public void BaseTestWithHeaderRow(string filename)
{
ParserContext context = new ParserContext(TestDataSample.GetExcelPath(filename));
ISpreadsheetParser parser = ParserFactory.CreateSpreadsheet(context);
parser.Context.Properties.Add("HasHeader", "1");
ToxySpreadsheet ss = parser.Parse();
Assert.AreEqual(1, ss.Tables[0].HeaderRows.Count);
Assert.AreEqual("A", ss.Tables[0].HeaderRows[0].Cells[0].Value);
Assert.AreEqual("B", ss.Tables[0].HeaderRows[0].Cells[1].Value);
Assert.AreEqual("C", ss.Tables[0].HeaderRows[0].Cells[2].Value);
Assert.AreEqual("D", ss.Tables[0].HeaderRows[0].Cells[3].Value);
Assert.AreEqual(3, ss.Tables[0].Rows.Count);
Assert.AreEqual("1", ss.Tables[0].Rows[0].Cells[0].Value);
Assert.AreEqual("2", ss.Tables[0].Rows[0].Cells[1].Value);
Assert.AreEqual("3", ss.Tables[0].Rows[0].Cells[2].Value);
Assert.AreEqual("4", ss.Tables[0].Rows[0].Cells[3].Value);
Assert.AreEqual("A1", ss.Tables[0].Rows[1].Cells[0].Value);
Assert.AreEqual("A2", ss.Tables[0].Rows[1].Cells[1].Value);
Assert.AreEqual("A3", ss.Tables[0].Rows[1].Cells[2].Value);
Assert.AreEqual("A4", ss.Tables[0].Rows[1].Cells[3].Value);
Assert.AreEqual("B1", ss.Tables[0].Rows[2].Cells[0].Value);
Assert.AreEqual("B2", ss.Tables[0].Rows[2].Cells[1].Value);
Assert.AreEqual("B3", ss.Tables[0].Rows[2].Cells[2].Value);
Assert.AreEqual("B4", ss.Tables[0].Rows[2].Cells[3].Value);
}
public void TestParseLineEvent()
{
string path = TestDataSample.GetTextPath("utf8.txt");
ParserContext context = new ParserContext(path);
PlainTextParser parser = (PlainTextParser)ParserFactory.CreateText(context);
parser.ParseLine += (sender, args) =>
{
if (args.LineNumber == 0)
{
Assert.AreEqual("hello world", args.Text);
}
else if (args.LineNumber == 1)
{
Assert.AreEqual("a2", args.Text);
}
else if (args.LineNumber == 2)
{
Assert.AreEqual("a3", args.Text);
}
else if (args.LineNumber == 3)
{
Assert.AreEqual("bbb4", args.Text);
}
};
string text = parser.Parse();
Assert.IsNotNullOrEmpty(text);
}
public void TestParseHtml()
{
string path = Path.GetFullPath(TestDataSample.GetHtmlPath("mshome.html"));
ParserContext context = new ParserContext(path);
IDomParser parser = (IDomParser)ParserFactory.CreateDom(context);
ToxyDom toxyDom = parser.Parse();
List <ToxyNode> metaNodeList = toxyDom.Root.SelectNodes("//meta");
Assert.AreEqual(7, metaNodeList.Count);
ToxyNode aNode = toxyDom.Root.SingleSelect("//a");
Assert.AreEqual(1, aNode.Attributes.Count);
Assert.AreEqual("href", aNode.Attributes[0].Name);
Assert.AreEqual("http://www.microsoft.com/en/us/default.aspx?redir=true", aNode.Attributes[0].Value);
ToxyNode titleNode = toxyDom.Root.ChildrenNodes[0].ChildrenNodes[0].ChildrenNodes[0];
Assert.AreEqual("title", titleNode.Name);
Assert.AreEqual("Microsoft Corporation", titleNode.ChildrenNodes[0].InnerText);
ToxyNode metaNode = toxyDom.Root.ChildrenNodes[0].ChildrenNodes[0].ChildrenNodes[7];
Assert.AreEqual("meta", metaNode.Name);
Assert.AreEqual(3, metaNode.Attributes.Count);
Assert.AreEqual("name", metaNode.Attributes[0].Name);
Assert.AreEqual("SearchDescription", metaNode.Attributes[0].Value);
Assert.AreEqual("scheme", metaNode.Attributes[2].Name);
Assert.AreEqual(string.Empty, metaNode.Attributes[2].Value);
}
public void TestReadExcelAndConvertToDataSet()
{
ParserContext c = new ParserContext(TestDataSample.GetExcelPath("Employee.xls"));
var parser = ParserFactory.CreateSpreadsheet(c);
var spreadsheet = parser.Parse();
DataSet ds = spreadsheet.ToDataSet();
Assert.AreEqual(3, ds.Tables.Count);
Assert.AreEqual("Sheet1", ds.Tables[0].TableName);
Assert.AreEqual("Sheet2", ds.Tables[1].TableName);
Assert.AreEqual("Sheet3", ds.Tables[2].TableName);
var s1 = ds.Tables[0];
Assert.AreEqual(System.DBNull.Value, s1.Rows[0][0]);
Assert.AreEqual(System.DBNull.Value, s1.Rows[0][1]);
Assert.AreEqual(System.DBNull.Value, s1.Rows[0][2]);
Assert.AreEqual("Employee Info", s1.Rows[1][1]);
Assert.AreEqual("Last name:", s1.Rows[3][1]);
Assert.AreEqual("lastName", s1.Rows[3][2]);
Assert.AreEqual("First name:", s1.Rows[4][1]);
Assert.AreEqual("firstName", s1.Rows[4][2]);
Assert.AreEqual("SSN:", s1.Rows[5][1]);
Assert.AreEqual("ssn", s1.Rows[5][2]);
}
public void TestParsePlainTextFromPDF()
{
string path = TestDataSample.GetPdfPath("Sample1.PDF");
var parser = new PDFTextParser(new ParserContext(path));
string result = parser.Parse();
Assert.IsTrue(result.StartsWith("LA MARCHE"));
}
public void TestReadRTF_Html()
{
string path = TestDataSample.GetRTFPath("htmlrtf2.rtf");
var parser = new RTFTextParser(new ParserContext(path));
string result = parser.Parse();
Assert.IsNotNullOrEmpty(result);
}
public void TestReadBigPDFFile()
{
string path = TestDataSample.GetPdfPath("Word97-2007BinaryFileFormat(doc)Specification.pdf");
var parser = new PDFTextParser(new ParserContext(path));
string result = parser.Parse();
Assert.IsTrue(true);
}
public void TestParseToxyDocumentFromPDF()
{
string path = TestDataSample.GetPdfPath("Sample1.PDF");
var parser = new PDFDocumentParser(new ParserContext(path));
var result = parser.Parse();
Assert.AreEqual(1474, result.Paragraphs.Count);
Assert.AreEqual("LA MARCHE MONDIALE DES FEMMES : UN MOUVEMENT IRRÉVERSIBLE", result.Paragraphs[0].Text);
}
public void TestParseOgg()
{
string path = Path.GetFullPath(TestDataSample.GetAudioPath("sample.ogg"));
ParserContext context = new ParserContext(path);
IMetadataParser parser = (IMetadataParser)ParserFactory.CreateMetadata(context);
ToxyMetadata x = parser.Parse();
Assert.AreEqual(15, x.Count);
}
public void TestParseMp3_Id3v1Only()
{
string path = Path.GetFullPath(TestDataSample.GetAudioPath("sample_v1_only.mp3"));
ParserContext context = new ParserContext(path);
IMetadataParser parser = (IMetadataParser)ParserFactory.CreateMetadata(context);
ToxyMetadata x = parser.Parse();
Assert.AreEqual(11, x.Count);
}
public void TestExcel2003TextParser()
{
ParserContext context = new ParserContext(TestDataSample.GetExcelPath("Employee.xls"));
ITextParser parser = ParserFactory.CreateText(context);
string result = parser.Parse();
Assert.IsNotNull(result);
Assert.IsTrue(result.IndexOf("Last name") > 0);
Assert.IsTrue(result.IndexOf("First name") > 0);
}
public void TestReadWholeText()
{
string path = TestDataSample.GetTextPath("utf8.txt");
ParserContext context = new ParserContext(path);
ITextParser parser = ParserFactory.CreateText(context);
string text = parser.Parse();
Assert.AreEqual("hello world" + Environment.NewLine + "a2" + Environment.NewLine + "a3" + Environment.NewLine + "bbb4" + Environment.NewLine, text);
}
public void PureTextMsg_ReadTextTest()
{
string path = TestDataSample.GetEmailPath("raw text mail demo.msg");
ParserContext context = new ParserContext(path);
var parser = ParserFactory.CreateText(context);
string result = parser.Parse();
Assert.IsNotNullOrEmpty(result);
}
public void TestPowerPoint()
{
string path = Path.GetFullPath(TestDataSample.GetOLE2Path("Test_Humor-Generation.ppt"));
ParserContext context = new ParserContext(path);
IMetadataParser parser = (IMetadataParser)ParserFactory.CreateMetadata(context);
ToxyMetadata x = parser.Parse();
Assert.AreEqual(8, x.Count);
Assert.AreEqual("Funny Factory", x.Get("Title").Value);
}
public void TestExcelFile()
{
string path = Path.GetFullPath(TestDataSample.GetExcelPath("comments.xls"));
ParserContext context = new ParserContext(path);
IMetadataParser parser = (IMetadataParser)ParserFactory.CreateMetadata(context);
ToxyMetadata x = parser.Parse();
Assert.AreEqual(8, x.Count);
Assert.AreEqual("Microsoft Excel", x.Get("ApplicationName").Value);
}
public void TestParseDirectoryFromZip()
{
ParserContext context = new ParserContext(TestDataSample.GetFilePath("toxy.zip", null));
ITextParser parser = ParserFactory.CreateText(context);
string list = parser.Parse();
Assert.IsNotNull(list);
string[] lines = list.Split(new string[] { Environment.NewLine }, StringSplitOptions.RemoveEmptyEntries);
Assert.AreEqual(68, lines.Length);
}
public void TestParsePlainTextFromSample5()
{
string path = TestDataSample.GetPdfPath("Sample5.PDF");
var parser = new PDFTextParser(new ParserContext(path));
string result = parser.Parse();
string[] results = result.Split('\n');
Assert.AreEqual("License income by market (%)", results[0]);
Assert.AreEqual("Philadelphia, Atlanta, Dallas, San Diego, and New", results[1]);
}
public void TestCorelDrawFile()
{
string path = Path.GetFullPath(TestDataSample.GetOLE2Path("TestCorel.shw"));
ParserContext context = new ParserContext(path);
IMetadataParser parser = (IMetadataParser)ParserFactory.CreateMetadata(context);
ToxyMetadata x = parser.Parse();
Assert.AreEqual(6, x.Count);
Assert.AreEqual("thorsteb", x.Get("Author").Value);
Assert.AreEqual("thorsteb", x.Get("LastAuthor").Value);
}
public void TestForeignNames()
{
string path = TestDataSample.GetVCardPath("PalmAgentSamples.vcf");
ParserContext context = new ParserContext(path);
VCardDocumentParser parser = new VCardDocumentParser(context);
var source = parser.Parse();
Assert.AreEqual(20, source.Cards.Count);
Assert.AreEqual("John Doe4", source.Cards[10].Name.FullName);
}
public void TestExcel2007TextParserWithoutSheetNames()
{
ParserContext context = new ParserContext(TestDataSample.GetExcelPath("WithVariousData.xlsx"));
context.Properties.Add("IncludeSheetNames", "0");
ITextParser parser = ParserFactory.CreateText(context);
string result = parser.Parse();
Assert.IsNotNull(result);
Assert.IsTrue(result.IndexOf("Sheet1") < 0);
}
public void TestExcel2007TextParserWithHeaderFooter()
{
ParserContext context = new ParserContext(TestDataSample.GetExcelPath("WithVariousData.xlsx"));
context.Properties.Add("IncludeHeaderFooter", "1");
ITextParser parser = ParserFactory.CreateText(context);
string result = parser.Parse();
Assert.IsNotNull(result);
Assert.IsTrue(result.IndexOf("This is the header") > 0);
}
public void TestPptx()
{
string path = Path.GetFullPath(TestDataSample.GetOOXMLPath("SampleShow.pptx"));
ParserContext context = new ParserContext(path);
IMetadataParser parser = (IMetadataParser)ParserFactory.CreateMetadata(context);
ToxyMetadata x = parser.Parse();
Assert.AreEqual(8, x.Count);
Assert.AreEqual("SlideShow Sample", x.Get("Title").Value);
Assert.AreEqual("Microsoft Office PowerPoint", x.Get("Application").Value);
}
public void TestDocx()
{
string path = Path.GetFullPath(TestDataSample.GetOOXMLPath("MultipleCoreProperties.docx"));
ParserContext context = new ParserContext(path);
IMetadataParser parser = (IMetadataParser)ParserFactory.CreateMetadata(context);
ToxyMetadata x = parser.Parse();
Assert.AreEqual(12, x.Count);
Assert.AreEqual("Format", x.Get("Title").Value);
Assert.AreEqual("Microsoft Macintosh Word", x.Get("Application").Value);
}
public void TestSolidWorksFile()
{
string path = Path.GetFullPath(TestDataSample.GetOLE2Path("TestSolidWorks.sldprt"));
ParserContext context = new ParserContext(path);
IMetadataParser parser = (IMetadataParser)ParserFactory.CreateMetadata(context);
ToxyMetadata x = parser.Parse();
Assert.AreEqual(10, x.Count);
Assert.AreEqual("{F29F85E0-4FF9-1068-AB91-08002B27B3D9}", x.Get("ClassID").Value);
Assert.AreEqual("scj", x.Get("LastAuthor").Value);
}
public void BaseTestWithoutHeader(string filename)
{
ParserContext context = new ParserContext(TestDataSample.GetExcelPath(filename));
ISpreadsheetParser parser = ParserFactory.CreateSpreadsheet(context);
ToxySpreadsheet ss = parser.Parse();
Assert.IsNull(ss.Tables[0].PageHeader);
Assert.AreEqual(0, ss.Tables[0].HeaderRows.Count);
Assert.AreEqual(9, ss.Tables[0].Rows.Count);
}
public void TestXlsx()
{
string path = Path.GetFullPath(TestDataSample.GetOOXMLPath("sample.xlsx"));
ParserContext context = new ParserContext(path);
IMetadataParser parser = (IMetadataParser)ParserFactory.CreateMetadata(context);
ToxyMetadata x = parser.Parse();
Assert.AreEqual(4, x.Count);
Assert.AreEqual("Microsoft Excel", x.Get("Application").Value);
Assert.AreEqual("12.0000", x.Get("AppVersion").Value);
}
public void TestParsePlainTextFromSample1()
{
string path = TestDataSample.GetPdfPath("Sample1.PDF");
var parser = new PDFTextParser(new ParserContext(path));
string result = parser.Parse();
Assert.IsTrue(result.StartsWith("LA MARCHE"));
ContainText(result, "Toute discussion stratégique sur nos actions nécessite un rappel de ce que nous avons fait en");
ContainText(result, "l’an 2000 et depuis. Au niveau mondial, en l’an 2000, nous avons mené une campagne de");
ContainText(result, "Une structure pour nous amener à 2005");
ContainText(result, "Lors de la 4e rencontre qui aura lieu en Inde, nous avons deux objectifs majeurs");
}
public void TestRead2Cards()
{
string path = TestDataSample.GetVCardPath("RfcAuthors.vcf");
ParserContext context = new ParserContext(path);
VCardDocumentParser parser = new VCardDocumentParser(context);
var cards = parser.Parse();
Assert.AreEqual(2, cards.Cards.Count);
ToxyBusinessCard tbc1 = cards.Cards[0];
Assert.AreEqual("Frank Dawson", tbc1.Name.FullName);
Assert.AreEqual(1, tbc1.Addresses.Count);
Assert.AreEqual(5, tbc1.Contacts.Count);
Assert.AreEqual("6544 Battleford Drive;Raleigh;NC;27613-3502;U.S.A.", tbc1.Addresses[0].ToString());
int i = 0;
Assert.AreEqual("+1-919-676-9515", tbc1.Contacts[i].Value);
Assert.AreEqual("MessagingService, WorkVoice", tbc1.Contacts[i].Name);
i++;
Assert.AreEqual("+1-919-676-9564", tbc1.Contacts[i].Value);
Assert.AreEqual("WorkFax", tbc1.Contacts[i].Name);
i++;
Assert.AreEqual("*****@*****.**", tbc1.Contacts[i].Value);
Assert.AreEqual("Internet", tbc1.Contacts[i].Name);
i++;
Assert.AreEqual("*****@*****.**", tbc1.Contacts[i].Value);
Assert.AreEqual("Internet", tbc1.Contacts[i].Name);
i++;
Assert.AreEqual("http://home.earthlink.net/~fdawson", tbc1.Contacts[i].Value);
Assert.AreEqual("Url-Default", tbc1.Contacts[i].Name);
Assert.AreEqual("Lotus Development Corporation", tbc1.Orgnization);
//2ed guy
ToxyBusinessCard tbc2 = cards.Cards[1];
Assert.AreEqual("Tim Howes", tbc2.Name.FullName);
Assert.AreEqual("Netscape Communications Corp.", tbc2.Orgnization);
Assert.AreEqual(1, tbc2.Addresses.Count);
Assert.AreEqual(3, tbc2.Contacts.Count);
Assert.AreEqual("501 E. Middlefield Rd.;Mountain View;CA;94043;U.S.A.", tbc2.Addresses[0].ToString());
i = 0;
Assert.AreEqual("+1-415-937-3419", tbc2.Contacts[i].Value);
Assert.AreEqual("MessagingService, WorkVoice", tbc2.Contacts[i].Name);
i++;
Assert.AreEqual("+1-415-528-4164", tbc2.Contacts[i].Value);
Assert.AreEqual("WorkFax", tbc2.Contacts[i].Name);
i++;
Assert.AreEqual("*****@*****.**", tbc2.Contacts[i].Value);
Assert.AreEqual("Internet", tbc2.Contacts[i].Name);
}
public void TestParseToxyDocumentFromPDF()
{
string path = TestDataSample.GetPdfPath("Sample1.PDF");
var parser = new PDFDocumentParser(new ParserContext(path));
var result = parser.Parse();
Assert.AreEqual(1474, result.Paragraphs.Count);
Assert.AreEqual("LA MARCHE MONDIALE DES FEMMES : UN MOUVEMENT IRRÉVERSIBLE", result.Paragraphs[0].Text);
Assert.AreEqual("DOCUMENT PRÉPARATOIRE", result.Paragraphs[1].Text);
Assert.AreEqual("e", result.Paragraphs[2].Text); //this is the superscript 'e'
Assert.AreEqual("4 Rencontre internationale de la Marche mondiale des femmes", result.Paragraphs[3].Text);
Assert.AreEqual("du 18-22 Mars 2003", result.Paragraphs[4].Text);
}
public void TestReadRTF_FormattedText()
{
string path = TestDataSample.GetRTFPath("Formated text.rtf");
var parser = new RTFTextParser(new ParserContext(path));
string result = parser.Parse();
string[] lines = result.Replace("\r\n", "\n").Split('\n');
Assert.AreEqual(lines.Length, 11);
Assert.AreEqual("11111111111", lines[0]);
Assert.AreEqual("22222222222", lines[1]);
Assert.AreEqual("张三李四王五", lines[2]);
Assert.AreEqual("RTF Sample , Author : yuans , contact : [email protected] , site : http://www.cnblogs.com/xdesigner .", lines[7]);
}
