/// <summary>
///
/// </summary>
/// <param name="mediaItem"></param>
/// <returns></returns>
public ContentCrawlProxy BuildSearchableAsset(MediaData mediaItem)
{
var searchableAsset = mediaItem as ISearchableAsset;
if (searchableAsset == null)
{
return(null);
}
var url = UrlResolver.Current.GetUrl(mediaItem.ContentLink);
if (url == null)
{
return(null);
}
searchableAsset._ContentID = string.Format("{0}", mediaItem.ContentGuid);
try
{
searchableAsset.AssetBlob = EpiHelper.ReadEpiBlob(mediaItem.BinaryData);
}
catch (Exception ex)
{
if (_logger != null)
{
_logger.Error(string.Format("Asset Crawler: Error Retrieving Asset {0} {1}", ex.Message, ex.StackTrace));
}
}
var pageCrawlParameters = new ContentCrawlProxy();
pageCrawlParameters.Content.Add(new CrawlerContent()
{
Name = "title",
Value = mediaItem.Name,
});
pageCrawlParameters.Content.Add(new CrawlerContent()
{
Name = "url",
Value = url.Replace("?epslanguage=en", string.Format("?epslanguage={0}", mediaItem.Language.Name)),
});
pageCrawlParameters.Content.Add(new CrawlerContent()
{
Name = "pagetype",
Value = "Media",
});
pageCrawlParameters.Content.Add(new CrawlerContent()
{
Name = "mimetype",
Value = mediaItem.MimeType,
});
pageCrawlParameters.Content.Add(new CrawlerContent()
{
Name = "contenttype",
Value = MimeType.GetDisplayName(mediaItem.MimeType),
});
pageCrawlParameters.Content.Add(new CrawlerContent()
{
Name = "language",
Value = Languages,
});
pageCrawlParameters.Content.Add(new CrawlerContent()
{
Name = "folder",
Value = EpiHelper.GetParentFolderName(mediaItem.ParentLink.ToPageReference()),
});
pageCrawlParameters.Content.Add(new CrawlerContent()
{
Name = "paths",
Value = EpiHelper.GetPageTreePaths(mediaItem.ParentLink.ToPageReference()),
});
pageCrawlParameters.Content.Add(new CrawlerContent()
{
Name = "path",
Value = EpiHelper.GetFolderPath(mediaItem.ParentLink.ToPageReference()),
});
//searchableAsset.CrawlProperties = pageCrawlMetadata;
pageCrawlParameters.ContentItem = searchableAsset;
return(pageCrawlParameters);
}
/// <summary>
/// Used to intialize default properties
/// </summary>
/// <param name="page"></param>
/// <returns></returns>
public ContentCrawlProxy BuildSearchablePage(PageData page)
{
var searchablePage = page as ISearchableContent;
if (searchablePage == null)
{
return(null);
}
searchablePage._ContentID = string.Format("{0}-{1}", page.ContentGuid, page.Language.Name);
var pageProps = new ContentCrawlProxy();
pageProps.Content.Add(new CrawlerContent()
{
Name = "title",
Value = page.Name,
});
pageProps.Content.Add(new CrawlerContent()
{
Name = "timestamp",
Value = page.Changed,
});
pageProps.Content.Add(new CrawlerContent()
{
Name = "url",
Value = page.LinkURL.Replace("epslanguage=en", string.Format("epslanguage={0}", page.Language.Name)),
});
pageProps.Content.Add(new CrawlerContent()
{
Name = "contentid",
Value = page.ContentLink.ID.ToString(),
});
pageProps.Content.Add(new CrawlerContent()
{
Name = "categories",
Value = EpiHelper.GetCategoryPaths(page.Category),
});
pageProps.Content.Add(new CrawlerContent()
{
Name = "hostname",
Value = EpiHelper.GetSitePath(page.ContentLink),
});
pageProps.Content.Add(new CrawlerContent()
{
Name = "folder",
Value = EpiHelper.GetParentName(page.ParentLink.ToPageReference()),
});
pageProps.Content.Add(new CrawlerContent()
{
Name = "path",
Value = EpiHelper.GetPageTreePath(page.ParentLink.ToPageReference()),
});
pageProps.Content.Add(new CrawlerContent()
{
Name = "paths",
Value = EpiHelper.GetPageTreePaths(page.ParentLink.ToPageReference()),
});
pageProps.Content.Add(new CrawlerContent()
{
Name = "pagetype",
Value = page.PageTypeName,
});
pageProps.Content.Add(new CrawlerContent()
{
Name = "mimetype",
Value = "text/html",
});
pageProps.Content.Add(new CrawlerContent()
{
Name = "contenttype",
Value = "HTML",
});
pageProps.Content.Add(new CrawlerContent()
{
Name = "language",
Value = new List <string>()
{
page.Language.Name,
},
});
// if enabled scrape page content
if (_crawlSettings.PageScrapper != null)
{
var scrapContent = _crawlSettings.PageScrapper.ScrapPage(EpiHelper.GetExternalURL(page));
pageProps.Content.Add(new CrawlerContent()
{
Name = MissionSearch.Global.ContentField,
Value = scrapContent,
});
}
// parse searchable block data and add to content
var parsedBlockText = ProcessContentReferences(page);
if (parsedBlockText.Any())
{
pageProps.Content.AddRange(parsedBlockText);
}
//searchablePage.CrawlProperties = pageCrawlMetadata;
pageProps.ContentItem = searchablePage;
return(pageProps);
}
