Fixed issue with media parsing

* Added missing folder string in GetExportFiles method * Modified logic in GetMessages method to process multiple links instead of one per message
marcin-przywoski · Feb 20, 2024 · 11ec95a · 11ec95a
1 parent 4c4e699
commit 11ec95a
Show file tree

Hide file tree

Showing 2 changed files with 57 additions and 25 deletions.
diff --git a/ExportViewer.Core/Services/DataParsingService.cs b/ExportViewer.Core/Services/DataParsingService.cs
@@ -71,7 +71,7 @@ public Task<IEnumerable<string>> GetExportFiles (string exportLocation , ExportT
                     Console.WriteLine($"Directory {subDirectoryLocation} does not exist.");
                 }
             }
-            foreach (var subDirectory in new[] { "archived_threads/" , "filtered_threads/" , "inbox/" })
+            foreach (var subDirectory in new[] { "archived_threads/" , "filtered_threads/" , "inbox/", "e2ee_cutover/" })
             {
                 var subDirectoryLocation = Path.Combine(exportLocation , "your_activity_across_facebook" , "messages" , subDirectory);
                 var fileExtension = fileExtensions[type];

diff --git a/ExportViewer.Core/Services/HtmlParsingService.cs b/ExportViewer.Core/Services/HtmlParsingService.cs
@@ -104,39 +104,71 @@ public async Task<IEnumerable<Message>> GetMessages (string filePath , CultureIn
                 Parallel.ForEach(divs , node =>
                 {
 
-                    var divImage = node.QuerySelector("img._a6_o._3-96");
-                    var divVideo = node.QuerySelector("video._a6_o._3-96");
+                    var divImages = node.QuerySelectorAll("img._a6_o._3-96");
+                    var divVideos = node.QuerySelectorAll("video._a6_o._3-96");
                     var divDate = node.QuerySelector("div._3-94._a6-o")?.QuerySelector("div._a72d");
 
-                    if (((divImage != null && divDate != null) || (divVideo != null && divDate != null)) && !string.IsNullOrEmpty(divDate.TextContent))
+                    if (((divImages != null && divDate != null) || (divVideos != null && divDate != null)) && !string.IsNullOrEmpty(divDate.TextContent))
                     {
-                        string href = divImage != null ? divImage.GetAttribute("src") : divVideo.GetAttribute("src");
-                        if ((!href.StartsWith("http") || !href.StartsWith("https")) && (href.EndsWith(".jpg") || href.EndsWith(".png") || href.EndsWith(".gif") || href.EndsWith(".mp4")))
+                        Parallel.ForEach(divImages , divImage =>
                         {
-                            if (Thread.CurrentThread.CurrentCulture.IsReadOnly || locale.IsReadOnly)
-                            {
-                                var clone = Thread.CurrentThread.CurrentCulture.Clone() as CultureInfo;
-                                clone.DateTimeFormat.PMDesignator = "po południu";
-                                clone.DateTimeFormat.AMDesignator = "rano";
-                                Thread.CurrentThread.CurrentCulture = clone;
-                                Thread.CurrentThread.CurrentUICulture = clone;
-                                locale = clone;
-                            }
-                            else
+                            string href = divImage.GetAttribute("src");
+                            if ((!href.StartsWith("http") || !href.StartsWith("https")) && (href.EndsWith(".jpg") || href.EndsWith(".png") || href.EndsWith(".gif")))
                             {
-                                locale.DateTimeFormat.PMDesignator = "po południu";
-                                locale.DateTimeFormat.AMDesignator = "rano";
+                                if (Thread.CurrentThread.CurrentCulture.IsReadOnly || locale.IsReadOnly)
+                                {
+                                    var clone = Thread.CurrentThread.CurrentCulture.Clone() as CultureInfo;
+                                    clone.DateTimeFormat.PMDesignator = "po południu";
+                                    clone.DateTimeFormat.AMDesignator = "rano";
+                                    Thread.CurrentThread.CurrentCulture = clone;
+                                    Thread.CurrentThread.CurrentUICulture = clone;
+                                    locale = clone;
+                                }
+                                else
+                                {
+                                    locale.DateTimeFormat.PMDesignator = "po południu";
+                                    locale.DateTimeFormat.AMDesignator = "rano";
+                                }
+
+
+                                DateTime parsedDate = DateTime.ParseExact(divDate.TextContent , "MMM dd, yyyy h:mm:sstt" , locale);
+
+                                if (File.Exists(exportLocation + href))
+                                {
+                                    messages.Add(new Message { Link = href , Date = parsedDate });
+                                }
                             }
+                        });
 
-
-
-                            DateTime parsedDate = DateTime.ParseExact(divDate.TextContent , "MMM dd, yyyy h:mm:sstt" , locale);
-
-                            if (File.Exists(exportLocation + href))
+                        Parallel.ForEach(divVideos , divVideo =>
+                        {
+                            string href = divVideo.GetAttribute("src");
+                            if ((!href.StartsWith("http") || !href.StartsWith("https")) && href.EndsWith(".mp4"))
                             {
-                                messages.Add(new Message { Link = href , Date = parsedDate });
+                                if (Thread.CurrentThread.CurrentCulture.IsReadOnly || locale.IsReadOnly)
+                                {
+                                    var clone = Thread.CurrentThread.CurrentCulture.Clone() as CultureInfo;
+                                    clone.DateTimeFormat.PMDesignator = "po południu";
+                                    clone.DateTimeFormat.AMDesignator = "rano";
+                                    Thread.CurrentThread.CurrentCulture = clone;
+                                    Thread.CurrentThread.CurrentUICulture = clone;
+                                    locale = clone;
+                                }
+                                else
+                                {
+                                    locale.DateTimeFormat.PMDesignator = "po południu";
+                                    locale.DateTimeFormat.AMDesignator = "rano";
+                                }
+
+
+                                DateTime parsedDate = DateTime.ParseExact(divDate.TextContent , "MMM dd, yyyy h:mm:sstt" , locale);
+
+                                if (File.Exists(exportLocation + href))
+                                {
+                                    messages.Add(new Message { Link = href , Date = parsedDate });
+                                }
                             }
-                        }
+                        });
                     }
 
                 });