public void TestAPIStructure()
{
bool foundType = false;
bool foundItem = false;
bool foundDetail = false;
Discoverer d = new Discoverer();
string testDoc = TestUtils.TestFileUtils.MakeRootPathAbsolute(@"Projects\Workshare.API\Workshare.API.Tests\TestDocs\test.doc");
var result = d.DiscoverDocument(testDoc);
IMetadataTypeCollection tc = result.GetMetadataTypesFound();
foreach (MetadataType mt in tc)
{
foundType = true;
var typeResult = result.GetItemsForType(mt);
foreach (var item in typeResult)
{
foundItem = true;
foreach (var detail in item.Details)
{
foundDetail = true;
}
}
}
Assert.AreEqual(true, foundType, "No metadataTypes found");
Assert.AreEqual(true, foundItem, "no items found");
Assert.AreEqual(true, foundDetail, "no details found");
}
public static void Main(string[] args)
{
// ensure path to file is absolute, not relative
string filename = args[0];
if (!Path.IsPathRooted(filename))
filename = Path.Combine(Environment.CurrentDirectory, filename);
Discoverer d = new Discoverer();
string xmlReport = d.DiscoverDocumentToXml(filename);
XmlDocument xmlDoc = new XmlDocument();
xmlDoc.Load(xmlReport);
// process XML report
string htmlReport = d.DiscoverDocumentToHtmlReport(filename);
// display HTML report
}
public void TestSimpleCleanRtf()
{
string testDoc = TestUtils.TestFileUtils.MakeRootPathAbsolute(@"Projects\Workshare.API\Workshare.API.Tests\TestDocs\test.rtf");
using (TempFile tf = new TempFile(testDoc))
{
string initialHash = tf.MD5Sum;
IOfficeCleaner c = new OfficeCleaner();
c.CleanFile(tf.FilePath);
string newHash = tf.MD5Sum;
Assert.AreNotEqual(initialHash, newHash, "We expected the Cleanion to change the file contents");
Discoverer d = new Discoverer();
using (var result = d.DiscoverDocument(tf.FilePath))
{
Assert.IsFalse(result.HasMetadataOfType(MetadataType.TrackedChange));
}
}
}
public static void Main(string[] args)
{
// ensure path to file is absolute, not relative
string filename = args[0];
if (!Path.IsPathRooted(filename))
filename = Path.Combine(Environment.CurrentDirectory, filename);
Discoverer d = new Discoverer();
using (IDiscoveryResult dr = d.DiscoverDocument(filename))
{
foreach (MetadataType mt in dr.GetMetadataTypesFound())
{
Console.WriteLine(mt.ToString() + " : ");
foreach (IMetadataItem item in dr.GetItemsForType(mt))
{
foreach (IDetail detail in item.Details)
{
Console.WriteLine("\t" + detail.Name + " : " + detail.Value);
}
}
}
}
}
public void TestDiscoverAndRedact()
{
Discoverer d = new Discoverer();
string testDoc = TestUtils.TestFileUtils.MakeRootPathAbsolute(@"Projects\Workshare.API\Workshare.API.Tests\TestDocs\test.doc");
using (var result = d.DiscoverDocument(testDoc))
{
Assert.IsTrue(result.HasMetadataOfType(MetadataType.HiddenText));
Assert.IsTrue(result.HasMetadataOfType(MetadataType.TrackedChange));
Assert.IsTrue(result.HasMetadataOfType(MetadataType.Header));
Assert.IsTrue(result.HasMetadataOfType(MetadataType.Footer));
Assert.IsTrue(result.HasMetadataOfType(MetadataType.DocumentStatistic));
Assert.IsTrue(result.HasMetadataOfType(MetadataType.BuiltInProperty));
Assert.IsTrue(result.HasMetadataOfType(MetadataType.Reviewer));
Assert.IsNotNull(result.GetItemsForType(MetadataType.HiddenText));
Assert.IsNull(result.GetItemsForType(MetadataType.HiddenSlide));
}
string cleanedDoc = Path.GetTempFileName();
Redactor r = new Redactor();
r.RedactFileTo(testDoc, cleanedDoc);
using (var result = d.DiscoverDocument(cleanedDoc))
{
Assert.IsFalse(result.HasMetadataOfType(MetadataType.HiddenText));
Assert.IsFalse(result.HasMetadataOfType(MetadataType.TrackedChange));
Assert.IsTrue(result.HasMetadataOfType(MetadataType.Header), "redact cannot clean headers");
Assert.IsTrue(result.HasMetadataOfType(MetadataType.Footer), "redact cannot clean footers");
Assert.IsFalse(result.HasMetadataOfType(MetadataType.DocumentStatistic));
Assert.IsFalse(result.HasMetadataOfType(MetadataType.BuiltInProperty));
Assert.IsFalse(result.HasMetadataOfType(MetadataType.Reviewer));
}
}
public void TestRelativePath3()
{
IDiscoverer c = new Discoverer();
c.DiscoverDocumentToXml("test.doc");
}
public void TestDiscoverPDF02()
{
Discoverer d = new Discoverer();
string testDoc = TestUtils.TestFileUtils.MakeRootPathAbsolute(@"Projects\Workshare.API\Workshare.API.Tests\TestDocs\Comments_Bookmarks_Attachments_Markups.pdf");
//Properties = 36, Attachments = 37, Bookmarks = 38, Markups = 39,
using (var result = d.DiscoverDocument(testDoc))
{
Assert.IsNull(result.GetItemsForType(MetadataType.HiddenText));
Assert.IsNull(result.GetItemsForType(MetadataType.HiddenSlide));
Assert.IsFalse(result.HasMetadataOfType(MetadataType.HiddenText));
Assert.IsFalse(result.HasMetadataOfType(MetadataType.TrackedChange));
Assert.IsFalse(result.HasMetadataOfType(MetadataType.Header));
Assert.IsFalse(result.HasMetadataOfType(MetadataType.Footer));
Assert.IsFalse(result.HasMetadataOfType(MetadataType.DocumentStatistic));
Assert.IsFalse(result.HasMetadataOfType(MetadataType.BuiltInProperty));
Assert.IsFalse(result.HasMetadataOfType(MetadataType.Reviewer));
Assert.IsTrue(result.HasMetadataOfType(MetadataType.Properties));
Assert.IsTrue(result.GetItemsForType(MetadataType.Properties).Count == 7);
Assert.IsTrue(result.HasMetadataOfType(MetadataType.Attachments));
Assert.IsTrue(result.GetItemsForType(MetadataType.Attachments).Count == 1);
Assert.IsTrue(result.HasMetadataOfType(MetadataType.Bookmarks));
Assert.IsTrue(result.GetItemsForType(MetadataType.Bookmarks).Count == 4);
Assert.IsTrue(result.HasMetadataOfType(MetadataType.Markups));
Assert.IsTrue(result.GetItemsForType(MetadataType.Markups).Count > 13);
}
}
public void TestDiscoverToXmlTwice()
{
Discoverer d = new Discoverer();
string testDoc = TestUtils.TestFileUtils.MakeRootPathAbsolute(@"Projects\Workshare.API\Workshare.API.Tests\TestDocs\test.ppt");
string report = d.DiscoverDocumentToXml(testDoc);
Assert.IsFalse(string.IsNullOrEmpty(report));
Assert.IsTrue(File.Exists(report));
string report2 = d.DiscoverDocumentToXml(testDoc);
Assert.IsFalse(string.IsNullOrEmpty(report2));
Assert.IsTrue(File.Exists(report2));
Assert.IsTrue(File.Exists(report));
Assert.AreNotEqual(report, report2);
File.Delete(report);
File.Delete(report2);
}
public void TestDiscoverPDFToXmlAgain()
{
Discoverer d = new Discoverer();
string testDoc = TestUtils.TestFileUtils.MakeRootPathAbsolute(@"Projects\Workshare.API\Workshare.API.Tests\TestDocs\Comments_Bookmarks_Attachments_Markups.pdf");
string report = d.DiscoverDocumentToXml(testDoc);
Assert.IsFalse(string.IsNullOrEmpty(report));
Assert.IsTrue(File.Exists(report));
string contents = File.ReadAllText(report);
File.Delete(report);
Assert.IsTrue(contents.StartsWith("<?")); // which proves nothing :)
Assert.IsTrue(contents.Length > 3000);
}
public void TestRelativePath2()
{
IDiscoverer c = new Discoverer();
c.DiscoverDocumentToHtmlReport("test.doc");
}
public void TestDiscoverPDFToXml()
{
Discoverer d = new Discoverer();
string testDoc = TestUtils.TestFileUtils.MakeRootPathAbsolute(@"Projects\Workshare.API\Workshare.API.Tests\TestDocs\test.pdf");
string report = d.DiscoverDocumentToXml(testDoc);
Assert.IsFalse(string.IsNullOrEmpty(report));
Assert.IsTrue(File.Exists(report));
string contents = File.ReadAllText(report);
File.Delete(report);
Assert.IsTrue(contents.StartsWith("<?"));
Assert.IsTrue(contents.Length > 1000);
}
public void TestDiscoverPDFToHtml()
{
Discoverer d = new Discoverer();
string testDoc = TestUtils.TestFileUtils.MakeRootPathAbsolute(@"Projects\Workshare.API\Workshare.API.Tests\TestDocs\Comments_Bookmarks_Attachments_Markups.pdf");
string report = d.DiscoverDocumentToHtmlReport(testDoc);
Assert.IsFalse(string.IsNullOrEmpty(report));
Assert.IsTrue(File.Exists(report));
string contents = File.ReadAllText(report);
File.Delete(report);
Assert.IsTrue(contents.StartsWith("<html "));
Assert.IsTrue(contents.Length > 7000);
// and can test for extant phrases of each type:
Assert.IsTrue(contents.IndexOf("TextAnnotation on page") > 0);
Assert.IsTrue(contents.IndexOf("Keywords: AcroTeX, AeB Pro, Acrobat annotations") > 0);
Assert.IsTrue(contents.IndexOf("AAA.docx") > 0);
Assert.IsTrue(contents.IndexOf("1 Introduction , links to page # 2") > 0);
}