crawler.js 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403
  1. /*
  2. * Mobifier, the gallery bot
  3. * Copyright (C) 2021 Daniele Bonini
  4. *
  5. * This program is free software: you can redistribute it and/or modify
  6. * it under the terms of the GNU General Public License as published by
  7. * the Free Software Foundation, either version 3 of the License, or
  8. * (at your option) any later version.
  9. *
  10. * This program is distributed in the hope that it will be useful,
  11. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13. * GNU General Public License for more details.
  14. *
  15. * You should have received a copy of the GNU General Public License
  16. * along with this program. If not, see <https://www.gnu.org/licenses/>.
  17. *
  18. */
  19. var recNum = 0;
  20. var nImgProcessed = 0;
  21. var output = 1;
  22. var verbose = 0;
  23. var host = "mbfy.it";
  24. var targetDomain = "example.com";
  25. var galTitle = "EXAMPLE";
  26. var landingPage = "Example.com";
  27. var crawlres = [];
  28. var logoFound = false;
  29. function loadDoc() {
  30. recNum++;
  31. bConnectionOK = false;
  32. if (verbose===0 && recNum===1) {
  33. document.write("Copyrights (C) 2021 Daniele Bonini. GNU General Public License ver3<br>");
  34. document.write("<br>");
  35. document.write("This is a bit of doc:<br>");
  36. document.write("- out: std (for standard output) or json<br>");
  37. document.write("- v: 0 (for standard verbosity) or 1 (for quite output)<br>");
  38. document.write("<br>");
  39. document.write("The service is running and trying to connect to the target server.<br>");
  40. document.write("The connection to the target web server can eventually fail.<br>");
  41. document.write("Cause could be found among the following reasons:<br>");
  42. document.write("- Internet connection lost.<br>");
  43. document.write("- DNS failure.<br>");
  44. document.write("- Target server is down.<br>");
  45. document.write("- Access-Control-Allow-Origin header missing on the target server.<br>");
  46. document.write("<br>");
  47. }
  48. var xhttp = new XMLHttpRequest();
  49. if (host == "mbfier.com") {
  50. xhttp.open("GET", "http://"+targetDomain+".mbfier.com/upload.php?url=http://"+targetDomain, true);
  51. } else {
  52. xhttp.open("GET", "http://"+targetDomain+".mbfy.it/upload.php?url=http://"+targetDomain, true);
  53. }
  54. xhttp.send();
  55. xhttp.onreadystatechange = function() {
  56. if (this.readyState == 4 && this.status == 200) {
  57. bConnectionOK = true;
  58. if (verbose===0) {
  59. document.write("Connection succeeded.<br><br>")
  60. }
  61. try {
  62. document.getElementById("grausi").value = this.responseText;
  63. }
  64. catch (error) {
  65. //console.error(error);
  66. if (!document.getElementById("grausi")) {
  67. document.write("<input id='grausi' type='hidden' value=''>");
  68. document.getElementById("grausi").value = this.responseText;
  69. }
  70. }
  71. finally {
  72. document.getElementById("grausi").addEventListener("change", launchParsing(), true);
  73. }
  74. } else {
  75. if ((this.readyState == 4 && this.status == 0) && (!bConnectionOK)) {
  76. if (verbose===0) {
  77. document.write("Final result:<br>");
  78. document.write("Connection problems detected.<br><br>");
  79. }
  80. }
  81. }
  82. }
  83. }
  84. function writeHeader() {
  85. //TITLE BAR
  86. if ((output == 1 && verbose == 1) && (nImgProcessed==0 && recNum==1)) {
  87. divTitle = "<div style='position:relative;margin-top:8px;top:0px;left:0px;width:100%'>";
  88. document.write(divTitle);
  89. divTitle = "<div style='height:50px;text-align:center;opacity:1.0;width:350px;clear:both;margin:auto;vertical-align:middle;border:0px solid green;background: rgba(3, 169, 245, 0.7);color:#FFFFFF;'><span style='position:relative;top:-15px;font-size:25px;font-weight:900;'><br>"+galTitle+"<br><br></span></div></div>";
  90. document.write(divTitle);
  91. }
  92. // END TITLE
  93. }
  94. function crawlImages() {
  95. var bTitle = false;
  96. var found = "||";
  97. var z = 0;
  98. var str = document.getElementById('grausi').value;
  99. str = str.replace("/\n/"," ");
  100. //var regexp = /<img.*?src="[^*?"<>|]+".*?>/gi;
  101. //var regexp = /("|')([^*?"<>|]+\.(png|gif|jpg|jpeg))("|')/gi;
  102. //var regexp = /<img\n?.*src=("|')?([^:*?"<>|]+\.(png|gif|jpg|jpeg))("|')?.*?\n?\/?>/gis;
  103. //var regexp = /<img.*\n?.*src=("|')?(?!(cid:|data:))([^*?"<>|]+\.(png|gif|jpg|jpeg))("|')?.*?\n?[^*?"<>|]*\/?>/gi;
  104. var regexp = /<img\n?.*?\n?\/?>/gis;
  105. var x = str.match(regexp);
  106. if (x) {
  107. for (i=0;i<x.length;i++) {
  108. str2 = x[i]; //.toLowerCase();
  109. str3 = str2.toLowerCase();
  110. if (str2.length <= 4) {
  111. continue;
  112. }
  113. if ((str2.indexOf("blank.gif")>-1 || str2.indexOf("pixel.gif")>-1 || str2.indexOf("pxl.gif")>-1 || str2.indexOf("pix.gif")>-1) && str2.indexOf("data-src")==-1) {
  114. continue;
  115. }
  116. if (str2.indexOf(" src=\"data:")>-1 || str2.indexOf(" src='data:")>-1 || str2.indexOf(" src=data:")>-1) {
  117. str2 = str2.replace(" src=", " zzz=");
  118. str2 = str2.replace(" data-src=", " src=");
  119. }
  120. if (str3.indexOf("credit-card")>-1 || str3.indexOf("loyalty")>-1) {
  121. continue;
  122. }
  123. oo = str2.indexOf(" src=");
  124. ii = str2.indexOf(" data-src=");
  125. jj = str2.indexOf(".jpg");
  126. yy = str2.indexOf(".jpeg");
  127. kk = str2.indexOf(".png");
  128. ww = str2.indexOf(".gif");
  129. if (((oo==-1) && (oo<ii)) && ii>-1 && ((jj>ii) || (yy>ii) || (kk>ii) || (ww>ii))) {
  130. //alert(str2);
  131. //var regexp3 = /<img\n?.*data-src=("|')([^*?"<>|]+\.(png|gif|jpg|jpeg))("|').*?\n?\/?>/gis;
  132. //var zz = str2.match(regexp3);
  133. //if (zz[0]) {
  134. //str3 = zz[0];
  135. str2 = str2.replace(" src=", " zzz=");
  136. str2 = str2.replace(" data-src=", " src=");
  137. //str2 = str3;
  138. //}
  139. }
  140. str2 = str2.trim();
  141. if (str2 !== "") {
  142. newurl = str2;
  143. //var regexp2 = /([\w\.-]+\.[a-z\.]{2,8})/gi;
  144. //var regexp2 = /([\w\.-]+\.(png|jpg|jpeg))/gi;
  145. //var regexp2 = /src="([\w\:\/\.-]+\.(png|jpg|jpeg))"/gi;
  146. //var regexp2 = /<img.*?src="[^*?"<>|]+".*?>/gi;
  147. //var regexp2 = /("|')([^*?"<>|]+\.(png|gif|jpg|jpeg))("|')/gi;
  148. //var regexp2 = /<img\n?.*src=("|')([^*?"<>|]+\.(png|gif|jpg|jpeg))("|').*?>/gi;
  149. var regexp2 = /<img\n?.*?\n?\/?>/gis;
  150. var y = newurl.match(regexp2);
  151. if (y[0]) {
  152. if (found.indexOf("|" + y[0] + "|") == -1) {
  153. crawlres[z] = y[0];
  154. z++;
  155. found = found + "|" + y[0] + "|";
  156. }
  157. }
  158. }
  159. }
  160. }
  161. /*
  162. * DUMP
  163. */
  164. dump = false;
  165. if (dump || (verbose == 0) || ((output==2) && (verbose == 1))) {
  166. for (i=0;i<crawlres.length;i++) {
  167. str = crawlres[i];
  168. str = str.replace("<", "&lt;");
  169. str = str.replace(">", "&gt;");
  170. crawlres[i] = str;
  171. }
  172. }
  173. if (output === 1) {
  174. if (verbose === 0) {
  175. document.write("Image list:<br>");
  176. }
  177. /*
  178. * PATH FIX
  179. */
  180. for (i=0;i<crawlres.length;i++) {
  181. str = crawlres[i];
  182. ipos = str.indexOf("src=//");
  183. if (ipos>-1) {
  184. str = str.replace("src=//", "src=http://");
  185. crawlres[i] = str;
  186. }
  187. ipos = str.indexOf("src='//");
  188. if (ipos>-1) {
  189. str = str.replace("src='//", "src='http://");
  190. crawlres[i] = str;
  191. }
  192. ipos = str.indexOf("src=\"//");
  193. if (ipos>-1) {
  194. str = str.replace("src=\"//", "src=\"http://");
  195. crawlres[i] = str;
  196. }
  197. ipos = str.indexOf("src=/");
  198. if (ipos>-1) {
  199. str = str.replace("src=/", "src=http://"+targetDomain+"/");
  200. crawlres[i] = str;
  201. }
  202. ipos = str.indexOf("src='/");
  203. if (ipos>-1) {
  204. str = str.replace("src='/", "src='http://"+targetDomain+"/");
  205. crawlres[i] = str;
  206. }
  207. ipos = str.indexOf("src=\"/");
  208. if (ipos>-1) {
  209. str = str.replace("src=\"/", "src=\"http://"+targetDomain+"/");
  210. crawlres[i] = str;
  211. }
  212. ipos = str.indexOf("src=http");
  213. if (ipos>-1) {
  214. continue;
  215. }
  216. ipos = str.indexOf("src='http");
  217. if (ipos>-1) {
  218. continue;
  219. }
  220. ipos = str.indexOf("src=\"http");
  221. if (ipos>-1) {
  222. continue;
  223. }
  224. str = str.replace("src=", "src=http://"+targetDomain+"/");
  225. crawlres[i] = str;
  226. str = str.replace("src='", "src='http://"+targetDomain+"/");
  227. crawlres[i] = str;
  228. str = str.replace("src=\"", "src=\"http://"+targetDomain+"/");
  229. crawlres[i] = str;
  230. }
  231. /*
  232. * SEARCH FOR LOGO
  233. */
  234. var newcrawlres = [];
  235. var found = -1;
  236. for (i=0;i<crawlres.length;i++) {
  237. str = crawlres[i];
  238. if (str.indexOf("\"logo\"")>-1 || str.indexOf("'logo'")>-1 || str.indexOf("logo.png")>-1 || str.indexOf("logo.gif")>-1 || str.indexOf("logo.jpg")>-1 || str.indexOf("logo.jpeg")>-1) {
  239. found=i;
  240. //alert("found="+found);
  241. break;
  242. }
  243. }
  244. if (found>-1) {
  245. newcrawlres[0] = crawlres[found];
  246. //alert(newcrawlres[0]);
  247. var j=1;
  248. for (i=0;i<found;i++) {
  249. newcrawlres[j] = crawlres[i];
  250. j++;
  251. }
  252. for (i=found+1;i<crawlres.length;i++) {
  253. newcrawlres[j] = crawlres[i];
  254. j++;
  255. }
  256. crawlres = newcrawlres;
  257. logoFound = true;
  258. //alert(newcrawlres[0]);
  259. }
  260. // END SEARCH
  261. for (i=0;i<crawlres.length;i++) {
  262. document.write(crawlres[i]);
  263. if (verbose === 0) {
  264. document.write("<br>");
  265. }
  266. }
  267. } else {
  268. document.write(JSON.stringify(crawlres));
  269. }
  270. }
  271. function postCrawling() {
  272. var colImages = document.images;
  273. var j = 0;
  274. var logoHeight = "300px";
  275. for(var i = 0; i < colImages.length; i++) {
  276. var oriWidth = colImages[i].width;
  277. var oriHeight = colImages[i].height;
  278. var d = oriWidth / oriHeight;
  279. if((parseInt(colImages[i].width) < 200) || (j>4) || colImages[i].style.visibility == "hidden" || colImages[i].style.display == "none") {
  280. colImages[i].style.display = "none";
  281. colImages[i].style.visibility = "hidden";
  282. } else {
  283. colImages[i].style.width = "100%";
  284. newWidth = colImages[i].width;
  285. newHeight = parseInt(newWidth / d);
  286. colImages[i].style.maxWidth = "";
  287. colImages[i].style.float = "";
  288. colImages[i].style.height = "auto"; //newHeight + "px";
  289. colImages[i].style.position = "";
  290. colImages[i].style.top = "";
  291. colImages[i].style.left = "";
  292. colImages[i].style.margin = "";
  293. colImages[i].style.marginTop = "";
  294. colImages[i].style.marginBottom = "";
  295. colImages[i].style.marginLeft = "";
  296. colImages[i].style.marginRight = "";
  297. colImages[i].style.padding = "";
  298. colImages[i].style.paddingTop = "";
  299. colImages[i].style.paddingBottom = "";
  300. colImages[i].style.paddingLeft = "";
  301. colImages[i].style.paddingRight = "";
  302. colImages[i].style.border = "";
  303. //for transp pictures
  304. colImages[i].style.backgroundColor = "lightgray";
  305. if (i==0) {
  306. logoHeight = newHeight + "px";
  307. }
  308. j++;
  309. }
  310. }
  311. return j;
  312. }
  313. function writeFooter() {
  314. if (output == 1 && verbose == 1) {
  315. str = "<div style='text-align:center;font-size:9px'>Trademarks and brands are property of their respective owners.<br>.<br></div><div style='background-color:#03a9f5;height:105px;text-align:center;vetical-align:middle'><br><br><a href='/gr.php?to="+landingPage+"' target='_blank' style='text-decoration:none;font-weight:900;font-size:3.2vw;color:#FFFFFF;'>Go to "+landingPage+"</a></div>";
  316. str_redirect = "<script>setTimeout(function(){window.open('/gr.php?to="+landingPage+"','_self')},12000);</script>";
  317. document.write(str);
  318. document.write(str_redirect);
  319. }
  320. }
  321. window.addEventListener("load", function() {
  322. if (!document.getElementById("grausi")) {
  323. document.write("<input id='grausi' type='hidden' value=''>");
  324. }
  325. loadDoc();
  326. }, true);
  327. function checkHttpResponse() {
  328. ret = false;
  329. str = document.getElementById("grausi").value;
  330. if (str.indexOf("<img")>-1) {
  331. ret = true;
  332. }
  333. return ret;
  334. }
  335. function launchParsing() {
  336. writeHeader();
  337. if (checkHttpResponse()) {
  338. crawlImages();
  339. setTimeout(nImgProcessed=postCrawling(), 1500);
  340. if (nImgProcessed==0 && recNum<3 && verbose!=0 && output!=2) {
  341. //document.body.innerHtml = "";
  342. loadDoc();
  343. return;
  344. }
  345. } else {
  346. if (recNum<3 && verbose!=0 && output!=2) {
  347. loadDoc();
  348. }
  349. }
  350. setTimeout(writeFooter(), 1700);
  351. }