Mercurial > hg > ywww

comparison xml/getAmazonInfo.php @ 32:8130865e9e82

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
refactor to try to improve error handling
author Charlie Root
date Fri, 04 Jan 2019 09:50:55 -0500
parents 4124f103b46b
children c9d9b76ecbf9
comparison
equal deleted inserted replaced
31:4124f103b46b 32:8130865e9e82
1 <?php 1 <?php
2 2
3 $lastReqTime=0;
4 function getAmazonDet($isbn,$go,$localeIn) 3 function getAmazonDet($isbn,$go,$localeIn)
5 { 4 {
5 $Adefault=array(
6 'language' =>'en', //what language to render the page in
7 'locale' =>$localeIn, //which server's products? available: ca,de,fr,jp,uk,us
8 //'mode' =>'books', //what product category?
9 'page' =>1, //first page to show (we are counting from 1 not 0)
10 //'search' =>'Machiavelli', //what to search for?
11 'operation' =>'ItemLookup', //what to do? //ItemSearch
12 // 'searchindex' =>'Books', //what product category for search?
13 'searchparameter' =>'ItemId', //what kind of search?
14 'searchparameterdata'=>$isbn, //what to search for?
15 //here some debugging flags you can put at the end of the URL to call this script with, like: '?show_array=true'
16 'show_array' =>false, //debug: show complete incoming array? You can use this to see what other information Amazon is sending
17 'show_url' =>false, //debug: show XML request url to be send to Amazon?
18 'show_xml' =>false, //debug: show incoming XML code from Amazon?
19 );
20 //change the debug options to true if you want to activate them or call the script with '?show_array=true' to see what actual information you're getting from Amazon and how little my standard script is actually showing of it
21 //for all parameters see if the user has overruled it or use the default
22 foreach ($Adefault as $i=>$d) {
23 $$i=isset($_GET[$i])?$_GET[$i]:$d;
24 }
25 switch($locale){
26 case "us":
27 $loc = 0;
28 break;
29 case "uk":
30 $loc = 1;
31 break;
32 case "ca":
33 $loc = 2;
34 break;
35 case "de":
36 $loc = 3;
37 break;
38 case "fr":
39 $loc = 4;
40 break;
41 };
42
43 for ($i=1; $i<=3; $i++) {
44 try {
45 return getAmazonDetRemote($go,$language,$locale,$loc,
46 $page,$operation,$searchparameter,
47 $searchparameterdata,
48 $show_array,$show_url,$show_xml);
49 }
50 catch (Exception $code) {
51 file_put_contents('/var/ywww/debug/phpDebug',
52 "Bang: $code $i\n",FILE_APPEND);
53 if ($code=='RequestThrottled') {
54 usleep(200000); // Try to reduce throttling until we get a
55 // principled solution in place
56 $bail=False;
57 }
58 else {
59 $bail=True;
60 }
61 }
62 if ($bail) { break; }
63 }
64 return getAmazonDetLocal($go,$language,$loc,
65 $page,$operation,$searchparameter,
66 $searchparameterdata,
67 $show_array,$show_url,$show_xml);
68 }
69
70 function getAmazonDetRemote($go,$language,$locale,$loc,
71 $page,$operation,$searchparameter,
72 $searchparameterdata,
73 $show_array,$show_url,$show_xml)
74 {
6 global $output; 75 global $output;
7 $Adefault=array( 76
8 'language' =>'en', //what language to render the page in 77 $Aassociates_id=array(
9 'locale' =>$localeIn, //which server's products? available: ca,de,fr,jp,uk,us 78 'uk' => 'bookwhack-21',
10 //'mode' =>'books', //what product category? 79 'us' => 'your02b-20',
11 'page' =>1, //first page to show (we are counting from 1 not 0) 80 'ca' => 'book009-20',
12 //'search' =>'Machiavelli', //what to search for? 81 'de' => 'book04c-21',
13 'operation' =>'ItemLookup', //what to do? //ItemSearch 82 'fr' => 'book07f-21',
14 // 'searchindex' =>'Books', //what product category for search? 83 );
15 'searchparameter' =>'ItemId', //what kind of search? 84
16 'searchparameterdata'=>$isbn, //what to search for? 85 $Aserver=array(
17 //here some debugging flags you can put at the end of the URL to call this script with, like: '?show_array=true' 86 'ca' => array(
18 'show_array' =>false, //debug: show complete incoming array? You can use this to see what other information Amazon is sending 87 'ext' => 'ca' , //Canadian normal server
19 'show_url' =>false, //debug: show XML request url to be send to Amazon? 88 'nor' => 'http://www.amazon.ca' , //Canadian normal server
20 'show_xml' =>false, //debug: show incoming XML code from Amazon? 89 'xml' => 'http://xml.amazon.com' , //Canadian xml server
21 ); 90 ),
22 //change the debug options to true if you want to activate them or call the script with '?show_array=true' to see what actual information you're getting from Amazon and how little my standard script is actually showing of it 91 'de' => array(
23 92 'ext' => 'de' , //German normal server
24 $Aassociates_id=array( 93 'nor' => 'http://www.amazon.de' , //German normal server
25 'uk' => 'bookwhack-21', 94 'xml' => 'http://xml-eu.amazon.com', //German xml server
26 'us' => 'your02b-20', 95 ),
27 'ca' => 'book009-20', 96 'fr' => array(
28 'de' => 'book04c-21', 97 'ext' => 'fr' , //French normal server
29 'fr' => 'book07f-21', 98 'nor' => 'http://www.amazon.fr' , //French normal server
30 ); 99 'xml' => 'http://xml-eu.amazon.com', //French xml server
31 100 ),
32 $Aserver=array( 101 'jp' => array(
33 'ca' => array( 102 'ext' => 'jp' , //Japanese normal server, not co.jp!
34 'ext' => 'ca' , //Canadian normal server 103 'nor' => 'http://www.amazon.co.jp' , //Japanese normal server
35 'nor' => 'http://www.amazon.ca' , //Canadian normal server 104 'xml' => 'http://xml.amazon.com' , //Japanese xml server
36 'xml' => 'http://xml.amazon.com' , //Canadian xml server 105 ),
37 ), 106 'uk' => array(
38 'de' => array( 107 'ext' => 'co.uk' , //UK normal server
39 'ext' => 'de' , //German normal server 108 'nor' => 'http://www.amazon.co.uk' , //UK normal server
40 'nor' => 'http://www.amazon.de' , //German normal server 109 'xml' => 'http://xml-eu.amazon.com', //UK xml server
41 'xml' => 'http://xml-eu.amazon.com', //German xml server 110 ),
42 ), 111 'us' => array(
43 'fr' => array( 112 'ext' => 'com' , //USA normal server
44 'ext' => 'fr' , //French normal server 113 'nor' => 'http://www.amazon.com' , //USA normal server
45 'nor' => 'http://www.amazon.fr' , //French normal server 114 'xml' => 'http://xml.amazon.com' , //USA xml server
46 'xml' => 'http://xml-eu.amazon.com', //French xml server 115 ),
47 ), 116 );
48 'jp' => array( 117
49 'ext' => 'jp' , //Japanese normal server, not co.jp! 118 //if(go != 1)
50 'nor' => 'http://www.amazon.co.jp' , //Japanese normal server 119 //include "aws_signed_request.php";
51 'xml' => 'http://xml.amazon.com' , //Japanese xml server 120 $public_key ="AKIAIHTNWC7L6LOUY4LQ";
52 ), 121 $private_key="zWQlIzndJDtXNfxEXH7K7YR7hzv3u77lOcqfqPde";
53 'uk' => array( 122 //this is the data that is used to form the request for AWS
54 'ext' => 'co.uk' , //UK normal server 123 //this is the part that is search specific
55 'nor' => 'http://www.amazon.co.uk' , //UK normal server 124 $parameters=array(
56 'xml' => 'http://xml-eu.amazon.com', //UK xml server 125 'Operation' =>$operation ,
57 ), 126 //'Keywords' =>urlencode($search) ,
58 'us' => array( 127 //'SearchIndex' =>$searchindex , //Books for example.
59 'ext' => 'com' , //USA normal server 128 "$searchparameter"=>$searchparameterdata ,
60 'nor' => 'http://www.amazon.com' , //USA normal server 129 'ItemPage' =>$page , //which page?
61 'xml' => 'http://xml.amazon.com' , //USA xml server 130 'AssociateTag' =>$Aassociates_id[$locale],
62 ), 131 'ResponseGroup' =>'ItemAttributes,Reviews,EditorialReview,OfferSummary,Offers,Images,AlternateVersions,SalesRank,BrowseNodes' , //Small, Medium, Large or SellerListing,'BrowseNodes',// ,
63 ); 132 'ReviewSort' =>'-HelpfulVotes'
64 133 );
65 //if(go != 1) 134
66 //include "aws_signed_request.php"; 135 $requestURI = $_SERVER['REQUEST_URI'];
67 $public_key ="AKIAIHTNWC7L6LOUY4LQ"; 136 $requestIP = $_SERVER['REMOTE_ADDR'];
68 $private_key="zWQlIzndJDtXNfxEXH7K7YR7hzv3u77lOcqfqPde"; 137 // if ($requestIP=="173.161.113.65" || $requestIP=="141.8.132.25") {
69 //for all parameters see if the user has overruled it or use the default 138 // $delay=60;
70 foreach ($Adefault as $i=>$d) { 139 // file_put_contents('/var/ywww/debug/phpDebug',
71 $$i=isset($_GET[$i])?$_GET[$i]:$d; 140 // "bad guy: $requestIP, $requestURI\n",
141 // FILE_APPEND);
142 // sleep($delay);
143 // # No, can't do this
144 // # return; # bomb!
145 // # 'Kung', sitting on my desk in the office while I'm at home,
146 // # is occasionally hitting xml/getAmazonInfo.php:
147 // # e.g. Losing: ItemLookupErrorResponse, RequestThrottled, 129.215.197.36, /xml/getAmazonInfo.php?searchparameterdata=075154454X&locale=uk
148 // # repeatedly, same params
149 // }
150 $ext=$Aserver[$locale]['ext'];
151 $file_data=$ext;
152 ksort($parameters);
153 foreach ($parameters as $i=>$d) {
154 $file_data.='&'.$i.'='.$d;
155 }
156 $gotit=0;
157 $url=aws_signed_request($ext,$parameters,$public_key,$private_key);
158 $crl = curl_init();
159 $timeout = 5;
160 curl_setopt ($crl, CURLOPT_URL,$url);
161 curl_setopt ($crl, CURLOPT_ENCODING , "gzip");
162 curl_setopt ($crl, CURLOPT_RETURNTRANSFER, 1);
163 curl_setopt ($crl, CURLOPT_CONNECTTIMEOUT, $timeout);
164 $semaphore = new SyncSemaphore("Amazon");
165 $gotit = $semaphore->lock(1000);
166 if (!$gotit) {
167 file_put_contents('/var/ywww/debug/phpDebug',
168 "Got it: ".(int)$gotit." $requestIP 1\n",FILE_APPEND);
169 $gotit=$semaphore->lock(1000);
170 file_put_contents('/var/ywww/debug/phpDebug',
171 "Got it: ".(int)$gotit." $requestIP 1a\n",FILE_APPEND);
172 }
173 $output = curl_exec($crl);
174 curl_close($crl);
175 usleep(500000);
176 if ($gotit) {
177 $semaphore->unlock();
178 }
179 else {
180 file_put_contents('/var/ywww/debug/phpDebug',
181 "W/o lock for $requestIP 1b\n",FILE_APPEND);
182 }
183 $review = "";
184 $review1 = "";
185 $review2 = "";
186 $review3 = "";
187 // HST added this
188 $mm=array();
189 if (preg_match("/<Error>/",$output,$mm)) {
190 $xml = new SimpleXMLElement($output);
191 $resName=$xml->getName();
192 $code=$xml->Error->Code;
193 if (!$code) {
194 $code=$xml->Items->Request->Errors->Error->Code;
195 }
196 file_put_contents('/var/ywww/debug/phpDebug',"Losing: ".$mm[0].", $resName, $code, $requestIP, $requestURI\n",FILE_APPEND);
197 if ($code!='RequestThrottled') {
198 file_put_contents('/var/ywww/debug/phpDebug',"parms: $go\nurl: $url\n".
199 print_r($parameters,TRUE)."\n",FILE_APPEND);
200 if ($code=="") {
201 file_put_contents('/var/ywww/debug/phpDebug',"error elt:\n$output\n",FILE_APPEND);
202 }
203 }
204 throw new Exception($code);
205 }
206 else {
207 $xml = new SimpleXMLElement($output);
208 set_error_handler(function () {
209 global $output;
210 file_put_contents('/var/ywww/debug/phpDebug',
211 "Caught one?: ".$searchparameterdata,
212 FILE_APPEND);
213 file_put_contents('/var/ywww/debug/phpDebug',
214 print_r($output, TRUE)."\n",
215 FILE_APPEND);
216 } );
217 $review = $xml->Items->Item->CustomerReviews->IFrameURL;
218 // The above is failing repeatedly --
219 //PHP Notice: Trying to get property of non-object in
220 // /var/ywww/xml/getAmazonInfo.php on line [109]
221 // See the dumped structure at the end of this file for the
222 // cause
223 restore_error_handler();
224 //echo $review;
225 }
226 if ($review != "")
227 {
228 $text = @file_get_contents($review . "&truncate=300");
229 $removeTop1 = preg_replace('~<div class="crIFrameHeaderLeftColumn">(.*?)<div class="crIFrameHeaderHistogram">~si', '', $text);
230 $removeTop2 = preg_replace('~<div class="crIFrameHeaderHistogram">(.*?)<div class="crIframeReviewList">~si', '', $text);
231 if (preg_match('~<body[^>]*>(.*?)</body>~si', $removeTop2, $body)){ $getBody = trim($body[1]); } else { $getBody = $removeTop2;}
232 $removeDiv = preg_replace('~<div[^>]*>(.*?)</div>~si', '', $getBody);
233 $removeCloseDivs = preg_replace('/<\/div>/','', $removeDiv);
234 $setBoundary = str_replace('<!-- BOUNDARY -->','BOTTOM-TOP', $removeCloseDivs);
235 //replace <!-- BOUNDARY --> with BOTTOM-TOP
236 $remove1 = '~<table cellpadding="0"(.*?)%">~s';
237 $setBoundary = preg_replace($remove1,'', $setBoundary);
238 $remove2 = '~</td><td bg(.*?)</table>~s';
239 $setBoundary = preg_replace($remove2,'', $setBoundary);
240 $remove3 = '~<a name=(.*?)</a>~s';
241 $setBoundary = preg_replace($remove3,'', $setBoundary);
242 $setBoundary2 = str_replace('<br />','', $setBoundary);
243 //remove all extra crap;
244 $setBoundary3 = str_replace('</td>','BOTTOM', $setBoundary2);
245 //replace </td> with BOTTOM
246
247 if (preg_match_all('~TOP(.*?)BOTTOM~s', $setBoundary3, $reviews))
248 {
249 $reviewContents = $reviews[1];
250 //print_r($reviewContents);
251 $review1 = trim($reviewContents[0]);
252 $review1 = str_replace("\n", "", $review1);
253 $review1 = str_replace("\r", "", $review1);
254 if (isset($reviewContents[1])) {
255 $review2 = trim($reviewContents[1]);
256 $review2 = str_replace("\n", "", $review2);
257 $review2 = str_replace("\r", "", $review2);
258 }
259 else {
260 $review2 = "";
261 }
262 if (isset($reviewContents[2])) {
263 $review3 = trim($reviewContents[2]);
264 $review3 = str_replace("\n", "", $review3);
265 $review3 = str_replace("\r", "", $review3);
266 }
267 else {
268 $review3 = "";
269 }
270 }
271 else
272 {
273 $review1 = "";
274 $review2 = "";
275 $review3 = "";
276 //echo "EPIC FAIL";
277 }
278
279 unset($xml->Items->Item->CustomerReviews);
280 $xdoc = new DomDocument;
281 $xdoc->loadXML($xml->asXML());
282
283 $cReviews = $xdoc ->createElement('CustomerReviews');
284 $cReviewHolder = $xdoc ->createElement('Review');
285 $cReview = $xdoc ->createElement('Content');
286 $cReviewHolder2 = $xdoc ->createElement('Review');
287 $cReview2 = $xdoc ->createElement('Content');
288 $cReviewHolder3 = $xdoc ->createElement('Review');
289 $cReview3 = $xdoc ->createElement('Content');
290
291 $txtNode = $xdoc ->createTextNode ($review1);
292 $cReview -> appendChild($txtNode);
293
294 $txtNode2 = $xdoc ->createTextNode ($review2);
295 $cReview2 -> appendChild($txtNode2);
296
297 $txtNode3 = $xdoc ->createTextNode ($review3);
298 $cReview3 -> appendChild($txtNode3);
299
300 $cReviewHolder -> appendChild($cReview);
301 $cReviewHolder2 -> appendChild($cReview2);
302 $cReviewHolder3 -> appendChild($cReview3);
303
304 $cReviews -> appendChild($cReviewHolder);
305 $cReviews -> appendChild($cReviewHolder2);
306 $cReviews -> appendChild($cReviewHolder3);
307
308 $xdoc->documentElement->childNodes->item(1)->childNodes->item(1)->appendChild($cReviews);
309 $newXML = simplexml_import_dom($xdoc);
310 $output = $newXML->asXml();
311 }
312
313 if($go == 1) {
314 $item = $xml->Items->Item[0];
315 if ($item && $item->ItemAttributes && $item->ItemAttributes->Title) {
316 $title = $item->ItemAttributes->Title; }
317 else { $title = "[no title]"; };
318 file_put_contents('/var/ywww/debug/phpDebug',"win: |$errorCode| ".
319 $title."\n",
320 FILE_APPEND);
321 return $output;
322 }
323 else
324 {
325 if ($xml->Items->Item) {
326 $title = $xml->Items->Item[0]->ItemAttributes->Title;
327 $author = $xml->Items->Item[0]->ItemAttributes->Author;
328 $binding = $xml->Items->Item[0]->ItemAttributes->Binding;
329 $dewey = $xml->Items->Item[0]->ItemAttributes->DeweyDecimalNumber;
330 if($dewey == "")
331 $dewey = "null";
332 $imageURL = $xml->Items->Item[0]->MediumImage->URL;
333 $salesRank = $xml->Items->Item[0]->SalesRank;
334 $pubDate = $xml->Items->Item[0]->ItemAttributes->PublicationDate;
335 if (strlen($pubDate)==4) { $pubDate=$pubDate."-01-01";}
336 if (strlen($pubDate)==7) { $pubDate=$pubDate."-01";}
337 if (strlen($pubDate)==0) {
338 $pubDate="null";
339 }
340 else {
341 $pubDate="\"$pubDate\"";
342 }
343 $publisher = $xml->Items->Item[0]->ItemAttributes->Publisher;
344 }
345 else {
346 $title = $salesRank = "";
347 $dewey = "null";
348 }
349
350 $genreID = "";
351 $genre = "";
352 $genArr = array();
353
354 if ($xml->Items->Item[0] && $xml->Items->Item[0]->BrowseNodes) {
355 for($i=0;$i<sizeof($xml->Items->Item->BrowseNodes->BrowseNode);$i++){
356 //sexy recursive function
357 findGenre($xml->Items->Item->BrowseNodes->BrowseNode[$i], $genreID, $genre);
358
359 if($genre != "")
360 $genArr[strval($genreID)] = strval($genre);
361 //$genArr[$i] = array(strval($genreID) => strval($genre));
362
363 //echo $genre;
364 //echo $genreID;
365
366 $genre = "";
367 $genreID = "";
368 }
369 }
370
371 $g1 = "null";
372 $g2 = "null";
373 $g3 = "null";
374 $loop = 1;
375
376 foreach ($genArr as $key => $value) {
377 //echo "$key => $value";
378 if ($key>2047) {
379 //HST added
380 break;
381 }
382 $queryG = "CALL b_addBrowseNode($key,\"$value\")"; //add the name value pair for genre to new table
383 //echo $queryG;
384 include "../../private/db.php";
385 $resG = mysqli_query($link, $queryG);
386 mysqli_close($link);
387
388 switch ($loop) {
389 case 1:
390 $g1 = $key;
391 break;
392 case 2:
393 $g2 = $key;
394 break;
395 case 3:
396 $g3 = $key;
397 break;
398 }
399
400 $loop++;
401 }
402
403 if($salesRank == "")
404 $salesRank = "null";
405
406 if($title != "")
407 {
408 include "../../private/db.php";
409 if ($publisher->count()==0) {
410 $publisher="null";
411 }
412 else {
413 $publisher=mysqli_real_escape_string($link,$publisher);
414 if (strlen($publisher)>30) {
415 $publisher=rtrim(substr($publisher,0,30),"\\");
416 }
417 $publisher="\"".$publisher."\"";
418 }
419 if ($author->count()==0) {
420 $author="unknown";
421 }
422 else {
423 $author=mysqli_real_escape_string($link,$author);
424 if (strlen($author)>30) {
425 $author=rtrim(substr($author,0,30),"\\");
426 }
427 }
428 $author="\"".$author."\"";
429 $title=mysqli_real_escape_string($link,$title);
430 if (strlen($title)>100) {
431 $title=rtrim(substr($title,0,100),"\\");
432 }
433 $title="\"".$title."\"";
434 $review1 = mysqli_real_escape_string($link,$review1);
435 if (strlen($review1)>500) { $review1=rtrim(substr($review1,0,500),"\\");}
436 $review2 = mysqli_real_escape_string($link,$review2);
437 if (strlen($review2)>500) { $review2=rtrim(substr($review2,0,500),"\\");}
438 $review3 = mysqli_real_escape_string($link,$review3);
439 if (strlen($review3)>500) { $review3=rtrim(substr($review3,0,500),"\\");}
440
441 $queryInsert = "CALL b_addNewBook(\"$searchparameterdata\",$title, $author,\"$binding\",\"$imageURL\", $dewey, $salesRank,$pubDate,$publisher,$g1,$g2,$g3,$loc)";
442 //echo $queryInsert;
443
444 $res = mysqli_query($link, $queryInsert);
445 if (!$res) {
446 $err=mysqli_error( $link );
447 mysqli_close($link);
448 file_put_contents('/var/ywww/debug/phpDebug',
449 "anb failed: $queryInsert\n$err\n",
450 FILE_APPEND);
451 exit($err);
452 }
453 $queryInsertReviews = "CALL b_insertReviews(\"$searchparameterdata\",\"$review1\",\"$review2\",\"$review3\")";
454 if($review1 != "") {
455 $res = mysqli_query($link, $queryInsertReviews);
456 if (!$res) {
457 $err=mysqli_error( $link );
458 mysqli_close($link);
459 file_put_contents('/var/ywww/debug/phpDebug',
460 "anr failed: $queryInsertReviews\n",
461 FILE_APPEND);
462 exit($err);
463 }
464 }
465 mysqli_close($link); //do not remove. reset is needed otherwise mysqli_fetch_array doesn't work after first loop
466 }
467
468 echo $output;
469 }
72 } 470 }
73 //this is the data that is used to form the request for AWS 471
74 //this is the part that is search specific 472 function getAmazonDetLocal($go,$language,$loc,
75 $parameters=array( 473 $page,$operation,$searchparameter,
76 'Operation' =>$operation , 474 $searchparameterdata,
77 //'Keywords' =>urlencode($search) , 475 $show_array,$show_url,$show_xml)
78 //'SearchIndex' =>$searchindex , //Books for example. 476 {
79 "$searchparameter"=>$searchparameterdata , 477 global $output;
80 'ItemPage' =>$page , //which page? 478 //look up info from db
81 'AssociateTag' =>$Aassociates_id[$locale], 479 include "../../private/db.php";
82 'ResponseGroup' =>'ItemAttributes,Reviews,EditorialReview,OfferSummary,Offers,Images,AlternateVersions,SalesRank,BrowseNodes' , //Small, Medium, Large or SellerListing,'BrowseNodes',// , 480 $query = "CALL b_getBookInfo('$searchparameterdata', $loc)";
83 'ReviewSort' =>'-HelpfulVotes' 481 //echo $query;
84 ); 482 $res = mysqli_query($link, $query) or exit( mysqli_error( $link ));
85
86 $requestURI = $_SERVER['REQUEST_URI'];
87 $requestIP = $_SERVER['REMOTE_ADDR'];
88 // if ($requestIP=="173.161.113.65" || $requestIP=="141.8.132.25") {
89 // $delay=60;
90 // file_put_contents('/var/ywww/debug/phpDebug',
91 // "bad guy: $requestIP, $requestURI\n",
92 // FILE_APPEND);
93 // sleep($delay);
94 // # No, can't do this
95 // # return; # bomb!
96 // # 'Kung', sitting on my desk in the office while I'm at home,
97 // # is occasionally hitting xml/getAmazonInfo.php:
98 // # e.g. Losing: ItemLookupErrorResponse, RequestThrottled, 129.215.197.36, /xml/getAmazonInfo.php?searchparameterdata=075154454X&locale=uk
99 // # repeatedly, same params
100 // }
101 $ext=$Aserver[$locale]['ext'];
102 $file_data=$ext;
103 ksort($parameters);
104 foreach ($parameters as $i=>$d) {
105 $file_data.='&'.$i.'='.$d;
106 }
107 $gotit=0;
108 $url=aws_signed_request($ext,$parameters,$public_key,$private_key);
109 $crl = curl_init();
110 $timeout = 5;
111 curl_setopt ($crl, CURLOPT_URL,$url);
112 curl_setopt ($crl, CURLOPT_ENCODING , "gzip");
113 curl_setopt ($crl, CURLOPT_RETURNTRANSFER, 1);
114 curl_setopt ($crl, CURLOPT_CONNECTTIMEOUT, $timeout);
115 $semaphore = new SyncSemaphore("Amazon");
116 $gotit = $semaphore->lock(1000);
117 if (!$gotit) {
118 file_put_contents('/var/ywww/debug/phpDebug',
119 "Got it: ".(int)$gotit." $requestIP 1\n",FILE_APPEND);
120 $gotit=$semaphore->lock(1000);
121 file_put_contents('/var/ywww/debug/phpDebug',
122 "Got it: ".(int)$gotit." $requestIP 1a\n",FILE_APPEND);
123 }
124 $output = curl_exec($crl);
125 curl_close($crl);
126 usleep(500000);
127 if ($gotit) {
128 $semaphore->unlock();
129 }
130 else {
131 file_put_contents('/var/ywww/debug/phpDebug',
132 "W/o lock for $requestIP 1b\n",FILE_APPEND);
133 }
134 $review = "";
135 $review1 = "";
136 $review2 = "";
137 $review3 = "";
138 // HST added this
139 $mm=array();
140 if (preg_match("/<Error>/",$output,$mm)) {
141 $xml = new SimpleXMLElement($output);
142 $resName=$xml->getName();
143 $code=$xml->Error->Code;
144 if (!$code) {
145 $code=$xml->Items->Request->Errors->Error->Code;
146 }
147 $errorCode=$code;
148 file_put_contents('/var/ywww/debug/phpDebug',"Losing: ".$mm[0].", $resName, $code, $requestIP, $requestURI\n",FILE_APPEND);
149 if ($code=='RequestThrottled') {
150 usleep(200000); // Try to reduce throttling until we get a
151 // principled solution in place
152 }
153 else {
154 file_put_contents('/var/ywww/debug/phpDebug',"parms: $go\nurl: $url\n".
155 print_r($parameters,TRUE)."\n",FILE_APPEND);
156 if ($code=="") {
157 file_put_contents('/var/ywww/debug/phpDebug',"error elt:\n$output\n",FILE_APPEND);
158 }
159 }
160 }
161 else {
162 $xml = new SimpleXMLElement($output);
163 set_error_handler(function () {
164 global $output;
165 file_put_contents('/var/ywww/debug/phpDebug',
166 "Caught one?: ".$searchparameterdata,
167 FILE_APPEND);
168 file_put_contents('/var/ywww/debug/phpDebug',
169 print_r($output, TRUE)."\n",
170 FILE_APPEND);
171 } );
172 $review = $xml->Items->Item->CustomerReviews->IFrameURL;
173 // The above is failing repeatedly --
174 //PHP Notice: Trying to get property of non-object in
175 // /var/ywww/xml/getAmazonInfo.php on line [109]
176 // See the dumped structure at the end of this file for the
177 // cause
178 restore_error_handler();
179 //echo $review;
180 }
181 if ($review != "")
182 {
183 $text = @file_get_contents($review . "&truncate=300");
184 $removeTop1 = preg_replace('~<div class="crIFrameHeaderLeftColumn">(.*?)<div class="crIFrameHeaderHistogram">~si', '', $text);
185 $removeTop2 = preg_replace('~<div class="crIFrameHeaderHistogram">(.*?)<div class="crIframeReviewList">~si', '', $text);
186 if (preg_match('~<body[^>]*>(.*?)</body>~si', $removeTop2, $body)){ $getBody = trim($body[1]); } else { $getBody = $removeTop2;}
187 $removeDiv = preg_replace('~<div[^>]*>(.*?)</div>~si', '', $getBody);
188 $removeCloseDivs = preg_replace('/<\/div>/','', $removeDiv);
189 $setBoundary = str_replace('<!-- BOUNDARY -->','BOTTOM-TOP', $removeCloseDivs);
190 //replace <!-- BOUNDARY --> with BOTTOM-TOP
191 $remove1 = '~<table cellpadding="0"(.*?)%">~s';
192 $setBoundary = preg_replace($remove1,'', $setBoundary);
193 $remove2 = '~</td><td bg(.*?)</table>~s';
194 $setBoundary = preg_replace($remove2,'', $setBoundary);
195 $remove3 = '~<a name=(.*?)</a>~s';
196 $setBoundary = preg_replace($remove3,'', $setBoundary);
197 $setBoundary2 = str_replace('<br />','', $setBoundary);
198 //remove all extra crap;
199 $setBoundary3 = str_replace('</td>','BOTTOM', $setBoundary2);
200 //replace </td> with BOTTOM
201
202 if (preg_match_all('~TOP(.*?)BOTTOM~s', $setBoundary3, $reviews))
203 {
204 $reviewContents = $reviews[1];
205 //print_r($reviewContents);
206 $review1 = trim($reviewContents[0]);
207 $review1 = str_replace("\n", "", $review1);
208 $review1 = str_replace("\r", "", $review1);
209 if (isset($reviewContents[1])) {
210 $review2 = trim($reviewContents[1]);
211 $review2 = str_replace("\n", "", $review2);
212 $review2 = str_replace("\r", "", $review2);
213 }
214 else {
215 $review2 = "";
216 }
217 if (isset($reviewContents[2])) {
218 $review3 = trim($reviewContents[2]);
219 $review3 = str_replace("\n", "", $review3);
220 $review3 = str_replace("\r", "", $review3);
221 }
222 else {
223 $review3 = "";
224 }
225 }
226 else
227 {
228 $review1 = "";
229 $review2 = "";
230 $review3 = "";
231 //echo "EPIC FAIL";
232 }
233
234 unset($xml->Items->Item->CustomerReviews);
235 $xdoc = new DomDocument;
236 $xdoc->loadXML($xml->asXML());
237
238 $cReviews = $xdoc ->createElement('CustomerReviews');
239 $cReviewHolder = $xdoc ->createElement('Review');
240 $cReview = $xdoc ->createElement('Content');
241 $cReviewHolder2 = $xdoc ->createElement('Review');
242 $cReview2 = $xdoc ->createElement('Content');
243 $cReviewHolder3 = $xdoc ->createElement('Review');
244 $cReview3 = $xdoc ->createElement('Content');
245
246 $txtNode = $xdoc ->createTextNode ($review1);
247 $cReview -> appendChild($txtNode);
248
249 $txtNode2 = $xdoc ->createTextNode ($review2);
250 $cReview2 -> appendChild($txtNode2);
251
252 $txtNode3 = $xdoc ->createTextNode ($review3);
253 $cReview3 -> appendChild($txtNode3);
254
255 $cReviewHolder -> appendChild($cReview);
256 $cReviewHolder2 -> appendChild($cReview2);
257 $cReviewHolder3 -> appendChild($cReview3);
258
259 $cReviews -> appendChild($cReviewHolder);
260 $cReviews -> appendChild($cReviewHolder2);
261 $cReviews -> appendChild($cReviewHolder3);
262
263 $xdoc->documentElement->childNodes->item(1)->childNodes->item(1)->appendChild($cReviews);
264 $newXML = simplexml_import_dom($xdoc);
265 $output = $newXML->asXml();
266 }
267 483
268 switch($locale){ 484 $output = "";
269 case "us": 485 $output .= "<?xml version=\"1.0\" encoding=\"UTF-8\"?>";
270 $loc = 0; 486 $output .= "<Details>";
271 break; 487 if ( mysqli_num_rows( $res ) > 0 )
272 case "uk": 488 {
273 $loc = 1; 489 $rows=mysqli_fetch_array($res, MYSQLI_ASSOC);
274 break; 490 mysqli_close($link); //do not remove. reset is needed otherwise mysqli_fetch_array doesn't work after first loop
275 case "ca": 491 $output .= "<ASIN>" . $searchparameterdata . "</ASIN>";
276 $loc = 2; 492 $output .= "<Title>" . htmlspecialchars($rows["Title"]) . "</Title>";
277 break; 493 $output .= "<Author>" . htmlspecialchars($rows["Author"]) . "</Author>";
278 case "de": 494 $output .= "<Binding>" . htmlspecialchars($rows["Binding"]) . "</Binding>";
279 $loc = 3; 495 $output .= "<Dewey>" . htmlspecialchars($rows["DeweyDecimal"]) . "</Dewey>";
280 break; 496 $output .= "<ImageURL>" . htmlspecialchars($rows["ImageURL"]) . "</ImageURL>";
281 case "fr": 497 $output .= "<SalesRank>" . htmlspecialchars($rows["SalesRank"]) . "</SalesRank>";
282 $loc = 4; 498 $output .= "<PublicationDate>" . htmlspecialchars($rows["PublicationDate"]) . "</PublicationDate>";
283 break; 499 $output .= "<Publisher>" . htmlspecialchars($rows["Publisher"]) . "</Publisher>";
284 }; 500 $output .= "<Genre1>" . htmlspecialchars($rows["Genre1"]) . "</Genre1>";
285 501 $output .= "<Genre2>" . htmlspecialchars($rows["Genre2"]) . "</Genre2>";
286 $errorCode = $xml->Error->Code; 502 $output .= "<Genre3>" . htmlspecialchars($rows["Genre3"]) . "</Genre3>";
287 //echo $errorCode; 503 $output .= "<ProductGroup>Book</ProductGroup>";
288 504 $output .= "<Error>AccountLimitExceeded</Error>";
289 if($errorCode != "AccountLimitExceeded")
290 {
291 if($go == 1) {
292 $item = $xml->Items->Item[0];
293 if ($item && $item->ItemAttributes && $item->ItemAttributes->Title) {
294 $title = $item->ItemAttributes->Title; }
295 else { $title = "[no title]"; };
296 file_put_contents('/var/ywww/debug/phpDebug',"win: |$errorCode| ".
297 $title."\n",
298 FILE_APPEND);
299 return $output;
300 }
301 else
302 {
303 if ($xml->Items->Item) {
304 $title = $xml->Items->Item[0]->ItemAttributes->Title;
305 $author = $xml->Items->Item[0]->ItemAttributes->Author;
306 $binding = $xml->Items->Item[0]->ItemAttributes->Binding;
307 $dewey = $xml->Items->Item[0]->ItemAttributes->DeweyDecimalNumber;
308 if($dewey == "")
309 $dewey = "null";
310 $imageURL = $xml->Items->Item[0]->MediumImage->URL;
311 $salesRank = $xml->Items->Item[0]->SalesRank;
312 $pubDate = $xml->Items->Item[0]->ItemAttributes->PublicationDate;
313 if (strlen($pubDate)==4) { $pubDate=$pubDate."-01-01";}
314 if (strlen($pubDate)==7) { $pubDate=$pubDate."-01";}
315 if (strlen($pubDate)==0) {
316 $pubDate="null";
317 }
318 else {
319 $pubDate="\"$pubDate\"";
320 }
321 $publisher = $xml->Items->Item[0]->ItemAttributes->Publisher;
322 }
323 else {
324 $title = $salesRank = "";
325 $dewey = "null";
326 }
327 505
328 $genreID = ""; 506 }
329 $genre = ""; 507 else {
330 $genArr = array(); 508 mysqli_close($link); //do not remove. reset is needed otherwise mysqli_fetch_array doesn't work after first loop
331 509 }
332 if ($xml->Items->Item[0] && $xml->Items->Item[0]->BrowseNodes) { 510 $output .= "</Details>";
333 for($i=0;$i<sizeof($xml->Items->Item->BrowseNodes->BrowseNode);$i++){ 511 echo $output;
334 //sexy recursive function
335 findGenre($xml->Items->Item->BrowseNodes->BrowseNode[$i], $genreID, $genre);
336
337 if($genre != "")
338 $genArr[strval($genreID)] = strval($genre);
339 //$genArr[$i] = array(strval($genreID) => strval($genre));
340
341 //echo $genre;
342 //echo $genreID;
343
344 $genre = "";
345 $genreID = "";
346 }
347 }
348
349 $g1 = "null";
350 $g2 = "null";
351 $g3 = "null";
352 $loop = 1;
353
354 foreach ($genArr as $key => $value) {
355 //echo "$key => $value";
356 if ($key>2047) {
357 //HST added
358 break;
359 }
360 $queryG = "CALL b_addBrowseNode($key,\"$value\")"; //add the name value pair for genre to new table
361 //echo $queryG;
362 include "../../private/db.php";
363 $resG = mysqli_query($link, $queryG);
364 mysqli_close($link);
365
366 switch ($loop) {
367 case 1:
368 $g1 = $key;
369 break;
370 case 2:
371 $g2 = $key;
372 break;
373 case 3:
374 $g3 = $key;
375 break;
376 }
377
378 $loop++;
379 }
380
381 if($salesRank == "")
382 $salesRank = "null";
383
384 if($title != "")
385 {
386 include "../../private/db.php";
387 if ($publisher->count()==0) {
388 $publisher="null";
389 }
390 else {
391 $publisher=mysqli_real_escape_string($link,$publisher);
392 if (strlen($publisher)>30) {
393 $publisher=rtrim(substr($publisher,0,30),"\\");
394 }
395 $publisher="\"".$publisher."\"";
396 }
397 if ($author->count()==0) {
398 $author="unknown";
399 }
400 else {
401 $author=mysqli_real_escape_string($link,$author);
402 if (strlen($author)>30) {
403 $author=rtrim(substr($author,0,30),"\\");
404 }
405 }
406 $author="\"".$author."\"";
407 $title=mysqli_real_escape_string($link,$title);
408 if (strlen($title)>100) {
409 $title=rtrim(substr($title,0,100),"\\");
410 }
411 $title="\"".$title."\"";
412 $review1 = mysqli_real_escape_string($link,$review1);
413 if (strlen($review1)>500) { $review1=rtrim(substr($review1,0,500),"\\");}
414 $review2 = mysqli_real_escape_string($link,$review2);
415 if (strlen($review2)>500) { $review2=rtrim(substr($review2,0,500),"\\");}
416 $review3 = mysqli_real_escape_string($link,$review3);
417 if (strlen($review3)>500) { $review3=rtrim(substr($review3,0,500),"\\");}
418
419 $queryInsert = "CALL b_addNewBook(\"$searchparameterdata\",$title, $author,\"$binding\",\"$imageURL\", $dewey, $salesRank,$pubDate,$publisher,$g1,$g2,$g3,$loc)";
420 //echo $queryInsert;
421
422 $res = mysqli_query($link, $queryInsert);
423 if (!$res) {
424 $err=mysqli_error( $link );
425 mysqli_close($link);
426 file_put_contents('/var/ywww/debug/phpDebug',
427 "anb failed: $queryInsert\n$err\n",
428 FILE_APPEND);
429 exit($err);
430 }
431 $queryInsertReviews = "CALL b_insertReviews(\"$searchparameterdata\",\"$review1\",\"$review2\",\"$review3\")";
432 if($review1 != "") {
433 $res = mysqli_query($link, $queryInsertReviews);
434 if (!$res) {
435 $err=mysqli_error( $link );
436 mysqli_close($link);
437 file_put_contents('/var/ywww/debug/phpDebug',
438 "anr failed: $queryInsertReviews\n",
439 FILE_APPEND);
440 exit($err);
441 }
442 }
443 mysqli_close($link); //do not remove. reset is needed otherwise mysqli_fetch_array doesn't work after first loop
444 }
445
446 echo $output;
447 }
448 }
449 else
450 {
451 //look up info from db
452 include "../../private/db.php";
453 $query = "CALL b_getBookInfo('$searchparameterdata', $loc)";
454 //echo $query;
455 $res = mysqli_query($link, $query) or exit( mysqli_error( $link ));
456
457 $output = "";
458 $output .= "<?xml version=\"1.0\" encoding=\"UTF-8\"?>";
459 $output .= "<Details>";
460 if ( mysqli_num_rows( $res ) > 0 )
461 {
462 $rows=mysqli_fetch_array($res, MYSQLI_ASSOC);
463 mysqli_close($link); //do not remove. reset is needed otherwise mysqli_fetch_array doesn't work after first loop
464 $output .= "<ASIN>" . $searchparameterdata . "</ASIN>";
465 $output .= "<Title>" . htmlspecialchars($rows["Title"]) . "</Title>";
466 $output .= "<Author>" . htmlspecialchars($rows["Author"]) . "</Author>";
467 $output .= "<Binding>" . htmlspecialchars($rows["Binding"]) . "</Binding>";
468 $output .= "<Dewey>" . htmlspecialchars($rows["DeweyDecimal"]) . "</Dewey>";
469 $output .= "<ImageURL>" . htmlspecialchars($rows["ImageURL"]) . "</ImageURL>";
470 $output .= "<SalesRank>" . htmlspecialchars($rows["SalesRank"]) . "</SalesRank>";
471 $output .= "<PublicationDate>" . htmlspecialchars($rows["PublicationDate"]) . "</PublicationDate>";
472 $output .= "<Publisher>" . htmlspecialchars($rows["Publisher"]) . "</Publisher>";
473 $output .= "<Genre1>" . htmlspecialchars($rows["Genre1"]) . "</Genre1>";
474 $output .= "<Genre2>" . htmlspecialchars($rows["Genre2"]) . "</Genre2>";
475 $output .= "<Genre3>" . htmlspecialchars($rows["Genre3"]) . "</Genre3>";
476 $output .= "<ProductGroup>Book</ProductGroup>";
477 $output .= "<Error>AccountLimitExceeded</Error>";
478
479 }
480 else {
481 mysqli_close($link); //do not remove. reset is needed otherwise mysqli_fetch_array doesn't work after first loop
482 }
483 $output .= "</Details>";
484 echo $output;
485 }
486 } 512 }
487 513
488 function findGenre($browseNode, &$ID, &$gen) 514 function findGenre($browseNode, &$ID, &$gen)
489 { 515 {
490 if($browseNode->Name == "Subjects") 516 if($browseNode->Name == "Subjects")