Subversion Repositories cheapmusic

<?php

error_reporting(E_ALL);

// Get itunes listings

function get_amazon_scrape($query, $searchCondition) {

    $arr = [];

    $products = [];

    libxml_use_internal_errors(true);

    $html = getUrl("https://www.amazon.com/s?k=" . rawurlencode($query) . "&sf=qz&unfiltered=1&ref=nb_sb_noss");

    $dom = new DOMDocument;

    $dom->loadHTML($html);

    $xpath = new DOMXPath($dom);

    $nodes = $xpath->query('//a/@href');

    foreach($nodes as $href) {

        if (strpos($href->nodeValue, "/gp/offer-listing/B") === 0) {

            $products[] = $href->nodeValue;

        }

    }

    $productCnt = 0;

    foreach($products as $product) {

        $url = "https://www.amazon.com" . $product . "&tag=uj024-20&language=en_US";

        $asin = explode('/', $product)[3];

        $html = getUrl("https://www.amazon.com/dp/" . $asin, "Mozilla/5.0 (Windows NT 10.0; …) Gecko/20100101 Firefox/70.0");

        $dom = new DOMDocument;

        $dom->loadHTML($html);

        $xpath = new DOMXPath($dom);

        $nodes = $xpath->query('//table[@id="productDetailsTable"]//ul/li');

        if ($nodes->length < 1 || ++$productCnt > 5) {

            continue;

        }

        $str = trim($nodes->item(0)->nodeValue);

        $p = strpos($str, " (");

        $format = ($p > 0 ? substr($str, 0, $p) : $str);

        foreach($nodes as $node) {

            $str = trim($node->nodeValue);

            if (strpos($str, "Amazon Best Sellers Rank:") === 0) {

                $p = strpos($str, " (");

                $rank = substr($str, 11, $p-11);

            }

        }

        if (strpos($format, "Audio CD") === 0 ||

            strpos($format, "Vinyl") === 0 ||

            strpos($format, "Hardcover") === 0 ||

            strpos($format, "Paperback") === 0) {

            if (strpos($format, "Audio CD") !== false) {

                $mediaType = "CD";

            } else if (strpos($format, "Vinyl") !== false) {

                $mediaType = "Record";

            } else if (strpos($format, "Paperback") !== false ||

                       strpos($format, "Sheet") !== false ||

                       strpos($format, "Hardcover") !== false) {

                $mediaType = "Book";

            }

            $str = "https://www.amazon.com" . $product;

            $html = getUrl($str, "Mozilla/5.0 (Windows NT 10.0; …) Gecko/20100101 Firefox/70.0");

            $dom = new DOMDocument;

            $dom->loadHTML($html);

            $xpath = new DOMXPath($dom);

            $nodes = $xpath->query('//div[@id="olpProductImage"]//img');

            $pic = $nodes->item(0)->getAttribute("src");

            $nodes = $xpath->query('//div[@id="olpProductDetails"]/h1');

            $title = trim($nodes->item(0)->nodeValue);

            $nodes = $xpath->query('//div[@id="olpProductByline"]');

            if ($nodes->length > 0) {

                $artists = trim($nodes->item(0)->nodeValue);

                $title .= " " . $artists;

            }

            $listings = $xpath->query('//div[contains(concat(" ", normalize-space(@class), " "), " olpOffer ")]');

            foreach($listings as $listing) {

                $nodes = $xpath->query('.//h3[contains(concat(" ", normalize-space(@class), " "), " olpSellerName ")]', $listing);

                $str = trim($nodes->item(0)->nodeValue);

                $sellerName = (empty($str) ? "Amazon" : $str);

                $merchantName = "Amazon";

                $feedbackPercent = -1;

                $feedbackScore = -1;

                if ($sellerName != "Amazon") {

                    $merchantName .= " Marketplace";

                    $nodes = $xpath->query('.//div[contains(concat(" ", normalize-space(@class), " "), " olpSellerColumn ")]//p', $listing);

                    if ($nodes->length > 0) {

                        $str = trim($nodes->item(0)->nodeValue);

                        $sellerrating = substr($str, 17);

                        $num = preg_match_all('/((?:[0-9]+,)*[0-9]+(?:\.[0-9]+)?)/', $sellerrating, $matches);

                        if ($num == 3) {

                            $feedbackPercent = (int)$matches[0][0];

                            $feedbackScore = (int)str_replace( ',', '', $matches[0][2]);

                        }

                    }

                }

                $nodes = $xpath->query('.//span[contains(concat(" ", normalize-space(@class), " "), " olpCondition ")]', $listing);

                $str = trim($nodes->item(0)->nodeValue);

                $pos = strpos($str, " - ");

                if ($pos !== false) {

                    $condition = trim(substr($str, 0, $pos));

                    $detailCondition = trim(substr($str, $pos+3));

                } else {

                    $condition = $str;

                    $detailCondition = $str;

                }

                if ($condition == "Collectible" || $condition == "Refurbished") {

                    $condition = 'Used';

                }

                $nodes = $xpath->query('.//div[contains(concat(" ", normalize-space(@class), " "), " olpConditionColumn ")]//div[contains(concat(" ", normalize-space(@class), " "), " comments ")]', $listing);

                if ($nodes->length > 0) {

                    $conditionComment = trim($nodes->item(0)->nodeValue);

                }

                $nodes = $xpath->query('.//div[contains(concat(" ", normalize-space(@class), " "), " olpPriceColumn ")]//span[contains(concat(" ", normalize-space(@class), " "), " olpOfferPrice ")]', $listing);

                $price = substr(trim($nodes->item(0)->nodeValue), 1);

                $currency = 'USD';

                $nodes = $xpath->query('.//div[contains(concat(" ", normalize-space(@class), " "), " olpPriceColumn ")]//span[contains(concat(" ", normalize-space(@class), " "), " olpShippingPrice ")]', $listing);

                if ($nodes->length > 0) {

                    $shippingCost = substr(trim($nodes->item(0)->nodeValue), 1);

                    $shippingCurrency = 'USD';

                    $freeShippingCap = 0;

                } else {

                    $shippingCost = 0.00;

                    $shippingCurrency = 'USD';

                    $nodes = $xpath->query('.//div[contains(concat(" ", normalize-space(@class), " "), " olpPriceColumn ")]//p[contains(concat(" ", normalize-space(@class), " "), " olpShippingInfo ")]', $listing);

                    $str= trim($nodes->item(0)->nodeValue);

                    if (strpos($str, "FREE Shipping") !== false) {

                        $freeShippingCap = 0.00;

                    }

                    if (strpos($str, "on orders over") !== false) {

                        $freeShippingCap = 25.00;

                    }

                }

                $country = 'US';

                $nodes = $xpath->query('.//div[contains(concat(" ", normalize-space(@class), " "), " olpDeliveryColumn ")]//ul/li', $listing);

                foreach($nodes as $node) {

                    $str = trim($node->nodeValue);

                    if (strpos($str, "Ships from") === 0) {

                        $p = strpos($str, ".");

                        $country = getCountryCode(substr($str, 11, $p-11));

                    }

                }

                $nodes = $xpath->query('.//div[contains(concat(" ", normalize-space(@class), " "), " olpPriceColumn ")]//i[contains(concat(" ", normalize-space(@class), " "), " a-icon-prime ")]', $listing);

                if ($nodes->length > 0) {

                    $sellerName .= " Prime";

                }

                $arr[] = array(

                    "Merchant" => $merchantName,

                    "Condition" => $condition,

                    "Title" => $title,

                    "Barcode" => "",

                    "BarcodeType" => "",

                    "Image" => $pic,

                    "URL" => $url,

                    "MediaType" => $mediaType,

                    "DetailCondition" => $detailCondition,

                    "Country" => $country,

                    "BestOffer" => false,

                    "TimeLeft" => 0,

                    "Price" => $price,

                    "Currency" => $currency,

                    "ListingType" => "Fixed",

                    "Location" => "US",

                    "Zip" => "",

                    "FeedbackScore" => $feedbackScore,

                    "FeedbackPercent" => $feedbackPercent,

                    "SellerName" => $sellerName,

                    "HandlingTime" => 1,

                    "ShippingCost" => $shippingCost,

                    "ShippingEstimated" => false,

                    "ShippingCurrency" => $shippingCurrency,

                    "FreeShippingCap" => $freeShippingCap,

                    "Show" => true

                );

            }

        }

    }

    return ($arr);

}