Split MarkdownLanguage types

Sources/SwiftParser/Languages/MarkdownLanguage+AutoLinkBuilder.swift

@@ -0,0 +1,24 @@
+import Foundation
+
+extension MarkdownLanguage {
+    public class AutoLinkBuilder: CodeElementBuilder {
+        public init() {}
+        public func accept(context: CodeContext, token: any CodeToken) -> Bool {
+            guard let tok = token as? Token else { return false }
+            if case .lessThan = tok { return true }
+            return false
+        }
+        public func build(context: inout CodeContext) {
+            context.index += 1
+            var text = ""
+            while context.index < context.tokens.count {
+                if let tok = context.tokens[context.index] as? Token {
+                    if case .greaterThan = tok { context.index += 1; break }
+                    else { text += tok.text; context.index += 1 }
+                } else { context.index += 1 }
+            }
+            context.currentNode.addChild(MarkdownAutoLinkNode(url: text))
+        }
+    }
+
+}

Sources/SwiftParser/Languages/MarkdownLanguage+BareAutoLinkBuilder.swift

@@ -0,0 +1,42 @@
+import Foundation
+
+extension MarkdownLanguage {
+    public class BareAutoLinkBuilder: CodeElementBuilder {
+        private static let regex: NSRegularExpression = {
+            let pattern = #"^((https?|ftp)://[^\s<>]+|www\.[^\s<>]+|[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Za-z]{2,})"#
+            return try! NSRegularExpression(pattern: pattern, options: [])
+        }()
+
+        public init() {}
+
+        public func accept(context: CodeContext, token: any CodeToken) -> Bool {
+            guard let tok = token as? Token else { return false }
+            let start = tok.range.lowerBound
+            let text = String(context.input[start...])
+            let range = NSRange(location: 0, length: text.utf16.count)
+            if let m = Self.regex.firstMatch(in: text, range: range), m.range.location == 0 {
+                return true
+            }
+            return false
+        }
+
+        public func build(context: inout CodeContext) {
+            guard let tok = context.tokens[context.index] as? Token else { return }
+            let start = tok.range.lowerBound
+            let text = String(context.input[start...])
+            let range = NSRange(location: 0, length: text.utf16.count)
+            guard let m = Self.regex.firstMatch(in: text, range: range) else { return }
+            let endPos = context.input.index(start, offsetBy: m.range.length)
+            let url = String(context.input[start..<endPos])
+            context.currentNode.addChild(MarkdownAutoLinkNode(url: url))
+            while context.index < context.tokens.count {
+                if let t = context.tokens[context.index] as? Token, t.range.upperBound <= endPos {
+                    context.index += 1
+                } else {
+                    break
+                }
+            }
+        }
+    }
+
+}

Sources/SwiftParser/Languages/MarkdownLanguage+BlockQuoteBuilder.swift

@@ -0,0 +1,39 @@
+import Foundation
+
+extension MarkdownLanguage {
+    public class BlockQuoteBuilder: CodeElementBuilder {
+        public init() {}
+        public func accept(context: CodeContext, token: any CodeToken) -> Bool {
+            guard let tok = token as? Token else { return false }
+            if case .greaterThan = tok {
+                if context.index == 0 { return true }
+                if let prev = context.tokens[context.index - 1] as? Token, case .newline = prev { return true }
+            }
+            return false
+        }
+        public func build(context: inout CodeContext) {
+            context.index += 1 // skip '>'
+            var text = ""
+            while context.index < context.tokens.count {
+                if let tok = context.tokens[context.index] as? Token {
+                    switch tok {
+                    case .newline:
+                        context.index += 1
+                        let node = MarkdownBlockQuoteNode(value: text.trimmingCharacters(in: .whitespaces))
+                        context.currentNode.addChild(node)
+                        return
+                    case .eof:
+                        let node = MarkdownBlockQuoteNode(value: text.trimmingCharacters(in: .whitespaces))
+                        context.currentNode.addChild(node)
+                        context.index += 1
+                        return
+                    default:
+                        text += tok.text
+                        context.index += 1
+                    }
+                } else { context.index += 1 }
+            }
+        }
+    }
+
+}

Sources/SwiftParser/Languages/MarkdownLanguage+CodeBlockBuilder.swift

@@ -0,0 +1,78 @@
+import Foundation
+
+extension MarkdownLanguage {
+    public class CodeBlockBuilder: CodeElementBuilder {
+        public init() {}
+        public func accept(context: CodeContext, token: any CodeToken) -> Bool {
+            guard let first = token as? Token else { return false }
+            let fenceKind: String
+            switch first {
+            case .backtick: fenceKind = "`"
+            case .tilde: fenceKind = "~"
+            default: return false
+            }
+            var idx = context.index
+            var count = 0
+            while idx < context.tokens.count, let t = context.tokens[idx] as? Token, t.kindDescription == fenceKind {
+                count += 1; idx += 1
+            }
+            guard count >= 3 else { return false }
+            if context.index == 0 { return true }
+            if let prev = context.tokens[context.index - 1] as? Token, case .newline = prev {
+                return true
+            }
+            return false
+        }
+        public func build(context: inout CodeContext) {
+            guard let startTok = context.tokens[context.index] as? Token else { return }
+            let fenceKind = startTok.kindDescription
+            var fenceLength = 0
+            while context.index < context.tokens.count, let t = context.tokens[context.index] as? Token, t.kindDescription == fenceKind {
+                fenceLength += 1
+                context.index += 1
+            }
+            // capture info string until end of line and trim whitespace
+            var info = ""
+            while context.index < context.tokens.count {
+                if let tok = context.tokens[context.index] as? Token {
+                    if case .newline = tok {
+                        context.index += 1
+                        break
+                    } else {
+                        info += tok.text
+                        context.index += 1
+                    }
+                } else {
+                    context.index += 1
+                }
+            }
+            info = info.trimmingCharacters(in: .whitespaces)
+            let lang = info.split(whereSeparator: { $0.isWhitespace }).first.map(String.init)
+
+            let blockStart = context.index
+            var text = ""
+            while context.index < context.tokens.count {
+                if let tok = context.tokens[context.index] as? Token {
+                    // check for closing fence at start of line
+                    if tok.kindDescription == fenceKind && (context.index == blockStart || (context.index > blockStart && (context.tokens[context.index - 1] as? Token)?.kindDescription == "newline")) {
+                        var idx = context.index
+                        var count = 0
+                        while idx < context.tokens.count, let t = context.tokens[idx] as? Token, t.kindDescription == fenceKind {
+                            count += 1; idx += 1
+                        }
+                        if count >= fenceLength {
+                            context.index = idx
+                            if context.index < context.tokens.count, let nl = context.tokens[context.index] as? Token, case .newline = nl { context.index += 1 }
+                            context.currentNode.addChild(MarkdownCodeBlockNode(lang: lang, content: text))
+                            return
+                        }
+                    }
+                    text += tok.text
+                    context.index += 1
+                } else { context.index += 1 }
+            }
+            context.currentNode.addChild(MarkdownCodeBlockNode(lang: lang, content: text))
+        }
+    }
+
+}

Sources/SwiftParser/Languages/MarkdownLanguage+Element.swift

@@ -0,0 +1,34 @@
+import Foundation
+
+extension MarkdownLanguage {
+    public enum Element: String, CodeElement {
+        case root
+        case paragraph
+        case heading
+        case text
+        case listItem
+        case orderedListItem
+        case unorderedList
+        case orderedList
+        case emphasis
+        case strong
+        case codeBlock
+        case inlineCode
+        case link
+        case blockQuote
+        case thematicBreak
+        case image
+        case html
+        case entity
+        case strikethrough
+        case table
+        case tableHeader
+        case tableRow
+        case tableCell
+        case autoLink
+        case linkReferenceDefinition
+        case footnoteDefinition
+        case footnoteReference
+    }
+
+}

Sources/SwiftParser/Languages/MarkdownLanguage+EmphasisBuilder.swift

@@ -0,0 +1,29 @@
+import Foundation
+
+extension MarkdownLanguage {
+    public class EmphasisBuilder: CodeElementBuilder {
+        public init() {}
+        public func accept(context: CodeContext, token: any CodeToken) -> Bool {
+            guard let tok = token as? Token else { return false }
+            if case .star = tok { return true }
+            if case .underscore = tok { return true }
+            return false
+        }
+        public func build(context: inout CodeContext) {
+            let snap = context.snapshot()
+            guard let open = context.tokens[context.index] as? Token else { return }
+            context.index += 1
+            let (children, ok) = MarkdownLanguage.parseInline(context: &context, closing: open, count: 1)
+            if ok {
+                let node = MarkdownEmphasisNode(value: "")
+                children.forEach { node.addChild($0) }
+                context.currentNode.addChild(node)
+            } else {
+                context.restore(snap)
+                context.currentNode.addChild(MarkdownTextNode(value: open.text))
+                context.index += 1
+            }
+        }
+    }
+
+}

Sources/SwiftParser/Languages/MarkdownLanguage+EntityBuilder.swift

@@ -0,0 +1,48 @@
+import Foundation
+
+extension MarkdownLanguage {
+    public class EntityBuilder: CodeElementBuilder {
+        public init() {}
+        public func accept(context: CodeContext, token: any CodeToken) -> Bool {
+            guard let tok = token as? Token else { return false }
+            if case .ampersand = tok { return true }
+            return false
+        }
+        public func build(context: inout CodeContext) {
+            context.index += 1
+            var text = ""
+            while context.index < context.tokens.count {
+                if let tok = context.tokens[context.index] as? Token {
+                    if case .semicolon = tok { context.index += 1; break }
+                    else { text += tok.text; context.index += 1 }
+                } else { context.index += 1 }
+            }
+            let decoded = decode(text)
+            context.currentNode.addChild(MarkdownEntityNode(value: decoded))
+        }
+
+        private func decode(_ entity: String) -> String {
+            switch entity {
+            case "amp": return "&"
+            case "lt": return "<"
+            case "gt": return ">"
+            case "quot": return "\""
+            case "apos": return "'"
+            default:
+                if entity.hasPrefix("#x") || entity.hasPrefix("#X") {
+                    let hex = entity.dropFirst(2)
+                    if let value = UInt32(hex, radix: 16), let scalar = UnicodeScalar(value) {
+                        return String(Character(scalar))
+                    }
+                } else if entity.hasPrefix("#") {
+                    let num = entity.dropFirst()
+                    if let value = UInt32(num), let scalar = UnicodeScalar(value) {
+                        return String(Character(scalar))
+                    }
+                }
+                return "&" + entity + ";"
+            }
+        }
+    }
+
+}

Sources/SwiftParser/Languages/MarkdownLanguage+FootnoteBuilder.swift

@@ -0,0 +1,59 @@
+import Foundation
+
+extension MarkdownLanguage {
+    public class FootnoteBuilder: CodeElementBuilder {
+        public init() {}
+        public func accept(context: CodeContext, token: any CodeToken) -> Bool {
+            guard let lb = token as? Token, case .lbracket = lb else { return false }
+            guard context.index + 2 < context.tokens.count else { return false }
+            guard let first = context.tokens[context.index + 1] as? Token else { return false }
+            if case .text(let s, _) = first, s.starts(with: "^") {
+                var idx = context.index + 2
+                while idx < context.tokens.count {
+                    if let t = context.tokens[idx] as? Token {
+                        if case .rbracket = t { return true }
+                        if case .text = t {
+                            idx += 1; continue
+                        }
+                        if case .number = t {
+                            idx += 1; continue
+                        }
+                    }
+                    break
+                }
+            }
+            return false
+        }
+        public func build(context: inout CodeContext) {
+            context.index += 1 // skip [
+            var id = ""
+            while context.index < context.tokens.count {
+                guard let tok = context.tokens[context.index] as? Token else { context.index += 1; continue }
+                if case .rbracket = tok { break }
+                id += tok.text
+                context.index += 1
+            }
+            if id.hasPrefix("^") { id.removeFirst() }
+            if context.index < context.tokens.count { context.index += 1 } // skip ]
+
+            if context.index < context.tokens.count,
+               let colon = context.tokens[context.index] as? Token,
+               case .text(let s, _) = colon,
+               s.trimmingCharacters(in: .whitespaces).hasPrefix(":") {
+                var text = s
+                context.index += 1
+                while context.index < context.tokens.count {
+                    if let tok = context.tokens[context.index] as? Token {
+                        if case .newline = tok { context.index += 1; break }
+                        else { text += tok.text; context.index += 1 }
+                    } else { context.index += 1 }
+                }
+                if text.hasPrefix(":") { text.removeFirst() }
+                let trimmed = text.trimmingCharacters(in: .whitespaces)
+                context.currentNode.addChild(MarkdownFootnoteDefinitionNode(identifier: id, text: trimmed))
+            } else {
+                context.currentNode.addChild(MarkdownFootnoteReferenceNode(identifier: id))
+            }
+        }
+    }
+}

Sources/SwiftParser/Languages/MarkdownLanguage+HTMLBlockBuilder.swift

@@ -0,0 +1,25 @@
+import Foundation
+
+extension MarkdownLanguage {
+    public class HTMLBlockBuilder: CodeElementBuilder {
+        public init() {}
+        public func accept(context: CodeContext, token: any CodeToken) -> Bool {
+            guard let tok = token as? Token else { return false }
+            if case .lessThan = tok, context.index == 0 {
+                let rest = String(context.input[tok.range.upperBound...]).lowercased()
+                return rest.hasPrefix("!doctype") || rest.hasPrefix("html")
+            }
+            return false
+        }
+        public func build(context: inout CodeContext) {
+            var text = ""
+            while context.index < context.tokens.count {
+                if let tok = context.tokens[context.index] as? Token { text += tok.text }
+                context.index += 1
+            }
+            let closed = MarkdownLanguage.isHTMLClosed(text)
+            context.currentNode.addChild(MarkdownHtmlNode(value: text, closed: closed))
+        }
+    }
+
+}