ACM: Fix proxy behavior for bibtex and PDF (zotero#1460)

Closes zotero#1449
We grab URLs of the page, but they default to http version there, which
causes all sorts of havoc for https version with hyphenated
proxy. Solving this simply by deproxifying standard pattern.

ACM Digital Library.js

@@ -9,7 +9,7 @@
 	"inRepository": true,
 	"translatorType": 4,
 	"browserSupport": "gcsibv",
-	"lastUpdated": "2017-01-26 12:12:58"
+	"lastUpdated": "2017-10-29 04:47:45"
 }
 
 /*
@@ -94,7 +94,8 @@ function scrape(doc) {
 
 	//compose bibtex URL
 	var bibtexstring = 'id=' + itemID + '&parent_id=' + parentID + '&expformat=bibtex';
-	var bibtexURL = url.replace(/citation\.cfm/, 'downformats.cfm')
+	var bibtexURL = url.replace(/dl[.-]acm[.-]org[^\/]*/, "dl.acm.org")  //deproxify the URL above.
+		.replace(/citation\.cfm/, 'downformats.cfm')
 		.replace(/([?&])id=[^&#]+/, '$1' + bibtexstring);
 	Zotero.debug('bibtex URL: ' + bibtexURL);
 
@@ -106,8 +107,10 @@ function scrape(doc) {
 			//get the URL for the pdf fulltext from the metadata
 			var pdfURL = ZU.xpath(doc, '//meta[@name="citation_pdf_url"]/@content')[0];
 			if (pdfURL) {
+				pdfURL = pdfURL.textContent.replace(/dl[.-]acm[.-]org[^\/]*/, "dl.acm.org"); //deproxify URL
+				Z.debug("pdfURL: " + pdfURL);
 				item.attachments = [{
-					url: pdfURL.textContent,
+					url: pdfURL,
 					title: "ACM Full Text PDF",
 					mimeType: "application/pdf"
 				}];