|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace frictionlessdata\datapackage\Resources; |
|
4
|
|
|
|
|
5
|
|
|
use frictionlessdata\datapackage\DataStreams\BaseDataStream; |
|
6
|
|
|
use frictionlessdata\datapackage\Registry; |
|
7
|
|
|
use frictionlessdata\datapackage\Validators\ResourceValidationError; |
|
8
|
|
|
use frictionlessdata\datapackage\Validators\ResourceValidator; |
|
9
|
|
|
use frictionlessdata\datapackage\Exceptions\ResourceValidationFailedException; |
|
10
|
|
|
use frictionlessdata\datapackage\Utils; |
|
11
|
|
|
|
|
12
|
|
|
abstract class BaseResource implements \Iterator |
|
13
|
|
|
{ |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* BaseResource constructor. |
|
17
|
|
|
* |
|
18
|
|
|
* @param object $descriptor |
|
19
|
|
|
* @param null|string $basePath |
|
20
|
|
|
* |
|
21
|
|
|
* @param bool $skipValidations |
|
22
|
|
|
* |
|
23
|
|
|
* @throws \frictionlessdata\datapackage\Exceptions\ResourceValidationFailedException |
|
24
|
|
|
*/ |
|
25
|
|
|
public function __construct($descriptor, $basePath, $skipValidations = false) |
|
26
|
|
|
{ |
|
27
|
|
|
$this->basePath = $basePath; |
|
28
|
|
|
$this->descriptor = Utils::objectify($descriptor); |
|
29
|
|
|
$this->skipValidations = $skipValidations; |
|
30
|
|
|
if (!$this->skipValidations) { |
|
31
|
|
|
$validationErrors = $this->validateResource(); |
|
32
|
|
|
if (count($validationErrors) > 0) { |
|
33
|
|
|
throw new ResourceValidationFailedException($validationErrors); |
|
34
|
|
|
} |
|
35
|
|
|
} |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
public static function handlesDescriptor($descriptor) |
|
39
|
|
|
{ |
|
40
|
|
|
return static::handlesProfile(Registry::getResourceValidationProfile($descriptor)); |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
public function read($readOptions = null) |
|
44
|
|
|
{ |
|
45
|
|
|
$limit = ($readOptions && isset($readOptions['limit'])) ? $readOptions['limit'] : null; |
|
46
|
|
|
$rows = []; |
|
47
|
|
|
foreach ($this->dataStreams() as $dataStream) { |
|
|
|
|
|
|
48
|
|
|
if (isset($dataStream->table)) { |
|
49
|
|
|
$readOptions['limit'] = $limit; |
|
50
|
|
|
foreach ($dataStream->table->read($readOptions) as $row) { |
|
51
|
|
|
$rows[] = $row; |
|
52
|
|
|
if ($limit !== null) { |
|
53
|
|
|
--$limit; |
|
54
|
|
|
if ($limit < 0) { |
|
55
|
|
|
break; |
|
56
|
|
|
} |
|
57
|
|
|
} |
|
58
|
|
|
} |
|
59
|
|
|
} else { |
|
60
|
|
|
foreach ($dataStream as $row) { |
|
61
|
|
|
$rows[] = $row; |
|
62
|
|
|
if ($limit !== null) { |
|
63
|
|
|
--$limit; |
|
64
|
|
|
if ($limit < 0) { |
|
65
|
|
|
break; |
|
66
|
|
|
} |
|
67
|
|
|
} |
|
68
|
|
|
} |
|
69
|
|
|
} |
|
70
|
|
|
if ($limit !== null && $limit < 0) { |
|
71
|
|
|
break; |
|
72
|
|
|
} |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
return $rows; |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
/** |
|
79
|
|
|
* Loads $this->dataStreams based on $this->path() and $this->data |
|
80
|
|
|
* @return BaseDataStream[]|null |
|
81
|
|
|
*/ |
|
82
|
|
|
public function dataStreams() |
|
83
|
|
|
{ |
|
84
|
|
|
if (is_null($this->dataStreams)) { |
|
85
|
|
|
$this->dataStreams = []; |
|
86
|
|
|
foreach ($this->path() as $path) { |
|
87
|
|
|
$this->dataStreams[] = $this->getDataStream($path); |
|
88
|
|
|
} |
|
89
|
|
|
$data = $this->data(); |
|
90
|
|
|
if ($data) { |
|
91
|
|
|
$this->dataStreams[] = $this->getInlineDataStream($data); |
|
92
|
|
|
} |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
return $this->dataStreams; |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* @return object |
|
100
|
|
|
*/ |
|
101
|
|
|
public function descriptor() |
|
102
|
|
|
{ |
|
103
|
|
|
return $this->descriptor; |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
|
|
/** |
|
107
|
|
|
* @return string |
|
108
|
|
|
*/ |
|
109
|
|
|
public function name() |
|
110
|
|
|
{ |
|
111
|
|
|
return $this->descriptor()->name; |
|
112
|
|
|
} |
|
113
|
|
|
|
|
114
|
|
|
public function path() |
|
115
|
|
|
{ |
|
116
|
|
|
if (isset($this->descriptor()->path)) { |
|
117
|
|
|
$path = $this->descriptor()->path; |
|
118
|
|
|
if (!is_array($path)) { |
|
119
|
|
|
$path = [$path]; |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
return $path; |
|
123
|
|
|
} else { |
|
124
|
|
|
return []; |
|
125
|
|
|
} |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
/** |
|
129
|
|
|
* If the resource's $path is local (non-http) |
|
130
|
|
|
* @return bool |
|
131
|
|
|
*/ |
|
132
|
|
|
public function isLocal() |
|
133
|
|
|
{ |
|
134
|
|
|
return !$this->isRemote(); |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
/** |
|
138
|
|
|
* If the resource's $path is remote (http) |
|
139
|
|
|
* @return bool |
|
140
|
|
|
*/ |
|
141
|
|
|
public function isRemote() |
|
142
|
|
|
{ |
|
143
|
|
|
$path = $this->path(); |
|
144
|
|
|
return Utils::isHttpSource(is_array($path) && count($path) > 0 ? $path[0] : $path); |
|
145
|
|
|
} |
|
146
|
|
|
|
|
147
|
|
|
public function data() |
|
148
|
|
|
{ |
|
149
|
|
|
return isset($this->descriptor()->data) ? $this->descriptor()->data : null; |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
// standard iterator functions - to iterate over the data sources |
|
153
|
|
|
public function rewind() |
|
154
|
|
|
{ |
|
155
|
|
|
$this->dataStreams = null; |
|
156
|
|
|
$this->currentDataStream = 0; |
|
157
|
|
|
foreach ($this->dataStreams() as $dataStream) { |
|
|
|
|
|
|
158
|
|
|
$dataStream->rewind(); |
|
159
|
|
|
} |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
public function current() |
|
163
|
|
|
{ |
|
164
|
|
|
return $this->dataStreams()[$this->currentDataStream]->current(); |
|
165
|
|
|
} |
|
166
|
|
|
|
|
167
|
|
|
public function key() |
|
168
|
|
|
{ |
|
169
|
|
|
return $this->dataStreams()[$this->currentDataStream]->key(); |
|
170
|
|
|
} |
|
171
|
|
|
|
|
172
|
|
|
public function next() |
|
173
|
|
|
{ |
|
174
|
|
|
return $this->dataStreams()[$this->currentDataStream]->next(); |
|
175
|
|
|
} |
|
176
|
|
|
|
|
177
|
|
|
public function valid() |
|
178
|
|
|
{ |
|
179
|
|
|
$dataStreams = $this->dataStreams(); |
|
180
|
|
|
if ($dataStreams[$this->currentDataStream]->valid()) { |
|
181
|
|
|
// current data stream is still valid |
|
182
|
|
|
return true; |
|
183
|
|
|
} else { |
|
184
|
|
|
++$this->currentDataStream; |
|
185
|
|
|
if (isset($dataStreams[$this->currentDataStream])) { |
|
186
|
|
|
// current data stream is done, but we have another data stream |
|
187
|
|
|
return true; |
|
188
|
|
|
} else { |
|
189
|
|
|
// no more data and no more data streams |
|
190
|
|
|
return false; |
|
191
|
|
|
} |
|
192
|
|
|
} |
|
193
|
|
|
} |
|
194
|
|
|
|
|
195
|
|
|
public function getFileExtension() |
|
196
|
|
|
{ |
|
197
|
|
|
return ''; |
|
198
|
|
|
} |
|
199
|
|
|
|
|
200
|
|
|
public function save($baseFilename) |
|
201
|
|
|
{ |
|
202
|
|
|
$dataStreams = $this->dataStreams(); |
|
203
|
|
|
$numDataStreams = count($dataStreams); |
|
204
|
|
|
$fileNames = []; |
|
205
|
|
|
$i = 0; |
|
206
|
|
|
foreach ($dataStreams as $dataStream) { |
|
|
|
|
|
|
207
|
|
|
if ($numDataStreams == 1) { |
|
208
|
|
|
$filename = $baseFilename.$this->getFileExtension(); |
|
209
|
|
|
} else { |
|
210
|
|
|
$filename = $baseFilename.'-data-'.$i.$this->getFileExtension(); |
|
211
|
|
|
} |
|
212
|
|
|
$fileNames[] = $filename; |
|
213
|
|
|
$dataStream->save($filename); |
|
214
|
|
|
++$i; |
|
215
|
|
|
} |
|
216
|
|
|
|
|
217
|
|
|
return $fileNames; |
|
218
|
|
|
} |
|
219
|
|
|
|
|
220
|
|
|
public static function validateDataSource($dataSource, $basePath = null) |
|
221
|
|
|
{ |
|
222
|
|
|
$errors = []; |
|
223
|
|
|
$dataSource = static::normalizeDataSource($dataSource, $basePath); |
|
224
|
|
|
if (!Utils::isHttpSource($dataSource) && !file_exists($dataSource)) { |
|
225
|
|
|
$errors[] = new ResourceValidationError( |
|
226
|
|
|
ResourceValidationError::SCHEMA_VIOLATION, |
|
227
|
|
|
"data source file does not exist or is not readable: {$dataSource}" |
|
228
|
|
|
); |
|
229
|
|
|
} |
|
230
|
|
|
|
|
231
|
|
|
return $errors; |
|
232
|
|
|
} |
|
233
|
|
|
|
|
234
|
|
|
/** |
|
235
|
|
|
* allows extending classes to add custom sources |
|
236
|
|
|
* used by unit tests to add a mock http source. |
|
237
|
|
|
* |
|
238
|
|
|
* @param string $dataSource |
|
239
|
|
|
* @param string|null $basePath |
|
240
|
|
|
* |
|
241
|
|
|
* @return string |
|
242
|
|
|
*/ |
|
243
|
|
|
public static function normalizeDataSource($dataSource, string $basePath = null) |
|
244
|
|
|
{ |
|
245
|
|
|
if (!empty($basePath) && !Utils::isHttpSource($dataSource)) { |
|
246
|
|
|
// TODO: support JSON pointers |
|
247
|
|
|
$absPath = $basePath.DIRECTORY_SEPARATOR.$dataSource; |
|
248
|
|
|
if (file_exists($absPath)) { |
|
249
|
|
|
$dataSource = $absPath; |
|
250
|
|
|
} |
|
251
|
|
|
} |
|
252
|
|
|
|
|
253
|
|
|
return $dataSource; |
|
254
|
|
|
} |
|
255
|
|
|
|
|
256
|
|
|
protected $descriptor; |
|
257
|
|
|
protected $basePath; |
|
258
|
|
|
protected $skipValidations = false; |
|
259
|
|
|
protected $currentDataPosition = 0; |
|
260
|
|
|
protected $currentDataStream = 0; |
|
261
|
|
|
protected $dataStreams = null; |
|
262
|
|
|
|
|
263
|
|
|
protected function validateResource() |
|
264
|
|
|
{ |
|
265
|
|
|
return ResourceValidator::validate($this->descriptor(), $this->basePath); |
|
|
|
|
|
|
266
|
|
|
} |
|
267
|
|
|
|
|
268
|
|
|
/** |
|
269
|
|
|
* @param string $dataSource |
|
270
|
|
|
* |
|
271
|
|
|
* @return BaseDataStream |
|
272
|
|
|
*/ |
|
273
|
|
|
abstract protected function getDataStream($dataSource); |
|
274
|
|
|
|
|
275
|
|
|
abstract protected function getInlineDataStream($data); |
|
276
|
|
|
|
|
277
|
|
|
protected static function handlesProfile($profile) |
|
278
|
|
|
{ |
|
279
|
|
|
return false; |
|
280
|
|
|
} |
|
281
|
|
|
} |
|
282
|
|
|
|
There are different options of fixing this problem.
If you want to be on the safe side, you can add an additional type-check:
If you are sure that the expression is traversable, you might want to add a doc comment cast to improve IDE auto-completion and static analysis:
Mark the issue as a false-positive: Just hover the remove button, in the top-right corner of this issue for more options.