// Check if Sister URL is available $SisterUrl = $SisterUrl_Ancher->getAttribute("href"); } // if echo "{$i} - {$Bk_ISBN} - {$BookTitle} - {$Author} - {$Edition} - {$BK_UsedPrice} - {$BK_NewPrice} - {$BK_DigitalPrice} - {$ImageUrl} <br /> {$SisterUrl} <br /><br />"; // Clearing Space unset($BookTitle); unset($SisterUrl); unset($Author); unset($Edition); unset($ImageUrl); unset($BK_UsedPrice); unset($BK_NewPrice); unset($BK_DigitalPrice); unset($Bk_ISBN); } // for } } // if } // Url Long books $url = "http://www.bkstr.com/webapp/wcs/stores/servlet/CourseMaterialsResultsView?catalogId=10001&categoryId=9604&storeId=10161&langId=-1&programId=562&termId=100014525&divisionDisplayName=Graduate%20School%20of%20Business&departmentDisplayName=BUS&courseDisplayName=GSB%20101§ionDisplayName=01&demoKey=null&purpose=browse"; // URL short Books $url = "http://www.bkstr.com/webapp/wcs/stores/servlet/CourseMaterialsResultsView?catalogId=10001&categoryId=9604&storeId=10161&langId=-1&programId=562&termId=100014525&divisionDisplayName=Stanford&departmentDisplayName=CHEM&courseDisplayName=130§ionDisplayName=01&demoKey=d&purpose=browse"; //$url = "http://www.bkstr.com/webapp/wcs/stores/servlet/CourseMaterialsResultsView?catalogId=10001&categoryId=9604&storeId=10161&langId=-1&programId=562&termId=100014525&divisionDisplayName=Graduate%20School%20of%20Business&departmentDisplayName=BUS&courseDisplayName=GSB%20101§ionDisplayName=01&demoKey=d&purpose=browse"; MainBookData($url); ?>
function ProcessDataDigging($file_name = "c:\\scrap\\book_data.csv") { //xdebug_start_trace(); $ProgramID = "647"; $TermID = "100013708"; $StoreId = "10742"; $output = fopen($file_name, 'w'); $row_data = "Program,Term,Department,Course,Section,Course URL,Book Title,BK Author,BK Edition,BK Image URL,BK Used Price,BK New Price,BK Digital Price,BK ISBN,Amazon List Price,Amazon Discount Price,Non Amazon New Price,Non Amazon Used Price,Amazon Detail Page URL,Detailed Link,Author(s),Edition,Publisher,ISBN (10),ISBN (13),ISBN (10) - Digi,ISBN (13) - Digi,List Price,You Pay Price\n"; fwrite($output, $row_data); $Division_Name = " "; $Division_Name_url = str_replace(" ", "%20", $Division_Name); // Corrects The URL Data, removes spaces $Department_arr = file_get_contents("http://www.bkstr.com/webapp/wcs/stores/servlet/LocateCourseMaterialsServlet?requestType=DEPARTMENTS&storeId={$StoreId}&programId={$ProgramID}&termId={$TermID}&divisionName={$Division_Name_url}&_="); $Department_arr = str_replace("<script>parent.doneLoaded('", "", $Department_arr); $Department_arr = str_replace("')</script>", "", $Department_arr); $Department_arr = json_decode($Department_arr, true); $Department_arr = $Department_arr['data'][0]; foreach ($Department_arr as $Department_Name => $Department_Value) { $Department_Name_url = str_replace(" ", "%20", $Department_Name); // Corrects The URL Data, removes spaces $Course_arr = file_get_contents("http://www.bkstr.com/webapp/wcs/stores/servlet/LocateCourseMaterialsServlet?requestType=COURSES&storeId={$StoreId}&programId={$ProgramID}&termId={$TermID}&divisionName={$Division_Name_url}&departmentName={$Department_Name_url}&_="); $Course_arr = str_replace("<script>parent.doneLoaded('", "", $Course_arr); $Course_arr = str_replace("')</script>", "", $Course_arr); $Course_arr = json_decode($Course_arr, true); $Course_arr = $Course_arr['data'][0]; foreach ($Course_arr as $Course_Name => $Course_Value) { $Course_Name_url = str_replace(" ", "%20", $Course_Name); // Corrects The URL Data, removes spaces $Section_arr = file_get_contents("http://www.bkstr.com/webapp/wcs/stores/servlet/LocateCourseMaterialsServlet?requestType=SECTIONS&storeId={$StoreId}&programId={$ProgramID}&termId={$TermID}&divisionName={$Division_Name_url}&departmentName={$Department_Name_url}&courseName={$Course_Name_url}&_="); $Section_arr = str_replace("<script>parent.doneLoaded('", "", $Section_arr); $Section_arr = str_replace("')</script>", "", $Section_arr); $Section_arr = json_decode($Section_arr, true); $Section_arr = $Section_arr['data'][0]; foreach ($Section_arr as $Section_Name => $Section_Value) { $Section_Name_url = str_replace(" ", "%20", $Section_Name); // $delay = rand(3, 5); // sleep($delay); $FinalUrl = "http://www.bkstr.com/webapp/wcs/stores/servlet/CourseMaterialsResultsView?catalogId=10001&categoryId=9604&storeId={$StoreId}&langId=-1&programId={$ProgramID}&termId={$TermID}&divisionDisplayName={$Division_Name_url}&departmentDisplayName={$Department_Name_url}&courseDisplayName={$Course_Name_url}§ionDisplayName={$Section_Name_url}&demoKey=null&purpose=browse"; $initial_csv_row_data = "Univ Of Illinois - Champaign,Spring 2010,{$Department_Name},{$Course_Name},{$Section_Name},{$FinalUrl}"; MainBookData($FinalUrl, $initial_csv_row_data, $output); echo "\n"; echo "Memory Usage = " . memory_get_usage() / (1024 * 1024) . "MB \n\n\n"; } // Section } // Course } // Department fclose($output); //xdebug_stop_trace(); }
function ProcessDataDigging_Generic($Store, $University_Name, $Campus_Name) { if (PHP_OS == "WINNT") { $file_name = "c:\\{$University_Name}({$Campus_Name}).csv"; } else { $file_name = "\\{$University_Name}({$Campus_Name}).csv"; } $output = fopen($file_name, 'w'); $row_data = "Program,Term,Department,Course,Section,Course URL,Book Title,BK Author,BK Edition,BK Image URL,BK Used Price,BK New Price,BK Digital Price,BK ISBN,Amazon List Price,Amazon Discount Price,Non Amazon New Price,Non Amazon Used Price,Amazon Detail Page URL,Detailed Link,Author(s),Edition,Publisher,ISBN (10),ISBN (13),ISBN (10) - Digi,ISBN (13) - Digi,List Price,You Pay Price\n"; fwrite($output, $row_data); $Program_arr = file_get_contents("http://www.bkstr.com/webapp/wcs/stores/servlet/LocateCourseMaterialsServlet?requestType=INITIAL&storeId={$Store}&demoKey=d&_="); $Program_arr = str_replace("<script>parent.doneLoaded('", "", $Program_arr); $Program_arr = str_replace("')</script>", "", $Program_arr); $Program_arr = json_decode($Program_arr, true); $Program_arr = $Program_arr['data'][0]; foreach ($Program_arr as $Program_Name => $Program_Value) { $Program_Name_url = str_replace(" ", "%20", $Program_Name); $term_arr = file_get_contents("http://www.bkstr.com/webapp/wcs/stores/servlet/LocateCourseMaterialsServlet?requestType=TERMS&storeId={$Store}&demoKey=d&programId={$Program_Value}&_="); $term_arr = str_replace("<script>parent.doneLoaded('", "", $term_arr); $term_arr = str_replace("')</script>", "", $term_arr); $term_arr = json_decode($term_arr, true); $term_arr = $term_arr['data'][0]; foreach ($term_arr as $term_Name => $term_Value) { $term_arr = str_replace(" ", "%20", $term_Name); $Division_arr = file_get_contents("http://www.bkstr.com/webapp/wcs/stores/servlet/LocateCourseMaterialsServlet?requestType=DIVISIONS&storeId={$Store}&demoKey=d&programId={$Program_Value}&termId={$term_Value}&_="); $Division_arr = str_replace("<script>parent.doneLoaded('", "", $Division_arr); $Division_arr = str_replace("')</script>", "", $Division_arr); $Division_arr = json_decode($Division_arr, true); $Division_arr = $Division_arr['data'][0]; if (!empty($Division_arr)) { foreach ($Division_arr as $Division_Name => $Division_Value) { $Division_Name_url = str_replace(" ", "%20", $Division_Name); // Corrects The URL Data, removes spaces $Department_arr = file_get_contents("http://www.bkstr.com/webapp/wcs/stores/servlet/LocateCourseMaterialsServlet?requestType=DEPARTMENTS&storeId={$Store}&programId={$Program_Value}&termId={$term_Value}&divisionName={$Division_Name_url}&_="); $Department_arr = str_replace("<script>parent.doneLoaded('", "", $Department_arr); $Department_arr = str_replace("')</script>", "", $Department_arr); $Department_arr = json_decode($Department_arr, true); $Department_arr = $Department_arr['data'][0]; foreach ($Department_arr as $Department_Name => $Department_Value) { $Department_Name_url = str_replace(" ", "%20", $Department_Name); // Corrects The URL Data, removes spaces $Course_arr = file_get_contents("http://www.bkstr.com/webapp/wcs/stores/servlet/LocateCourseMaterialsServlet?requestType=COURSES&storeId={$Store}&programId={$Program_Value}&termId={$term_Value}&divisionName={$Division_Name_url}&departmentName={$Department_Name_url}&_="); $Course_arr = str_replace("<script>parent.doneLoaded('", "", $Course_arr); $Course_arr = str_replace("')</script>", "", $Course_arr); $Course_arr = json_decode($Course_arr, true); $Course_arr = $Course_arr['data'][0]; foreach ($Course_arr as $Course_Name => $Course_Value) { $Course_Name_url = str_replace(" ", "%20", $Course_Name); // Corrects The URL Data, removes spaces $Section_arr = file_get_contents("http://www.bkstr.com/webapp/wcs/stores/servlet/LocateCourseMaterialsServlet?requestType=SECTIONS&storeId={$Store}&programId={$Program_Value}&termId={$term_Value}&divisionName={$Division_Name_url}&departmentName={$Department_Name_url}&courseName={$Course_Name_url}&_="); $Section_arr = str_replace("<script>parent.doneLoaded('", "", $Section_arr); $Section_arr = str_replace("')</script>", "", $Section_arr); $Section_arr = json_decode($Section_arr, true); $Section_arr = $Section_arr['data'][0]; foreach ($Section_arr as $Section_Name => $Section_Value) { $Section_Name_url = str_replace(" ", "%20", $Section_Name); // $delay = rand(3, 5); // sleep($delay); $FinalUrl = "http://www.bkstr.com/webapp/wcs/stores/servlet/CourseMaterialsResultsView?catalogId=10001&categoryId=9604&storeId={$Store}&langId=-1&programId={$Program_Value}&termId={$term_Value}&divisionDisplayName={$Division_Name_url}&departmentDisplayName={$Department_Name_url}&courseDisplayName={$Course_Name_url}§ionDisplayName={$Section_Name_url}&demoKey=null&purpose=browse"; $initial_csv_row_data = "{$Program_Name},{$term_Name},{$Division_Name},{$Department_Name},{$Course_Name},{$Section_Name},{$FinalUrl}"; MainBookData($FinalUrl, $initial_csv_row_data, $output); echo "\n"; echo "Memory Usage = " . memory_get_usage() / (1024 * 1024) . "MB \n\n\n"; } // Section } // Course } // Department } } else { $Division_Name = " "; $Division_Name_url = str_replace(" ", "%20", $Division_Name); // Corrects The URL Data, removes spaces $Department_arr = file_get_contents("http://www.bkstr.com/webapp/wcs/stores/servlet/LocateCourseMaterialsServlet?requestType=DEPARTMENTS&storeId={$Store}&programId={$Program_Value}&termId={$term_Value}&divisionName={$Division_Name_url}&_="); $Department_arr = str_replace("<script>parent.doneLoaded('", "", $Department_arr); $Department_arr = str_replace("')</script>", "", $Department_arr); $Department_arr = json_decode($Department_arr, true); $Department_arr = $Department_arr['data'][0]; foreach ($Department_arr as $Department_Name => $Department_Value) { $Department_Name_url = str_replace(" ", "%20", $Department_Name); // Corrects The URL Data, removes spaces $Course_arr = file_get_contents("http://www.bkstr.com/webapp/wcs/stores/servlet/LocateCourseMaterialsServlet?requestType=COURSES&storeId={$Store}&programId={$Program_Value}&termId={$term_Value}&divisionName={$Division_Name_url}&departmentName={$Department_Name_url}&_="); $Course_arr = str_replace("<script>parent.doneLoaded('", "", $Course_arr); $Course_arr = str_replace("')</script>", "", $Course_arr); $Course_arr = json_decode($Course_arr, true); $Course_arr = $Course_arr['data'][0]; foreach ($Course_arr as $Course_Name => $Course_Value) { $Course_Name_url = str_replace(" ", "%20", $Course_Name); // Corrects The URL Data, removes spaces $Section_arr = file_get_contents("http://www.bkstr.com/webapp/wcs/stores/servlet/LocateCourseMaterialsServlet?requestType=SECTIONS&storeId={$Store}&programId={$Program_Value}&termId={$term_Value}&divisionName={$Division_Name_url}&departmentName={$Department_Name_url}&courseName={$Course_Name_url}&_="); $Section_arr = str_replace("<script>parent.doneLoaded('", "", $Section_arr); $Section_arr = str_replace("')</script>", "", $Section_arr); $Section_arr = json_decode($Section_arr, true); $Section_arr = $Section_arr['data'][0]; foreach ($Section_arr as $Section_Name => $Section_Value) { $Section_Name_url = str_replace(" ", "%20", $Section_Name); // $delay = rand(3, 5); // sleep($delay); $FinalUrl = "http://www.bkstr.com/webapp/wcs/stores/servlet/CourseMaterialsResultsView?catalogId=10001&categoryId=9604&storeId={$Store}&langId=-1&programId={$Program_Value}&termId={$term_Value}&divisionDisplayName={$Division_Name_url}&departmentDisplayName={$Department_Name_url}&courseDisplayName={$Course_Name_url}§ionDisplayName={$Section_Name_url}&demoKey=null&purpose=browse"; $initial_csv_row_data = "{$Program_Name},{$term_Name},Spring 2010,{$Department_Name},{$Course_Name},{$Section_Name},{$FinalUrl}"; MainBookData($FinalUrl, $initial_csv_row_data, $output); echo "\n"; echo "Memory Usage = " . memory_get_usage() / (1024 * 1024) . "MB \n\n\n"; } // Section } // Course } // Department } // ELSE to check If Division is Zero } } fclose($output); }