Updated toc parsing

Sources/ToucanSDK/Extensions/TocElement+Tree.swift

@@ -0,0 +1,53 @@
+//
+//  File.swift
+//  toucan
+//
+//  Created by Viasz-Kádi Ferenc on 2024. 10. 14..
+//
+
+import Foundation
+
+extension [TocElement] {
+
+    /// Builds a tree of Table of Content (ToC) nodes based on the hierarchy levels of the elements.
+    ///
+    /// - Returns: An array of `ToCNode` objects representing the hierarchical structure of the ToC.
+    func buildToCTree() -> [ToCNode] {
+        var result: [ToCNode] = []
+        var stack: [ToCNode] = []
+
+        for element in self {
+            let newNode = ToCNode(
+                level: element.level,
+                text: element.text,
+                fragment: element.fragment
+            )
+
+            // Find the correct parent for the current node
+            while let last = stack.last, last.level >= element.level {
+                stack.removeLast()
+            }
+
+            if let parent = stack.last {
+                // Append new node as a child of the last node in the stack
+                var updatedParent = parent
+                updatedParent.children.append(newNode)
+                stack[stack.count - 1] = updatedParent
+                if let index = result.firstIndex(where: {
+                    $0.fragment == parent.fragment && $0.level == parent.level
+                }) {
+                    result[index] = updatedParent
+                }
+            }
+            else {
+                // Add the new node to the result if it has no parent
+                result.append(newNode)
+            }
+
+            // Add the new node to the stack
+            stack.append(newNode)
+        }
+
+        return result
+    }
+}

Sources/ToucanSDK/Markdown/MarkdownRenderer.swift

@@ -51,58 +51,4 @@ public struct MarkdownRenderer {
         var htmlVisitor = MarkupToHTMLVisitor(delegate: delegate)
         return htmlVisitor.visitDocument(document)
     }
-
-    /// Render a Table of Contents
-    public func renderToC(
-        markdown: String
-    ) -> [ToC] {
-        let document = Document(
-            parsing: markdown
-        )
-        var headingsVisitor = MarkupToHXVisitor()
-        return Self.buildToC(headingsVisitor.visitDocument(document))
-    }
-
-    // MARK: - private
-
-    static func buildToC(
-        _ headings: [MarkupToHXVisitor.HX]
-    ) -> [ToC] {
-        var result: [ToC] = []
-        var stack: [ToC] = []
-
-        for heading in headings {
-            let newNode = ToC(
-                level: heading.level,
-                text: heading.text,
-                fragment: heading.fragment
-            )
-
-            // Find the correct parent for the current node
-            while let last = stack.last, last.level >= heading.level {
-                stack.removeLast()
-            }
-
-            if let parent = stack.last {
-                // Append new node as a child of the last node in the stack
-                var updatedParent = parent
-                updatedParent.children.append(newNode)
-                stack[stack.count - 1] = updatedParent
-                if let index = result.firstIndex(where: {
-                    $0.fragment == parent.fragment && $0.level == parent.level
-                }) {
-                    result[index] = updatedParent
-                }
-            }
-            else {
-                // Add the new node to the result if it has no parent
-                result.append(newNode)
-            }
-
-            // Add the new node to the stack
-            stack.append(newNode)
-        }
-
-        return result
-    }
 }

Sources/ToucanSDK/Markdown/MarkupHeadingVisitor.swift

@@ -0,0 +1,36 @@
+//
+//  File.swift
+//
+//
+//  Created by Tibor Bodecs on 03/05/2024.
+//
+
+import Markdown
+
+struct MarkupHeadingVisitor: MarkupVisitor {
+
+    typealias Result = [TocElement]
+
+    let levels: [Int]
+
+    init(levels: [Int] = [2, 3]) {
+        self.levels = levels
+    }
+
+    // MARK: - visitor functions
+
+    mutating func defaultVisit(_ markup: any Markup) -> Result {
+        markup.children.flatMap { visit($0) }
+    }
+
+    // MARK: - elements
+
+    mutating func visitHeading(_ heading: Heading) -> Result {
+        guard levels.contains(heading.level) else {
+            return []
+        }
+        return [
+            TocElement(heading)
+        ]
+    }
+}

Sources/ToucanSDK/Parsers/ToC/HTMLToCParser.swift

@@ -0,0 +1,59 @@
+//
+//  File.swift
+//  toucan
+//
+//  Created by Viasz-Kádi Ferenc on 2024. 10. 14..
+//
+
+import Foundation
+import Logging
+import SwiftSoup
+
+/// A struct that parses HTML content to extract table of contents (ToC) elements from heading tags.
+/// This struct conforms to the `ToCElementParser` protocol.
+struct HTMLToCParser: ToCElementParser {
+
+    /// Logger instance used to log errors encountered during parsing.
+    let logger: Logger
+
+    /// Parses the provided HTML string and extracts table of contents elements from `<h2>` and `<h3>` tags.
+    ///
+    /// - Parameter value: The HTML string to parse.
+    /// - Returns: An array of `TocElement` objects or `nil` if an error occurs during parsing.
+    func parse(from value: String) -> [TocElement]? {
+        do {
+            let document: SwiftSoup.Document = try SwiftSoup.parse(value)
+            let headings = try document.select("h2, h3")
+            return headings.compactMap { TocElement($0) }
+        }
+        catch Exception.Error(_, let message) {
+            logger.error("\(message)")
+            return nil
+        }
+        catch {
+            logger.error("\(error.localizedDescription)")
+            return nil
+        }
+    }
+}
+
+extension TocElement {
+
+    /// Initializes a new instance with the provided SwiftSoup.Element.
+    /// Attempts to extract text, level, and fragment from the element.
+    /// If the fragment attribute is empty or an error occurs during extraction, initialization fails.
+    init?(_ element: SwiftSoup.Element) {
+        do {
+            text = try element.text()
+            level = element.nodeName().hasSuffix("2") ? 2 : 3
+            fragment = try element.attr("id")
+
+            guard !fragment.isEmpty else {
+                return nil
+            }
+        }
+        catch {
+            return nil
+        }
+    }
+}

Sources/ToucanSDK/Parsers/ToC/MarkdownToCParser.swift

@@ -0,0 +1,35 @@
+//
+//  File.swift
+//  toucan
+//
+//  Created by Viasz-Kádi Ferenc on 2024. 10. 14..
+//
+
+import Foundation
+import Markdown
+
+/// A parser that extracts table of contents elements from a Markdown document.
+struct MarkdownToCParser: ToCElementParser {
+
+    /// Parses a string containing Markdown content and returns an array of `TocElement` objects.
+    ///
+    /// - Parameter value: A string containing Markdown content.
+    /// - Returns: An array of `TocElement` objects if headings are found, otherwise `nil`.
+    func parse(from value: String) -> [TocElement]? {
+        let document = Markdown.Document(parsing: value)
+        var headingsVisitor = MarkupHeadingVisitor()
+        return headingsVisitor.visitDocument(document)
+    }
+}
+
+extension TocElement {
+
+    /// Initializes a `TocElement` from a `Markdown.Heading`.
+    ///
+    /// - Parameter element: A `Markdown.Heading` from which to initialize the `TocElement`.
+    init(_ element: Markdown.Heading) {
+        level = element.level
+        text = element.plainText
+        fragment = text.lowercased().slugify()
+    }
+}

Sources/ToucanSDK/Parsers/ToC/ToCParser.swift

@@ -0,0 +1,18 @@
+//
+//  File.swift
+//  toucan
+//
+//  Created by Viasz-Kádi Ferenc on 2024. 10. 11..
+//
+
+import Foundation
+
+/// A protocol that defines a method for parsing table of contents elements from a string value.
+protocol ToCElementParser {
+
+    /// Parses the given string value into an array of `TocElement` objects.
+    ///
+    /// - Parameter value: The string representation of the table of contents elements.
+    /// - Returns: An array of `TocElement` objects if parsing is successful, or `nil` if parsing fails.
+    func parse(from value: String) -> [TocElement]?
+}

Sources/ToucanSDK/Site/SiteRenderer.swift

@@ -62,6 +62,10 @@ struct SiteRenderer {
     let destinationUrl: URL
 
     let fileManager: FileManager = .default
+
+    let htmlToCParser: HTMLToCParser
+    let markdownToCParser: MarkdownToCParser
+
     let logger: Logger
 
     let templateRenderer: MustacheToHTMLRenderer
@@ -79,6 +83,8 @@ struct SiteRenderer {
         self.templatesUrl = templatesUrl
         self.overridesUrl = overridesUrl
         self.destinationUrl = destinationUrl
+        self.htmlToCParser = .init(logger: logger)
+        self.markdownToCParser = .init()
         self.logger = logger
 
         let calendar = Calendar(identifier: .gregorian)
@@ -314,7 +320,7 @@ struct SiteRenderer {
         //        let run = transformers.array("run", as: [String: Any].self)
 
         let markdown = pageBundle.markdown.dropFrontMatter()
-        var toc: [ToC]? = nil
+        var toc: [ToCNode]? = nil
         var time: Int? = nil
         var contents = ""
 
@@ -360,38 +366,7 @@ struct SiteRenderer {
             try? fileManager.delete(at: fileURL)
 
             time = readingTime(contents)
-
-            do {
-                let doc: Document = try SwiftSoup.parse(contents)
-
-                var tocList: [MarkupToHXVisitor.HX] = []
-                let headings = try doc.select("h2, h3")
-                for h in headings {
-                    let n = h.nodeName()
-                    let attr = try h.attr("id")
-                    guard !attr.isEmpty else { continue }
-                    let val = try h.text()
-
-                    let level = n.hasSuffix("2") ? 2 : 3
-
-                    tocList.append(
-                        .init(
-                            level: level,
-                            text: val,
-                            fragment: attr
-                        )
-                    )
-                }
-
-                toc = MarkdownRenderer.buildToC(tocList)
-
-            }
-            catch Exception.Error(_, let message) {
-                logger.error("\(message)")
-            }
-            catch {
-                logger.error("\(error.localizedDescription)")
-            }
+            toc = htmlToCParser.parse(from: contents)?.buildToCTree()
         }
 
         if renderFallback {
@@ -400,7 +375,8 @@ struct SiteRenderer {
 
         var context: [String: Any] = [:]
         context["readingTime"] = time ?? readingTime(markdown)
-        context["toc"] = toc ?? renderer.renderToC(markdown: markdown)
+        context["toc"] =
+            toc ?? markdownToCParser.parse(from: markdown)?.buildToCTree()
         context["contents"] = contents
 
         return context