first commit

models/logics/commands/Base/BaseCommandSpiderLogic.php

@@ -0,0 +1,38 @@
+<?php
+namespace app\models\logics\commands\base;
+
+use app\common\CurlApp;
+
+class BaseCommandSpiderLogic
+{
+
+    protected ?CurlApp $curl = null;
+
+    protected string $brandName = '';
+
+    protected string $baseUrl = '';
+
+    public function __construct()
+    {
+        ini_set('pcre.backtrack_limit', '-1');
+        $this->curl = new CurlApp();
+    }
+
+    public function exec()
+    {
+
+        $url = rtrim($this->baseUrl, '/') . '/fashion-shows/designer/' . $this->brandName;
+//        var_dump($url);
+//        list($request, $httpCode) = $curl->setUrl($url)->exec();
+        $request = $this->curl->setUrl($url)->exec();
+
+        if ($request) {
+            preg_match_all('/window.__PRELOADED_STATE__ = ([\s\S]*?);<\/script>/', $request, $matches);
+            $val = json_decode(current(end($matches)), true);
+            return $val['transformed']['runwayDesignerContent']['designerCollections'] ?? [];
+        } else {
+//            $this->logger->info('未找到数据.');
+            return [];
+        }
+    }
+}

models/logics/commands/SpiderVogue.php

@@ -0,0 +1,190 @@
+<?php
+
+namespace app\models\logics\commands;
+
+use app\common\CommonHelper;
+use app\enums\SourceEnum;
+use app\models\Brand;
+use app\models\BrandRunway;
+use app\models\BrandRunwayImages;
+use app\models\BrandSource;
+use app\models\logics\commands\base\BaseCommandSpiderLogic;
+use yii\helpers\ArrayHelper;
+
+class SpiderVogue extends BaseCommandSpiderLogic
+{
+    protected string $baseUrl = 'https://www.vogue.com/';
+
+    public function __construct()
+    {
+        parent::__construct();
+//        $this->curl->setProxy('127.0.0.1', 58591);
+    }
+
+    public function setBrandName(string $name = ''): SpiderVogue
+    {
+        $this->brandName = $name;
+        return $this;
+    }
+
+    public function start()
+    {
+        $source = BrandSource::find()->where(['source' => SourceEnum::VOGEU->value()])->asArray()->all();
+//        var_dump(ArrayHelper::getColumn($source, 'brand_id'));
+        $i = 0;
+        $brandNameContainer = [];
+        $brandIdContainer = [];
+        foreach (Brand::find()->where(
+            ['>','id', 5747] // 5897
+        )->each() as $item) {
+            $item['name'] = strtr($item['name'], [
+               ' ' => '-',
+               '.' => '-'
+            ]);
+//            var_dump(strtolower($item['name']));
+//            $runwayList = $this->setBrandName($item['name'])->_getBrandRunwayList(strtolower($item['name']));
+            echo $item['name'] . PHP_EOL;
+
+                $brandNameContainer[] = strtolower($item['name']);
+                $brandIdContainer[] = $item['id'];
+                $i++;
+
+                if ($i != 5) {
+                    continue;
+                }
+
+
+            $runwayList = $this->_getBrandRunwayListMulti($brandNameContainer);
+
+            foreach ($runwayList as $index => $listItem) {
+//                var_dump('id---', $brandIdContainer[$index]);
+//                var_dump($listItem);
+                if (!$listItem) {
+                    continue;
+                }
+                if (!$sourceModel = BrandSource::find()->where(['brand_id' => $brandIdContainer[$index], 'is_deleted' => 0])->one()) {
+                    $sourceModel = new BrandSource();
+                    $sourceModel->brand_id = $brandIdContainer[$index];
+                    $sourceModel->source = SourceEnum::VOGEU->value();
+                    $sourceModel->save();
+                    var_dump($sourceModel->errors);
+                }
+            }
+
+            $brandNameContainer = $brandIdContainer = [];
+            $i = 0;
+
+            if (!$runwayList) {
+                continue;
+            }
+
+
+            var_dump($runwayList);
+            foreach ($runwayList as $runwayItem) {
+                var_dump($runwayItem['hed']);
+                if (!$model = BrandRunway::find()->where(['brand_id' => $item['id'], 'title' => $runwayItem['hed']])->one()) {
+                    echo $runwayItem['hed'] . PHP_EOL;
+                    $model = new BrandRunway();
+                    $model->title = $runwayItem['hed'];
+                    $model->brand_id = $item['id'];
+                    $model->year = CommonHelper::getYear($runwayItem['hed']);
+                    $model->cover = '';
+                    $model->save();
+                    var_dump($model->errors);
+                    if (!$sourceModel = BrandSource::find()->where(['brand_id' => $item['id'], 'is_deleted' => 0])->one()) {
+                        $sourceModel = new BrandSource();
+                        $sourceModel->brand_id = $item['id'];
+                        $sourceModel->source = SourceEnum::VOGEU->value();
+                        $sourceModel->save();
+                        var_dump($sourceModel->errors);
+                    }
+
+                    $pageUri = $runwayItem['url'];
+                    $requestUrl = rtrim($this->baseUrl, '/') . $pageUri . '/slideshow/collection';
+                    $detailData = $this->_getDetail($requestUrl);
+
+                    preg_match_all('/window\.__PRELOADED_STATE__ = (.*?);</s', $detailData, $matches);
+
+                    $saveUrl = $detailUrl = [];
+                    if (count($matches) > 1) {
+                        $val = json_decode(current($matches[1]), true);
+                        $images = $val['transformed']['runwayGalleries']['galleries'][0]['items'] ?? false;
+
+                        if ($images === false) {
+                            echo '获取图片失败' . PHP_EOL;
+//                            $this->logger->warning($requestUrl . '获取图片失败.');
+                            return;
+                        }
+
+                        foreach (is_array($images) ? $images : [] as $img) {
+                            $saveUrl[] = [
+                                'src' => $img['image']['sources']['xxl']['url']
+                            ];
+                            foreach ($img['details'] ?? [] as $detail) {
+                                $detailUrl[] = ['src' => $detail['image']['sources']['xxl']['url']];
+                                $brandModel = new BrandRunwayImages();
+                                $brandModel->image = $detail['image']['sources']['xxl']['url'];
+                                $brandModel->runway_id = $model->id;
+                                $brandModel->brand_id = $item['id'];
+                                $brandModel->save();
+                                var_dump($brandModel->errors);
+                            }
+                        }
+
+//                        $model->images = json_encode($saveUrl);
+                        $model->cover = '';
+                        $model->save();
+                    }
+                }
+            }
+        }
+    }
+
+    public function getDetail($url): bool|string
+    {
+        return $this->curl->setUrl($url)->exec();
+    }
+
+
+    public function getBrandRunwayList(string $url)
+    {
+        $request = $this->curl->setUrl($url)->exec();
+
+        if ($request) {
+            preg_match_all('/window.__PRELOADED_STATE__ = ([\s\S]*?);<\/script>/', $request, $matches);
+            $val = json_decode(current(end($matches)), true);
+            return $val['transformed']['runwayDesignerContent']['designerCollections'] ?? [];
+        } else {
+//            $this->logger->info('未找到数据.');
+            return [];
+        }
+    }
+
+    private function _getBrandRunwayListMulti(array $brandNames = []): array
+    {
+//        $brandNames = [];
+        $resp = [];
+        foreach ($brandNames as $brandName) {
+            $url[] = rtrim($this->baseUrl, '/') . '/fashion-shows/designer/' . $brandName;
+//        var_dump($url);
+//        list($request, $httpCode) = $curl->setUrl($url)->exec();
+
+        }
+
+        $request = $this->curl->execMulti($url);
+//        var_dump($request);die;
+        foreach ($request as $item) {
+            if ($item) {
+                preg_match_all('/window.__PRELOADED_STATE__ = ([\s\S]*?);<\/script>/', $item, $matches);
+                $val = json_decode(current(end($matches)), true);
+                $resp[] = $val['transformed']['runwayDesignerContent']['designerCollections'] ?? [];
+            } else {
+//            $this->logger->info('未找到数据.');
+                $resp[] = [];
+            }
+        }
+
+        return $resp;
+    }
+
+}