API tools faq
paste
Advertisement
someone_

threaddl

someone_
Jan 21st, 2015
305
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
Bash 20.69 KB | None | 0 0
raw download clone embed print report diff
  1. #!/bin/bash
  2.  
  3. name=$(basename $0)
  4.  
  5. usage () {
  6.     echo "$name - 4chan thread downloader"
  7.     echo ""
  8.     echo "Usage: $name [4chan thread URL] <time value and/or return target>"
  9.     echo ""
  10.     echo "  <time value>"
  11.     echo "     1: run script once"
  12.     echo "     10 - 999: Set the initial"
  13.     echo "      waiting time between runs"
  14.     echo "     Default: 10 seconds"
  15.     echo ""
  16.     echo "  <return target>"
  17.     echo "     Set the target path of the 'Return'"
  18.     echo "      link"
  19.     echo "     Default: ./"
  20.     exit 0
  21. }
  22.  
  23. [ $# -eq 0 ] && usage
  24.  
  25. for arg in $@; do
  26.     urltest=$(echo "$arg" | sed -e "s_^https_http_" | egrep -o "^http://boards.4chan.org/[a-z0-9]+/thread/[0-9]+")
  27.     timetest=$(echo "$arg" |egrep "^[0-9]{1,3}$")
  28.     if [ $urltest ]; then
  29.         URL="$urltest"
  30.     elif [ $timetest ]; then
  31.         SLP="$arg"
  32.     else
  33.         RET="$arg"
  34.     fi
  35. done
  36.  
  37. [ ! $URL ] && usage
  38.  
  39. echo "4chan downloader"
  40.  
  41. LOC=$(echo "$URL" | sed 's_.*/\([^/]\+\)/thread/\([0-9]\+\)_\1\_\2_')
  42.  
  43. if [ ! $LOC ]; then
  44.     echo "Can't determine the thread's number"
  45.     echo "Use valid URL without hash or search tags"
  46.     echo ""
  47.     usage
  48. fi
  49.  
  50. ST="s.4cdn.org"
  51. [ ! $SLP ] && SLP="10"
  52. [ ! $RET ] && RET="./"
  53. SLAP=$SLP
  54. NO=$(echo "$LOC" | grep -o '[0-9]\+$')
  55. BO=$(echo "$LOC" | grep -o '^[^_]\+')
  56. LM=""
  57. alias wget="wget --referer=\"http://boards.4chan.org/"$BO"\""
  58.  
  59. thejob () {
  60.     if [ ! -d $LOC ]; then
  61.         mkdir $LOC
  62.     fi
  63.  
  64.     if [ ! -d $LOC/misc ]; then
  65.         mkdir $LOC/misc
  66.     fi
  67.  
  68.     cd $LOC/misc
  69.  
  70.     touch images_list
  71.  
  72.     egrep "File: <a[^>]*>[^<]*</a>[^<]*" ../../$LOC.html -o | sed -e 's_^File: <a title="\([^"]*\)" href="[^"]*/\([0-9]\+\....\)"[^/]*/a> (\([^)]\+\))$_\2|\3|\1_g' -e 's_^File: <a href="[^"]*/\([0-9]\+\....\)[^>]*>\([^<]*\)</a> (\([^)]\+\))$_\1|\3|\2_g' -e '/^$/d' -e '$ s_$_\n_' > a
  73.  
  74.     cat images_list a | sed -e '$ s@$@\n'$(ls|grep spoiler)'@' | sort | uniq | sed -e '/^\s*$/d' > images_list
  75.  
  76.     rm a
  77.  
  78.     if [ ! -s gallery.html ]; then
  79.         cat <<EOF > gallery.html
  80. <!DOCTYPE html><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=utf-8" /><meta name="viewport" content="width=device-width, user-scalable=yes, initial-scale=1.0" /><title>Image gallery</title><style type="text/css">/*<![CDATA[*/.loading{display:inline-block;width:0;height:0;border-right:20px solid #39f;border-top:20px solid red;border-left:20px solid yellow;border-bottom:20px solid green;border-radius:20px;-moz-border-radius:20px;-webkit-border-radius:20px;animation:bganim 1.5s ease 0s infinite;-moz-animation:bganim 1.5s ease 0s infinite;-webkit-animation:bganim 1.5s ease 0s infinite}@keyframes bganim{from{transform:rotate(0deg)}to{transform:rotate(360deg)}}@-moz-keyframes bganim{from{-moz-transform:rotate(0deg)}to{-moz-transform:rotate(360deg)}}@-webkit-keyframes bganim{from{-webkit-transform:rotate(0deg)}to{-webkit-transform:rotate(360deg)}}html,body{width:100%;height:100%;margin:0;font-family:Helvetica,Arial,Verdana,sans-serif;text-align:center}div:not(.loading){text-align:center;max-width:250px;padding:5px;border:1px solid black;clear:both;display:inline-block;margin-top:5px}img{max-width:250px;max-height:250px}s{margin-top:3px;display:block;word-wrap:break-word;text-decoration:none}table{border:0;width:100%;height:100%}td{width:50%;text-align:left}td:first-of-type{text-align:right}.middle{vertical-align:middle}/*]]>*/</style></head><body><table><tr><td><div class="loading"></div></td><td class="middle">Loading...</td></tr></table><script type="text/javascript">/*<![CDATA[*/var d=document,\$=function(a){return d.querySelector(a)},html="<h1>Image gallery</h1>",x=new XMLHttpRequest(),td=\$(".middle"),r;x.onreadystatechange=function(){if(x.readyState==4){if(x.status==200){td.innerHTML="Parsing gallery...";var b = x.responseText.split("\n");b.some(function(a){if(b.lastIndexOf(a)==0&&a.indexOf('|')==-1)return;a=a.split("|");if(a.length>1){html+='<div><a href="../'+a[0]+'"><img src="'+a[1].indexOf('Spoiler')==0?b[b.length-1]:(a[0].split(".")[0]+"s.jpg")+'" alt="Image" /></a><s>'+a[1]+'</s><s>'+a[2]+'</s></div>'}return});\$("body").innerHTML=html}else{\$("td").innerHTML="Er<br>Are you trying to";td.innerHTML="ror<br>open this offline?"}}};x.open("GET","images_list",true);x.setRequestHeader("Cache-Control","no-cache");td.innerHTML="Fetching images list...";x.send(null)//]]></script></body></html>
  81. EOF
  82.     fi
  83.  
  84.     cd ../..
  85.  
  86.     egrep "//.\.t\.4cdn\.org/[^.]+\.jpg" $LOC.html -o | sed 's_^//_http:&_g' > $LOC/misc/misc
  87.  
  88.     egrep "//${ST}/image/[^.]+\...." $LOC.html -o | sed 's_^//_http:&_g' | uniq >> $LOC/misc/misc
  89.  
  90.     egrep "//${ST}/image/country/[^.]+\...." $LOC.html -o | sed 's_^//_http:&_g' >> $LOC/misc/misc
  91.  
  92.     egrep "//${ST}/css/[a-z]+\.[0-9]+\.css" $LOC.html -o | sed -e 's_^//_http:&_' | head -n1 > $LOC/misc/css
  93.  
  94.     egrep "//${ST}/css/[a-z]+\.[0-9]+\.css" $LOC.html -o | sed -e 's_^//_http:&_' | egrep 'tomorrow|prettify' >> $LOC/misc/css
  95.  
  96.     egrep 'data-src="[^.]+\.[^"]+' $LOC.html -o | sed 's_^data.src."_http://'$ST'/image/title/_' | head -n1 > $LOC/misc/logo
  97.  
  98.     egrep "//i\.4cdn\.org/[^.]+\.(jpg|png|gif|webm)" $LOC.html -o | sed 's_^//_http:&_g' > $LOC/images
  99.  
  100.     sed -i -e 's@\(</head>\)@\n\1@' $LOC.html
  101.  
  102.     mv $LOC.html a
  103.  
  104.     head -n1 a > $LOC.html
  105.  
  106.     cat << EOF >> $LOC.html
  107. <!<script>/*<![CDATA[*/var d=document,dE=d.documentElement,lS=localStorage,z,i,u="threaddl_arch_theme",e;Element.prototype.gA=function(a){return this.getAttribute(a);};Element.prototype.pN=function(){return this.parentNode;};String.prototype.iO=function(a){return this.indexOf(a);};function s(a,b,c){a.setAttribute(b,c);}function sT(b){lS.setItem(u,b);for(i=0;(z=d.getElementsByTagName("link")[i]);i++){if(z.gA("rel").iO("style")!=1&&z.gA("title")){z.disabled=true;if(z.gA("title")==(b?"switch":"Tomorrow"))z.disabled=false;}}}sT(parseInt(lS.getItem(u))||0);(function(){var a,b,c,e,f,g,h,i,j,k,l=parseInt(lS.qp_opt),m,n=parseInt(lS.img_hover),o;String.prototype.reverse=function(){return this.split("").reverse().join("")};k={hover:function(a){var b,c,d,e,f,g,h;d=a.clientX,e=a.clientY;g=k.el.style;b=dE.clientHeight,c=dE.clientWidth;f=k.el.offsetHeight;h=e-120;g.top=b<=f||h<=0?"0px":h+f>=b?b-f+"px":h+"px";if(d<=c-400){g.left=d+45+"px";return g.right=null}else{g.left=null;return g.right=c-d+45+"px"}},hoverend:function(){a.rm(k.el);return delete k.el}};a=function(a,b){if(b==null){b=d.body}return b.querySelector(a)};a.extend=function(a,b){var c;for(c in b){a[c]=b[c]}};a.extend(a,{id:function(a){return d.getElementById(a)},addStyle:function(b){var c;c=a.el("style",{textContent:b});a.add(d.head,c);return c},x:function(a,b){b===null&&(b=d.body);return d.evaluate(a,b,null,8,null).singleNodeValue},addClass:function(a,b){return a.classList.add(b)},rmClass:function(a,b){return a.classList.remove(b)},rm:function(a){return a.pN().removeChild(a)},tn:function(a){return d.createTextNode(a)},nodes:function(a){var b,c,e=0;if(!(a instanceof Array)){return a}b=d.createDocumentFragment();for(;e<a.length;e++){c=a[e];b.appendChild(c)}return b},add:function(b,c){return b.appendChild(a.nodes(c))},after:function(b,c){return b.pN().insertBefore(a.nodes(c),b.nextSibling)},el:function(b,c){var e=d.createElement(b);if(c){a.extend(e,c)}return e},on:function(a,b,c){var d,e=0,f=b.split(" ");for(;e<f.length;e++){d=f[e];a.addEventListener(d,c,false)}},off:function(a,b,c){var d,e=0,f=b.split(" ");for(;e<f.length;e++){d=f[e];a.removeEventListener(d,c,false)}},visible:function(a){var b = a.getBoundingClientRect();return (b.top+b.height>=0&&(dE.clientHeight-b.bottom)+b.height>=0)}});b=function(a,b){if(b==null){b=d.body}return [].slice.call(b.querySelectorAll(a))};c={post:function(b,d){var e=a.id("pc"+b);e&&a.add(d,c.cleanPost(e.cloneNode(true)))},cleanPost:function(c){var d,e,f,g,h,i=Date.now(),j=a(".post",c),k=_j=_k=_l=0,l=[].slice.call(c.childNodes),m=b(".inline",j),n=b(".inlined",j);for(;k<l.length;k++){if((d=l[k])!==j){a.rm(d)}}for(;_j<m.length;_j++){a.rm(m[_j])}for(;_k<n.length;_k++){a.rmClass(n[_k],"inlined")}(f=b("[id]",c)).push(c);for(;_l<f.length;_l++){(e=f[_l]).id=""+i+"_"+e.id}a.rmClass(c,"forwarded");a.rmClass(c,"qphl");a.rmClass(j,"highlight");a.rmClass(j,"qphl");return c}};j={init:function(){return e.callbacks.push(this.node)},node:function(b){var c=a(".postInfo > .dateTime",b.el);if(b.isInlined){return}j.date=new Date(c.dataset.utc*1e3);c.title="4chan time: "+c.textContent;return c.textContent=j.zeroPad(j.date.getMonth()+1)+"/"+j.zeroPad(j.date.getDate())+"/"+(j.date.getFullYear()-2e3)+"("+j.day[j.date.getDay()]+")"+j.zeroPad(j.date.getHours())+":"+j.zeroPad(j.date.getMinutes())},day:["Sun","Mon","Tue","Wed","Thu","Fri","Sat"],zeroPad:function(a){if(10>a){return"0"+a}else{return a}}};g={init:function(){return e.callbacks.push(this.node)},node:function(b){var c,d,e,f,g,j,k={},n=0,m=b.quotes;if(b.isInlined){return}for(;m.length>n;n++){j=m[n];if(g=j.hash.slice(2)){k[g]=true}}c=a.el("a",{href:"#p"+b.ID,className:"backlink",textContent:">>"+b.ID});for(g in k){if(!(e=a.id("pi"+g))||/\bop\b/.test(e.pN().className)){continue}f=c.cloneNode(true);l&&a.on(f,"mouseover",i.mouseover);a.on(f,"click",h.toggle);if(!(d=a.id("blc"+g))){d=a.el("span",{className:"container",id:"blc"+g});a.add(e,d)}a.add(d,[a.tn(" "),f])}}};h={init:function(){return e.callbacks.push(this.node)},node:function(b){var c,d=_j=0,e=b.quotes,f=b.backlinks;for(;e.length>d;d++){c=e[d];if(!c.hash){continue}c.removeAttribute("onclick");a.on(c,"click",h.toggle)}for(;f.length>_j;_j++){a.on(f[_j],"click",h.toggle)}},toggle:function(b){if(b.shiftKey||b.altKey||b.ctrlKey||b.metaKey||b.button!==0){return}b.preventDefault();var c=this.dataset.id||this.hash.slice(2);if(/\binlined\b/.test(this.className)){h.rm(this,c)}else{if(a.x("ancestor::div[contains(@id,'p"+c+"')]",this)){return}h.add(this,c)}return this.classList.toggle("inlined")},add:function(b,d){var e=d,f=a.id("p"+e),g=a.el("div",{id:"i"+e,className:"inline"}),h=/\bbacklink\b/.test(b.className),i=h?b.pN():a.x("ancestor-or-self::*[parent::blockquote][1]",b);a.after(i,g);c.post(e,g);if(!f){return}if(h){a.addClass(f.pN(),"forwarded");++f.dataset.forwarded||(f.dataset.forwarded=1)}},rm:function(c,d){var e,f,g,h,i;e=a.x("following::div[@id='i"+d+"']",c);a.rm(e);i=b(".backlink.inlined",e);for(g=0,h=i.length;h>g;g++){f=i[g];e=a.id(f.hash.slice(1));if(!--e.dataset.forwarded){a.rmClass(e.pN(),"forwarded")}}if(/\bbacklink\b/.test(c.className)){e=a.id("p"+d);if(!--e.dataset.forwarded){return a.rmClass(e.pN(),"forwarded")}}}};i={init:function(){return e.callbacks.push(this.node)},node:function(b){var c,d=_j=0,e=b.quotes,f=b.backlinks;for(;e.length>d;d++){c=e[d];if(c.hash||/\bdeadlink\b/.test(c.className)){a.on(c,"mouseover",i.mouseover)}}for(;f.length>_j;_j++){a.on(f[_j],"mouseover",i.mouseover)}},mouseover:function(e){var f,g=this.hash.slice(2),h,j,l,m=0,n;if(/\binlined\b/.test(this.className)){return}if(h=a.id("qp")){if(h===k.el){delete k.el}a.rm(h)}if(k.el){return}h=k.el=a.el("div",{id:"qp",className:"reply dialog"});k.hover(e);a.add(d.body,h);f=a.id("p"+g);c.post(g,h);a.on(this,"mousemove",k.hover);a.on(this,"mouseout click",i.mouseout);if(!f){return}if(/\bop\b/.test(f.className)){a.addClass(f.pN(),"qphl")}else{a.addClass(f,"qphl")}l=a.x("ancestor::*[@id][1]",this).id.match(/\d+$/)[0];n=b(".quotelink, .backlink",h);for(;n.length>m;m++){j=n[m];if(j.hash.slice(2)===l){a.addClass(j,"forwardlink")}}},mouseout:function(b){var c;k.hoverend();if(c=a.id(this.hash.slice(1))){a.rmClass(c,"qphl");a.rmClass(c.pN(),"qphl")}a.off(this,"mousemove",k.hover);return a.off(this,"mouseout click",i.mouseout)}};m={init:function(){return e.callbacks.push(this.node)},node:function(b){if(!b.img){return}return a.on(b.img,"mouseover",m.mouseover)},mouseover:function(){var b,c;if(b=a.id("ihover")){if(b===k.el){delete k.el}a.rm(b)}if(k.el){return}c={id:'ihover',src:this.pN().href};c.src.reverse().iO('m')==0&&a.extend(c,{autoplay:true,loop:true});b=k.el=a.el(c.loop?'video':'img',c);a.add(d.body,b);a.on(b,"load",m.load);a.on(this,"mousemove",k.hover);return a.on(this,"mouseout",m.mouseout)},load:function(){var a;if(!this.pN()){return}a=this.style;return k.hover({clientX:-45+parseInt(a.left),clientY:120+parseInt(a.top)})},mouseout:function(){k.hoverend();a.off(this,"mousemove",k.hover);return a.off(this,"mouseout",m.mouseout)}};o={videos:[],init:function(){return e.callbacks.push(this.node);a.on(window,'resize scroll visibilitychange',o.videoHandler)},node:function(b){if(!b.img){return}return a.on(b.img.pN(),'click',o.cb.toggle)},cb:{toggle:function(e){if(e.shiftKey||e.altKey||e.ctrlKey||e.metaKey||e.button!==0){return}e.preventDefault();return o.toggle(this)}},toggle:function(b){var c,e=b.firstChild;if(e.hidden){c=b.getBoundingClientRect();if(d.body.gA("class")!=='i'){if(c.top<0){d.body.scrollTop+=c.top-42}if(c.left<0){d.body.scrollLeft+=c.left}}else{if(c.top<0){dE.scrollTop+=c.top-42}if(c.left<0){dE.scrollLeft+=c.left}}return o.contract(e)}else{return o.expand(e)}},contract:function(b){var c=b.nextSibling,d;b.hidden=false;c.hidden=true;if(c.loop){d=o.videos;d=d.splice(d.indexOf(a.rm(c)),1)}},expand:function(b){var c,d,e,f;if(a.x('ancestor-or-self::*[@hidden]',b)){return}b.hidden=true;if((d=b.nextSibling)&&(d.nodeName=='IMG'||(e=d.nodeName =='VIDEO'))){d.hidden=false;e&&d.play();return}c=b.pN();f=c.href;d=f.reverse().iO('m')==0?a.el('video',{src:f,autoplay:!0,loop:!0}):a.el('img',{src:f,style: 'width:100px;height:100px;'});if(d.loop){o.videos.push(d);a.on(d,'canplay',o.videoHandler)}else{!d.naturalWidth?a.on(d,'load',function(){d.removeAttribute("style")}):d.removeAttribute("style")}return a.add(c,d)},videoHandler:function(){var b,c=o.videos,v;for(b=0;c.length>b;b++){v=c[b];a.visible(v)&&!d.hidden?v.play():v.pause()}}};e={init:function(){(function(){var c=b('a.fileThumb'),i=0;/WebKit/.test(navigator.userAgent)||s(d.body,'class','i');for(;c.length>i;i++){var n=a.nodes([a.tn(' '),a.el('a',{textContent:'google',href:'http://www.google.com/searchbyimage?image_url='+c[i].href,target:'_blank'})]);-1!==location.protocol.iO('http')&&c[i].previousElementSibling.appendChild(n)}for(c=b('span.abbr a'),i=0;c.length>i;i++){s(c[i],'onclick','javascript:e=this.pN().pN().lastChild.style;e.display=e.display=="block"?"none":"block"')}}());a.add(a("div.navLinks"),[a.tn(" ["),a.el("a",{href:"javascript:;",textContent:"QuotePreview is "+(l?"ON":"OFF"),onclick:function(){lS.qp_opt=l?0:1;window.location.reload()}}),a.tn("] ["),a.el("a",{href:"javascript:;",textContent:"ImageHover is "+(n?"ON":"OFF"),onclick:function(){lS.img_hover=n?0:1;window.location.reload()}}),a.tn("] ["),a.el("a",{href:"./"+window.location.href.split("/").pop().split(".")[0]+"/misc/gallery.html",textContent:"Image only view"}),a.tn("]")]);j.init();h.init();l&&i.init();n&&m.init();o.init();g.init();e.ready()},ready:function(){var c=[],d=0,f=b(".postContainer",a.id("delform"));for(;f.length>d;d++){c.push(e.preParse(f[d]))}e.node(c);if(MutationObserver=window.MutationObserver||window.WebKitMutationObserver||window.OMutationObserver){observer=new MutationObserver(e.observer);observer.observe(a(".board"),{childList:true,subtree:true})}else{a.on(a(".board"),"DOMNodeInserted",e.listener)}},preParse:function(c){var d=a(".post",c),e=c.pN().className,f={root:c,el:d,"class":d.className,ID:d.id.match(/\d+$/)[0],threadID:a.x("ancestor::div[parent::div[@class='board']]",c).id.match(/\d+$/)[0],isInlined:/\binline\b/.test(e),blockquote:d.lastElementChild,quotes:b("a.quotelink[href^='#p']",d),backlinks:d.getElementsByClassName("backlink"),img:false},g;if(g=a('img[data-md5]',d)){f.img=g}return f},node:function(a){for(var b=0,c=e.callbacks;c.length>b;b++){var d=c[b];try{for(var f=0;a.length>f;f++){d(a[f])}}catch(g){alert("Error: "+g.message+"\nReport the bug to HandyAnon@Steam\n\nURL: "+window.location+"\n"+g.stack)}}},observer:function(a){var b,c,d,f,g,h,i,j;d=[];for(f=0,h=a.length;h>f;f++){c=a[f];j=c.addedNodes;for(g=0,i=j.length;i>g;g++){b=j[g];if(/\bpostContainer\b/.test(b.className)){d.push(e.preParse(b))}}}if(d.length){return e.node(d)}},listener:function(a){var b;b=a.target;if(/\bpostContainer\b/.test(b.className)){return e.node([e.preParse(b)])}},callbacks:[]};a.on(d,"DOMContentLoaded",e.init)}).call(this)//]]></script><style>/*<![CDATA[*/img[data-md5]+*{max-width:100%!important;}.i img[data-md5]+*{width:100%!important;}.op:after{clear:both;content:'';display:block;}#qp{padding:2px 2px 5px;position:fixed;border:1px solid rgba(128,128,128,0.5);}#qp .post{border:none;margin:0;padding:0;}#qp img,#qp video{max-height:300px;max-width:500px;}.qphl{outline:2px solid rgba(216,94,49,.7);}.inlined{opacity:.5;}.inline{border:1px solid rgba(128,128,128,0.5);display:table;margin:2px;padding:2px;}.inline .post{background:none;border:none;margin:0;padding:0;}.forwarded{display:none;}.quotelink.forwardlink,.backlink.forwardlink{text-decoration:none;border-bottom:1px dashed;}#ihover{max-height:97%;max-width:75%;padding-bottom:18px;position:fixed;} blockquote{word-wrap:break-word;min-width:120px;}/*]]>*/</style>
  108. EOF
  109.  
  110.     tail -n1 a >> $LOC.html
  111.  
  112.     rm a
  113.  
  114.     sed -i -e '1 {s_<script_!>\n&_
  115.         s_<link [^>]*RSS feed[^>]*>__
  116.         s@//'$ST'/image/\(favicon[^.]*\.ico\)@'$LOC'/misc/\1@
  117.         s_<link rel="alternate style[^-]*\(<link[^>]*tomorrow\.[^>]*>\)<link[^>]*>_\1_
  118.         s@//'$ST'/css/\([^.]\+\.[^.]\+\.css\)@'$LOC'/misc/\1@g}' -e '$ {s_</head>_<!&_
  119.         s_<div id="boardNavDesktop" class="desktop">_\n_
  120.         s_<div class="boardBanner"_\n<!&_
  121.  s@ data-src="[^.]\+\.\([^"]\+\)">@><img alt="4chan" src="'$LOC'/misc/logo.\1" />@
  122.         s_<hr class="abovePost_\n_
  123.         s_ .<a[^>]*>Catalog</a>.__g
  124.         s_\(<div class="navLinks desktop">.<a href="/[^/]\+/[^#]*\)#bottom\(">Bottom</a>.\)_\n<!\1javascript:dE.scrollIntoView(false)\2</div><hr>\n_
  125.         s_\(<form name="delform" id="delform"\)[^>]*_\n<!\1_
  126.         s@//.\.t\.4cdn\.org/[^/]*/\([0-9]*s\.jpg\)@'$LOC'/misc/\1@g
  127.         s@//i\.4cdn\.org/[^/]*/\([0-9]*\....\)@'$LOC'/\1@g
  128.         s@//'$ST'/image/title/[a-z0-9-]*\.\(...\)@'$LOC'/misc/logo.\1@g
  129.         s@//'$ST'/image/\(spoiler[^.]*\....\)@'$LOC'/misc/\1@g
  130.         s@//'$ST'/image/\(filedeleted-res\.gif\)@'$LOC'/misc/\1@g
  131.         s@//'$ST'/image/country/\([^.]*\....\)@'$LOC'/misc/\1@g
  132.         s@//'$ST'/image/\([a-z]*icon.gif\)@'$LOC'/misc/\1@g
  133.         s@//'$ST'/image/\(archived.gif\)@'$LOC'/misc/\1@g
  134.         s_<div data.tip[^>]*>[^<]*<div>__g
  135.         s_\(<a href="\)'$NO'#p_\1#p_g
  136.         s_<a href="#p'$NO'"[^>]*>&gt;&gt;'$NO'_& (OP)_g
  137.         s:\(<a href="\)\([0-9]\+\)\(#p[0-9]*\)\([^<]*\):\1'$BO'_\2.html" target="_blank" \3 (Cross-thread):g
  138.         s_\(<div class="navLinks navLinksBot desktop">.<a href="/[^/]\+/"[^>]*>Return</a>. .<a href="\)#top\(">Top</a>.\)_\n<!<span style="float:right;">Style: [ <a href="javascript:sT(1);dE.scrollIntoView(false)">Default</a> | <a href="javascript:sT(0);dE.scrollIntoView(false)">Tomorrow</a> ]</span>\1javascript:dE.scrollIntoView()\2\n_
  139.         s@</body>@\n<!</div></div></form>&@}' -e '{s:\(<a href="\)/./\([^>]*>Return</a>\):\1'"$RET"'\2:g
  140.         s_\(<a[^>]*href="\)//_\1http://_g
  141.         s_\(<a[^>]*\) onclick="replyhl[^"]*"_\1_g
  142.         s_\(<a href="javascript:\)quote[^>]*_\1;"_g
  143.         s_<div class="postInfoM_\n_g
  144.         s_<div class="file"_\n<!&_g
  145.         s_<div class="postInfo _\n<!&_g
  146.         s_<a href="/ic\?/anim\.php?file=[0-9]*" target="\_blank">_<a title="View is supported only on 4chan">_g
  147.         s_\(<input type="checkbox"\)[^>]*_\1_g
  148.         s_<a onclick="toggle..exif[^"]*"_<a_g
  149.         s_<div class="mFileInfo mobile">[^<]*</div>__g
  150.         s_<wbr>__g}' $LOC.html
  151.  
  152.     grep '^<!' $LOC.html > a
  153.  
  154.     sed -i -e '2,$ s_^<!__' a
  155.  
  156.     tr -d '\n' < a > $LOC.html
  157.  
  158.     sed -i -e 's_!>_\n_g' $LOC.html
  159.  
  160.     rm a
  161.  
  162.     cd $LOC/misc
  163.  
  164.     CSS=$(basename `head -n1 css`)
  165.  
  166.     if [ "$CSSt" != "$CSS" ]; then
  167.         CSSt=CSS
  168.         if [ "$(ls|grep '.css')" ]; then
  169.             rm *.css
  170.         fi
  171.         wget -q -nc -i css
  172.         wget -q -nc "$(grep -o 'fade[^.]*\.png' $CSS | sed -e 's_.*_http://'$ST'/image/&_')"
  173.         sed -i -e 's_/image/\(fade[^.]*\.png\)_\1_g' $CSS
  174.     fi
  175.  
  176.     if [ "$(ls|grep logo.)" ]; then
  177.         rm "$(ls|grep logo.|head -n1)"
  178.     fi
  179.  
  180.     wget -q -i logo -O "logo.$(sed 's_\._\n_g' logo|tail -n1)"
  181.  
  182.     rm logo css
  183.  
  184.     touch .nomedia
  185.  
  186.     cd ..
  187.  
  188.     for image in $(cat images); do
  189.         wget -q -nc $image
  190.     done
  191.  
  192.     rm images
  193.  
  194.     cd misc
  195.  
  196.     for misc in $(cat misc); do
  197.         wget -q -nc $misc
  198.     done
  199.  
  200.     rm misc
  201.  
  202.     cd ../..
  203. }
  204.  
  205. exito () {
  206.     echo "Session completed. Exiting"
  207.     exit 0
  208. }
  209.  
  210. echo ""
  211. echo "Downloading to $LOC"
  212. echo "------------------------------"
  213.  
  214. while true; do
  215.     trap exito 1 2 3 15
  216.  
  217.     stat="$(wget -S --spider "$URL" 2>&1)"
  218.  
  219.     if [ "$(echo "$stat" |grep '404 Not Found')" ]; then
  220.         if [ -s $LOC.html ]; then
  221.             echo "Thread has 404'd or 4chan is down. Stopping script"
  222.         else
  223.             echo "Thread does not exist. Stopping script"
  224.         fi
  225.         exit 0
  226.     fi
  227.  
  228.     if [ "$LM" != "$(echo "$stat" |grep Last-Modified)" ]; then
  229.         LM="$(echo "$stat" |grep Last-Modified)"
  230.         if [ $SLP -gt 1 ] && [ $SLP -lt 10 ]; then
  231.             SLP="10"
  232.         elif [ $SLP -gt 999 ]; then
  233.             SLP="999"
  234.         fi
  235.         SLAP=$SLP
  236.         wget -np -nd -nH -q -erobots=off $URL -O $LOC.html
  237.         if [ "$(egrep -o -e '<div class=\"closed\">' "$LOC".html)" ]; then
  238.             echo "Thread is archived on 4chan. Downloading once"
  239.             SLP="1"
  240.         fi
  241.         thejob
  242.     else
  243.         SLAP=`expr "$SLAP" + "5"`
  244.     fi
  245.  
  246.     trap - 1 2 3 15
  247.  
  248.     if [ $SLP = "1" ]; then
  249.         exito
  250.     fi
  251.  
  252.     echo -ne OK
  253.  
  254.     sleep $SLAP
  255.  
  256.     echo -ne "\b\b  \b\b"
  257. done
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement
Public Pastes
Advertisement
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!