EDIT: New Suggestion HTML Renderer for PDF using PdfSharp

(After trying wkhtmltopdf and suggesting to avoid it)

HtmlRenderer.PdfSharp is a 100% fully C# managed code, easy to use, thread safe and most importantly FREE (New BSD License) solution.

Usage

1. Download HtmlRenderer.PdfSharp nuget package.

2. Use Example Method.

   public static Byte[] PdfSharpConvert(String html)
   {
       Byte[] res = null;
       using (MemoryStream ms = new MemoryStream())
       {
           var pdf = TheArtOfDev.HtmlRenderer.PdfSharp.PdfGenerator.GeneratePdf(html, PdfSharp.PageSize.A4);
           pdf.Save(ms);
           res = ms.ToArray();
       }
       return res;
   }
   

A very Good Alternate Is a Free Version of iTextSharp

Until version 4.1.6 iTextSharp was licensed under the LGPL licence and versions until 4.16 (or there may be also forks) are available as packages and can be freely used. Of course someone can use the continued 5+ paid version.

I tried to integrate wkhtmltopdf solutions on my project and had a bunch of hurdles.

I personally would avoid using wkhtmltopdf - based solutions on Hosted Enterprise applications for the following reasons.

1. First of all wkhtmltopdf is C++ implemented not C#, and you will experience various problems embedding it within your C# code, especially while switching between 32bit and 64bit builds of your project. Had to try several workarounds including conditional project building etc. etc. just to avoid "invalid format exceptions" on different machines.
2. If you manage your own virtual machine its ok. But if your project is running within a constrained environment like (Azure (Actually is impossible withing azure as mentioned by the TuesPenchin author) , Elastic Beanstalk etc) it's a nightmare to configure that environment only for wkhtmltopdf to work.
3. wkhtmltopdf is creating files within your server so you have to manage user permissions and grant "write" access to where wkhtmltopdf is running.
4. Wkhtmltopdf is running as a standalone application, so its not managed by your IIS application pool. So you have to either host it as a service on another machine or you will experience processing spikes and memory consumption within your production server.
5. It uses temp files to generate the pdf, and in cases Like AWS EC2 which has really slow disk i/o it is a big performance problem.
6. The most hated "Unable to load DLL 'wkhtmltox.dll'" error reported by many users.

--- PRE Edit Section ---

For anyone who want to generate pdf from html in simpler applications / environments I leave my old post as suggestion.

TuesPechkin

https://www.nuget.org/packages/TuesPechkin/

or Especially For MVC Web Applications (But I think you may use it in any .net application)

Rotativa

https://www.nuget.org/packages/Rotativa/

They both utilize the wkhtmtopdf binary for converting html to pdf. Which uses the webkit engine for rendering the pages so it can also parse css style sheets.

They provide easy to use seamless integration with C#.

Rotativa can also generate directly PDFs from any Razor View.

Additionally for real world web applications they also manage thread safety etc...

After reading this issue, it seems that we should use absolute path with UserDataDir, so this works fine:

browser = await Puppeteer.LaunchAsync(new LaunchOptions
{
    Headless = true,
    UserDataDir = Path.Combine(AppDomain.CurrentDomain.BaseDirectory, ".local-chromium", "Win64-884014", "chrome-win", "UserData").Replace(@"\", @"\\")
});

But another problem appeared because browser doesn't save any cookie. Again, some issues on github said that the browser should be closed in order to save the cookies. I tried this:

await page.CloseAsync();
await browser.CloseAsync();

But it doesn't works as expected because the browser is still running in the background. The final solution for my was this:

var chromes = System.Diagnostics.Process.GetProcesses().Where(x => x.ProcessName == "chrome").ToList();

foreach (var ch in chromes)
{
    try
    {
        ch.Kill();
    }
    catch
    {
    }
}

The above code will close all chrome browsers, so it may need some changes in order to close the current browser only.

You can call EvaluateFunction passing that ElementHandle as an argument

var content = await Page.EvaluateFunctionAsync<string>("e => e.textContent", elementHandle);

If you have many scenarios like that, you can build an extension method to solve that for you ;)

You can wait for some selector that might tell you that the page is ready. e.g.:

await page.WaitForSelectorAsync(".someSelector");

For anybody still encountering this issue, the answer by jamie-tillman on this github issue solved it for me:

FROM mcr.microsoft.com/dotnet/aspnet:6.0 AS base
#####################
#PUPPETEER RECIPE based on https://github.com/hardkoded/puppeteer-sharp/issues/1180#issuecomment-1015532968
#####################
RUN apt-get update && apt-get -f install && apt-get -y install wget gnupg2 apt-utils
RUN wget -q -O - https://dl.google.com/linux/linux_signing_key.pub | apt-key add -
RUN echo 'deb [arch=amd64] http://dl.google.com/linux/chrome/deb/ stable main' >> /etc/apt/sources.list
RUN apt-get update \
&& apt-get install -y google-chrome-stable --no-install-recommends --allow-downgrades fonts-ipafont-gothic fonts-wqy-zenhei fonts-thai-tlwg fonts-kacst fonts-freefont-ttf
######################
#END PUPPETEER RECIPE
######################
ENV PUPPETEER_EXECUTABLE_PATH "/usr/bin/google-chrome-stable"
WORKDIR /app
EXPOSE 80
EXPOSE 443

The path to the executable is automatically detected by Puppeteer. However, I still had to disable sandboxmode as seen in the snippet below:

var browser = await Puppeteer.LaunchAsync(new LaunchOptions
{
    Headless = true,
    Args = new [] {
      "--disable-gpu",
      "--disable-dev-shm-usage",
      "--disable-setuid-sandbox",
      "--no-sandbox"}
});

More useful information is found in the issue thread so I recommend skimming through it if you still have trouble.

So my problem was happening because errors are apparently not serializable in javascript land, so we need to get the browser to run a function to extract the message and any other details we need from the error and return that.

This is what's working for me right now:

_page.Console += async (sender, args) =>
{
    switch (args.Message.Type)
    {
        case ConsoleType.Error:
            try
            {
                var errorArgs = await Task.WhenAll(args.Message.Args.Select(arg => arg.ExecutionContext.EvaluateFunctionAsync("(arg) => arg instanceof Error ? arg.message : arg", arg)));
                _logger.LogError($"{args.Message.Text} args: [{string.Join<object>(", ", errorArgs)}]");
            }
            catch { }
            break;
        case ConsoleType.Warning:
            _logger.LogWarning(args.Message.Text);
            break;
        default:
            _logger.LogInformation(args.Message.Text);
            break;
    }
};

I got the idea from a comment from one of the puppeteer contributors here and ported it to pupeteer-sharp.

The way it's supposed to be done atm is like this:

page.on('console', async msg => {
  // serialize my args the way I want
  const args = await Promise.all(msg.args.map(arg => arg.executionContext().evaluate(arg => {
    // I'm in a page context now. If my arg is an error - get me its message.
    if (arg instanceof Error)
      return arg.message;
    // return arg right away. since we use `executionContext.evaluate`, it'll return JSON value of
    // the argument if possible, or `undefined` if it fails to stringify it.
    return arg;
  }, arg)));
  console.log(...args);
});

If anyone needs it, here is the code I used below :

        Base64Image img = new Base64Image();            
        using (Page page = await parentPage.Browser.NewPageAsync())
        {                
            var tcs = new TaskCompletionSource();
            page.Response += async (sender, e) =>
            {
                string contentType = null;
                string[] imageMimeTypes = { "image/jpeg", "image/png", "image/webp", "image/gif" };

                if (e.Response.Headers.ContainsKey("content-type"))
                    contentType = (string)e.Response.Headers["content-type"].ToLower();

                if (e.Response.Headers.ContainsKey("Content-Type"))
                    contentType = (string)e.Response.Headers["Content-Type"].ToLower();

                if (imageMimeTypes.Any(t => t == contentType))
                {
                    img.ContentType = contentType;
                    

                    await e.Response.BufferAsync().AsTask().ContinueWith(async (data) => img.Bytes = await data);   
                    tcs.TrySetResult();
                }
            };
            await Task.WhenAll(page.GoToAsync(uri.AbsoluteUri), tcs.Task);
            await page.CloseAsync();
        }
        return img;

You don't need multiple browsers, you can use one browser with multiple Tabs (or Pages as Puppeteer calls them). Here's my sample code of that solves the same thing you do (convert HTML to PDF). It creates one browser instance which is handed down to four processes (could be more) which each create and remove their own Page.

public class PuppeteerSharpSample {

    public async Task CreatePdfBatch(IEnumerable<string> urlList)
    {
        await using var browser = await Puppeteer.LaunchAsync( new LaunchOptions { Headless = true, ExecutablePath ="PathToChromeOrChromium.exe"};).ConfigureAwait(false);

        await urlList.ForEachAsync(4, async url =>
            {
                await PrintPdf(url, browser).ConfigureAwait(false);
            })
            .ContinueWith(t =>
                {
                    if (t.Exception != null)
                    {
                        throw t.Exception;
                    }
                })
            .ConfigureAwait(false);
    }

    private async Task PrintPdf(Browser browser, string Url)
    {
        await using var page = await browser.NewPageAsync().ConfigureAwait(false);

        await page.GoToAsync(url).ConfigureAwait(false);

        await page.PdfAsync("pdfFileNameMustBeMadeUniqueOfCourse.pdf").ConfigureAwait(false);  
    }
}

public static class HelperClass
{
    //taken from https://scatteredcode.net/parallel-foreach-async-in-c/
    public static Task ForEachAsync<T>(this IEnumerable<T> source, int dop, Func<T, Task> body)
    {
        async Task AwaitPartition(IEnumerator<T> partition)
        {
            using (partition)
            {
                while (partition.MoveNext())
                {
                    await body(partition.Current).ContinueWith(t =>
                        {
                            if (t.IsFaulted && t.Exception != null)
                            {
                                throw t.Exception;
                            }
                        })
                        .ConfigureAwait(false);
                }
            }
        }

        return Task.WhenAll(
            Partitioner
                .Create(source)
                .GetPartitions(dop)
                .AsParallel()
                .Select(AwaitPartition));
    }
}

As a side note: you also don't need to use GetBrowser() if Chromium (can be Chrome, Chromium or the new Edge) is installed on your machine. Then you can just point to the .exe as is shown in the above code.

I solved the problem like this.

const links = await page.evaluate(() =>
              Array.from(document.querySelectorAll('a span'), e => 
              e.textContent));