1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
declare(strict_types=1); |
4
|
|
|
|
5
|
|
|
namespace PhpCfdi\RfcLinc\Updater; |
6
|
|
|
|
7
|
|
|
use EngineWorks\ProgressStatus\NullProgress; |
8
|
|
|
use EngineWorks\ProgressStatus\ProgressInterface; |
9
|
|
|
use PhpCfdi\RfcLinc\DataGateway\FactoryInterface; |
10
|
|
|
use PhpCfdi\RfcLinc\Domain\Catalog; |
11
|
|
|
use PhpCfdi\RfcLinc\Domain\VersionDate; |
12
|
|
|
use PhpCfdi\RfcLinc\Downloader\DownloaderInterface; |
13
|
|
|
use PhpCfdi\RfcLinc\Downloader\PhpDownloader; |
14
|
|
|
use PhpCfdi\RfcLinc\Util\ReaderInterface; |
15
|
|
|
use PhpCfdi\RfcLinc\Util\TemporaryFilename; |
16
|
|
|
use Psr\Log\LoggerInterface; |
17
|
|
|
use Psr\Log\NullLogger; |
18
|
|
|
|
19
|
|
|
class Updater |
20
|
|
|
{ |
21
|
|
|
const URL_BLOBS_LIST = 'https://cfdisat.blob.core.windows.net/lco?restype=container&comp=list'; |
22
|
|
|
|
23
|
|
|
/** @var VersionDate */ |
24
|
|
|
private $date; |
25
|
|
|
|
26
|
|
|
/** @var FactoryInterface */ |
27
|
|
|
private $gateways; |
28
|
|
|
|
29
|
|
|
/** @var DownloaderInterface */ |
30
|
|
|
private $downloader; |
31
|
|
|
|
32
|
|
|
/** @var IndexInterpreter */ |
33
|
|
|
private $indexInterpreter; |
34
|
|
|
|
35
|
|
|
/** @var Importer|null */ |
36
|
|
|
private $importer; |
37
|
|
|
|
38
|
|
|
/** @var LoggerInterface */ |
39
|
|
|
private $logger; |
40
|
|
|
|
41
|
|
|
/** @var ProgressInterface */ |
42
|
|
|
private $progress; |
43
|
|
|
|
44
|
12 |
|
public function __construct(VersionDate $date, FactoryInterface $gateways) |
45
|
|
|
{ |
46
|
12 |
|
$this->date = $date; |
47
|
12 |
|
$this->gateways = $gateways; |
48
|
12 |
|
$this->downloader = new PhpDownloader(); |
49
|
12 |
|
$this->indexInterpreter = new IndexInterpreter(); |
50
|
12 |
|
$this->logger = new NullLogger(); |
51
|
12 |
|
$this->progress = new NullProgress(); |
52
|
12 |
|
} |
53
|
|
|
|
54
|
1 |
|
public function hasImporter(): bool |
55
|
|
|
{ |
56
|
1 |
|
return ($this->importer instanceof Importer); |
57
|
|
|
} |
58
|
|
|
|
59
|
3 |
|
public function importer(): Importer |
60
|
|
|
{ |
61
|
3 |
|
if ($this->importer instanceof Importer) { |
62
|
2 |
|
return $this->importer; |
63
|
|
|
} |
64
|
1 |
|
throw new \LogicException('There is no importer, did you call run() method?'); |
65
|
|
|
} |
66
|
|
|
|
67
|
2 |
|
public function catalog(): Catalog |
68
|
|
|
{ |
69
|
2 |
|
return $this->importer()->catalog(); |
70
|
|
|
} |
71
|
|
|
|
72
|
3 |
|
public function progress(): ProgressInterface |
73
|
|
|
{ |
74
|
3 |
|
return $this->progress; |
75
|
|
|
} |
76
|
|
|
|
77
|
1 |
|
public function date(): VersionDate |
78
|
|
|
{ |
79
|
1 |
|
return $this->date; |
80
|
|
|
} |
81
|
|
|
|
82
|
1 |
|
public function gateways(): FactoryInterface |
83
|
|
|
{ |
84
|
1 |
|
return $this->gateways; |
85
|
|
|
} |
86
|
|
|
|
87
|
2 |
|
public function downloader(): DownloaderInterface |
88
|
|
|
{ |
89
|
2 |
|
return $this->downloader; |
90
|
|
|
} |
91
|
|
|
|
92
|
2 |
|
public function indexInterpreter(): IndexInterpreter |
93
|
|
|
{ |
94
|
2 |
|
return $this->indexInterpreter; |
95
|
|
|
} |
96
|
|
|
|
97
|
2 |
|
public function logger(): LoggerInterface |
98
|
|
|
{ |
99
|
2 |
|
return $this->logger; |
100
|
|
|
} |
101
|
|
|
|
102
|
2 |
|
public function setDownloader(DownloaderInterface $downloader) |
103
|
|
|
{ |
104
|
2 |
|
$this->downloader = $downloader; |
105
|
2 |
|
} |
106
|
|
|
|
107
|
1 |
|
public function setIndexInterpreter(IndexInterpreter $indexInterpreter) |
108
|
|
|
{ |
109
|
1 |
|
$this->indexInterpreter = $indexInterpreter; |
110
|
1 |
|
} |
111
|
|
|
|
112
|
2 |
|
public function setLogger(LoggerInterface $logger) |
113
|
|
|
{ |
114
|
2 |
|
$this->logger = $logger; |
115
|
2 |
|
} |
116
|
|
|
|
117
|
2 |
|
public function setProgress(ProgressInterface $progress) |
118
|
|
|
{ |
119
|
2 |
|
$this->progress = $progress; |
120
|
2 |
|
} |
121
|
|
|
|
122
|
1 |
|
public function run(): int |
123
|
|
|
{ |
124
|
1 |
|
$indexUrl = $this->indexUrl(); |
125
|
1 |
|
$this->logger->info("Processing {$indexUrl}..."); |
126
|
|
|
|
127
|
1 |
|
$this->logger->debug("Downloading {$indexUrl}..."); |
128
|
1 |
|
$indexContents = $this->downloader->download($indexUrl); |
129
|
|
|
|
130
|
1 |
|
$this->logger->debug('Obtaining blobs...'); |
131
|
1 |
|
$blobs = $this->indexInterpreter->obtainBlobs($indexContents); |
132
|
1 |
|
$blobsCount = count($blobs); |
133
|
|
|
|
134
|
1 |
|
$this->logger->debug("Processing $blobsCount blobs..."); |
135
|
1 |
|
$processedLines = $this->runBlobs(...$blobs); |
136
|
|
|
|
137
|
1 |
|
$this->logger->info(sprintf('Processed %s lines', number_format($processedLines))); |
138
|
1 |
|
return $processedLines; |
139
|
|
|
} |
140
|
|
|
|
141
|
1 |
|
public function runBlobs(Blob ...$blobs): int |
142
|
|
|
{ |
143
|
1 |
|
$processedLines = 0; |
144
|
1 |
|
$this->processBegin(); |
145
|
1 |
|
foreach ($blobs as $blob) { |
146
|
1 |
|
$processedLines = $processedLines + $this->processBlob($blob); |
147
|
|
|
} |
148
|
1 |
|
$this->processEnd(); |
149
|
|
|
|
150
|
1 |
|
return $processedLines; |
151
|
|
|
} |
152
|
|
|
|
153
|
2 |
|
public function processBegin() |
154
|
|
|
{ |
155
|
2 |
|
$this->logger->notice('Starting general process...'); |
156
|
|
|
|
157
|
|
|
// obtain or create version |
158
|
2 |
|
$gwCatalogs = $this->gateways->catalog(); |
159
|
2 |
|
if ($gwCatalogs->exists($this->date)) { |
160
|
|
|
throw new \RuntimeException('The version is already in the catalog, it was not expected to exists'); |
161
|
|
|
} |
162
|
|
|
// start optimizations |
163
|
2 |
|
$this->gateways->optimizer()->prepare(); |
164
|
|
|
|
165
|
|
|
// create and store version |
166
|
2 |
|
$catalog = new Catalog($this->date, 0, 0, 0, 0); |
167
|
2 |
|
$gwCatalogs->insert($catalog); |
168
|
|
|
|
169
|
|
|
// create importer |
170
|
2 |
|
$this->importer = new Importer($catalog, $this->gateways, $this->progress()); |
171
|
|
|
|
172
|
|
|
// set all records as deleted |
173
|
2 |
|
$this->gateways->listedRfc()->markAllAsDeleted(); |
174
|
|
|
|
175
|
2 |
|
$this->logger->debug('General process started'); |
176
|
2 |
|
} |
177
|
|
|
|
178
|
1 |
|
public function processBlob(Blob $blob): int |
179
|
|
|
{ |
180
|
|
|
// create temp file |
181
|
1 |
|
$downloaded = new TemporaryFilename(); |
182
|
1 |
|
$filename = (string) $downloaded; |
183
|
1 |
|
$url = $blob->url(); |
184
|
1 |
|
$expectedMd5 = $blob->md5(); |
185
|
|
|
|
186
|
1 |
|
$this->logger->info("Downloading $url..."); |
187
|
|
|
|
188
|
|
|
// download the resourse |
189
|
1 |
|
$this->logger->debug("Downloading $url into $filename..."); |
190
|
1 |
|
$downloadStart = time(); |
191
|
1 |
|
$this->downloader->downloadAs($url, $filename); |
192
|
1 |
|
$downloadElapsed = time() - $downloadStart; |
193
|
1 |
|
$this->logger->debug("Download $url takes $downloadElapsed seconds"); |
194
|
|
|
|
195
|
|
|
// check the md5 checksum |
196
|
1 |
|
if ('' !== $expectedMd5) { |
197
|
1 |
|
$this->logger->debug("Checking $expectedMd5 on $filename..."); |
198
|
1 |
|
$this->checkFileMd5($filename, $expectedMd5); |
199
|
|
|
} |
200
|
|
|
|
201
|
|
|
// process file |
202
|
1 |
|
$this->logger->debug("Opening $filename (as packed data)..."); |
203
|
1 |
|
$reader = $this->createReaderForPackedFile($filename); |
204
|
1 |
|
$processedLines = $this->processReader($reader); |
205
|
1 |
|
$this->logger->debug("Closing $filename..."); |
206
|
1 |
|
$reader->close(); |
207
|
|
|
|
208
|
1 |
|
$this->logger->notice(sprintf('Blob %s process %s lines', $url, number_format($processedLines))); |
209
|
|
|
|
210
|
|
|
// clear the resource |
211
|
1 |
|
$this->logger->debug("Removing $filename..."); |
212
|
1 |
|
$downloaded->unlink(); |
213
|
|
|
|
214
|
1 |
|
return $processedLines; |
215
|
|
|
} |
216
|
|
|
|
217
|
2 |
|
public function processReader(ReaderInterface $reader): int |
218
|
|
|
{ |
219
|
2 |
|
return $this->importer()->importReader($reader); |
220
|
|
|
} |
221
|
|
|
|
222
|
2 |
|
public function processEnd() |
223
|
|
|
{ |
224
|
2 |
|
$importer = $this->importer(); |
225
|
2 |
|
$catalog = $importer->catalog(); |
226
|
2 |
|
$gwRfc = $this->gateways->listedRfc(); |
227
|
|
|
|
228
|
|
|
// count how many were deleted and log |
229
|
2 |
|
$this->logger->debug('Checking deletes...'); |
230
|
2 |
|
foreach ($gwRfc->eachDeleted() as $rfc) { |
231
|
1 |
|
$importer->performDelete($rfc); |
232
|
|
|
} |
233
|
2 |
|
$this->logger->debug(sprintf('Found %s lines deleted', number_format($catalog->deleted()))); |
234
|
|
|
|
235
|
2 |
|
$active = $gwRfc->countDeleted(false); |
236
|
2 |
|
$catalog->setRecords($active); |
237
|
2 |
|
$this->logger->info(sprintf('Found %s RFC active', number_format($active))); |
238
|
|
|
|
239
|
|
|
// store current version |
240
|
2 |
|
$this->logger->debug('Saving version...'); |
241
|
2 |
|
$this->gateways->catalog()->update($catalog); |
242
|
|
|
|
243
|
|
|
// end optimizations |
244
|
2 |
|
$this->gateways->optimizer()->finish(); |
245
|
2 |
|
$this->logger->notice('General process finish'); |
246
|
2 |
|
} |
247
|
|
|
|
248
|
2 |
|
public function checkFileMd5(string $filename, string $expectedMd5) |
249
|
|
|
{ |
250
|
|
|
// check md5 |
251
|
2 |
|
$md5file = (string) md5_file($filename); |
252
|
2 |
|
if ($md5file !== $expectedMd5) { |
253
|
1 |
|
throw new \RuntimeException(sprintf( |
254
|
1 |
|
'The MD5 from file "%s" does not match with "%s"', |
255
|
1 |
|
$md5file, |
256
|
1 |
|
$expectedMd5 |
257
|
|
|
)); |
258
|
|
|
} |
259
|
2 |
|
} |
260
|
|
|
|
261
|
1 |
|
public function createReaderForPackedFile(string $filename): ReaderInterface |
262
|
|
|
{ |
263
|
1 |
|
$reader = new PackedBlobReader(); |
264
|
1 |
|
$reader->open($filename); |
265
|
1 |
|
return $reader; |
266
|
|
|
} |
267
|
|
|
|
268
|
1 |
|
public function indexUrl(): string |
269
|
|
|
{ |
270
|
1 |
|
return $this->buildIndexUrl($this->date); |
271
|
|
|
} |
272
|
|
|
|
273
|
1 |
|
public static function buildIndexUrl(VersionDate $date): string |
274
|
|
|
{ |
275
|
1 |
|
return static::URL_BLOBS_LIST . '&prefix=l_RFC_' . $date->format('_'); |
276
|
|
|
} |
277
|
|
|
} |
278
|
|
|
|