www

commit 6d1e4471548b77bab99ed219fe95817b33804336
parent 90ff41cb40c3f80b92bf266249ce91e68c75f769
Author: Simon Kornblith <simon@simonster.com>
Date:   Wed, 21 Jun 2006 15:18:18 +0000

- Remove load eventListener after it has been called once
- Capture editors from Google Books



Diffstat:
M
chrome/chromeFiles/content/scholar/xpcom/ingester.js
 | 
14
+++++++++-----
M
scrapers.sql
 | 
12
+++++++++++-

2 files changed, 20 insertions(+), 6 deletions(-)
diff --git a/chrome/chromeFiles/content/scholar/xpcom/ingester.js b/chrome/chromeFiles/content/scholar/xpcom/ingester.js
@@ -147,7 +147,7 @@ Scholar.Ingester.Utilities.prototype.processDocuments = function(browser, firstD
 		var onLoad = function() {
 			Scholar.debug("onLoad called");
 			if(hiddenBrowser.id == "scholar-hidden-browser") {
-				hiddenBrowser.removeEventListener("DOMContentLoaded", onLoad, true);
+				hiddenBrowser.removeEventListener("load", onLoad, true);
 				try {
 					var newHiddenBrowser = new Object();
 					Scholar.debug("new hidden browser");
@@ -586,20 +586,18 @@ Scholar.Ingester.Document.prototype.scrapePage = function(callback) {
 	
 	Scholar.debug("Scraping "+this.browser.contentDocument.location.href);
 	
-	Scholar.debug(this.scraper.scraperJavaScript);
-	
 	var scraperSandbox = this._sandbox;
 	try {
 		Components.utils.evalInSandbox(this.scraper.scraperJavaScript, scraperSandbox);
 	} catch(e) {
 		Scholar.debug(e+' in scraperJavaScript for '+this.scraper.label);
 		this._scrapePageComplete();
+		return;
 	}
 	
-	Scholar.debug("scraping complete");
-	
 	// If synchronous, call _scrapePageComplete();
 	if(!this._waitForCompletion) {
+		Scholar.debug("is asynch");
 		this._scrapePageComplete();
 	}
 }
@@ -739,6 +737,12 @@ Scholar.Ingester.Document.prototype._updateDatabase = function() {
 					creatorIndex++;
 				}
 			}
+			if(this.model.data[uri][prefixDummy + 'editor']) {
+				for(i in this.model.data[uri][prefixDummy + 'editor']) {
+					newItem.setCreator(creatorIndex, null, this.model.data[uri][prefixDummy + 'editor'][i], 3);
+					creatorIndex++;
+				}
+			}
 			
 			// Handle years, extracting from date if necessary
 			if(Scholar.ItemFields.isValidForType(Scholar.ItemFields.getID("year"), typeID)) {
diff --git a/scrapers.sql b/scrapers.sql
@@ -1412,11 +1412,21 @@ utilities.loadDocument(newUri, browser, function(newBrowser) {
 				for(j in authors) {
 					model.addStatement(uri, prefixDC + ''creator'', authors[j]);
 				}
+			} else if(field == "Editor(s)") {
+				var authors = value.split(", ");
+				for(j in authors) {
+					model.addStatement(uri, prefixDummy + ''editor'', authors[j]);
+				}
 			} else if(field == "Publisher") {
 				model.addStatement(uri, prefixDC + ''publisher'', value);
 			} else if(field == "Publication Date") {
+				var date = value;
+				
 				jsDate = new Date(value);
-				var date = utilities.dateToISO(jsDate);
+				if(!isNaN(jsDate.valueOf())) {
+					date = utilities.dateToISO(jsDate);
+				}
+				
 				model.addStatement(uri, prefixDC + ''date'', date);
 			} else if(field == "Format") {
 				model.addStatement(uri, prefixDC + ''medium'', value);