public void TestRegexLookupCount()
{
Workshare.Policy.Condition.ContentAnalysis analysis = new Workshare.Policy.Condition.ContentAnalysis();
string filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\TestDocument1.doc");
string description = "Originally a test document used by one of FCS Lite's tests.";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
string regEx = @"(The quick brown fox jumped over the lazy dog.){1}";
string[] context = { "Paragraph", "HeaderOrFooter", "Comment", "Footnote", "Endnote", "TrackChange", "TextBox", "Reviewer", "HiddenText", "SmallText", "WhiteText", "AttachedTemplate", "SmartTag", "Version", "AutoVersion", "Field", "Hyperlink", "RoutingSlip", "Variable", "HiddenSlide", "SpeakerNote" };
int regexCount = analysis.RegExLookupCount(testFile, regEx, context);
Assert.AreEqual(1, regexCount, "Expected the value of regexCount to be [1], but was [" + regexCount + "]");
}
}
public void TestTextLookupCount()
{
Workshare.Policy.Condition.ContentAnalysis analysis = new Workshare.Policy.Condition.ContentAnalysis();
string filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\TestDocument1.doc");
string description = "Originally a test document used by one of FCS Lite's tests.";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
string[] content = { "A4CC013A-BBE2-4a21-8651-F4511D689EDA" };
string[] context = { "Paragraph", "Header", "Footer", "Comment", "Footnote", "Endnote", "TrackChange", "TextBox", "Reviewer", "HiddenText", "SmallText", "WhiteText", "AttachedTemplate", "SmartTag", "Version", "AutoVersion", "Field", "Hyperlink", "RoutingSlip", "Variable", "HiddenSlide", "SpeakerNote" };
int textCount = analysis.TextLookupCount(testFile, content, context);
Assert.AreEqual(6, textCount, "Expected value of textCount to be [5], but was [" + textCount + "]");
}
}
public void TestTextLookupCountCorruptedWordDoc()
{
Workshare.Policy.Condition.ContentAnalysis analysis = new Workshare.Policy.Condition.ContentAnalysis();
string filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\TestDocument1.doc");
string description = "Originally a test document used by one of FCS Lite's tests.";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
string[] content = { "A4CC013A-BBE2-4a21-8651-F4511D689EDA" };
string[] context = { "Paragraph", "Header", "Footer", "Comment", "Footnote", "Endnote", "TrackChange", "TextBox", "Reviewer", "HiddenText", "SmallText", "WhiteText", "AttachedTemplate", "SmartTag", "Version", "AutoVersion", "Field", "Hyperlink", "RoutingSlip", "Variable", "HiddenSlide", "SpeakerNote" };
try
{
int textCount = analysis.TextLookupCount(testFile, content, context);
Assert.Fail("Should have thrown an exception on reading corrupted WORD document");
}
catch (Exception e)
{
string msg = e.Message;
}
}
}
public void TestHtmlHTMLTextLookupCount()
{
Workshare.Policy.Condition.ContentAnalysis analysis = new Workshare.Policy.Condition.ContentAnalysis();
string filename = m_testunicodehtmldoc;
string description = "An HTML file";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
string[] content = { "testing" };
string[] context = { "Paragraph" };
int textCount = analysis.TextLookupCount(testFile, content, context);
Assert.AreEqual(1, textCount, "Expected value of textCount to be [1], but was [" + textCount + "]");
// this removed until we can look at HtmlAgilityPack's use of codepages
//string[] content2 = { "正規表現" };
//textCount = analysis.TextLookupCount(testFile, content2, context);
//Assert.AreEqual(1, textCount, "Expected value of textCount to be [1], but was [" + textCount + "]");
}
}
public void TestCannotOpenContentInFileType()
{
Workshare.Policy.Condition.ContentAnalysis analysis = new Workshare.Policy.Condition.ContentAnalysis();
string[] fileTypes = { "RTFDocument", "XMLDocument" };
string filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\TestDocument1.doc");
string description = "Originally a test document used by one of FCS Lite's tests.";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
//file type is incorrect, not password protected
Assert.IsFalse(analysis.CannotOpenContentInFileType(testFile, fileTypes), "unexpected result in 'incorrect file type, not password protected'");
//file type is correct, not password protected
fileTypes = new string[] { "XMLDocument", "WordDocument", "RTFDocument"};
Assert.IsFalse(analysis.CannotOpenContentInFileType(testFile, fileTypes), "unexpected result in 'correct file type, not password protected'");
}
filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\PasswordProtectedDoc.doc");
description = "A password protected file";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
//file type is correct and document is password protected - nominal .doc case
Assert.IsTrue(analysis.CannotOpenContentInFileType(testFile, fileTypes), "unexpected result in 'incorrect file type, not password protected'");
//file type is incorrect, password protected
fileTypes = new string[] { "RTFDocument", "ExcelSpreadsheet" };
Assert.IsFalse(analysis.CannotOpenContentInFileType(testFile, fileTypes), "unexpected result in 'incorrect file type, not password protected'");
}
//nominal xls
filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\PasswordProtected.xls");
description = "A password protected file";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
//file type is correct and document is password protected
fileTypes = new string[] { "ExcelSheet", "PowerPoint", "PDFDocument", "ZIP", "RTFDocument" };
Assert.IsTrue(analysis.CannotOpenContentInFileType(testFile, fileTypes), "unexpected result on xls");
}
//nominal ppt
filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\PasswordProtected.ppt");
description = "A password protected file";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
//file type is correct and document is password protected
Assert.IsTrue(analysis.CannotOpenContentInFileType(testFile, fileTypes), "unexpected result on ppt");
}
//nominal pdf
filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\PasswordProtected.pdf");
description = "A password protected file";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
//file type is correct and document is password protected
Assert.IsTrue(analysis.CannotOpenContentInFileType(testFile, fileTypes), "unexpected result on pdf");
}
//nominal zip
filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\PasswordProtected.zip");
description = "A password protected file";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
//file type is correct and document is password protected
Assert.IsTrue(analysis.CannotOpenContentInFileType(testFile, fileTypes), "unexpected result on zip");
}
//nominal rtf
filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\profanity.rtf");
description = "A password protected file";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
//file type is correct and document is password protected
Assert.IsFalse(analysis.CannotOpenContentInFileType(testFile, fileTypes), "unexpected result on rtf");
}
//nominal text
filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\TestDocument.txt");
description = "A password protected file";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
//file type is correct and document is password protected
Assert.IsFalse(analysis.CannotOpenContentInFileType(testFile, fileTypes), "unexpected result on rtf");
}
}
public void TestRegexLookupInFileType()
{
Workshare.Policy.Condition.ContentAnalysis analysis = new Workshare.Policy.Condition.ContentAnalysis();
string regEx = @"(The quick brown fox jumped over the lazy dog.){1}";
string[] context = { "Paragraph", "HeaderOrFooter", "Comment", "Footnote", "Endnote", "TrackChange", "TextBox", "Reviewer", "HiddenText", "SmallText", "WhiteText", "AttachedTemplate", "SmartTag", "Version", "AutoVersion", "Field", "Hyperlink", "RoutingSlip", "Variable", "HiddenSlide", "SpeakerNote" };
string[] fileTypes = { "WordDocument" };
string filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\TestDocument1.doc");
string description = "Originally a test document used by one of FCS Lite's tests.";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
System.Collections.DictionaryEntry[] foundRegex = analysis.RegExLookupInFileType(testFile, fileTypes, regEx, context);
Assert.AreEqual(1, foundRegex.Length);
}
//right file type, wrong content
filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\testlookup.doc");
description = "File doesn't contain any of the stuff we're searching for";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
System.Collections.DictionaryEntry[] foundRegex = analysis.RegExLookupInFileType(testFile, fileTypes, regEx, context);
Assert.AreEqual(0, foundRegex.Length);
}
//wrong file type, right content
filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\TestDocument.txt");
description = "File contains things we're looking for";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
System.Collections.DictionaryEntry[] foundRegex = analysis.RegExLookupInFileType(testFile, fileTypes, regEx, context);
Assert.AreEqual(0, foundRegex.Length);
}
//now change file type and repeat with previous doc - right file type, right content
fileTypes[0] = "TextDocument";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
System.Collections.DictionaryEntry[] foundRegex = analysis.RegExLookupInFileType(testFile, fileTypes, regEx, context);
Assert.AreEqual(1, foundRegex.Length);
}
}
public void TestTextLookupInFileTypeMatchWholeWord()
{
Workshare.Policy.Condition.ContentAnalysis analysis = new Workshare.Policy.Condition.ContentAnalysis();
string[] content = { "A4CC013A-BBE2-4a21-8651-F4511D689EDA" };
bool[] matchWholeWords = { false };
string[] context = { "Paragraph", "Header", "Footer", "Comment", "Footnote", "Endnote", "TrackChange", "TextBox", "Reviewer", "HiddenText", "SmallText", "WhiteText", "AttachedTemplate", "SmartTag", "Version", "AutoVersion", "Field", "Hyperlink", "RoutingSlip", "Variable", "HiddenSlide", "SpeakerNote" };
string[] fileTypes = { "WordDocument" };
//right file type, right content
string filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\TestDocument1.doc");
string description = "Originally a test document used by one of FCS Lite's tests.";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
System.Collections.DictionaryEntry[] foundContents = analysis.TextLookupInFileType(testFile, fileTypes, content, context, matchWholeWords);
Assert.AreEqual(6, foundContents.Length);
}
matchWholeWords = new bool[] { true };
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
System.Collections.DictionaryEntry[] foundContents = analysis.TextLookupInFileType(testFile, fileTypes, content, context, matchWholeWords);
Assert.AreEqual(5, foundContents.Length);
}
//right file type, wrong content
filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\testlookup.doc");
description = "File doesn't contain any of the stuff we're searching for";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
System.Collections.DictionaryEntry[] foundContents = analysis.TextLookupInFileType(testFile, fileTypes, content, context, matchWholeWords);
Assert.AreEqual(0, foundContents.Length);
}
//wrong file type, right content
filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\TestDocument.txt");
description = "File contains things we're looking for";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
System.Collections.DictionaryEntry[] foundContents = analysis.TextLookupInFileType(testFile, fileTypes, content, context, matchWholeWords);
Assert.AreEqual(0, foundContents.Length);
}
//now change file type and repeat with previous doc - right file type, right content
fileTypes[0] = "TextDocument";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
System.Collections.DictionaryEntry[] foundContents = analysis.TextLookupInFileType(testFile, fileTypes, content, context, matchWholeWords);
Assert.AreEqual(1, foundContents.Length);
}
}
public void TestFallbackTextLookupCountBadType()
{
try
{
Workshare.Policy.Condition.ContentAnalysis analysis = new Workshare.Policy.Condition.ContentAnalysis();
string filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\unrecognized_type.xyz");
string description = "A random chunk of binary rubbish";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
string[] content = { "text" };
string[] context = { "Paragraph" };
int textCount = analysis.TextLookupCount(testFile, content, context);
}
}
catch(Exception)
{
Assert.Fail("Fallback should not cause an exception when analysing a corrupt document");
}
}
public void TestFallbackTextLookupCount()
{
Workshare.Policy.Condition.ContentAnalysis analysis = new Workshare.Policy.Condition.ContentAnalysis();
string filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\lotus-wordpro.lwp");
string description = "A lotus wordpro document";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
string[] content = { "text" };
string[] context = { "Paragraph" };
int textCount = analysis.TextLookupCount(testFile, content, context);
Assert.AreEqual(69, textCount, "Expected value of textCount to be [69], but was [" + textCount + "]");
}
}
public void TestPdfLookupCountOnCorruptedPdf()
{
Workshare.Policy.Condition.ContentAnalysis analysis = new Workshare.Policy.Condition.ContentAnalysis();
string filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\test2corrupted.pdf");
string description = "A corrupted pdf document";
try
{
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
string[] content = { "point" };
string[] context = { "Paragraph" };
int textCount = analysis.TextLookupCount(testFile, content, context);
Assert.Fail("Should have thrown an exception on reading corrupt document");
}
}
catch (Exception e)
{
string message = e.Message;
}
}
public void TestPdfTextLookupCountOverLineBreaks2()
{
Workshare.Policy.Condition.ContentAnalysis analysis = new Workshare.Policy.Condition.ContentAnalysis();
string filename = Workshare.TestUtils.TestFileUtils.MakeRootPathAbsolute(@"\projects\Hygiene\src\ContentAnalysis.Tests\docs\testlookup_adobe.pdf");
string description = "Test pdf document";
using (IFile testFile = Workshare.Policy.Engine.FileFactory.Create(filename, description))
{
string[] content = { "This first paragraph extends over multiple" };// multiple lines" };
string[] context = { "Paragraph" };
int textCount = analysis.TextLookupCount(testFile, content, context);
Assert.AreEqual(1, textCount, "Expected value of textCount to be [1], but was [" + textCount + "]");
}
}
