public static async Task <string> ProcessVideo(
[ActivityTrigger] DurableActivityContext ctx,
TraceWriter log
)
{
var url = ctx.GetInput <string>();
var asm = Assembly.Load("Microsoft.Extensions.Options, Culture=neutral, PublicKeyToken=adb9793829ddae60");
PuppeteerSharp.Browser browser = null;
try
{
browser = await Puppeteer.ConnectAsync(new ConnectOptions { BrowserWSEndpoint = Constants.BrowserWSEndpoint });
var page = await browser.NewPageAsync();
await page.GoToAsync(url, new NavigationOptions { Timeout = 4000 });
}
catch (Exception ex)
{
return(ex.Message);
}
finally
{
if (browser != null)
{
browser.Disconnect();
}
}
return($"Processed {url} shit all.");
}
static async System.Threading.Tasks.Task StartWebBrowser()
{
var browserRevision = PuppeteerSharp.BrowserFetcher.DefaultChromiumRevision;
var browserFetcher = new PuppeteerSharp.BrowserFetcher(new PuppeteerSharp.BrowserFetcherOptions
{
Path = BrowserDownloadDirPath()
});
await browserFetcher.DownloadAsync(browserRevision);
browser = await PuppeteerSharp.Puppeteer.LaunchAsync(new PuppeteerSharp.LaunchOptions
{
Headless = false,
UserDataDir = UserDataDirPath("default"),
DefaultViewport = null,
ExecutablePath = browserFetcher.RevisionInfo(browserRevision).ExecutablePath,
});
browserPage = (await browser.PagesAsync()).FirstOrDefault() ?? await browser.NewPageAsync();
await browserPage.ExposeFunctionAsync("____callback____", (string returnValue) =>
{
callbackFromBrowserDelegate?.Invoke(returnValue);
return(0);
});
}
public PuppeteerBrowser(IOptionsMonitor <BrowserConfiguration> optionsAccessor, ILogger <PuppeteerBrowser> logger)
{
_logger = logger;
_options = optionsAccessor;
_outputDir = _options.CurrentValue.OutputDirectory;
// This can definitely fail and probably shouldn't be in a constructor, but...
_browser = Puppeteer.LaunchAsync(new LaunchOptions
{
Headless = true,
Args = new [] { "--no-sandbox", "--disable-setuid-sandbox" }
}).Result;
}
public static async Task <bool> LoginToTiqa(PuppeteerSharp.Browser browser)
{
try
{
var pages = await browser.PagesAsync();
var page = pages[0];
await page.GoToAsync(AppConfig.tiqaUrl);
var un = await page.WaitForSelectorAsync(AppConfig.unFld);
var pw = await page.WaitForSelectorAsync(AppConfig.pwFld);
var lgn = await page.WaitForSelectorAsync(AppConfig.lgnBtn);
await un.TypeAsync(AppConfig.un);
await pw.TypeAsync(AppConfig.pw);
await lgn.ClickAsync();
await page.WaitForNavigationAsync();
var cks = await page.GetCookiesAsync();
AppConfig.cks = cks.ToList();//.Where(c => c.Domain == new Uri(page.Url).Host).ToList();
AppConfig.sessId = AppConfig.cks.Where(c => c.Name == AppConfig.sessIdName).FirstOrDefault().Value;
AppConfig.WriteOut($">> session id for crawl set to {AppConfig.sessId}");
AppConfig.crawlStartUri = page.Url;
return(true);
}
catch (NavigationException ex)
{
AppConfig.ErrHand(ex, $"XX failed to navigate to url");
return(false);
}
catch (Exception ex)
{
AppConfig.ErrHand(ex);
return(false);
}
}
public async Task <byte[]> ScreenshotPNG(string url, IEnumerable <CookieParam> cookies, ViewPortOptions viewport, ScreenshotOptions options, RevisionInfo revision)
{
LaunchOptions launchOptions = new LaunchOptions()
{
ExecutablePath = revision.ExecutablePath,
Args = BrowserArgs,
Headless = true,
DefaultViewport = viewport,
Timeout = 0
};
browser = await Puppeteer.LaunchAsync(launchOptions);
try {
var page = await browser.NewPageAsync();
try {
NavigationOptions navigationOptions = new NavigationOptions()
{
Timeout = 0
};
string originalUserAgent = await browser.GetUserAgentAsync();
await page.SetUserAgentAsync($"{originalUserAgent} {USER_AGENT_SUFFIX}");
if (cookies.Any())
{
await page.SetCookieAsync(cookies.ToArray());
}
await page.GoToAsync(url, navigationOptions);
return(await page.ScreenshotDataAsync(options));
} finally {
await Cleanup(page);
}
} finally {
Cleanup(browser);
}
}
public async Task StartBrowserAsync()
{
var fetcher = new BrowserFetcher();
var revisions = fetcher.LocalRevisions();
if (!revisions.Any())
{
_logger.LogInformation("Downloading Chromium browser");
await fetcher.DownloadAsync(BrowserFetcher.DefaultRevision);
}
_logger.LogInformation("Launching Chromium browser");
Browser = await Puppeteer.LaunchAsync(new LaunchOptions
{
DefaultViewport = new ViewPortOptions()
{
Width = 1366,
Height = 768
},
Headless = !Debugger.IsAttached
});
}
public async Task <byte[]> PrintPDF(string url, IEnumerable <CookieParam> cookies, ViewPortOptions viewport, PdfOptions options, RevisionInfo revision)
{
LaunchOptions launchOptions = new LaunchOptions()
{
ExecutablePath = "C:\\Program Files (x86)\\Google\\Chrome\\Application\\chrome.exe",
Args = BrowserArgs,
Headless = true,
DefaultViewport = viewport,
Timeout = 0
};
browser = await Puppeteer.LaunchAsync(launchOptions);
try {
var page = await browser.NewPageAsync();
try {
WaitUntilNavigation[] waitUntilNavigations = { WaitUntilNavigation.Networkidle0 };
NavigationOptions navigationOptions = new NavigationOptions()
{
Timeout = 0,
WaitUntil = waitUntilNavigations
};
string originalUserAgent = await browser.GetUserAgentAsync();
await page.SetUserAgentAsync($"{originalUserAgent} {USER_AGENT_SUFFIX}");
if (cookies.Any())
{
await page.SetCookieAsync(cookies.ToArray());
}
await page.GoToAsync(url, navigationOptions);
await InjectCustomStylesAsync(page, ref options);
bool hasPageNumbers = await page.EvaluateFunctionAsync <bool>("(window.UltimatePDF? window.UltimatePDF.hasPageNumbers : function() { return false; })");
if (hasPageNumbers)
{
/*
* When the layout has page numbers, we first retrieve a
* first blank pdf to calculate the number of pages.
* Then, knowing how many pages, we can layout the headers and footers,
* and retrieve the final pdf.
*/
byte[] blankContents = await page.PdfDataAsync(options);
using (var blankPdf = new PDFUtils(blankContents)) {
await page.EvaluateFunctionAsync("window.UltimatePDF.layoutPages", blankPdf.Pages);
return(await page.PdfDataAsync(options));
}
}
else
{
return(await page.PdfDataAsync(options));
}
} finally {
await Cleanup(page);
}
} finally {
Cleanup(browser);
}
}
private void Cleanup(PuppeteerSharp.Browser browser)
{
browser.Disconnect();
Dispose();
}
