Skip to content

Instantly share code, notes, and snippets.

@jacobk

jacobk/menu_scrape.js

Created February 3, 2011 22:10
Show Gist options
  • Save jacobk/810331 to your computer and use it in GitHub Desktop.
Save jacobk/810331 to your computer and use it in GitHub Desktop.
Download ZIP
Raw
menu_scrape.js
var hp = "<html xmlns=\x22http://www.w3.org/1999/xhtml\x22><head><link href=\x22css/style.css\x22 type=\x22text/css\x22 rel=\x22stylesheet\x22><link href=\x22css/fontstyle.css\x22 type=\x22text/css\x22 rel=\x22stylesheet\x22><title>Meny</title><meta name=\x22robots\x22 content=\x22index,follow\x22><meta name=\x22googlebot\x22 content=\x22index, follow\x22><meta http-equiv=\x22Content-Language\x22 content=\x22SE\x22><meta name=\x22description\x22 content=\x22\x22><meta name=\x22keywords\x22 content=\x22\x22><meta name=\x22distribution\x22 content=\x22Global\x22><meta name=\x22rating\x22 content=\x22general\x22><meta name=\x22copyright\x22 content=\x224D Profile\x22><meta name=\x22author\x22 content=\x22Marks & Bleeds\x22><script src=\x22jscripts/AC_RunActiveContent.js\x22 language=\x22javascript\x22></script><script type=\x22text/javascript\x22 src=\x22jscripts/prototype.js\x22></script><script type=\x22text/javascript\x22 src=\x22jscripts/scriptaculous.js?load=effects,builder\x22></script><script type=\x22text/javascript\x22 src=\x22jscripts/lightbox.js\x22></script><link rel=\x22stylesheet\x22 href=\x22css/lightbox.css\x22 type=\x22text/css\x22 media=\x22screen\x22 /></head><body><div align=\x22center\x22><br /><br /><div style=\x22width: 100%; align: center;\x22><div style=\x22width: 800px; height:15px;\x22 align=\x22right\x22><div class=\x22menutext\x22 style=\x22\x22 onmouseover=\x22this.className=\x27menutexth\x27\x22 onmouseout=\x22this.className=\x27menutext\x27\x22><a href=\x22Kontakt\x22>Kontakt</a> &nbsp; &nbsp; &nbsp; &nbsp; </div><div class=\x22menutext\x22 style=\x22\x22 onmouseover=\x22this.className=\x27menutexth\x27\x22 onmouseout=\x22this.className=\x27menutext\x27\x22><a href=\x22Oppettider\x22>÷ppettider</a> &nbsp; &nbsp; &nbsp; &nbsp; </div><div class=\x22menutext\x22 style=\x22\x22 onmouseover=\x22this.className=\x27menutexth\x27\x22 onmouseout=\x22this.className=\x27menutext\x27\x22><a href=\x22Fest\x22>Fest</a> &nbsp; &nbsp; &nbsp; &nbsp; </div><div class=\x22menutext\x22 style=\x22\x22 onmouseover=\x22this.className=\x27menutexth\x27\x22 onmouseout=\x22this.className=\x27menutext\x27\x22><a href=\x22Catering\x22>Catering</a> &nbsp; &nbsp; &nbsp; &nbsp; </div><div class=\x22menutexth\x22 style=\x22\x22 onmouseover=\x22this.className=\x27menutexth\x27\x22 onmouseout=\x22this.className=\x27menutext\x27\x22><a href=\x22Meny\x22>Meny</a> &nbsp; &nbsp; &nbsp; &nbsp; </div><div class=\x22menutext\x22 style=\x22\x22 onmouseover=\x22this.className=\x27menutexth\x27\x22 onmouseout=\x22this.className=\x27menutext\x27\x22><a href=\x22Index\x22>V‰lkommen</a> &nbsp; &nbsp; &nbsp; &nbsp; </div></div></div><div style=\x22min-height: 600px; width: 800px; border: solid 0px #000;\x22 align=\x22left\x22><p>&nbsp;</p><p style=\x22text-align: center;\x22><table border=\x220\x22 cellspacing=\x2215\x22><tbody><tr><td valign=\x22top\x22><img src=\x22http://www.restauranghojdpunkten.se/images/folders/Logo/side3.jpg\x22 alt=\x22\x22 width=\x22300\x22 height=\x221000\x22 /></td><td valign=\x22top\x22><table class=\x22holderSide\x22 border=\x220\x22 cellspacing=\x2215\x22><tbody><tr><td style=\x22width: 50%;\x22 valign=\x22top\x22><h2><strong>Meny vecka 5</strong></h2><br /><p style=\x22text-align: justify;\x22><strong>Sallad, hembakat br&ouml;d, dryck, kaffe och te ing&aring;r.</strong></p><p style=\x22text-align: left;\x22>&nbsp;</p><p style=\x22text-align: left;\x22><br /><br /><strong>M&aring;ndag</strong></p><p style=\x22text-align: left;\x22>&nbsp;</p><p style=\x22text-align: left;\x22><strong></strong></p><p>Karl - Johans gryta med paprikaris.</p><p>&nbsp;</p><p><br />Isterband med persiljestuvad potatis, saltgurka och r&ouml;dbetor.</p><p>&nbsp;<br />Kycklingsallad med stekt bacon <br />(avh. 50 kr/ serve.60kr)<br />&nbsp;</p><p>&nbsp;</p><p>&nbsp;</p><p>&nbsp;</p><p>&nbsp;</p><p>&nbsp;</p><p>&nbsp;</p><p><strong>Onsdag<br /></strong></p><p>&nbsp;</p><p style=\x22text-align: left;\x22>Wallenbergare med kokt potatis, gr&auml;dds&aring;s och gr&ouml;na &auml;rtor</p><p style=\x22text-align: left;\x22><br />Potatis och purjol&ouml;kssoppa med strimlad r&ouml;kt skinka.</p><p style=\x22text-align: left;\x22><br />Laxsallad med dill och citron <br />(avh. 50 kr/ serve.60kr) <br /><br />&nbsp;</p><p style=\x22text-align: left;\x22><br /><strong>&nbsp;</strong></p><p><strong></strong></p><p><strong>Fredag</strong></p><p>&nbsp;</p><p>Helstekt Kotlettrad med potatisgrat&auml;ng</p><p><br />Nasigoreng</p><p><br />Ost och skinka sallad <br />(avh. 50 kr/ serve.60kr). <br />&nbsp;<br />&nbsp;</p><p>&nbsp;</p><p>&nbsp;</p></td><td style=\x22width: 50%;\x22 valign=\x22top\x22><h2>&nbsp;70: -</h2><p>&nbsp;</p><p style=\x22text-align: left;\x22>&nbsp;</p><p><strong></strong>&nbsp;<strong>&nbsp;</strong></p><p style=\x22text-align: left;\x22><strong>&nbsp;</strong></p><p style=\x22text-align: left;\x22><strong>&nbsp;</strong></p><p style=\x22text-align: left;\x22>&nbsp;</p><p style=\x22text-align: left;\x22><strong></strong></p><p style=\x22text-align: left;\x22><strong></strong></p><p style=\x22text-align: left;\x22><strong></strong></p><p style=\x22text-align: left;\x22><strong></strong></p><p style=\x22text-align: left;\x22><strong></strong></p><p style=\x22text-align: left;\x22><strong></strong></p><p style=\x22text-align: left;\x22><strong>Tisdag</strong></p><p style=\x22text-align: left;\x22>&nbsp;</p><p style=\x22text-align: left;\x22>Stekt sill med potatismos skirat sm&ouml;r och lingon.</p><p style=\x22text-align: left;\x22><br />Pasta med oxk&ouml;tt, soltorkad tomat och champinjoner.</p><p style=\x22text-align: left;\x22><br />Grekisksallad med tzatziki <br />(avh. 50 kr/ serve.60kr) <br />&nbsp;</p><p>&nbsp;</p><p>&nbsp;</p><p>&nbsp;</p><p>&nbsp;</p><p>&nbsp;</p><p>&nbsp;</p><p><strong>Torsdag<br /><br /></strong></p><p>Kycklingfil&eacute; med paprika crem&eacute; s&aring;s och ris.</p><p><br />Pasta med tomat, f&auml;nk&aring;l, r&auml;kor och r&ouml;kt lax.</p><p><br />Paj med stekt bacon och purjol&ouml;k (avh. 50 kr/ serve.60kr).</p></td></tr><tr><td>&nbsp;</td><td>&nbsp;</td></tr></tbody></table><p>&nbsp;</p></td></tr><tr><td>&nbsp;</td><td>&nbsp; <table class=\x22holderSide\x22 border=\x220\x22 cellspacing=\x2215\x22><tbody><tr><td style=\x22width: 100%;\x22 valign=\x22top\x22><h2>&nbsp;&Agrave; la carte</h2><p>&nbsp; <table style=\x22width: 429px; height: 94px;\x22 border=\x220\x22><tbody><tr><td>L&ouml;vbiff med idahopotatis, stekta gr&ouml;nsaker, r&ouml;dvins.&nbsp;&amp; persiljes.</td><td style=\x22text-align: right;\x22>80 kr</td></tr><tr><td>&nbsp;Schnitzel med idahopotatis, stekta gr&ouml;nsaker,&nbsp;bruns&aring;s och citron</td><td style=\x22text-align: right;\x22>75 kr</td></tr><tr><td>&nbsp;Kycklingfil&eacute; med idahopotatis, stekta gr&ouml;nsaker och s&aring;s</td><td style=\x22text-align: right;\x22>75 kr</td></tr></tbody></table></p><p>&nbsp;</p><h2>Hamburgare</h2><p><table style=\x22width: 429px; height: 94px;\x22 border=\x220\x22><tbody><tr><td>Originalburgare med pommes och aioli</td><td style=\x22text-align: right;\x22>70 kr</td></tr><tr><td>Tigerburgare med ost och bacon</td><td style=\x22text-align: right;\x22>75 kr</td></tr><tr><td>Mexikansk burgare med guacamole, och salsa</td><td style=\x22text-align: right;\x22>80 kr</td></tr><tr><td>Kycklingburgare med bacon samt pepparrot &amp; Dijondressing</td><td style=\x22text-align: right;\x22>80 kr</td></tr></tbody></table></p><p>&nbsp;</p><h2>Varma pastar&auml;tter</h2><p><table style=\x22width: 429px; height: 94px;\x22 border=\x220\x22><tbody><tr><td>Kyckling pasta med soltorkad tomat och basilika</td><td style=\x22text-align: right;\x22>70 kr</td></tr><tr><td>Pasta med lax och Skaldjur</td><td style=\x22text-align: right;\x22>70 kr</td></tr><tr><td>Pasta bolognese</td><td style=\x22text-align: right;\x22>70 kr</td></tr><tr><td>Pasta med s&auml;songens gr&ouml;nsaker</td><td style=\x22text-align: right;\x22>70 kr</td></tr></tbody></table></p><p>&nbsp;</p><h2>Bakad potatis</h2><p><table style=\x22width: 429px; height: 94px;\x22 border=\x220\x22><tbody><tr><td>Kyckling och bacon</td><td style=\x22text-align: right;\x22>&nbsp;55 kr</td></tr><tr><td>Skagen</td><td style=\x22text-align: right;\x22>&nbsp;55 kr</td></tr><tr><td>Vegetarisk</td><td style=\x22text-align: right;\x22>&nbsp;55 kr</td></tr></tbody></table></p></td></tr></tbody></table></td></tr></tbody></table></p><p>&nbsp;</p><p>&nbsp;</p></div><div style=\x22height: 50px; width: 800px;\x22 class=\x22content\x22>&nbsp;</div></div></body></html>"
var ideon = "<!DOCTYPE HTML PUBLIC \x22-//W3C//DTD HTML 4.01 Transitional//EN\x22 \x22http://www.w3.org/TR/html4/loose.dtd\x22><html><head><title></title><meta name=\x22keywords\x22 content=\x22\x22><meta name=\x22description\x22 content=\x22\x22><meta http-equiv=\x22Content-Type\x22 content=\x22text/html; charset=ISO-8859-1\x22><link rel=\x22stylesheet\x22 type=\x22text/css\x22 href=\x22/global/styleTempl/garsila2/structure-0908.css\x22><link rel=\x22stylesheet\x22 type=\x22text/css\x22 href=\x22/global/styleTempl/garsila2/style.css\x22><link rel=\x22stylesheet\x22 type=\x22text/css\x22 href=\x22/global/styleTempl/garsila2/menuTabs.css\x22><style type=\x22text/css\x22>BODY, TD{color: #505659;font-family: verdana,geneva;}A, #mainContent A, #sideEditor A{color: #786755;}#mainContent A, #sideEditor A{color: #786755 !important; /*important is for editor*/text-decoration:none !important;}#mainContent A:hover, #sideEditor A:hover{text-decoration:underline !important;}H1, H2, H3, H4, H5, H6{font-family: verdana,geneva;color: #4D2A33;}.pageBgColor{background-color: #FFFFFF;}#mainContent, #sideEditor, #breakingMsg,#passwordBoxSite /*some duplicate from body for the editor*/{color: #505659;font-size: 0.8125em ;background-color: #FFFFFF !important;background-image:none !important;line-height: 1.4em ;}#mainContent TD, #sideEditor TD /*editor, IE*/{line-height: 1.4em ;}#mainContent HR, #sideEditor HR{border: 1px solid #505659;}#mainContent INPUT, #mainContent SELECT, #mainContent TEXTAREA, #sideEditor INPUT{color: #505659;border: 1px solid #505659;background-color: #FFFFFF;font-family:verdana,geneva;}#mainContent BLOCKQUOTE, #sideEditor BLOCKQUOTE{background-color: #FFFFFF;font-family: verdana,geneva;background-image: url(\x27/global/styleTempl/garsila2/images/_defaults/quote_FFFFFF.gif\x27);}#mainContent IMG, #sideEditor IMG{border: 1px solid #4A452B;}.menuTabs A{background-image: url(/global/images/site/menu/poptab/tabL.gif);}.menuTabs A SPAN{background-image: url(/global/images/site/menu/poptab/tabR.gif);}#leftAd, #rightAd{color: #FFFFFF;}#blog .infoTitle{border-top: 1px solid #BBBBBB;border-bottom: 1px solid #BBBBBB;}#blog .bPublicised{color: #505659;}#blog .blogComments, #blog .permaLink{background-color: #FFFFFF;border: 5px double #BBBBBB;}.gBookEntryDate{color: #505659;}.fadedUcase{color: #505659;}.msg{color: #BBBBBB;}#gedcomHeader{background: url(\x27/global/styleTempl/garsila2/images/_defaults/sideShadeLine_777777_normalHeight.png\x27) 0px -20px;}.sa-gedcom .facts_label{background: url(\x27/global/styleTempl/garsila2/images/_defaults/sideShadeLine_777777_normalHeight.png\x27) 0px -30px;border: 1px solid #786755;}.sa-headerBorder{border-left:1px solid #786755;border-top:1px solid #786755;border-right:1px solid #786755;}#body.sa-gedcom #container{border:1px solid #786755;}#body.sa-gedcom .facts_value{border:1px solid #786755;}#passwordBox{border: 1px solid #BBBBBB;}#previewMode{border: 1px solid #4D2A33;background-color: #FFFFFF;color: #505659;}#previewMode A{color: #4D2A33;}#sideColLeft, #sideColRight, #gBookSideCol{background: url(\x27/global/styleTempl/garsila2/images/_defaults/sidecol_FFFFFF.png\x27) no-repeat center top;}#blogRightCol{background: url(\x27/global/styleTempl/garsila2/images/_defaults/sidecolNarrow_FFFFFF.png\x27) no-repeat center top;}.sideColor{background: url(\x27/global/styleTempl/garsila2/images/_defaults/sideShadeLine_777777_normalHeight.png\x27);}.colorShade{background: url(\x27/global/styleTempl/garsila2/images/_defaults/sideShade_777777_normalHeight.png\x27);}#header #mainTitle{left: 0px;}#header #subTitle{left: 0px;}#header H1{font-family: arial black,avant garde;color: #4D2A33;}#header H2{font-family: verdana,geneva;text-align: left;}#header H2{color: #4D2A33;}#header {background-image:url(\x27img/banner/4/__4__91283_normalHeight.jpg\x27);}#footer{font-size: 0.8125em ;}#footer SPAN.visitors, #footer SPAN.footerTxt{color: #7C3380;}#subFooter A, #leftAd, #rightAd, #topAd{color: #666666;font-size: 12px;}.vSpacer{background-image: none !important;}#mainContent TEXTAREA{font-size: 12px;}.centerCell{width: 868px;}.shade{background: url(/global/images/site/shade/892-sides-30.png) repeat-y center;}#bottomShade{background-image: url(/global/images/site/shade/892-bottom-30.png);}#header{height: 261px; }#mainColRight, #mainColLeft{width: 562px;min-width: 562px; }#fullPage{width: 800px;}#gBook{width: 793px;}#gBookBody{width: 522px;}#blog{width: 803px;}#blogBody{width: 580px;}</style><script src=\x22/global/site/jquery-1.2.6-min.js\x22 type=\x22text/javascript\x22></script><script src=\x22/global/site/jQueryPlugins/jqModal/jqModal.js\x22 type=\x22text/javascript\x22></script><script src=\x22/global/site/jQueryPlugins/jquery.cookie.js\x22 type=\x22text/javascript\x22></script><script src=\x22/global/site/site-0321-src.js\x22 type=\x22text/javascript\x22></script><link type=\x22text/css\x22 rel=\x22stylesheet\x22 media=\x22all\x22 href=\x22/global/site/jQueryPlugins/jqModal/jqModal.css\x22></head><body><table align=\x22center\x22 cellpadding=\x220\x22 cellspacing=\x220\x22 border=\x220\x22 id=\x22structure\x22 width=\x22100%\x22><tr><td valign=\x22top\x22 align=\x22right\x22><div id=\x22flags\x22><table border=\x220\x22 cellspacing=\x220\x22 cellpadding=\x220\x22><tr><td><!-- --></td></tr></table></div></td><td><div class=\x22shade\x22><div class=\x22centerCell\x22><div id=\x22mainMenu\x22 class=\x22menuTabs\x22><ul><li class=\x22active\x22><a href=\x22default.asp\x22><span>Veckans meny</span></a></li><li><a href=\x22gastmatsalen.asp\x22><span>G&#228;stmatsalen</span></a></li><li><a href=\x22smaratter.asp\x22><span>Sm&#229;r&#228;tter</span></a></li><li><a href=\x22roomservice.asp\x22><span>Catering & Utskick</span></a></li><li><a href=\x22buffe.asp\x22><span>Buff&#233;</span></a></li><li><a href=\x22sasong.asp\x22><span>S&#228;song</span></a></li><li><a href=\x22erbjudanden.asp\x22><span>Erbjudanden</span></a></li><li><a href=\x22kontakt.asp\x22><span>Kontakt</span></a></li></ul></div></div></div></td><td></td></tr><tr><td class=\x22sideColor\x22 width=\x2250%\x22 valign=\x22top\x22 align=\x22right\x22></td><td><div class=\x22colorShade\x22><div id=\x22header\x22 class=\x22centerCell\x22 style=\x22position:relative\x22><div class=\x22centerCell\x22><div id=\x22mainTitle\x22><h1>Restaurang Ideon</h1></div><div id=\x22subTitle\x22><h2></h2></div></div></div></div></td><td class=\x22sideColor\x22 width=\x2250%\x22></td></tr><tr><td align=\x22right\x22 valign=\x22top\x22></td><td valign=\x22top\x22> <div class=\x22shade\x22><div id=\x22mainContentWrapper\x22 class=\x22centerCell pageBgColor\x22><div id=\x22mainContent\x22><div id=\x22fullPage\x22><h1>Veckans Meny&#160; vecka&#160;5&#160;&#160;&#160;&#160;&#160;</h1> <p><span style=\x22color: #032f52; font-size: 12px;\x22><strong>Varmr&#228;tt, sm&#246;r &amp; hembakat br&#246;d, m&#229;ltidsdryck i glas, sallad och kaffe&#160; &#160; Pris&#160; 70 kr</strong>&#160; </span></p> <p>&#160;</p> <p><span style=\x22color: #032f52; font-size: 12px;\x22><span style=\x22color: #1e3332;\x22><strong>&#214;ppettider: M&#229;ndag - fredag 11.00 - 14.00</strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 12px;\x22><span style=\x22color: #1e3332;\x22><strong>Bordsbokning p&#229; Hyllan 046-286 20 30 </strong></span></span></p> <p><strong><span style=\x22color: #000000;\x22>Vi s&#228;ljer kupongh&#228;ften, <span style=\x22text-decoration: underline;\x22>630:-</span> f&#246;r 10 luncher</span></strong>&#160;</p> <h2><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><span style=\x22text-decoration: underline;\x22><span style=\x22color: #800000;\x22>M&#197;NDAG&#160;&#160;31 JANUARI</span></span></span></span></h2> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong><span style=\x22color: #000000;\x22>Dagens R&#228;tt: Pocherad hokifil&#233; med pesto och bulgur samt apelsin-och basilikakesella</span></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong><span style=\x22color: #000000;\x22>Dagens R&#228;tt: K&#246;ttbullar med gr&#228;dds&#229;s,lingonsylt och gurka</span></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong><span style=\x22color: #008000;\x22>Vegetarisk:</span>&#160;<span style=\x22color: #339966;\x22>&#160; Vegetarisk thaigryta</span></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <h3><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><span style=\x22text-decoration: underline;\x22><span style=\x22color: #800000;\x22>Tisdag&#160;&#160;&#160;&#160;1 Februari</span></span></span></span></h3> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong><span style=\x22color: #000000;\x22>Dagens R&#228;tt:</span> <span style=\x22color: #000000;\x22>Fl&#228;skkotlett ossobucco med tagliatelle</span></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong><span style=\x22color: #000000;\x22>Dagens R&#228;tt: Rostbiff med ljummen potatissallad</span></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong><span style=\x22color: #008000;\x22>Vegetarisk:</span>&#160;&#160;<span style=\x22color: #339966;\x22> Fylld paprika samt couscous och chilistuvad majs</span></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <h3><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><span style=\x22text-decoration: underline;\x22><span style=\x22color: #800000;\x22>Onsdag&#160;&#160;&#160;2 Februari</span></span></span></span></h3> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong><span style=\x22color: #000000;\x22>Dagens R&#228;tt: Timjankryddad skinkstek med &#228;delosts&#229;s</span></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong><span style=\x22color: #000000;\x22>Dagens R&#228;tt: Fiskgrat&#228;ng med vittvinss&#229;s,sparris och r&#228;kor</span></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong><span style=\x22color: #008000;\x22>Vegetarisk:</span>&#160;<span style=\x22color: #339966;\x22>&#160;&#160;Oliv-och potatisomelett med kron&#228;rtskocka och ratataouille</span></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <p>&#160;</p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><span style=\x22color: #800000;\x22><strong><span style=\x22text-decoration: underline;\x22>TORSDAG&#160;&#160;3 FEBRUARI</span>&#160;</strong></span></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong><span style=\x22color: #000000;\x22>Dagens R&#228;tt: F&#228;rsbiff med fetaost och soltorkad tomat samt gurkyoghurt</span></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong><span style=\x22color: #000000;\x22>Dagens R&#228;tt: &#196;rtsoppa med fl&#228;skt&#228;rningar samt pannkakor med sy&#246;t och gr&#228;dde</span></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong><span style=\x22color: #008000;\x22>Vegatarisk:&#160;&#160; Tomat och spenatgryta med matvete</span></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <h3><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><span style=\x22text-decoration: underline;\x22><span style=\x22color: #800000;\x22>Fredag&#160;&#160;&#160;&#160;4 FEBRUARI</span></span></span></span></h3> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><span style=\x22color: #008000;\x22><strong></strong></span></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #505659;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #000000;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #000000;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #000000;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #000000;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #000000;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #000000;\x22><strong></strong></span></span></p> <p><span style=\x22color: #032f52; font-size: 9pt;\x22><span style=\x22color: #000000;\x22><strong></strong></span></span></p> <p><span style=\x22color: #000000;\x22><strong>Dagens R&#228;tt:&#160; Revbensspj&#228;ll med bbq-s&#229;s och stekt kulpotatis</strong></span></p> <p><span style=\x22color: #000000;\x22><strong>Dagens R&#228;tt:&#160; Fiskf&#228;rsbiff med dill och r&#228;ks&#229;s</strong></span></p> <p><span style=\x22color: #339966;\x22><strong>Vegetarisk:</strong>&#160;&#160;&#160;<strong>Linsbiff a la Lindstr&#246;m med creme fraish&#233;&#160;&#160; </strong></span></p> <p>&#160;&#160;</p> <p>&#160;</p> <h2 style=\x22TEXT-ALIGN: center\x22><span style=\x22color: #000000;\x22>V&#196;LKOMNA</span></h2> <h2 style=\x22TEXT-ALIGN: center\x22><span style=\x22color: #000000;\x22>Vi finns i Alfahuset</span></h2> <p style=\x22TEXT-ALIGN: center\x22>&#160;</p> <p style=\x22TEXT-ALIGN: center\x22>&#160;</p> <p style=\x22TEXT-ALIGN: center\x22>&#160;</p> <p style=\x22TEXT-ALIGN: center\x22>&#160;</p> <p>&#160;</p> <p>&#160;</p></div></div> <div style=\x22clear:both;\x22>&#160;</div><div id=\x22footer\x22><span class=\x22visitors\x22>Antal bes&#246;k:&nbsp;&nbsp;<strong>50†722</strong></span><span class=\x22footerTxt\x22></span></div></div></div><div id=\x22bottomShade\x22></div> <div id=\x22subFooter\x22><a href=\x22http://www.snabber.se/\x22 title=\x22Gˆr egen hemsida &ndash; Snabber\x22 target=\x22_blank\x22>Gˆr egen hemsida &ndash; Snabber</a> </div></td><td valign=\x22top\x22><div id=\x22rightAd\x22></div></td></tr></table><div class=\x22jqmWindow\x22 id=\x22dialog\x22><div id=\x22dialogBx\x22><a href=\x22#\x22 class=\x22jqmClose\x22 style=\x22float: right\x22><img src=\x22/global/images/site/close.gif\x22 alt=\x22\x22 border=\x220\x22></a><a href=\x22#\x22 class=\x22jqmClose\x22>st&auml;ng&nbsp;</a></div><p style=\x22margin-top:30px;margin-bottom:30px\x22></p></div><script language=\x22JavaScript\x22 type=\x22text/javascript\x22>$(document).ready(function(){$(\x22img\x22, \x22#mainContent\x22).mouseover(function(){if ($(this).is(\x27.pop\x27) ){ if($(this).attr(\x27src\x27).indexOf(\x22http://\x22) == -1){ $(this).css(\x22cursor\x22,\x22pointer\x22);}}});$(\x22img\x22, \x22#mainContent\x22).click(function(){var origSrc = $(this).attr(\x27src\x27);if($(this).parent().attr(\x27href\x27)){//do nothing. Image is wrapped in a link (open link instead of popup)}else{if ($(this).is(\x27.pop\x27) ){if(origSrc.indexOf(\x22http://\x22) == -1){var imgName = origSrc.substr(origSrc.lastIndexOf(\x22/\x22)+1,origSrc.length);theWidth = $(this).attr(\x27width\x27);theHeight = $(this).attr(\x27height\x27);popImg(\x22imgPopup.asp?d=restaurang&imgName=\x22 + imgName, theWidth, theHeight);}}}});$(\x22a\x22, \x22#mainContent\x22).click(function(){var theHref = $(this).attr(\x27href\x27);if(theHref.indexOf(\x22http://\x22) < 0){if(theHref.indexOf(\x22javascript:\x22) < 0){if ( ($(this).is(\x27.pop\x27)) && (theHref.indexOf(\x22img/\x22) > -1) ){ $(this).removeAttr(\x22href\x22);var imgName = theHref.substr(theHref.lastIndexOf(\x22/\x22)+1,theHref.length);theWidth = $(this).attr(\x27width\x27);theHeight = $(this).attr(\x27height\x27);popImg(\x22imgPopup.asp?d=restaurang&imgName=\x22 + imgName, theWidth, theHeight);}}}});$(\x22img[src*=\x27xxlarge\x27]\x22).css(\x22clear\x22,\x22both\x22);$(\x22img[src*=\x27xxxlarge\x27]\x22).css(\x22clear\x22,\x22both\x22);$(\x22img[src*=\x27xxxlarge\x27]\x22).css(\x22display\x22,\x22block\x22);$(\x22img[src*=\x27xxxlarge\x27]\x22).css(\x22margin\x22,\x220 auto\x22);$(\x22img[src*=\x27xxxlarge\x27]\x22).css(\x22float\x22,\x22none\x22);$(\x22a[href$=doc]\x22).addClass(\x22doc\x22); $(\x22a[href$=docx]\x22).addClass(\x22doc\x22);$(\x22a[href$=rtf]\x22).addClass(\x22doc\x22);$(\x22a[href$=xls]\x22).addClass(\x22xls\x22);$(\x22a[href$=xlsx]\x22).addClass(\x22xls\x22);$(\x22a[href$=ppt]\x22).addClass(\x22ppt\x22);$(\x22a[href$=pps]\x22).addClass(\x22ppt\x22);$(\x22a[href$=pdf]\x22).addClass(\x22pdf\x22); $(\x22.sm2Container\x22).mouseout(function(){ $(\x27.sm2Container\x27).css(\x22display\x22,\x22none\x22); }); $(\x22#mainMenu\x22).mouseover(function(){ $(\x27.sm2Container\x27).css(\x22display\x22,\x22none\x22); }); });</script><script src=\x22http://www.google-analytics.com/ga.js\x22 type=\x22text/javascript\x22></script><script language=\x22JavaScript\x22 type=\x22text/javascript\x22>var pageTracker = _gat._getTracker(\x22UA-3012821-1\x22);pageTracker._initData();pageTracker._trackPageview(\x22/customerSites/restaurang/pages.html\x22);pageTracker._addIgnoredOrganic(\x22snabber.se\x22);pageTracker._addIgnoredRef(\x22snabber.se\x22);pageTracker._addIgnoredRef(\x22www.snabber.se\x22);</script></body></html>"
// TODO: Make it properly case insensitive
$(function() {
console.log("Scraping Höjdpunkten");
HPScraper.init(hp);
console.log(HPScraper.scrape("måndag"));
console.log(HPScraper.scrape("tisdag"));
console.log(HPScraper.scrape("onsdag"));
console.log(HPScraper.scrape("torsdag"));
console.log(HPScraper.scrape("fredag"));
console.log("Scraping IDEON");
IDEONScraper.init(ideon);
console.log(IDEONScraper.scrape("måndag"));
console.log(IDEONScraper.scrape("tisdag"));
console.log(IDEONScraper.scrape("onsdag"));
console.log(IDEONScraper.scrape("torsdag"));
console.log(IDEONScraper.scrape("fredag"));
});
var IDEONScraper = {
init: function(html) {
this.html = $(html.toLowerCase());
this.html.find("strong:contains('måndag - fredag')").remove();
},
scrape: function(day) {
switch(day)
{
case "måndag":
return this._scrape_bounded("mÅndag");
break;
case "tisdag":
case "onsdag":
case "torsdag":
return this._scrape_bounded(day);
break;
case "fredag":
return this._scrape_last(day);
break;
default:
console.log("Don't know how to scrape " + day);
}
},
_scrape_bounded: function(day) {
var boundaries = {
"mÅndag": "tisdag",
"tisdag": "onsdag",
"onsdag": "torsdag",
"torsdag": "fredag"
};
// Ideon arbitirariliy have days all upcase or cap case
var hsel = "#fullpage :contains('" + day + "')",
bsel = "#fullpage :contains('" + boundaries[day] + "')",
h = this.html.find(hsel).slice(0,1),
b = this.html.find(bsel).slice(0,1),
i = h.index() + 1,
j = b.index() - 1,
elems = h.parent().children().slice(i,j);
return HPScraper._extract_meals(elems);
},
_scrape_last: function(day) {
var hsel = "#fullpage :contains('" + day + "')",
h = this.html.find(hsel).slice(0,1),
i = h.index() + 1,
elems = h.parent().children().slice(i,i+15);
return HPScraper._extract_meals(elems);
}
}
var HPScraper = {
init: function(html) {
this.html = $(html.toLowerCase());
},
scrape: function(day) {
switch(day)
{
case "måndag":
case "tisdag":
case "onsdag":
return this._scrape_bounded(day);
break;
case "torsdag":
case "fredag":
return this._scrape_last(day);
break;
default:
console.log("Don't know how to scrape " + day);
}
},
_scrape_bounded: function(day) {
var boundaries = {
"måndag": "onsdag",
"tisdag": "torsdag",
"onsdag": "fredag"
};
var h = this.html.find(".holderside p:contains('" + day + "')"),
b = this.html.find(".holderside p:contains('" + boundaries[day] + "')"),
i = h.index() + 1,
j = b.index() - 1,
elems = h.parent().children().slice(i,j);
return this._extract_meals(elems);
},
_scrape_last: function(day) {
var h = this.html.find(".holderside p:contains('" + day + "')"),
i = h.index() + 1,
elems = h.parent().children().slice(i);
return this._extract_meals(elems);
},
_extract_meals: function(elems) {
var meals = [];
elems.each(function(i,e) {
var meal = $.trim($(this).text());
if (meal) meals.push(meal);
});
return meals;
}
}
Raw
output.js
Scraping Höjdpunkten
scrape.js:8["karl - johans gryta med paprikaris.", "isterband med persiljestuvad potatis, saltgurka och rödbetor.", "kycklingsallad med stekt bacon (avh. 50 kr/ serve.60kr)"]
scrape.js:9["stekt sill med potatismos skirat smör och lingon.", "pasta med oxkött, soltorkad tomat och champinjoner.", "grekisksallad med tzatziki (avh. 50 kr/ serve.60kr)"]
scrape.js:10["wallenbergare med kokt potatis, gräddsås och gröna ärtor", "potatis och purjolökssoppa med strimlad rökt skinka.", "laxsallad med dill och citron (avh. 50 kr/ serve.60kr)"]
scrape.js:11["kycklingfilé med paprika cremé sås och ris.", "pasta med tomat, fänkål, räkor och rökt lax.", "paj med stekt bacon och purjolök (avh. 50 kr/ serve.60kr)."]
scrape.js:12["helstekt kotlettrad med potatisgratäng", "nasigoreng", "ost och skinka sallad (avh. 50 kr/ serve.60kr)."]
scrape.js:14Scraping IDEON
scrape.js:16["dagens rätt: pocherad hokifilé med pesto och bulgur samt apelsin-och basilikakesella", "dagens rätt: köttbullar med gräddsås,lingonsylt och gurka", "vegetarisk: vegetarisk thaigryta"]
scrape.js:17["dagens rätt: fläskkotlett ossobucco med tagliatelle", "dagens rätt: rostbiff med ljummen potatissallad", "vegetarisk: fylld paprika samt couscous och chilistuvad majs"]
scrape.js:18["dagens rätt: timjankryddad skinkstek med ädelostsås", "dagens rätt: fiskgratäng med vittvinssås,sparris och räkor", "vegetarisk: oliv-och potatisomelett med kronärtskocka och ratataouille"]
scrape.js:19["dagens rätt: färsbiff med fetaost och soltorkad tomat samt gurkyoghurt", "dagens rätt: Ärtsoppa med fläsktärningar samt pannkakor med syöt och grädde", "vegatarisk: tomat och spenatgryta med matvete"]
scrape.js:20["dagens rätt: revbensspjäll med bbq-sås och stekt kulpotatis", "dagens rätt: fiskfärsbiff med dill och räksås", "vegetarisk: linsbiff a la lindström med creme fraishé"]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment