public static void Run()
{
var configuration = new Configuration(Common.MyAppSid, Common.MyAppKey);
var apiInstance = new ParseApi(configuration);
try
{
var options = new TextOptions
{
FileInfo = new FileInfo
{
FilePath = "cells/two-worksheets.xlsx",
StorageName = Common.MyStorage
},
StartPageNumber = 1,
CountPagesToExtract = 1
};
var request = new TextRequest(options);
var response = apiInstance.Text(request);
foreach (var page in response.Pages)
{
Console.WriteLine($"PageIndex: {page.PageIndex}. Text: {page.Text}");
}
}
catch (Exception e)
{
Console.WriteLine("Exception while calling ParseApi: " + e.Message);
}
}
public void TestParse_RawTemplate()
{
var testFile = TestFiles.TemplateDocumentDocx;
var options = new ParseOptions
{
FileInfo = testFile.ToFileInfo(),
Template = GetTemplate()
};
var request = new ParseRequest(options);
var result = ParseApi.Parse(request);
Assert.IsNotNull(result);
Assert.IsNotEmpty(result.FieldsData);
Assert.AreEqual(4, result.Count);
var dataFieldNames = new[] { "FIELD1", "RELATEDFIELD2", "REGEX", "TABLECELLS" };
foreach (var field in result.FieldsData)
{
Assert.IsTrue(dataFieldNames.Contains(field.Name));
}
var table = result.FieldsData.First(x => string.Equals(x.Name, "TABLECELLS")).PageArea.PageTableArea;
if (table != null)
{
Assert.AreEqual(4, table.ColumnCount);
Assert.AreEqual(3, table.RowCount);
Assert.AreEqual(
"Cell 12",
table.PageTableAreaCells.First(x => x.ColumnIndex == 2 && x.RowIndex == 1).PageArea.PageTextArea.Text);
}
}
public void Analyse(IOutputType outputType)
{
var trackId = _interaction.GetTrackId();
var trackResults = ParseApi.GetResultsFor(trackId);
outputType.Print(trackResults);
}
public void TestExtractPages()
{
var testFile = TestFiles.FourPages;
var options = new TextOptions
{
FileInfo = testFile.ToFileInfo(),
StartPageNumber = 0,
CountPagesToExtract = 4,
FormattedTextOptions = new FormattedTextOptions
{
Mode = "PlainText"
}
};
var request = new TextRequest(options);
var result = ParseApi.Text(request);
Assert.IsNotNull(result.Pages);
Assert.AreEqual(0, result.Pages[0].PageIndex);
Assert.AreEqual(
"Text inside bookmark 0\r\n\r\nPage 0 heading\r\n\r\nPage Text - Page 0\r\n\r\n\fText inside bookmark 1\r\n\r\n",
result.Pages[0].Text);
Assert.AreEqual(3, result.Pages[3].PageIndex);
Assert.AreEqual("\fText inside bookmark 3\r\n\r\nPage 3 heading\r\n\r\nPage Text - Page 3\r\n\r\n", result.Pages[3].Text);
}
public static void Run()
{
var configuration = new Configuration(Common.MyAppSid, Common.MyAppKey);
var apiInstance = new ParseApi(configuration);
try
{
var options = new TextOptions
{
FileInfo = new FileInfo
{
FilePath = "pdf/PDF with attachements.pdf",
Password = "******",
StorageName = Common.MyStorage
},
ContainerItemInfo = new ContainerItemInfo {
RelativePath = "template-document.pdf"
},
StartPageNumber = 2,
CountPagesToExtract = 1
};
var request = new TextRequest(options);
var response = apiInstance.Text(request);
Console.WriteLine($"Text: {response.Pages[0].Text}");
}
catch (Exception e)
{
Console.WriteLine("Exception while calling ParseApi: " + e.Message);
}
}
public static void Run()
{
var configuration = new Configuration(Common.MyAppSid, Common.MyAppKey);
var apiInstance = new ParseApi(configuration);
try
{
var options = new ImagesOptions()
{
FileInfo = new FileInfo
{
FilePath = "slides/three-slides.pptx",
StorageName = Common.MyStorage
},
StartPageNumber = 1,
CountPagesToExtract = 2
};
var request = new ImagesRequest(options);
var response = apiInstance.Images(request);
foreach (var page in response.Pages)
{
Console.WriteLine($"Images from {page.PageIndex} page.");
foreach (var image in page.Images)
{
Console.WriteLine($"Image path in storage: {image.Path}. Download url: {image.DownloadUrl}");
Console.WriteLine($"File format: {image.FileFormat}. Page index: {image.PageIndex}");
}
}
}
catch (Exception e)
{
Console.WriteLine("Exception while calling ParseApi: " + e.Message);
}
}
public static void Run()
{
var configuration = new Configuration(Common.MyAppSid, Common.MyAppKey);
var apiInstance = new ParseApi(configuration);
try
{
var options = new TextOptions
{
FileInfo = new FileInfo
{
FilePath = "words-processing/docx/formatted-document.docx",
StorageName = Common.MyStorage
},
FormattedTextOptions = new FormattedTextOptions
{
Mode = "Markdown"
}
};
var request = new TextRequest(options);
var response = apiInstance.Text(request);
Console.WriteLine($"Text: {response.Text}");
}
catch (Exception e)
{
Console.WriteLine("Exception while calling ParseApi: " + e.Message);
}
}
public static void Run()
{
var configuration = new Configuration(Common.MyAppSid, Common.MyAppKey);
var apiInstance = new ParseApi(configuration);
try
{
var options = new TextOptions
{
FileInfo = new FileInfo
{
FilePath = "email/eml/embedded-image-and-attachment.eml",
StorageName = Common.MyStorage
}
};
var request = new TextRequest(options);
var response = apiInstance.Text(request);
Console.WriteLine($"Text: {response.Text}");
}
catch (Exception e)
{
Console.WriteLine("Exception while calling ParseApi: " + e.Message);
}
}
public void Analyse(IOutputType outputType)
{
int maxTrackId = int.Parse(ConfigurationManager.AppSettings["NumTracks"]);
for (int trackId = 0; trackId < maxTrackId; trackId++)
{
var results = ParseApi.GetResultsFor(trackId);
outputType.Print(results);
_ui.WaitForInteraction();
}
}
public void TestText_NotSupportedFile()
{
var testFile = TestFiles.JpegFile;
var options = new TextOptions
{
FileInfo = testFile.ToFileInfo(),
};
var request = new TextRequest(options);
var ex = Assert.Throws <ApiException>(() => { ParseApi.Text(request); });
Assert.AreEqual($"The specified file '{testFile.FullName}' has type which is not currently supported.", ex.Message);
}
public void TestParse_WithoutOptions()
{
var testFile = TestFiles.JpegFile;
var options = new ParseOptions
{
FileInfo = testFile.ToFileInfo(),
};
var request = new ParseRequest(options);
var ex = Assert.Throws <ApiException>(() => { ParseApi.Parse(request); });
Assert.AreEqual("Request parameters missing or have incorrect format", ex.Message);
}
public void TestGetImage_FileNotFoundResult()
{
var testFile = TestFiles.NotExist;
var options = new ImagesOptions
{
FileInfo = testFile.ToFileInfo(),
};
var request = new ImagesRequest(options);
var ex = Assert.Throws <ApiException>(() => { ParseApi.Images(request); });
Assert.AreEqual($"Can't find file located at '{testFile.FullName}'.", ex.Message);
}
public void TestText()
{
var testFile = TestFiles.OnePage;
var options = new TextOptions
{
FileInfo = testFile.ToFileInfo(),
};
var request = new TextRequest(options);
var result = ParseApi.Text(request);
Assert.IsNotNull(result.Text);
Assert.AreEqual("First Page\r\r\f", result.Text);
}
public void TestText_IncorrectPassword()
{
var testFile = TestFiles.PasswordProtected;
var options = new TextOptions
{
FileInfo = testFile.ToFileInfo(),
};
options.FileInfo.Password = "******";
var request = new TextRequest(options);
var ex = Assert.Throws <ApiException>(() => { ParseApi.Text(request); });
Assert.AreEqual($"Password provided for file '{testFile.FullName}' is incorrect.", ex.Message);
}
public void Analyse(IOutputType outputType)
{
int maxTrackId = int.Parse(ConfigurationManager.AppSettings["NumTracks"]);
var allData = new List <TrackData>();
var loadingBar = new LoadingBar("Fetching Data", maxTrackId);
loadingBar.Start();
for (int trackId = 0; trackId < maxTrackId; trackId++)
{
allData.Add(ParseApi.GetResultsFor(trackId));
loadingBar.Next();
}
outputType.Print(allData);
}
public void TestParse_FileNotFoundResult()
{
var testFile = TestFiles.NotExist;
var options = new ParseOptions
{
FileInfo = testFile.ToFileInfo(),
TemplatePath = "templates/document-template.json",
Template = GetTemplate()
};
var request = new ParseRequest(options);
var ex = Assert.Throws <ApiException>(() => { ParseApi.Parse(request); });
Assert.AreEqual($"Can't find file located at '{testFile.FullName}'.", ex.Message);
}
public void TestParse_NotSupportedFile()
{
var testFile = TestFiles.JpegFile;
var options = new ParseOptions
{
FileInfo = testFile.ToFileInfo(),
TemplatePath = "templates/document-template.json",
Template = GetTemplate()
};
var request = new ParseRequest(options);
var ex = Assert.Throws <ApiException>(() => { ParseApi.Parse(request); });
Assert.AreEqual($"The specified file '{testFile.FullName}' has type which is not currently supported.", ex.Message);
}
public void ImageExtractTest_Pdf_Container_FromPages_Error()
{
var testFile = TestFiles.Zip;
var options = new ImagesOptions
{
FileInfo = testFile.ToFileInfo(),
StartPageNumber = 1,
CountPagesToExtract = 2
};
var request = new ImagesRequest(options);
var ex = Assert.Throws <ApiException>(() => { ParseApi.Images(request); });
Assert.AreEqual($"The specified file '{testFile.FullName}' has type which is not currently supported.", ex.Message);
}
public void ImageExtractTest_Pdf_FromPages_OutOfThePageRange()
{
var testFile = TestFiles.Pdf;
var options = new ImagesOptions
{
FileInfo = testFile.ToFileInfo(),
StartPageNumber = 3,
CountPagesToExtract = 5
};
var request = new ImagesRequest(options);
var ex = Assert.Throws <ApiException>(() => { ParseApi.Images(request); });
Assert.AreEqual("Request parameters missing or have incorrect format", ex.Message);
}
public void TestGetImage_IncorrectPassword()
{
var testFile = TestFiles.PasswordProtected;
var options = new ImagesOptions
{
FileInfo = new FileInfo {
FilePath = testFile.FullName, Password = "******"
},
};
var request = new ImagesRequest(options);
var ex = Assert.Throws <ApiException>(() => { ParseApi.Images(request); });
Assert.AreEqual($"Password provided for file '{testFile.FullName}' is incorrect.", ex.Message);
}
public void BeforeAllTests()
{
var config = new Configuration(_appSid, _appKey)
{
ApiBaseUrl = _apiBaseUrl
};
ParseApi = new ParseApi(config);
InfoApi = new InfoApi(config);
TemplateApi = new TemplateApi(config);
FileApi = new FileApi(config);
FolderApi = new FolderApi(config);
StorageApi = new StorageApi(config);
UploadTestFiles();
}
public void TestParse_IncorrectPassword()
{
var testFile = TestFiles.PasswordProtected;
var options = new ParseOptions
{
FileInfo = testFile.ToFileInfo(),
TemplatePath = "templates/document-template.json",
Template = GetTemplate()
};
options.FileInfo.Password = "******";
var request = new ParseRequest(options);
var ex = Assert.Throws <ApiException>(() => { ParseApi.Parse(request); });
Assert.AreEqual($"Password provided for file '{testFile.FullName}' is incorrect.", ex.Message);
}
public static void Run()
{
// For example purposes create template if not exists.
TemplateUtils.CreateIfNotExist("templates/companies.json");
var configuration = new Configuration(Common.MyAppSid, Common.MyAppKey);
var apiInstance = new ParseApi(configuration);
try
{
var options = new ParseOptions
{
FileInfo = new FileInfo {
FilePath = "words-processing/docx/companies.docx", StorageName = Common.MyStorage
},
TemplatePath = "templates/companies.json"
};
var request = new ParseRequest(options);
var response = apiInstance.Parse(request);
foreach (var data in response.FieldsData)
{
if (data.PageArea.PageTextArea != null)
{
Console.WriteLine($"Field name: {data.Name}. Text : {data.PageArea.PageTextArea.Text}");
}
if (data.PageArea.PageTableArea != null)
{
Console.WriteLine($"Table name: {data.Name}.");
foreach (var cell in data.PageArea.PageTableArea.PageTableAreaCells)
{
Console.WriteLine(
$"Table cell. Row {cell.RowIndex} column {cell.ColumnIndex}. Text: {cell.PageArea.PageTextArea.Text}");
}
}
}
}
catch (Exception e)
{
Console.WriteLine("Exception while calling ParseApi: " + e.Message);
}
}
public static void Run()
{
var configuration = new Configuration(Common.MyAppSid, Common.MyAppKey);
var apiInstance = new ParseApi(configuration);
try
{
var options = new ParseOptions
{
FileInfo = new FileInfo {
FilePath = "containers/archive/companies.zip", StorageName = Common.MyStorage
},
ContainerItemInfo = new ContainerItemInfo {
RelativePath = "companies.docx"
},
Template = TemplateUtils.GetTemplate()
};
var request = new ParseRequest(options);
var response = apiInstance.Parse(request);
foreach (var data in response.FieldsData)
{
if (data.PageArea.PageTextArea != null)
{
Console.WriteLine($"Field name: {data.Name}. Text : {data.PageArea.PageTextArea.Text}");
}
if (data.PageArea.PageTableArea != null)
{
Console.WriteLine($"Table name: {data.Name}.");
foreach (var cell in data.PageArea.PageTableArea.PageTableAreaCells)
{
Console.WriteLine(
$"Table cell. Row {cell.RowIndex} column {cell.ColumnIndex}. Text: {cell.PageArea.PageTextArea.Text}");
}
}
}
}
catch (Exception e)
{
Console.WriteLine("Exception while calling ParseApi: " + e.Message);
}
}
public void TestExtractFormattedPage()
{
var testFile = TestFiles.FormattedDocument;
var options = new TextOptions
{
FileInfo = testFile.ToFileInfo(),
FormattedTextOptions = new FormattedTextOptions
{
Mode = "Markdown"
},
StartPageNumber = 1,
CountPagesToExtract = 1
};
var request = new TextRequest(options);
var result = ParseApi.Text(request);
Assert.IsNotEmpty(result.Pages);
Assert.IsTrue(result.Pages[0].Text.Contains("**Second page bold text**"));
Assert.IsTrue(result.Pages[0].Text.Contains("# Second page heading"));
}
public void TestGetImage_Docx()
{
var testFile = TestFiles.FourPages;
var options = new ImagesOptions
{
FileInfo = testFile.ToFileInfo(),
};
var request = new ImagesRequest(options);
var result = ParseApi.Images(request);
Assert.IsNotNull(result);
int i = 0;
foreach (var image in result.Images)
{
Assert.AreEqual($"parser/images/words/docx/four-pages_docx/image_{i}.jpeg", image.Path);
Assert.NotNull(image.DownloadUrl);
i++;
}
}
public void TestGetImage_Pdf_FromPages()
{
var testFile = TestFiles.Pdf;
var options = new ImagesOptions
{
FileInfo = testFile.ToFileInfo(),
StartPageNumber = 1,
CountPagesToExtract = 2
};
var request = new ImagesRequest(options);
var result = ParseApi.Images(request);
Assert.IsNotNull(result);
Assert.IsNotEmpty(result.Pages);
Assert.AreEqual(2, result.Pages.Count);
Assert.AreEqual("parser/images/pdf/template-document_pdf/page_1/image_0.jpeg", result.Pages[0].Images[0].Path);
Assert.AreEqual("parser/images/pdf/template-document_pdf/page_2/image_0.jpeg", result.Pages[1].Images[0].Path);
}
public void TestGetImage_Email()
{
var testFile = TestFiles.ImageAndAttachment;
var options = new ImagesOptions
{
FileInfo = testFile.ToFileInfo(),
};
var request = new ImagesRequest(options);
var result = ParseApi.Images(request);
Assert.IsNotNull(result);
var paths = new[]
{
"parser/images/email/eml/embedded-image-and-attachment_eml/",
};
foreach (var image in result.Images)
{
Assert.IsTrue(paths.Any(image.Path.Contains));
}
}
public void TestText_WithPassword()
{
var testFile = TestFiles.PasswordProtected;
var options = new TextOptions
{
FileInfo = testFile.ToFileInfo(),
StartPageNumber = 0,
CountPagesToExtract = 1,
FormattedTextOptions = new FormattedTextOptions
{
Mode = "PlainText"
}
};
var request = new TextRequest(options);
var result = ParseApi.Text(request);
Assert.IsNull(result.Text);
Assert.AreEqual(
"Text inside a bookmark 1\r\n\r\nPage 1 heading!\r\n\r\nSample test text - Page 1!\r\n\r\n\fText inside a bookmark 2\r\n\r\n",
result.Pages[0].Text);
}
public void TestExtractFormatted()
{
var testFile = TestFiles.FormattedDocument;
var options = new TextOptions
{
FileInfo = testFile.ToFileInfo(),
FormattedTextOptions = new FormattedTextOptions
{
Mode = "Html"
}
};
var request = new TextRequest(options);
var result = ParseApi.Text(request);
Assert.IsNotNull(result);
Assert.IsTrue(result.Text.Contains("<b>Bold text</b>"));
Assert.IsTrue(result.Text.Contains("<i>Italic text</i>"));
Assert.IsTrue(result.Text.Contains("<h1>Heading 1</h1>"));
Assert.IsTrue(result.Text.Contains("<tr><td><p>table</p></td>"));
Assert.IsTrue(result.Text.Contains("<ol><li><i>First element</i>"));
Assert.IsTrue(result.Text.Contains("<a href=\"http://targetwebsite.domain\">Hyperlink </a>"));
}