124: Added media parser from markdown files

jarmatys · Dec 22, 2024 · 12472f9 · 12472f9
1 parent 84f1f27
commit 12472f9
Show file tree

Hide file tree

Showing 10 changed files with 446 additions and 30 deletions.
diff --git a/API/ASSISTENTE.Infrastructure.MarkDownParser/Contracts/FilePath.cs b/API/ASSISTENTE.Infrastructure.MarkDownParser/Contracts/FilePath.cs
@@ -7,7 +7,6 @@ public sealed class FilePath : ValueObject
 {
     public string Path { get; }
     public string FileName { get; } 
-
     private string Extension { get; }
 
     private FilePath(string path, string extension, string fileName)

diff --git a/API/ASSISTENTE.Infrastructure.MarkDownParser/Contracts/IMarkDownParser.cs b/API/ASSISTENTE.Infrastructure.MarkDownParser/Contracts/IMarkDownParser.cs
@@ -5,4 +5,5 @@ namespace ASSISTENTE.Infrastructure.MarkDownParser.Contracts;
 public interface IMarkDownParser
 {
     public Result<FileContent> Parse(FilePath filePath);
+    public Result<List<MediaUrl>> GetMediaUrls(FilePath filePath);
 }
diff --git a/API/ASSISTENTE.Infrastructure.MarkDownParser/Contracts/MediaUrl.cs b/API/ASSISTENTE.Infrastructure.MarkDownParser/Contracts/MediaUrl.cs
@@ -0,0 +1,46 @@
+using CSharpFunctionalExtensions;
+using SOFTURE.Results;
+
+namespace ASSISTENTE.Infrastructure.MarkDownParser.Contracts;
+
+public sealed class MediaUrl : ValueObject
+{
+    private static readonly List<string> _supportedExtensions =
+    [
+        "png",
+        "jpg",
+        "mp3"
+    ];
+
+    public string Url { get; }
+    public string Extension { get; }
+
+    private MediaUrl(string url, string extension)
+    {
+        Url = url;
+        Extension = extension;
+    }
+
+    public static Result<MediaUrl> Create(string url)
+    {
+        if (string.IsNullOrEmpty(url))
+            return Result.Failure<MediaUrl>(CommonErrors.EmptyParameter.Build());
+
+        var extension = url.Split('.').Last();
+
+        return _supportedExtensions.Contains(extension)
+            ? new MediaUrl(url, extension)
+            : Result.Failure<MediaUrl>(MediaUrlErrors.InvalidFileExtension.Build());
+    }
+
+    protected override IEnumerable<IComparable> GetEqualityComponents()
+    {
+        yield return Url;
+    }
+}
+
+public static class MediaUrlErrors
+{
+    public static readonly Error InvalidFileExtension = new(
+        "MediaUrl.InvalidFileExtension", "Invalid file extension - only '.png, .jpg, .mp3' file types are allowed");
+}
diff --git a/API/ASSISTENTE.Infrastructure.MarkDownParser/Errors/MarkDownParserErrors.cs b/API/ASSISTENTE.Infrastructure.MarkDownParser/Errors/MarkDownParserErrors.cs
@@ -12,4 +12,7 @@ internal static class MarkDownParserErrors
 
     public static readonly Error OnlyHeadersNotAllowed = new(
         "MarkDownParser.OnlyHeadersNotAllowed", "Notes with only headers are not allowed");
+
+    public static readonly Error UnsupportedMedia = new(
+        "MarkDownParser.UnsupportedMedia", "Media type is not supported");
 }
diff --git a/API/ASSISTENTE.Infrastructure.MarkDownParser/Extensions/MarkDownExtensions.cs b/API/ASSISTENTE.Infrastructure.MarkDownParser/Extensions/MarkDownExtensions.cs
@@ -27,17 +27,18 @@ public static string GetContent(this LeafBlock leaf)
         return string.Join(" ", contentsList);
     }
 
-    public static string GetUrls(this LeafBlock leaf)
+    public static List<string> GetUrls(this LeafBlock leaf)
     {
-        if (leaf.Inline == null) return string.Empty;
+        if (leaf.Inline == null) return [];
 
         var urls = leaf.Inline
             .Descendants()
             .OfType<LinkInline>()
             .Select(x => x.Url)
+            .Where(url => url != string.Empty)
             .ToList();
 
-        return string.Join(" ", urls);
+        return urls!;
     }
 
     public static string GetCodeLine(this LeafBlock leaf)
@@ -91,8 +92,8 @@ public static string GetListContent(this ListBlock block)
 
                 var line = $"{counter}. {content}";
 
-                if (!string.IsNullOrEmpty(urls))
-                    line += $" | Links associated with this point: {urls}";
+                if (urls.Count != 0)
+                    line += $" | Links associated with this point: {string.Join(" ", urls)}";
 
                 listContent.Add(line);
             }

diff --git a/API/ASSISTENTE.Infrastructure.MarkDownParser/MarkDownParser.cs b/API/ASSISTENTE.Infrastructure.MarkDownParser/MarkDownParser.cs
@@ -46,6 +46,30 @@ public Result<FileContent> Parse(FilePath filePath)
         return FileContent.Create(filePath.FileName, elements);
     }
 
+    public Result<List<MediaUrl>> GetMediaUrls(FilePath filePath)
+    {
+        var content = File.ReadAllText(filePath.Path);
+        var parsedMarkdown = Markdown.Parse(content);
+
+        var mediaUrls = parsedMarkdown
+            .Descendants<ParagraphBlock>()
+            .Select(paragraphBlock => paragraphBlock.GetUrls())
+            .Where(mediaUrl => mediaUrl.Count != 0)
+            .SelectMany(mediaUrl => mediaUrl)
+            .Select(MediaUrl.Create)
+            .ToList();
+
+        if (mediaUrls.Count == 0)
+            return Result.Failure<List<MediaUrl>>(MarkDownParserErrors.EmptyContent.Build());
+
+        if (mediaUrls.Any(x => x.IsFailure))
+            return Result.Failure<List<MediaUrl>>(MarkDownParserErrors.UnsupportedMedia.Build());
+
+        return mediaUrls
+            .Select(x => x.Value)
+            .ToList();
+    }
+
     private static Result<ElementBase> GetElement(IMarkdownObject block)
     {
         switch (block)
@@ -88,7 +112,7 @@ private static Result<ElementBase> GetElement(IMarkdownObject block)
             {
                 var paragraphContent = paragraph.GetContent();
                 var paragraphUrls = paragraph.GetUrls();
-                
+
                 var paragraphElement = new Paragraph(paragraphContent, paragraphUrls) as ElementBase;
 
                 return Result.Success(paragraphElement);

diff --git a/API/ASSISTENTE.Infrastructure.MarkDownParser/Models/Paragraph.cs b/API/ASSISTENTE.Infrastructure.MarkDownParser/Models/Paragraph.cs
@@ -2,16 +2,16 @@
 
 namespace ASSISTENTE.Infrastructure.MarkDownParser.Models;
 
-internal sealed record Paragraph(string Content, string Urls) : ElementBase(Content)
+internal sealed record Paragraph(string Content, List<string> Urls) : ElementBase(Content)
 {
-    public string Urls { get; } = Urls;
+    public List<string> Urls { get; } = Urls;
 
     public override string GetContent()
     {
         var content = $"Paragraph content: {Content}";
 
-        if (!string.IsNullOrEmpty(Urls))
-            content += $" | Urls connected with this paragraph: {Urls}";
+        if (Urls.Count != 0)
+            content += $" | Urls connected with this paragraph: {string.Join(" ", Urls)}";
 
         return content;
     }