From 5d17ff3f5324ef74a1937a113995c6bf202645ef Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Adomas=20Ven=C4=8Dkauskas?= <adomas.ven@gmail.com>
Date: Tue, 27 Sep 2016 17:43:56 +0300
Subject: [PATCH] Implements #1021 on Z4fx extension

---
 .../zotero/xpcom/connector/translator.js      |   5 +-
 .../zotero/xpcom/translation/translate.js     |   9 +-
 .../zotero/xpcom/translation/translator.js    |   5 +-
 .../zotero/xpcom/translation/translators.js   | 139 ++++++++++--------
 test/content/support.js                       |  12 +-
 test/tests/translateTest.js                   |   2 +-
 test/tests/translatorsTest.js                 |  89 +++++++++++
 7 files changed, 189 insertions(+), 72 deletions(-)
 create mode 100644 test/tests/translatorsTest.js

diff --git a/chrome/content/zotero/xpcom/connector/translator.js b/chrome/content/zotero/xpcom/connector/translator.js
index c9cd682106..56acb305fb 100644
--- a/chrome/content/zotero/xpcom/connector/translator.js
+++ b/chrome/content/zotero/xpcom/connector/translator.js
@@ -425,7 +425,10 @@ Zotero.Translator.prototype.init = function(info) {
 	 
 	if(this.translatorType & TRANSLATOR_TYPES["web"]) {
 		// compile web regexp
-		this.webRegexp = this.target ? new RegExp(this.target, "i") : null;
+		this.webRegexp = {
+			root: this.target ? new RegExp(this.target, "i") : null,
+			all: this.targetAll ? new RegExp(this.targetAll, "i") : null
+		};
 	} else if(this.hasOwnProperty("webRegexp")) {
 		delete this.webRegexp;
 	}
diff --git a/chrome/content/zotero/xpcom/translation/translate.js b/chrome/content/zotero/xpcom/translation/translate.js
index eb1be0c14d..0766677a51 100644
--- a/chrome/content/zotero/xpcom/translation/translate.js
+++ b/chrome/content/zotero/xpcom/translation/translate.js
@@ -1883,7 +1883,8 @@ Zotero.Translate.Web.prototype.Sandbox = Zotero.Translate.Sandbox._inheritFromBa
  */
 Zotero.Translate.Web.prototype.setDocument = function(doc) {
 	this.document = doc;
-	this.setLocation(doc.location.href);
+	this.rootDocument = doc.defaultView.top.document || doc;
+	this.setLocation(doc.location.href, this.rootDocument.location.href);
 }
 
 /**
@@ -1900,9 +1901,11 @@ Zotero.Translate.Web.prototype.setCookieSandbox = function(cookieSandbox) {
  * Sets the location to operate upon
  *
  * @param {String} location The URL of the page to translate
+ * @param {String} rootLocation The URL of the root page, within which `location` is embedded
  */
-Zotero.Translate.Web.prototype.setLocation = function(location) {
+Zotero.Translate.Web.prototype.setLocation = function(location, rootLocation) {
 	this.location = location;
+	this.rootLocation = rootLocation || location;
 	this.path = this.location;
 }
 
@@ -1910,7 +1913,7 @@ Zotero.Translate.Web.prototype.setLocation = function(location) {
  * Get potential web translators
  */
 Zotero.Translate.Web.prototype._getTranslatorsGetPotentialTranslators = function() {
-	return Zotero.Translators.getWebTranslatorsForLocation(this.location);
+	return Zotero.Translators.getWebTranslatorsForLocation(this.location, this.rootLocation);
 }
 
 /**
diff --git a/chrome/content/zotero/xpcom/translation/translator.js b/chrome/content/zotero/xpcom/translation/translator.js
index 5bddc14a06..1b6a754812 100644
--- a/chrome/content/zotero/xpcom/translation/translator.js
+++ b/chrome/content/zotero/xpcom/translation/translator.js
@@ -115,7 +115,10 @@ Zotero.Translator.prototype.init = function(info) {
 	if(this.translatorType & TRANSLATOR_TYPES["web"]) {
 		// compile web regexp
 		this.cacheCode |= !this.target;
-		this.webRegexp = this.target ? new RegExp(this.target, "i") : null;
+		this.webRegexp = {
+			root: this.target ? new RegExp(this.target, "i") : null,
+			all: this.targetAll ? new RegExp(this.targetAll, "i") : null
+		};
 	} else if(this.hasOwnProperty("webRegexp")) {
 		delete this.webRegexp;
 	}
diff --git a/chrome/content/zotero/xpcom/translation/translators.js b/chrome/content/zotero/xpcom/translation/translators.js
index 04c94c9d2a..f063a20278 100644
--- a/chrome/content/zotero/xpcom/translation/translators.js
+++ b/chrome/content/zotero/xpcom/translation/translators.js
@@ -46,7 +46,7 @@ Zotero.Translators = new function() {
 		Zotero.debug("Initializing translators");
 		var start = new Date;
 		
-		_cache = {"import":[], "export":[], "web":[], "search":[]};
+		_cache = {"import":[], "export":[], "web":[], "webWithTargetAll":[], "search":[]};
 		_translators = {};
 		
 		var sql = "SELECT fileName, metadataJSON, lastModifiedTime FROM translatorCache";
@@ -152,6 +152,9 @@ Zotero.Translators = new function() {
 					for (let type in TRANSLATOR_TYPES) {
 						if (translator.translatorType & TRANSLATOR_TYPES[type]) {
 							_cache[type].push(translator);
+							if ((translator.translatorType & TRANSLATOR_TYPES.web) && translator.targetAll) {
+								_cache.webWithTargetAll.push(translator);
+							}
 						}
 					}
 					
@@ -267,76 +270,92 @@ Zotero.Translators = new function() {
 	/**
 	 * Gets web translators for a specific location
 	 * @param {String} uri The URI for which to look for translators
+	 * @param {String} rootUri The root URI of the page, different from `uri` if running in an iframe
 	 */
-	this.getWebTranslatorsForLocation = function(uri) {
-		return this.getAllForType("web").then(function(allTranslators) {
+	this.getWebTranslatorsForLocation = function(URI, rootURI) {
+		var isFrame = URI !== rootURI;
+		var type = isFrame ? "webWithTargetAll" : "web";
+		
+		return this.getAllForType(type).then(function(allTranslators) {
 			var potentialTranslators = [];
+			var translatorConverterFunctions = [];
 			
-			var properHosts = [];
-			var proxyHosts = [];
+			var rootSearchURIs = this.getSearchURIs(rootURI);
+			var frameSearchURIs = isFrame ? this.getSearchURIs(URI) : rootSearchURIs;
 			
-			var properURI = Zotero.Proxies.proxyToProper(uri);
-			var knownProxy = properURI !== uri;
-			if(knownProxy) {
-				// if we know this proxy, just use the proper URI for detection
-				var searchURIs = [properURI];
-			} else {
-				var searchURIs = [uri];
-				
-				// if there is a subdomain that is also a TLD, also test against URI with the domain
-				// dropped after the TLD
-				// (i.e., www.nature.com.mutex.gmu.edu => www.nature.com)
-				var m = /^(https?:\/\/)([^\/]+)/i.exec(uri);
-				if(m) {
-					// First, drop the 0- if it exists (this is an III invention)
-					var host = m[2];
-					if(host.substr(0, 2) === "0-") host = host.substr(2);
-					var hostnames = host.split(".");
-					for(var i=1; i<hostnames.length-2; i++) {
-						if(TLDS[hostnames[i].toLowerCase()]) {
-							var properHost = hostnames.slice(0, i+1).join(".");
-							searchURIs.push(m[1]+properHost+uri.substr(m[0].length));
-							properHosts.push(properHost);
-							proxyHosts.push(hostnames.slice(i+1).join("."));
+			Zotero.debug("Translators: Looking for translators for "+Object.keys(frameSearchURIs).join(', '));
+			
+			for (let translator of allTranslators) {
+				translatorLoop:
+				for (let rootSearchURI in rootSearchURIs) {
+					let isGeneric = (!translator.webRegexp.root && translator.runMode === Zotero.Translator.RUN_MODE_IN_BROWSER);
+					if (!isGeneric && !translator.webRegexp.root) {
+						continue;
+					}
+					let rootURIMatches = isGeneric || rootSearchURI.length < 8192 && translator.webRegexp.root.test(rootSearchURI);
+					if (translator.webRegexp.all && rootURIMatches) {
+						for (let frameSearchURI in frameSearchURIs) {
+							let frameURIMatches = frameSearchURI.length < 8192 && translator.webRegexp.all.test(frameSearchURI);
+								
+							if (frameURIMatches) {
+								potentialTranslators.push(translator);
+								translatorConverterFunctions.push(frameSearchURIs[frameSearchURI]);
+								// prevent adding the translator multiple times
+								break translatorLoop;
+							}
 						}
 					}
-				}
-			}
-			
-			Zotero.debug("Translators: Looking for translators for "+searchURIs.join(", "));
-			
-			var converterFunctions = [];
-			for(var i=0; i<allTranslators.length; i++) {
-				for(var j=0; j<searchURIs.length; j++) {
-					if((!allTranslators[i].webRegexp
-							&& allTranslators[i].runMode === Zotero.Translator.RUN_MODE_IN_BROWSER)
-							|| (uri.length < 8192 && allTranslators[i].webRegexp.test(searchURIs[j]))) {
-						// add translator to list
-						potentialTranslators.push(allTranslators[i]);
-						
-						if(j === 0) {
-							if(knownProxy) {
-								converterFunctions.push(Zotero.Proxies.properToProxy);
-							} else {
-								converterFunctions.push(null);
-							}
-						} else {
-							converterFunctions.push(new function() {
-								var re = new RegExp('^https?://(?:[^/]+\\.)?'+Zotero.Utilities.quotemeta(properHosts[j-1])+'(?=/)', "gi");
-								var proxyHost = proxyHosts[j-1].replace(/\$/g, "$$$$");
-								return function(uri) { return uri.replace(re, "$&."+proxyHost) };
-							});
-						}
-						
-						// don't add translator more than once
+					else if(!isFrame && (isGeneric || rootURIMatches)) {
+						potentialTranslators.push(translator);
+						translatorConverterFunctions.push(rootSearchURIs[rootSearchURI]);
 						break;
 					}
 				}
 			}
 			
-			return [potentialTranslators, converterFunctions];
-		});
-	}
+			return [potentialTranslators, translatorConverterFunctions];
+		}.bind(this));
+	},
+
+	/**
+	 * Get the array of searchURIs and related proxy converter functions
+	 * 
+	 * @param {String} URI to get searchURIs and converterFunctions for
+	 */
+	this.getSearchURIs = function(URI) {
+		var properURI = Zotero.Proxies.proxyToProper(URI);
+		if (properURI !== URI) {
+			// if we know this proxy, just use the proper URI for detection
+			let obj = {};
+			obj[properURI] = Zotero.Proxies.properToProxy;
+			return obj;
+		}
+			
+		var searchURIs = {};
+		searchURIs[URI] = null;
+		
+		// if there is a subdomain that is also a TLD, also test against URI with the domain
+		// dropped after the TLD
+		// (i.e., www.nature.com.mutex.gmu.edu => www.nature.com)
+		var m = /^(https?:\/\/)([^\/]+)/i.exec(URI);
+		if (m) {
+			// First, drop the 0- if it exists (this is an III invention)
+			var host = m[2];
+			if(host.substr(0, 2) === "0-") host = host.substr(2);
+			var hostnames = host.split(".");
+			for (var i=1; i<hostnames.length-2; i++) {
+				if (TLDS[hostnames[i].toLowerCase()]) {
+					var properHost = hostnames.slice(0, i+1).join(".");
+					searchURIs[m[1]+properHost+URI.substr(m[0].length)] = new function() {
+						var re = new RegExp('^https?://(?:[^/]+\\.)?'+Zotero.Utilities.quotemeta(properHost)+'(?=/)', "gi");
+						var proxyHost = hostnames.slice(i+1).join(".").replace(/\$/g, "$$$$");
+						return function(uri) { return uri.replace(re, "$&."+proxyHost) };
+					};
+				}
+			}
+		}
+		return searchURIs;
+	},
 	
 	/**
 	 * Gets import translators for a specific location
diff --git a/test/content/support.js b/test/content/support.js
index 40c1f5eadc..827177a975 100644
--- a/test/content/support.js
+++ b/test/content/support.js
@@ -757,10 +757,11 @@ var generateTranslatorExportData = Zotero.Promise.coroutine(function* generateTr
 /**
  * Build a dummy translator that can be passed to Zotero.Translate
  */
-function buildDummyTranslator(translatorType, code, translatorID="dummy-translator") {
-	let info = {
-		"translatorID":translatorID,
-		"translatorType":translatorType,
+function buildDummyTranslator(translatorType, code, info={}) {
+	const TRANSLATOR_TYPES = {"import":1, "export":2, "web":4, "search":8};
+	info = Object.assign({
+		"translatorID":"dummy-translator",
+		"translatorType":TRANSLATOR_TYPES[translatorType],
 		"label":"Dummy Translator",
 		"creator":"Simon Kornblith",
 		"target":"",
@@ -768,10 +769,9 @@ function buildDummyTranslator(translatorType, code, translatorID="dummy-translat
 		"browserSupport":"g",
 		"inRepository":false,
 		"lastUpdated":"0000-00-00 00:00:00",
-	};
+	}, info);
 	let translator = new Zotero.Translator(info);
 	translator.code = code;
-	translator.getCode = function() {return Promise.resolve(code)};
 	return translator;
 }
 
diff --git a/test/tests/translateTest.js b/test/tests/translateTest.js
index c81cef0814..5258361022 100644
--- a/test/tests/translateTest.js
+++ b/test/tests/translateTest.js
@@ -561,7 +561,7 @@ describe("Zotero.Translate", function() {
 					item.title = "The Definitive Guide of Owls";
 					item.tags = ['owl', 'tag'];
 					item.complete();
-				}`, 'child-dummy-translator'
+				}`, {translatorID: 'child-dummy-translator'}
 			);
 			sinon.stub(Zotero.Translators, 'get').withArgs('child-dummy-translator').returns(childTranslator);
 			
diff --git a/test/tests/translatorsTest.js b/test/tests/translatorsTest.js
new file mode 100644
index 0000000000..b9b1ffe0d7
--- /dev/null
+++ b/test/tests/translatorsTest.js
@@ -0,0 +1,89 @@
+"use strict";
+
+describe("Zotero.Translators", function () {
+	describe("#getWebTranslatorsForLocation()", function () {
+		var genericTranslator, topLevelTranslator, frameTranslator;
+		var noMatchURL = 'http://notowls.com/citation/penguin-migration-patterns';
+		var topMatchURL = 'http://www.owl.com/owl_page/snowy_owl';
+		var frameMatchURL = 'http://iframe.owl.com/citation/owl-migration-patterns';
+		
+		before(function* (){
+			genericTranslator = buildDummyTranslator('web', `function doDetect() {}; function doWeb(); {}`, {
+				translatorID: 'generic-translator'
+			});
+			topLevelTranslator = buildDummyTranslator('web', `function doDetect() {}; function doWeb(); {}`, {
+				translatorID: 'top-level-translator',
+				target: "https?://www\\.owl\\.com/(citation|owl_page)/.+"
+			});
+			frameTranslator = buildDummyTranslator('web', `function doDetect() {}; function doWeb(); {}`, {
+				translatorID: 'frame-translator',
+				target: "https?://([^.]+\\.)?owl\\.com/(citation|owl_page)/.+",
+				targetAll: "https?://iframe.owl\\.com/(citation|owl_page)/.+"
+			});	
+			
+			let getAllForType = sinon.stub(Zotero.Translators, 'getAllForType');
+			getAllForType.withArgs('web').resolves([genericTranslator, topLevelTranslator, frameTranslator]);
+			getAllForType.withArgs('webWithTargetAll').resolves([frameTranslator]);
+			
+			let regexp = new RegExp(topLevelTranslator.target, 'i');
+			assert.isFalse(regexp.test(noMatchURL));
+			assert.isTrue(regexp.test(topMatchURL));
+			assert.isFalse(regexp.test(frameMatchURL));
+
+			regexp = new RegExp(frameTranslator.target, 'i');
+			assert.isFalse(regexp.test(noMatchURL));
+			assert.isTrue(regexp.test(topMatchURL));
+			assert.isTrue(regexp.test(frameMatchURL));
+
+			regexp = new RegExp(frameTranslator.targetAll, 'i');
+			assert.isFalse(regexp.test(noMatchURL));
+			assert.isFalse(regexp.test(topMatchURL));
+			assert.isTrue(regexp.test(frameMatchURL));
+		});
+		
+		after(function* (){
+			Zotero.Translators.getAllForType.restore();
+		});
+		
+		describe("when called from a root document", function() {
+			it("should return generic translators when not matching any translator `target`", function* () {
+				var translators = yield Zotero.Translators.getWebTranslatorsForLocation(noMatchURL, noMatchURL);
+				assert.equal(translators[0].length, 1);
+				assert.equal(translators[0][0].translatorID, 'generic-translator');
+			});
+			
+			it("should return all matching translators without `targetAll` property", function* () {
+				var translators = yield Zotero.Translators.getWebTranslatorsForLocation(topMatchURL, topMatchURL);
+				assert.equal(translators[0].length, 2);
+				assert.equal(translators[0][0].translatorID, 'generic-translator');
+				assert.equal(translators[0][1].translatorID, 'top-level-translator');
+			});
+			
+			it("should return translators that match both `target` and `targetAll` when both properties present", function* () {
+				var translators = yield Zotero.Translators.getWebTranslatorsForLocation(frameMatchURL, frameMatchURL);
+				assert.equal(translators[0].length, 2);
+				assert.equal(translators[0][0].translatorID, 'generic-translator');
+				assert.equal(translators[0][1].translatorID, 'frame-translator');
+			});
+		
+		});
+		
+		describe("when called from an iframe", function() {
+			it("should not return generic translators or translators without `targetAll` property", function* () {
+				var translators = yield Zotero.Translators.getWebTranslatorsForLocation(frameMatchURL, noMatchURL);
+				assert.equal(translators[0].length, 0);
+			});
+		
+			it("should not return translators that match `target` but not `targetAll", function* () {
+				var translators = yield Zotero.Translators.getWebTranslatorsForLocation(noMatchURL, topMatchURL);
+				assert.equal(translators[0].length, 0);
+			});
+			
+			it("should return translators that match both `target` and `targetAll`", function* () {
+				var translators = yield Zotero.Translators.getWebTranslatorsForLocation(frameMatchURL, topMatchURL);
+				assert.equal(translators[0].length, 1);
+				assert.equal(translators[0][0].translatorID, 'frame-translator');
+			});
+		});
+	});
+});