public static IReadOnlyList <IAFile> GetItemThumbnailFiles(
this IAItem @this)
{
return(IAFileInterpreter.ScrapeArchiveThumbnailFiles(@this)
.Where(t => t.FilePathUrl.EndsWith(".jpg"))
.ToArray());
}
public static string GetItemDownloadPageUrl(
this IAItem @this)
{
return($"https://www.archive.org/" +
$"download/" +
$"{@this.Identifier}/");
}
public void TestGetItem()
{
IAItem item = userMethods.GetItem(1);
var expected = 1;
var actual = item.PartNo;
Assert.True(expected.Equals(actual));
}
public void CreateItemTest()
{
IFactory factory = new Factory();
List <string> mockData = new List <string>()
{
"7", "testpart", "I AM A TEST PART", "7.77", "7.77", "../"
};
IAItem expected = new AItem(mockData, new ConsoleLogger());
IAItem actual = factory.CreateItem(mockData, new ConsoleLogger());
Assert.True(expected.PartDescription.Equals(actual.PartDescription));
}
public virtual void DeepCopyIn(
IAItem item,
IAItemGetter rhs,
ErrorMaskBuilder?errorMask,
TranslationCrystal?copyMask,
bool deepCopy)
{
base.DeepCopyIn(
(IOblivionMajorRecord)item,
(IOblivionMajorRecordGetter)rhs,
errorMask,
copyMask,
deepCopy: deepCopy);
}
public static IEnumerable <IAFile> ScrapeArchiveFiles(
IAItem internetArchiveItem)
{
var context = BrowsingContext.New(
Configuration.Default.WithDefaultLoader());
var downloadPageUrl = internetArchiveItem.GetItemDownloadPageUrl();
using (var document = context
.OpenAsync(downloadPageUrl)
.GetAwaiter()
.GetResult())
{
var maincontent = document
.GetElementById("maincontent");
var container = maincontent
.GetElementsByClassName("container-ia")
.First();
var directoryListing = container
.GetElementsByClassName("download-directory-listing")
.First();
var tbody = directoryListing
.GetElementsByTagName("tbody")
.First();
var fileNodeList = tbody
.GetElementsByTagName("tr")
.Skip(1);
var itemIndex = 0;
foreach (var fileNode in fileNodeList)
{
var fileLinkElement = fileNode
.GetElementsByTagName("td")
.First()
.GetElementsByTagName("a")
.First();
var fileLinkPath = fileLinkElement
.GetAttribute("href");
var fileTitle = fileLinkElement
.TextContent;
var fileDate = fileNode
.GetElementsByTagName("td")
.Skip(1)
.First()
.TextContent;
var fileSize = fileNode
.GetElementsByTagName("td")
.Skip(2)
.First()
.TextContent;
var fileKind = DetermineIAFileKind(fileTitle);
//var airDate = DetermineArchiveFileAirDate(
// postShowStr)
// .GetValueOrDefault();
var approximateBytes = DetermineArchiveFileSizeBytes(fileSize);
if (!DateTime.TryParseExact(
fileDate,
"dd-MMM-yyyy ss:mm",
DateTimeFormatInfo.CurrentInfo,
DateTimeStyles.None,
out var lastModifiedDate))
{
throw new FormatException(
$"Cannot parse dateTime from string {fileDate.Quote()}.");
}
yield return(new IAFile(
internetArchiveItem,
fileLinkPath,
fileKind,
fileTitle,
lastModifiedDate,
approximateBytes,
itemIndex));
itemIndex++;
}
}
}
public static IReadOnlyList <IAFile> GetItemFiles(
this IAItem @this)
{
return(IAFileInterpreter.ScrapeArchiveFiles(@this)
.ToArray());
}
private IARoot GetInstantArticleCount(int maxDays, int maxEntries)
{
EntryCollection entries = null;
//We only build the entries if blogcore doesn't exist and we'll need them later...
if (dataService.GetLastEntryUpdate() == DateTime.MinValue)
{
entries = BuildEntries(null, maxDays, maxEntries);
}
//Try to get out as soon as possible with as little CPU as possible
if (inASMX)
{
DateTime lastModified = SiteUtilities.GetLatestModifedEntryDateTime(dataService, entries);
if (SiteUtilities.GetStatusNotModified(lastModified))
{
return(null);
}
}
if (inASMX)
{
string referrer = Context.Request.UrlReferrer != null ? Context.Request.UrlReferrer.AbsoluteUri : "";
if (ReferralBlackList.IsBlockedReferrer(referrer))
{
if (siteConfig.EnableReferralUrlBlackList404s)
{
return(null);
}
}
else
{
loggingService.AddReferral(
new LogDataItem(
Context.Request.RawUrl,
referrer,
Context.Request.UserAgent,
Context.Request.UserHostName));
}
}
//not-modified didn't work, do we have this in cache?
string CacheKey = string.Format("InstantArticle:{0}:{1}", maxDays, maxEntries);
IARoot documentRoot = cache[CacheKey] as IARoot;
if (documentRoot == null) //we'll have to build it...
{
//However, if we made it this far, the not-modified check didn't work, and we may not have entries...
if (entries == null)
{
entries = BuildEntries(null, maxDays, maxEntries);
}
documentRoot = new IARoot();
documentRoot.Namespaces.Add("content", "http://purl.org/rss/1.0/modules/content/");
IAChannel ch = new IAChannel();
ch.Title = siteConfig.Title;
ch.Link = SiteUtilities.GetBaseUrl(siteConfig);
if (siteConfig.Description == null || siteConfig.Description.Trim().Length == 0)
{
ch.Description = siteConfig.Subtitle;
}
else
{
ch.Description = siteConfig.Description;
}
ch.PubDate = DateTime.UtcNow.ToString();
ch.LastBuildDate = DateTime.UtcNow.ToString();
if (siteConfig.RssLanguage != null && siteConfig.RssLanguage.Length > 0)
{
ch.Language = siteConfig.RssLanguage;
}
//generator
ch.Docs = string.Empty;
documentRoot.Channels.Add(ch);
foreach (Entry entry in entries)
{
if (entry.IsPublic == false || entry.Syndicated == false)
{
continue;
}
IAItem item = new IAItem();
List <XmlElement> anyElements = new List <XmlElement>();
XmlDocument xmlDoc = new XmlDocument();
item.Title = entry.Title;
item.PubDate = entry.CreatedUtc.ToString("R");
if (ch.LastBuildDate == null || ch.LastBuildDate.Length == 0)
{
ch.LastBuildDate = item.PubDate;
}
item.Link = SiteUtilities.GetPermaLinkUrl(siteConfig, (ITitledEntry)entry);
item.Guid = entry.EntryId;
if (!siteConfig.AlwaysIncludeContentInRSS &&
entry.Description != null &&
entry.Description.Trim().Length > 0)
{
item.Description = PreprocessItemContent(entry.EntryId, entry.Description);
}
else
{
item.Description = ContentFormatter.FormatContentAsHTML(PreprocessItemContent(entry.EntryId, entry.Content));
}
XmlElement contentEncoded = xmlDoc.CreateElement("content", "encoded", "http://purl.org/rss/1.0/modules/content/");
string encData = string.Format("<!doctype html>" +
"<html lang=\"en\" prefix=\"op: http://media.facebook.com/op#\">" +
"<head>" +
"<meta charset=\"utf-8\">" +
"<link rel=\"canonical\" href=\"{3}\">" +
"<meta property=\"op:markup_version\" content=\"v1.0\">" +
"</head>" +
"<body><article>" +
"<header>{0}</header>" +
"{1}" +
"<footer>{2}</footer>" +
"</article></body></html>",
entry.Title,
ContentFormatter.FormatContentAsHTML(PreprocessItemContent(entry.EntryId, entry.Content)),
string.Empty, item.Link);
XmlCDataSection cdata = xmlDoc.CreateCDataSection(encData);
contentEncoded.AppendChild(cdata);
anyElements.Add(contentEncoded);
item.anyElements = anyElements.ToArray();
ch.Items.Add(item);
}
cache.Insert(CacheKey, documentRoot, DateTime.Now.AddMinutes(5));
}
return(documentRoot);
}
public void RemapLinks(IAItem obj, IReadOnlyDictionary <FormKey, FormKey> mapping)
{
base.RemapLinks(obj, mapping);
}
