internal static IEnumerable <string> ReadDatasetURLs(string dataset)
{
if (Datasets.Contains(dataset))
{
foreach (var line in FilesManager.ReadLinesFromEmbeddedFile("Dataset." + dataset + "_urls.csv"))
{
if (line.StartsWith('"'))
{
yield return(line.Substring(1, line.Length - 2));
}
else
{
yield return(line);
}
}
}
}
public MainPage()
{
this.InitializeComponent();
// Output folder link => clipboard
var outputPath = FilesManager.GetOutputFolderPath();
var cbData = new DataPackage();
cbData.SetText(outputPath);
Clipboard.SetContent(cbData);
// Initialize URLs list and navigate to next URL
urlsToCapture = URLsSource.ReadDatasetURLs(DATASET).ToList();
counter = STARTCOUNTER - 1;
if (STARTCOUNTER == 1)
{
counter = FilesManager.GetLastSavedCounter();
}
NavigateToNextUrl(this, null);
}
private async void CaptureScreenshotsAndNavigateToNextURL(object sender, RoutedEventArgs e)
{
do
{
try
{
await DoCaptureScreenshots();
await DoNavigateToNextUrl();
}
catch (Exception ex)
{
if (INTERACTIVE)
{
var md = new MessageDialog(ex.Message);
await md.ShowAsync();
}
else
{
FilesManager.WriteTextToFileAsync(counter.ToString("D5") + "_error.log", ex.Message);
}
}
}while (!INTERACTIVE);
}
private async Task DoCaptureScreenshots()
{
// Get unique file name for the current URL
var fileName = await JavascriptInterop.GetUniqueFileNameFromURLAsync(webview);
fileName = counter.ToString("D5") + "_" + fileName;
if (currentErrorStatus != WebErrorStatus.Unknown)
{
var errorString = Enum.GetName(typeof(WebErrorStatus), currentErrorStatus);
var errorMessage = "Error navigating to " + currentURL + " => " + errorString;
if (INTERACTIVE)
{
var md = new MessageDialog(errorMessage);
await md.ShowAsync();
}
else
{
FilesManager.WriteTextToFileAsync(fileName + "_error.log", errorMessage);
}
}
else
{
// Get view and content dimensions
var viewDimensions = ScreenCapture.GetViewDimensions(webview);
var contentDimensions = await ScreenCapture.GetContentDimensionsAsync(webview);
// Resize view to content size
ScreenCapture.SetViewDimensions(webview, contentDimensions);
// Capture a screenshot
await ScreenCapture.CreateAndSaveScreenshotAsync(webview, capture, null, warmup : true); // necessary because sometimes screenshot alters layout
await ScreenCapture.CreateAndSaveScreenshotAsync(webview, capture, fileName);
try
{
// Capture a description of all chars/words/lines/blocks bounding boxes
// Draw all these bounding boxes on the screen
var pageElementsTree = await ScreenCapture.CreateAndSaveTextBoundingBoxes(webview, fileName);
// Capture a new screenshot
await ScreenCapture.CreateAndSaveScreenshotAsync(webview, captureBoxes, fileName, "boxes");
}
catch (Exception e)
{
if (INTERACTIVE)
{
var md = new MessageDialog(e.Message);
await md.ShowAsync();
}
else
{
FilesManager.WriteTextToFileAsync(fileName + "_error.log", e.Message);
}
}
if (INTERACTIVE)
{
// Reset view to its original size
ScreenCapture.SetViewDimensions(webview, viewDimensions);
}
else
{
// Choose random width for next image
viewDimensions.width = ScreenCapture.GetRandowWidth();
ScreenCapture.SetViewDimensions(webview, viewDimensions);
}
}
}
static JavascriptInterop()
{
javascriptDefinitions = FilesManager.ReadStringFromEmbeddedFile(JS_FILE_PATH);
}
