1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace PhpTek\Exodus\Model; |
4
|
|
|
|
5
|
|
|
use Page; |
|
|
|
|
6
|
|
|
use PhpTek\Exodus\Tool\StaticSiteMimeProcessor; |
7
|
|
|
use SilverStripe\Core\ClassInfo; |
8
|
|
|
use SilverStripe\ORM\DataObject; |
9
|
|
|
use SilverStripe\Forms\LiteralField; |
10
|
|
|
use SilverStripe\Forms\GridField\GridFieldAddNewButton; |
11
|
|
|
use SilverStripe\Forms\GridField\GridFieldAddExistingAutocompleter; |
12
|
|
|
use SilverStripe\Assets\File; |
13
|
|
|
use SilverStripe\ORM\FieldType\DBInt; |
14
|
|
|
use SilverStripe\ORM\FieldType\DBText; |
15
|
|
|
use SilverStripe\ORM\FieldType\DBVarchar; |
16
|
|
|
use SilverStripe\Forms\TextareaField; |
17
|
|
|
use SilverStripe\Forms\DropdownField; |
18
|
|
|
use SilverStripe\ORM\ValidationResult; |
19
|
|
|
use SilverStripe\Dev\Debug; |
20
|
|
|
|
21
|
|
|
/** |
22
|
|
|
* Represents a single import-rule that applies to some or all of the content to be imported. |
23
|
|
|
*/ |
24
|
|
|
class StaticSiteContentSourceImportSchema extends DataObject |
25
|
|
|
{ |
26
|
|
|
/** |
27
|
|
|
* Default |
28
|
|
|
* |
29
|
|
|
* @var string |
30
|
|
|
*/ |
31
|
|
|
private static $default_applies_to = '.*'; |
32
|
|
|
|
33
|
|
|
/** |
34
|
|
|
* @var string |
35
|
|
|
*/ |
36
|
|
|
private static $table_name = 'StaticSiteContentSourceImportSchema'; |
|
|
|
|
37
|
|
|
|
38
|
|
|
/** |
39
|
|
|
* |
40
|
|
|
* @var array |
41
|
|
|
*/ |
42
|
|
|
private static $db = [ |
|
|
|
|
43
|
|
|
"DataType" => DBVarchar::class, |
44
|
|
|
"Order" => DBInt::class, |
45
|
|
|
"AppliesTo" => DBVarchar::class, |
46
|
|
|
"MimeTypes" => DBText::class, |
47
|
|
|
"Notes" => DBText::class, // Purely informational. Not used in imports. |
48
|
|
|
]; |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* |
52
|
|
|
* @var array |
53
|
|
|
*/ |
54
|
|
|
private static $summary_fields = [ |
|
|
|
|
55
|
|
|
"AppliesTo", |
56
|
|
|
"DataType", |
57
|
|
|
"Order", |
58
|
|
|
]; |
59
|
|
|
|
60
|
|
|
/** |
61
|
|
|
* |
62
|
|
|
* @var array |
63
|
|
|
*/ |
64
|
|
|
private static $field_labels = [ |
|
|
|
|
65
|
|
|
"AppliesTo" => "URL Pattern", |
66
|
|
|
"DataType" => "Data Type", |
67
|
|
|
"Order" => "Priority", |
68
|
|
|
"MimeTypes" => "Mime-types", |
69
|
|
|
]; |
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* |
73
|
|
|
* @var string |
74
|
|
|
*/ |
75
|
|
|
private static $default_sort = "Order"; |
|
|
|
|
76
|
|
|
|
77
|
|
|
/** |
78
|
|
|
* |
79
|
|
|
* @var array |
80
|
|
|
*/ |
81
|
|
|
private static $has_one = [ |
|
|
|
|
82
|
|
|
"ContentSource" => StaticSiteContentSource::class, |
83
|
|
|
]; |
84
|
|
|
|
85
|
|
|
/** |
86
|
|
|
* |
87
|
|
|
* @var array |
88
|
|
|
*/ |
89
|
|
|
private static $has_many = [ |
|
|
|
|
90
|
|
|
"ImportRules" => StaticSiteContentSourceImportRule::class, |
91
|
|
|
]; |
92
|
|
|
|
93
|
|
|
/** |
94
|
|
|
* Used as the title in the CMS. |
95
|
|
|
* |
96
|
|
|
* @return string |
97
|
|
|
*/ |
98
|
|
|
public function getTitle(): string |
99
|
|
|
{ |
100
|
|
|
$type = ClassInfo::shortName($this->DataType); |
101
|
|
|
|
102
|
|
|
return sprintf('%s (%s)', $type, $this->AppliesTo); |
103
|
|
|
} |
104
|
|
|
|
105
|
|
|
/** |
106
|
|
|
* |
107
|
|
|
* @return FieldList |
|
|
|
|
108
|
|
|
*/ |
109
|
|
|
public function getCMSFields() |
110
|
|
|
{ |
111
|
|
|
$fields = parent::getCMSFields(); |
112
|
|
|
$fields->removeFieldFromTab('Root.Main', 'DataType'); |
113
|
|
|
$fields->removeByName('ContentSourceID'); |
114
|
|
|
$dataObjects = ClassInfo::subclassesFor(DataObject::class); |
115
|
|
|
|
116
|
|
|
array_shift($dataObjects); |
117
|
|
|
natcasesort($dataObjects); |
118
|
|
|
|
119
|
|
|
$fields->insertBefore('Order', LiteralField::create( |
120
|
|
|
'ImportIntro', |
121
|
|
|
'' |
122
|
|
|
. '<p class="message notice">Map MIME-Types to Silverstripe Data Types and' |
123
|
|
|
. ' relate them to one or more Import Rules (See below).</p>' |
124
|
|
|
)); |
125
|
|
|
$fields->dataFieldbyName('Order') |
126
|
|
|
->setDescription('Schema are assigned a higher priority through lower numbers.') |
127
|
|
|
->setAttribute('style', 'width: 100px'); |
128
|
|
|
|
129
|
|
|
$fields->dataFieldByName('AppliesTo') |
130
|
|
|
->setDescription('A full or partial URI whose content is suited to the Data Type selected below. Supports regular expressions.'); |
131
|
|
|
$fields->addFieldToTab('Root.Main', DropdownField::create('DataType', 'Data Type', $dataObjects)); |
132
|
|
|
$mimes = TextareaField::create('MimeTypes', 'Mime-types') |
133
|
|
|
->setRows(3) |
134
|
|
|
->setDescription('Be sure to pick a Mime-type that the above Data Type supports' |
135
|
|
|
. ' e.g. text/html for <strong>SiteTree</strong>,' |
136
|
|
|
. ' image/png, mage/jpeg, image/webp etc for <strong>Image</strong>' |
137
|
|
|
. ' or application/pdf, text/csv etc for <strong>File</strong>.' |
138
|
|
|
. ' Separate multiple Mimes by a newline.'); |
139
|
|
|
$fields->addFieldToTab('Root.Main', $mimes); |
140
|
|
|
$notes = TextareaField::create('Notes', 'Notes') |
141
|
|
|
->setDescription('Use this field to add any notes about this schema.' |
142
|
|
|
. ' (Purely informational. Data is not used in imports).'); |
143
|
|
|
$fields->addFieldToTab('Root.Main', $notes); |
144
|
|
|
|
145
|
|
|
$importRules = $fields->dataFieldByName('ImportRules'); |
146
|
|
|
$fields->removeFieldFromTab('Root', 'ImportRules'); |
147
|
|
|
$fields->dataFieldByName('DataType')->setDescription('' |
148
|
|
|
. 'The Silverstripe content class with which content from the selected' |
149
|
|
|
. ' Mime-Types will be associated.'); |
150
|
|
|
|
151
|
|
|
// Don't show for File subclasses, these obviously don't require CSS-based import rules |
152
|
|
|
if ($this->DataType && in_array(File::class, ClassInfo::ancestry($this->DataType))) { |
153
|
|
|
return $fields; |
|
|
|
|
154
|
|
|
} |
155
|
|
|
|
156
|
|
|
if ($importRules) { |
157
|
|
|
$conf = $importRules->getConfig(); |
158
|
|
|
$conf->removeComponentsByType([ |
159
|
|
|
GridFieldAddExistingAutocompleter::class, |
160
|
|
|
GridFieldAddNewButton::class |
161
|
|
|
]); |
162
|
|
|
$addNewButton = (new GridFieldAddNewButton('before'))->setButtonName("Add Rule"); |
163
|
|
|
$conf->addComponent($addNewButton); |
164
|
|
|
$fields->addFieldToTab('Root.Main', $importRules); |
165
|
|
|
$fields->insertBefore('ImportRules', LiteralField::create( |
166
|
|
|
'ImportRuleIntro', |
167
|
|
|
'' |
168
|
|
|
. '<p class="message notice">An import rule determines how content located at crawled URLs' |
169
|
|
|
. ' should be imported into a Data Type\'s fields with the use of CSS selectors.' |
170
|
|
|
. ' Where more than one schema exists for the same Data Type, they\'ll be processed in the order of Priority,' |
171
|
|
|
. ' where the first one to match a URI and Mime combination is the one that will be used.<p>' |
172
|
|
|
)); |
173
|
|
|
} |
174
|
|
|
|
175
|
|
|
return $fields; |
|
|
|
|
176
|
|
|
} |
177
|
|
|
|
178
|
|
|
/** |
179
|
|
|
* |
180
|
|
|
* @return void |
181
|
|
|
*/ |
182
|
|
|
public function requireDefaultRecords() |
183
|
|
|
{ |
184
|
|
|
foreach (StaticSiteContentSource::get() as $source) { |
185
|
|
|
if (!$source->Schemas()->count()) { |
186
|
|
|
Debug::message("Making a schema for $source->ID"); |
187
|
|
|
$defaultSchema = StaticSiteContentSourceImportSchema::create(); |
188
|
|
|
$defaultSchema->Order = 1000000; |
189
|
|
|
$defaultSchema->AppliesTo = self::$default_applies_to; |
190
|
|
|
$defaultSchema->DataType = Page::class; |
191
|
|
|
$defaultSchema->ContentSourceID = $source->ID; |
192
|
|
|
$defaultSchema->MimeTypes = "text/html"; |
193
|
|
|
$defaultSchema->write(); |
194
|
|
|
|
195
|
|
|
foreach (StaticSiteContentSourceImportRule::get()->filter(['SchemaID' => 0]) as $rule) { |
196
|
|
|
$rule->SchemaID = $defaultSchema->ID; |
197
|
|
|
$rule->write(); |
198
|
|
|
} |
199
|
|
|
} |
200
|
|
|
} |
201
|
|
|
} |
202
|
|
|
|
203
|
|
|
/** |
204
|
|
|
* Return the import rules in a format suitable for configuring StaticSiteContentExtractor. |
205
|
|
|
* |
206
|
|
|
* @return array $output. A map of field name => [CSS selector, CSS selector, ...] |
207
|
|
|
*/ |
208
|
|
|
public function getImportRules(): array |
209
|
|
|
{ |
210
|
|
|
$output = []; |
211
|
|
|
|
212
|
|
|
foreach ($this->ImportRules() as $rule) { |
|
|
|
|
213
|
|
|
if (!isset($output[$rule->FieldName])) { |
214
|
|
|
$output[$rule->FieldName] = []; |
215
|
|
|
} |
216
|
|
|
|
217
|
|
|
$ruleArray = [ |
218
|
|
|
'selector' => trim((string) $rule->CSSSelector), |
219
|
|
|
'attribute' => $rule->Attribute, |
220
|
|
|
'plaintext' => $rule->PlainText, |
221
|
|
|
'excludeselectors' => preg_split("#\n#", trim((string) $rule->ExcludeCSSSelector)), |
222
|
|
|
'outerhtml' => $rule->OuterHTML, |
223
|
|
|
]; |
224
|
|
|
|
225
|
|
|
$output[$rule->FieldName][] = $ruleArray; |
226
|
|
|
} |
227
|
|
|
|
228
|
|
|
return $output; |
229
|
|
|
} |
230
|
|
|
|
231
|
|
|
/** |
232
|
|
|
* |
233
|
|
|
* @return \ValidationResult |
|
|
|
|
234
|
|
|
*/ |
235
|
|
|
public function validate() |
236
|
|
|
{ |
237
|
|
|
$result = ValidationResult::create(); |
238
|
|
|
$mime = $this->validateMimes(); |
239
|
|
|
$appliesTo = $this->validateUrlPattern(); |
240
|
|
|
|
241
|
|
|
if (!is_bool($mime)) { |
242
|
|
|
$result->addError('Invalid Mime-type "' . $mime . '" for DataType "' . $this->DataType . '"'); |
243
|
|
|
} |
244
|
|
|
|
245
|
|
|
if (!is_bool($appliesTo)) { |
246
|
|
|
$result->addError('Invalid PCRE expression "' . $appliesTo . '"'); |
247
|
|
|
} |
248
|
|
|
|
249
|
|
|
return $result; |
250
|
|
|
} |
251
|
|
|
|
252
|
|
|
/** |
253
|
|
|
* |
254
|
|
|
* Validate user-inputted mime-types until we use some sort of multi-select list in the CMS to select from (@todo). |
255
|
|
|
* |
256
|
|
|
* @return mixed boolean|string Boolean true if all is OK, otherwise the invalid mimeType to be shown in the CMS UI |
257
|
|
|
*/ |
258
|
|
|
public function validateMimes() |
259
|
|
|
{ |
260
|
|
|
$selectedMimes = StaticSiteMimeProcessor::get_mimetypes_from_text($this->MimeTypes); |
|
|
|
|
261
|
|
|
|
262
|
|
|
$dt = $this->DataType ?? $_POST['DataType']; // @todo |
263
|
|
|
if (!$dt) { |
264
|
|
|
return true; // probably just creating |
265
|
|
|
} |
266
|
|
|
|
267
|
|
|
/* |
268
|
|
|
* This is v.sketch. It relies on the name of user-entered DataTypes containing |
269
|
|
|
* the string we want to match on in its classname = bad |
270
|
|
|
* @todo prolly just replace this wih a regex.. |
271
|
|
|
*/ |
272
|
|
|
switch ($dt) { |
273
|
|
|
case stristr($dt, 'image') !== false: |
274
|
|
|
$type = 'image'; |
275
|
|
|
break; |
276
|
|
|
case stristr($dt, 'file') !== false: |
277
|
|
|
$type = 'file'; |
278
|
|
|
break; |
279
|
|
|
case stristr($dt, 'page') !== false: |
280
|
|
|
default: |
281
|
|
|
$type = 'sitetree'; |
282
|
|
|
break; |
283
|
|
|
} |
284
|
|
|
|
285
|
|
|
$mimesForSSType = StaticSiteMimeProcessor::get_mime_for_ss_type($type); |
286
|
|
|
$mimes = $mimesForSSType ? $mimesForSSType : []; |
287
|
|
|
|
288
|
|
|
foreach ($mimes as $mime) { |
289
|
|
|
if (!in_array($mime, $mimesForSSType)) { |
290
|
|
|
return $mime; |
291
|
|
|
} |
292
|
|
|
} |
293
|
|
|
|
294
|
|
|
return true; |
295
|
|
|
} |
296
|
|
|
|
297
|
|
|
/** |
298
|
|
|
* |
299
|
|
|
* Prevent ugly CMS console errors if user-defined regex's are not 100% PCRE compatible. |
300
|
|
|
* |
301
|
|
|
* @return mixed string | boolean |
302
|
|
|
*/ |
303
|
|
|
public function validateUrlPattern() |
304
|
|
|
{ |
305
|
|
|
// Basic check uses negative lookbehind and checks if glob chars exist which are _not_ preceeded by a '.' char |
306
|
|
|
if (preg_match("#(?<!.)(\+|\*)#", $this->AppliesTo)) { |
307
|
|
|
return $this->AppliesTo; |
308
|
|
|
} |
309
|
|
|
|
310
|
|
|
return true; |
311
|
|
|
} |
312
|
|
|
} |
313
|
|
|
|
The issue could also be caused by a filter entry in the build configuration. If the path has been excluded in your configuration, e.g.
excluded_paths: ["lib/*"]
, you can move it to the dependency path list as follows:For further information see https://scrutinizer-ci.com/docs/tools/php/php-scrutinizer/#list-dependency-paths