Ticket #379: amazonr2new.txt

REPLACE INTO "translators" VALUES ('96b9f483-c44d-5784-cdad-ce21b984fe01', '2006-11-24 17:00:00', 1, 100, 4, 'Amazon.com', 'Simon Kornblith', '^http://www\.amazon\.com/', 
'function detectWeb(doc, url) {
        var searchRe = new RegExp(''^http://(?:www\.)?amazon\.com/(gp/search/|exec/obidos/search-handle-url/|s/)'');
        if(searchRe.test(doc.location.href)) {
                return "multiple";
        } else {
                var namespace = doc.documentElement.namespaceURI;
                var nsResolver = namespace ? function(prefix) {
                        if (prefix == ''x'') return namespace; else return null;
                } : null;
                
                var xpath = ''//table/tbody/tr/td[@class="bucket"]/div[@class="content"]/ul/li'';
                if(doc.evaluate(xpath, doc, nsResolver, XPathResult.ANY_TYPE, null).iterateNext()) {
                        return "book";
                }
        }
}
',
'function scrape(doc) { 
        var namespace = doc.documentElement.namespaceURI;
        var nsResolver = namespace ? function(prefix) {
                if (prefix == ''x'') return namespace; else return null;
        } : null;

        var newItem = new Zotero.Item("book");
        
        // Retrieve authors
        try {
                var xpath = ''//table/tbody/tr/td[2]/form/div[@class="buying"]/a/text()[1]'';
                var elmts = doc.evaluate(xpath, doc, nsResolver, XPathResult.ANY_TYPE, null);
                var elmt;
                while(elmt = elmts.iterateNext()) {
                        newItem.creators.push(Zotero.Utilities.cleanAuthor(elmt.nodeValue, "author"));
                }
        } catch(ex) {Zotero.Utilities.debug(ex);}
        
        // Retrieve data from "Product Details" box
        var xpath = ''//table/tbody/tr/td[@class="bucket"]/div[@class="content"]/ul/li'';
        var elmts = doc.evaluate(xpath, doc, nsResolver, XPathResult.ANY_TYPE, null);
        var elmt;
        
        newItem.extra = "";
        while(elmt = elmts.iterateNext()) {
                try {
                        var attribute = Zotero.Utilities.cleanString(doc.evaluate(''./B[1]/text()[1]'', elmt, nsResolver, XPathResult.ANY_TYPE, null).iterateNext().nodeValue);
                        var value = Zotero.Utilities.getNodeString(doc, elmt, ''./descendant-or-self::*[name() != "B"]/text()'', nsResolver);
                        if(value) {
                                value = Zotero.Utilities.cleanString(value);
                                
                                if(attribute == "Publisher:") {
                                        if(value.lastIndexOf("(") != -1) {
                                                newItem.date = value.substring(value.lastIndexOf("(")+1, value.length-1);
                                                
                                                value = value.substring(0, value.lastIndexOf("(")-1);
                                        }
                                        if(value.lastIndexOf(";") != -1) {
                                                newItem.edition = value.substring(value.lastIndexOf(";")+2, value.length);
                                                
                                                value = value.substring(0, value.lastIndexOf(";"));
                                        }
                                        newItem.publisher = value;
                                } else if(attribute == "ISBN:") {
                                        newItem.ISBN = value;
                                } else if(value.substring(value.indexOf(" ")+1, value.length) == "pages") {
                                        newItem.pages = value.substring(0, value.indexOf(" "));
                                } else if(attribute != "Average Customer Review:") {
                                        if(attribute == "In-Print Editions:") {
                                                value = value.replace(" | All Editions", "");
                                        } else {
                                                value = value.replace(/\([^)]*\)/g, "");
                                        }
                                        
                                        newItem.extra += attribute+" "+value+"\n";
                                }
                        }
                } catch(ex) {}
        }
        
        if(newItem.extra) {
                newItem.extra = newItem.extra.substr(0, newItem.extra.length-1);
        }
        
        newItem.attachments.push({title:"Amazon.com Product Page", document:doc});
        
        var xpath = ''//table/tbody/tr/td[2]/form/div[@class="buying"]/b[@class="sans"]/text()[1]'';
        var title = doc.evaluate(xpath, doc, nsResolver, XPathResult.ANY_TYPE, null).iterateNext().nodeValue;
        title = Zotero.Utilities.cleanString(title);
        if(title.lastIndexOf("(") != -1 && title.lastIndexOf(")") == title.length-1) {
                title = title.substring(0, title.lastIndexOf("(")-1);
        }
        newItem.title = title;
        
        newItem.complete();
}

function doWeb(doc, url) {
        var searchRe = new RegExp(''^http://www\.amazon\.com/(gp/search/|exec/obidos/search-handle-url/|s/)'');
        var m = searchRe.exec(doc.location.href)
        if(m) {
                var namespace = doc.documentElement.namespaceURI;
                var nsResolver = namespace ? function(prefix) {
                        if (prefix == ''x'') return namespace; else return null;
                } : null;
                
                var xpath = ''//table/tbody/tr/td/a/span[@class="srTitle"]/../..'';
                
                var searchresults = Zotero.Utilities.gatherElementsOnXPath(doc, doc, xpath, nsResolver);
                var items = Zotero.Utilities.getItemArray(doc, searchresults, ''^http://www\.amazon\.com/(gp/product/|exec/obidos/tg/detail/|[^/]+/dp/)'', ''^(Buy new|Hardcover|Paperback|Digital)$'');
                items = Zotero.selectItems(items);
                
                if(!items) {
                        return true;
                }
                
                var uris = new Array();
                for(var i in items) {
                        uris.push(i);
                }
                
                Zotero.Utilities.processDocuments(uris, function(doc) { scrape(doc) },
                        function() { Zotero.done(); }, null);
                
                Zotero.wait();
        } else {
                scrape(doc);
        }
}');