public void Teardown()
{
_indexer = null;
_nOffset = 0;
_sFragments = null;
_nOffsets = null;
}
public void Setup()
{
_indexer = new HtmlIndexer();
_nOffset = 0;
_sFragments = new ArrayList();
_nOffsets = new ArrayList();
}
private void Indexing()
{
_indexer = new HtmlIndexer(_webSite.WebSiteIndex, _webSite.WebSiteUrl); //c:\index http://blog.codinghorror.com/
if (WebSite.NewIndex)
{
_indexer.AddDirectory(new DirectoryInfo(_webSite.WebSiteLocation), "*.htm*"); //\blog.codinghorror.com
}
_indexer.Close();
_searcher = new Searcher(_webSite.WebSiteIndex);
SearchEnabled = true;
BusyIndicator = false;
}
static bool ProcessResourceTextImpl(IResource res, IResourceTextConsumer consumer)
{
OutlookProcessor.CheckState();
try
{
OutlookProcessor processor = OutlookSession.OutlookProcessor;
if (processor != null)
{
MailBodyDescriptorDelegate myDelegate = CreateMailBodyDescriptor;
MailBodyDescriptor mailBody = (MailBodyDescriptor)processor.RunUniqueJob(myDelegate, res);
if (mailBody != null && Core.State != CoreState.ShuttingDown)
{
// Order of sections: Source, Subject, Body.
IResource resPerson = res.GetLinkProp(Core.ContactManager.Props.LinkFrom);
IResource resAccount = res.GetLinkProp(PROP.EmailAccountFrom);
if (resPerson != null)
{
// Construct [From] section out of contact name and its account
string fromText = resPerson.DisplayName;
if (resAccount != null)
{
fromText += " " + resAccount.DisplayName;
}
consumer.AddDocumentFragment(res.Id, fromText, DocumentSection.SourceSection);
}
consumer.AddDocumentHeading(res.Id, mailBody.Subject);
consumer.RestartOffsetCounting();
if (mailBody.IsHTML)
{
HtmlIndexer.IndexHtml(res, mailBody.Body, consumer, DocumentSection.BodySection);
}
else
{
consumer.AddDocumentFragment(res.Id, mailBody.Body.Replace("\r\n", "\n"));
}
}
}
}
catch (OutlookThreadTimeoutException)
{
if (consumer.Purpose == TextRequestPurpose.Indexing)
{
// retry indexing of the email later
Guard.QueryIndexingWithCheckId(res);
}
return(false);
}
return(true);
}
bool IResourceTextProvider.ProcessResourceText(IResource res, IResourceTextConsumer consumer)
{
if (res != null)
{
int id = res.Id;
// if (res.Type == NntpPlugin._newsArticle || res.Type == NntpPlugin._newsLocalArticle)
if (NntpPlugin.IsNntpType(res.Type))
{
string text = res.GetPropText(Core.Props.LongBody);
if (text.Trim().Length > 0)
{
consumer.AddDocumentFragment(id, text);
}
else
{
HtmlIndexer.IndexHtml(res, res.GetPropText(NntpPlugin._propHtmlContent), consumer, DocumentSection.BodySection);
}
consumer.RestartOffsetCounting();
consumer.AddDocumentHeading(id, res.GetPropText(Core.Props.Subject));
IResource author = res.GetLinkProp(Core.ContactManager.Props.LinkFrom);
IResource account = res.GetLinkProp(Core.ContactManager.Props.LinkEmailAcctFrom);
if (author != null)
{
// Construct [From] section out of contact name and its account
string fromText = author.DisplayName;
if (account != null)
{
fromText += " " + account.DisplayName;
}
consumer.AddDocumentFragment(id, fromText + " ", DocumentSection.SourceSection);
}
IResourceList groups = res.GetLinksOfType(NntpPlugin._newsGroup, NntpPlugin._propTo);
foreach (IResource group in groups)
{
consumer.AddDocumentFragment(id, group.GetPropText(Core.Props.Name) + " ", DocumentSection.SourceSection);
}
}
else
{
IResource article = res.GetLinkProp(NntpPlugin._propAttachment);
if (article != null && NntpPlugin.IsNntpType(article.Type))
{
consumer.AddDocumentHeading(id, res.GetPropText(Core.Props.Name));
}
}
}
return(true);
}
public void CheckOffsetsTags()
{
string sBody = "<html>Welcome</br>to <b>hell</b> </html>";
HtmlIndexer.IndexHtml(0xCD, sBody, this, null);
Assert.AreEqual(_sFragments.Count, 4);
Assert.AreEqual(_sFragments[0], "Welcome");
Assert.AreEqual(_sFragments[1], "to ");
Assert.AreEqual(_sFragments[2], "hell");
Assert.AreEqual(_sFragments[3], " ");
Assert.AreEqual(_nOffsets[0], sBody.IndexOf("Welcome"));
Assert.AreEqual(_nOffsets[1], sBody.IndexOf("to"));
Assert.AreEqual(_nOffsets[2], sBody.IndexOf("hell"));
Assert.AreEqual(_nOffsets[3], sBody.IndexOf(" "));
}
public void CheckOffsetsSimple()
{
string sBody = "Here comes the <nite>!";
HtmlIndexer.IndexHtml(0xCD, sBody, this, null);
Assert.AreEqual(_sFragments.Count, 4);
Assert.AreEqual(_sFragments[0], "Here ");
Assert.AreEqual(_sFragments[1], "comes ");
Assert.AreEqual(_sFragments[2], "the <");
Assert.AreEqual(_sFragments[3], "nite>!");
Assert.AreEqual(_nOffsets[0], sBody.IndexOf("Here"));
Assert.AreEqual(_nOffsets[1], sBody.IndexOf("comes"));
Assert.AreEqual(_nOffsets[2], sBody.IndexOf("the"));
Assert.AreEqual(_nOffsets[3], sBody.IndexOf("nite"));
}
internal static void RunConverter(string fileName, IResourceTextConsumer consumer, IResource res)
{
Process process = CreateConverterProcess(fileName, true);
try
{
if (!process.Start())
{
throw new Exception();
}
}
catch
{
return;
}
try
{
Encoding utf8 = new UTF8Encoding(false, false);
StreamReader outputReader = new StreamReader(process.StandardOutput.BaseStream, utf8);
string content = Utils.StreamReaderReadToEnd(outputReader);
consumer.RestartOffsetCounting(); // Just in case ;)
HtmlIndexer.IndexHtml(res, content, consumer, DocumentSection.BodySection);
}
finally
{
string stderr = Utils.StreamReaderReadToEnd(process.StandardError);
process.WaitForExit();
if (process.ExitCode != 0)
{
throw new Exception(_converterName + " (Excel-to-HTML) has performed an invalid operation while converting \"" + fileName + "\". " + stderr);
}
}
return;
}
private static void ProcessHTMLFragment(IResource res, IResourceTextConsumer consumer)
{
HtmlIndexer.IndexHtml(res, res.GetPropText(Core.Props.LongBody), consumer, null);
}
