null [種類, 大きさ]
133 */
134 public function info(string $id): ?array {
135 $loc = $this->locate($id);
136 if ($loc === null) return null;
137 if ($loc[0] === 'loose') {
138 $ctx = inflate_init(ZLIB_ENCODING_DEFLATE);
139 $fh = fopen($loc[1], 'rb');
140 $head = '';
141 while (!str_contains($head, "\0") && !feof($fh)) $head .= (string)@inflate_add($ctx, (string)fread($fh, 256), ZLIB_SYNC_FLUSH);
142 fclose($fh);
143 if (!preg_match('/^(commit|tree|blob|tag) (\d+)\0/', $head, $m)) throw new GitException("壊れたオブジェクト：{$id}");
144 return [$m[1], (int)$m[2]];
145 }
146 [$type, $size, $dataAt, $baseOff, $baseId] = $loc[1]->header($loc[2]);
147 if (isset(Pack::TYPES[$type])) return [Pack::TYPES[$type], $size];
148 // 差分：大きさは差分の頭、種類は元を辿る
149 $head = $loc[1]->inflate($dataAt, $size);
150 $target = Delta::targetSize($head);
151 $pack = $loc[1];
152 $off = $loc[2];
153 for ($depth = 0; $depth < 10000; $depth++) {
154 [$t, , , $bo, $bi] = $pack->header($off);
155 if (isset(Pack::TYPES[$t])) return [Pack::TYPES[$t], $target];
156 if ($bo !== null) {
157 $off = $bo;
158 continue;
159 }
160 $l = $this->locate(bin2hex($bi));
161 if ($l === null) throw new GitException('差分の元が有りません：'.bin2hex($bi));
162 if ($l[0] === 'loose') return [$this->info(bin2hex($bi))[0], $target];
163 [$pack, $off] = [$l[1], $l[2]];
164 }
165 throw new GitException("差分が深過ぎます：{$id}");
166 }
167
168 private function readLoose(string $path, string $id): array {
169 $raw = @file_get_contents($path);
170 $data = $raw === false ? false : @zlib_decode($raw);
171 if ($data === false) throw new GitException("壊れたオブジェクト：{$id}");
172 $nul = strpos($data, "\0");
173 if ($nul === false || !preg_match('/^(commit|tree|blob|tag) (\d+)$/', substr($data, 0, $nul), $m) || (int)$m[2] !== strlen($data) - $nul - 1) {
174 throw new GitException("壊れたオブジェクト：{$id}");
175 }
176 return [$m[1], substr($data, $nul + 1)];
177 }
178
179 /**
180 * パックの中のオブジェクト（差分は元を辿って当てる）
181 *
182 * @return array [種類, データ]
183 */
184 public function readPacked(Pack $pack, int $offset): array {
185 $key = $pack->path.':'.$offset;
186 if (isset($this->cache[$key])) return $this->cache[$key];
187 // 差分の連鎖を集める
188 $chain = [];
189 $p = $pack;
190 $off = $offset;
191 while (true) {
192 $k = $p->path.':'.$off;
193 if (isset($this->cache[$k])) {
194 $base = $this->cache[$k];
195 break;
196 }
197 [$type, $size, $dataAt, $baseOff, $baseId] = $p->header($off);
198 if (isset(Pack::TYPES[$type])) {
199 $base = [Pack::TYPES[$type], $p->inflate($dataAt, $size)];
200 $this->remember($k, $base);
201 break;
202 }
203 $chain[] = [$p, $off, $dataAt, $size];
204 if (count($chain) > 10000) throw new GitException('差分が深過ぎます');
205 if ($baseOff !== null) {
206 $off = $baseOff;
207 continue;
208 }
209 $hex = bin2hex($baseId);
210 $o = $p->find($baseId);
211 if ($o !== null) {
212 $off = $o;
213 continue;
214 }
215 $l = $this->locate($hex);
216 if ($l === null) throw new GitException("差分の元が有りません：{$hex}");
217 if ($l[0] === 'loose') {
218 $base = $this->readLoose($l[1], $hex);
219 break;
220 }
221 [$p, $off] = [$l[1], $l[2]];
222 }
223 for ($i = count($chain) - 1; $i >= 0; $i--) {
224 [$cp, $coff, $dataAt, $size] = $chain[$i];
225 $base = [$base[0], Delta::apply($base[1], $cp->inflate($dataAt, $size))];
226 $this->remember($cp->path.':'.$coff, $base);
227 }
228 return $base;
229 }
230
231 private function remember(string $key, array $obj): void {
232 $n = strlen($obj[1]);
233 if ($n > self::CACHE_LIMIT / 4) return;
234 while ($this->cacheBytes + $n > self::CACHE_LIMIT && !empty($this->cache)) {
235 $k = array_key_first($this->cache);
236 $this->cacheBytes -= strlen($this->cache[$k][1]);
237 unset($this->cache[$k]);
238 }
239 $this->cache[$key] = $obj;
240 $this->cacheBytes += $n;
241 }
242
243 /**
244 * ルーズオブジェクトとして書く（既に有れば何もしない）
245 *
246 * @return string ID
247 */
248 public function write(string $type, string $data): string {
249 if (!in_array($type, ['commit', 'tree', 'blob', 'tag'], true)) throw new GitException("不正な種類：{$type}");
250 $id = $this->repo->hashObject($type, $data);
251 if ($this->exists($id)) return $id;
252 $dir = $this->dirs[0].'/'.substr($id, 0, 2);
253 $this->repo->mkdir($dir);
254 $level = $this->repo->config->getInt('core.loosecompression', $this->repo->config->getInt('core.compression', -1));
255 $this->repo->writeFile($dir.'/'.substr($id, 2), (string)gzcompress($type.' '.strlen($data)."\0".$data, max(-1, min(9, $level))), true);
256 return $id;
257 }
258
259 /**
260 * 短いIDを解決する
261 *
262 * @return string|null 一意に決まれば完全なID
263 */
264 public function resolvePrefix(string $prefix): ?string {
265 $prefix = strtolower($prefix);
266 if (!preg_match('/^[0-9a-f]{4,}$/', $prefix) || strlen($prefix) > $this->repo->hexLen) return null;
267 if (strlen($prefix) === $this->repo->hexLen) return $this->exists($prefix) ? $prefix : null;
268 $found = [];
269 foreach ($this->dirs as $dir) {
270 foreach (glob($dir.'/'.substr($prefix, 0, 2).'/'.substr($prefix, 2).'*') ?: [] as $f) $found[substr($prefix, 0, 2).basename($f)] = true;
271 }
272 foreach ($this->packs(true) as $pack) foreach ($pack->findPrefix($prefix) as $id) $found[$id] = true;
273 return count($found) === 1 ? (string)array_key_first($found) : null;
274 }
275
276 /** 全部のID（重複あり得る） */
277 public function ids(): \Generator {
278 foreach ($this->dirs as $dir) {
279 foreach (glob($dir.'/[0-9a-f][0-9a-f]/*') ?: [] as $f) {
280 $id = basename(dirname($f)).basename($f);
281 if ($this->repo->isId($id)) yield $id;
282 }
283 }
284 foreach ($this->packs(true) as $pack) yield from $pack->ids();
285 }
286
287 public function dir(): string {
288 return $this->dirs[0];
289 }
290}