[+] Process html files

[+] Convert html to pdf

Author: SolidProgramming

.gitignore

@@ -360,4 +360,5 @@ MigrationBackup/
 .ionide/
 
 # Fody - auto-generated XML schema
-FodyWeavers.xsd
+FodyWeavers.xsd
+/PuppeteerPDFAPI/Binaries

PuppeteerPDFAPI/Misc/Helpers.cs

@@ -27,7 +27,7 @@ internal static string GetBrowserPath()
             if (!Directory.Exists(path))
                 Directory.CreateDirectory(path);
 
-            if (browserName.ToLower() == "chromium")
+            if (browserName.Equals("chromium", StringComparison.CurrentCultureIgnoreCase))
                 browserName = "chrome";
 
             if (RuntimeInformation.IsOSPlatform(OSPlatform.Windows))

PuppeteerPDFAPI/Models/FileModel.cs

@@ -3,15 +3,15 @@
     public class FileModel
     {
         public int Id { get; init; }
-        public string Guid { get; set; } = default!;
-        public string FileExtension { get; set; } = default!;
+        public string? Guid { get; set; }
+        public string? FileExtension { get; set; }
         public string Filename
         {
             get
             {
                 return Guid + FileExtension;
             }
         }
-        public IFormFile File { get; set; } = default!;
+        public string? Content { get; set; }
     }
 }

PuppeteerPDFAPI/Program.cs

@@ -1,7 +1,6 @@
 global using PuppeteerPDFAPI.Models;
 using PuppeteerPDFAPI.Services;
 
-
 WebApplicationBuilder builder = WebApplication.CreateSlimBuilder(args);
 
 builder.Services.AddSingleton<IFileService, FileService>();
@@ -14,25 +13,30 @@
 
 WebApplication app = builder.Build();
 
-app.MapPost("pdf", async (IFileService fileService, HttpRequest request) =>
+app.MapPost("pdf", async (IFileService fileService, IPuppeteerService puppeteerService, HttpRequest request) =>
 {
     if (!request.Form.Files.Any())
         return Results.BadRequest("No uploaded files!");
 
     if (request.Form.Files.Count > 1)
-        return Results.Problem("Too many files uploaded. Upload limit = 1");
+        return Results.BadRequest("Too many files uploaded. Upload limit = 1");
+
+    IFormFile file = request.Form.Files[0];
+
+    (FileModel? processedFile, string? ErrorMessage) = await fileService.ProcessAsync(file);
+
+    if (processedFile is null)
+        return Results.BadRequest(ErrorMessage);
 
-    FileModel file = new()
-    {
-        Guid = Guid.NewGuid().ToString(),
-        FileExtension = request.Form.Files[0].FileName,
-        File = request.Form.Files[0]
-    };
+    if (string.IsNullOrEmpty(processedFile.Content))
+        return Results.BadRequest("Empty file!");
 
+    (Stream? pdfStream, ErrorMessage) = await puppeteerService.ConvertAsync(processedFile.Content);
 
-    await fileService.UploadAsync(file);
+    if (pdfStream is null)
+        return Results.Problem(ErrorMessage);
 
-    return Results.Ok();
+    return Results.File(pdfStream, contentType: "application/pdf");
 });
 
 IPuppeteerService puppeteerService = app.Services.GetRequiredService<IPuppeteerService>();

PuppeteerPDFAPI/Properties/launchSettings.json

@@ -4,8 +4,8 @@
     "http": {
       "commandName": "Project",
       "dotnetRunMessages": true,
-      "launchBrowser": true,
-      "launchUrl": "todos",
+      "launchBrowser": false,
+      "launchUrl": "",
       "applicationUrl": "http://localhost:5252",
       "environmentVariables": {
         "ASPNETCORE_ENVIRONMENT": "Development"

PuppeteerPDFAPI/Services/FileService.cs

@@ -1,52 +1,57 @@
 using HtmlAgilityPack;
-using PuppeteerPDFAPI.Models;
+using PuppeteerSharp;
+using System.IO;
+using System.Text;
 
 namespace PuppeteerPDFAPI.Services
 {
     public interface IFileService
     {
-        Task<(bool Success, string? ErrorMessage)> UploadAsync(FileModel file);
-        bool IsValidHtml();
+        Task<(FileModel?, string? ErrorMessage)> ProcessAsync(IFormFile file);
     }
 
     public class FileService(ILogger<FileService> logger) : IFileService
     {
-        public bool IsValidHtml()
+        public async Task<(FileModel?, string? ErrorMessage)> ProcessAsync(IFormFile formFile)
         {
-            throw new NotImplementedException();
-        }
+            string? extension = Path.GetExtension(formFile.FileName);
 
-        public async Task<(bool Success, string? ErrorMessage)> UploadAsync(FileModel file)
-        {
-            return await Task.Run<(bool, string?)>(() =>
+            if (extension == null)
+                return (default, "No file extension found!");
+
+            if (!extension.Equals(".html"))
+                return (default, "File has to be of type HTML!");
+
+            using StreamReader sr = new(formFile.OpenReadStream(), Encoding.UTF8);
+
+            string content = await sr.ReadToEndAsync();
+
+            if (!IsValidHtml(content))
+                return (default, "Invalid HTML!");
+
+            FileModel file = new()
             {
-                try
-                {
-                    using Stream stream = new FileStream(file.Guid + file.FileExtension, FileMode.Create);
-                    file.File.CopyTo(stream);
-                    return (true, default);
-                }
-                catch (Exception ex)
-                {
-                    return (false, ex.ToString());
-                }
-            });
+                Guid = Guid.NewGuid().ToString(),
+                FileExtension = extension,
+                Content = content,
+            };
+
+            return (file, default);
         }
 
-        private static bool IsValidHTML(string html)
+        private bool IsValidHtml(string html)
         {
             try
             {
                 HtmlDocument? doc = new();
                 doc.LoadHtml(html);
                 return !doc.ParseErrors.Any();
             }
-            catch (Exception)
+            catch (Exception ex)
             {
+                logger.LogError($"{DateTime.Now} [{nameof(FileService)}] | {ex}");
                 return false;
             }
-
         }
-
     }
 }

PuppeteerPDFAPI/Services/PuppeteerService.cs

@@ -1,58 +1,94 @@
 using PuppeteerPDFAPI.Misc;
 using PuppeteerSharp;
 using PuppeteerSharp.BrowserData;
-using System.IO;
-using System.Reflection;
 
 namespace PuppeteerPDFAPI.Services
 {
     public interface IPuppeteerService
     {
         Task InitializeAsync();
+        Task<(Stream?, string? ErrorMessage)> ConvertAsync(string html);
     }
 
-    public class PuppeteerService(ILogger<PuppeteerService> logger) : IPuppeteerService
+    public class PuppeteerService(ILogger<PuppeteerService> logger, IHostApplicationLifetime appLifetime) : IPuppeteerService
     {
-        const SupportedBrowser Browser = SupportedBrowser.Chrome;
+        const SupportedBrowser TargetBrowser = SupportedBrowser.Chrome;
+        IBrowser? Browser = default;
 
         public async Task InitializeAsync()
         {
+            appLifetime.ApplicationStopping.Register(async () =>
+            {
+                await Shutdown();
+            });
+
             string browserPath = Helpers.GetBrowserPath();
 
-            BrowserFetcherOptions browserFetcherOptions = new() { Browser = Browser, Path = browserPath };
+            BrowserFetcherOptions browserFetcherOptions = new() { Browser = TargetBrowser, Path = browserPath };
             BrowserFetcher? browserFetcher = new(browserFetcherOptions);
 
             InstalledBrowser? installedBrowser = browserFetcher.GetInstalledBrowsers()
-                      .Where(_ => _.Browser == Browser).MaxBy(_ => _.BuildId);
+                      .Where(_ => _.Browser == TargetBrowser).MaxBy(_ => _.BuildId);
 
             if (installedBrowser == null)
             {
-                logger.LogWarning($"{DateTime.Now} [{nameof(PuppeteerService)}] | {Browser} is not installed! Trying to download latest version...");
+                logger.LogWarning($"{DateTime.Now} [{nameof(PuppeteerService)}] | {TargetBrowser} is not installed! Trying to download latest version...");
 
                 installedBrowser = await browserFetcher.DownloadAsync(BrowserTag.Latest);
 
-                logger.LogInformation($"{DateTime.Now} [{nameof(PuppeteerService)}] | {Browser} finished downloading...");
+                logger.LogInformation($"{DateTime.Now} [{nameof(PuppeteerService)}] | {TargetBrowser} finished downloading...");
             }
 
             try
             {
-                logger.LogInformation($"{DateTime.Now} [{nameof(PuppeteerService)}] | Trying to startup {Browser}...");
-                using IBrowser? browser = await Puppeteer.LaunchAsync(
+                logger.LogInformation($"{DateTime.Now} [{nameof(PuppeteerService)}] | Trying to startup {TargetBrowser}...");
+
+                Browser = await Puppeteer.LaunchAsync(
                     new LaunchOptions()
                     {
-                        ExecutablePath = Helpers.GetBrowserBinPath(Browser.ToString())
+                        ExecutablePath = Helpers.GetBrowserBinPath(TargetBrowser.ToString())
                     });
 
-                await browser.NewPageAsync();
-                await browser.CloseAsync();
-
-                logger.LogInformation($"{DateTime.Now} [{nameof(PuppeteerService)}] | {Browser} startup finished.");
+                logger.LogInformation($"{DateTime.Now} [{nameof(PuppeteerService)}] | {TargetBrowser} startup finished.");
             }
             catch (Exception ex)
             {
-                logger.LogInformation($"{DateTime.Now} [{nameof(PuppeteerService)}] | {Browser} startup failed with error: {ex}.");
+                logger.LogInformation($"{DateTime.Now} [{nameof(PuppeteerService)}] | {TargetBrowser} startup failed with error: {ex}.");
                 throw;
             }
         }
+
+        public async Task<(Stream?, string? ErrorMessage)> ConvertAsync(string html)
+        {
+            if (Browser is null)
+                return (default, "Browser not initialized!");
+
+            try
+            {
+                using IPage page = await Browser.NewPageAsync();
+                await page.SetContentAsync(html);
+
+                Stream pdfStream = await page.PdfStreamAsync();
+
+                await page.CloseAsync();
+
+                return (pdfStream, default);
+            }
+            catch (Exception ex)
+            {
+                logger.LogError($"{DateTime.Now} [{nameof(PuppeteerService)}] | {ex}");
+                return (null, ex.ToString());
+            }
+        }
+
+        public async Task Shutdown()
+        {
+            if (Browser is null)
+                return;
+
+            await Browser.CloseAsync();
+            await Browser.DisposeAsync();
+
+        }
     }
 }