Add NorgLexerHEAD1.1.0main

1 files changed, 187 insertions, 0 deletions

diff --git a/Sources/NorgKit/Parsers/InlineScanner.swift b/Sources/NorgKit/Parsers/InlineScanner.swift
new file mode 100644
index 0000000..881a069
--- /dev/null
+++ b/Sources/NorgKit/Parsers/InlineScanner.swift
@@ -0,0 +1,187 @@
+/// Scans for inline norg constructs.
+enum InlineScanner {
+
+  struct Bracketed: Equatable {
+    let open: Int
+    let body: Range<Int>
+    let close: Int
+  }
+
+  enum AnchorSuffix: Equatable {
+    case target(Bracketed)
+    case description(Bracketed)
+  }
+
+  enum InlineObject: Equatable {
+    /// `\x` — `escaped` is the index of the literal scalar, or `nil` for a
+    /// trailing backslash at the end of the range.
+    case escape(escaped: Int?, end: Int)
+    /// `%…%`, dropped from rendered output.
+    case comment(open: Int, body: Range<Int>, close: Int, end: Int)
+    /// Verbatim / math: `` `…` `` or `$…$`, inner content taken literally.
+    case verbatim(style: InlineStyle, open: Int, body: Range<Int>, close: Int, end: Int)
+    /// An attached modifier (`*…*`, `/…/`, …) whose body is parsed recursively.
+    case modifier(style: InlineStyle, open: Int, body: Range<Int>, close: Int, end: Int)
+    /// `{target}` optionally followed by `[description]`.
+    case link(open: Int, target: Bracketed, description: Bracketed?, end: Int)
+    /// `[name]` optionally followed by `{target}` or `[description]`.
+    case anchor(open: Int, name: Bracketed, suffix: AnchorSuffix?, end: Int)
+
+    var end: Int {
+      switch self {
+      case .escape(_, let end), .comment(_, _, _, let end), .verbatim(_, _, _, _, let end),
+        .modifier(_, _, _, _, let end), .link(_, _, _, let end), .anchor(_, _, _, let end):
+        return end
+      }
+    }
+  }
+
+  static let modifiers: [Unicode.Scalar: InlineStyle] = [
+    "*": .bold,
+    "/": .italic,
+    "_": .underline,
+    "-": .strikethrough,
+    "^": .superscript,
+    ",": .subscript,
+    "!": .spoiler,
+  ]
+
+  static let literalModifiers: [Unicode.Scalar: InlineStyle] = [
+    "`": .verbatim,
+    "$": .math,
+  ]
+
+  static let significant = ASCIIByteSet("*/_-^,!\u{60}$%{[\\")
+
+  static func object(in chars: [Unicode.Scalar], at i: Int, to hi: Int) -> InlineObject? {
+    let c = chars[i]
+    let prev: Unicode.Scalar? = i > 0 ? chars[i - 1] : nil
+    let next: Unicode.Scalar? = i + 1 < chars.count ? chars[i + 1] : nil
+
+    // Escapes: the next scalar is taken literally.
+    if c == "\\" {
+      return i + 1 < hi ? .escape(escaped: i + 1, end: i + 2) : .escape(escaped: nil, end: i + 1)
+    }
+
+    // Comments are dropped from the rendered output.
+    if c == "%", isOpener(prev: prev, next: next),
+      let close = literalClose(chars, from: i + 1, to: hi, char: "%") {
+      return .comment(open: i, body: (i + 1)..<close, close: close, end: close + 1)
+    }
+
+    // Verbatim / math objects: literal inner content.
+    if let style = literalModifiers[c], isOpener(prev: prev, next: next),
+      let close = literalClose(chars, from: i + 1, to: hi, char: c) {
+      return .verbatim(style: style, open: i, body: (i + 1)..<close, close: close, end: close + 1)
+    }
+
+    // Links: {location} optionally followed by [description].
+    if c == "{", let close = bracketClose(chars, from: i + 1, to: hi, char: "}") {
+      let target = Bracketed(open: i, body: (i + 1)..<close, close: close)
+      var j = close + 1
+      var description: Bracketed?
+      if j < hi, chars[j] == "[", let dclose = bracketClose(chars, from: j + 1, to: hi, char: "]") {
+        description = Bracketed(open: j, body: (j + 1)..<dclose, close: dclose)
+        j = dclose + 1
+      }
+      return .link(open: i, target: target, description: description, end: j)
+    }
+
+    // Anchors: [name] (declaration), [name]{location} (definition), or
+    // [name][description] (declaration with a custom description).
+    if c == "[", let close = bracketClose(chars, from: i + 1, to: hi, char: "]") {
+      let name = Bracketed(open: i, body: (i + 1)..<close, close: close)
+      var j = close + 1
+      var suffix: AnchorSuffix?
+      if j < hi, chars[j] == "{", let tclose = bracketClose(chars, from: j + 1, to: hi, char: "}") {
+        suffix = .target(Bracketed(open: j, body: (j + 1)..<tclose, close: tclose))
+        j = tclose + 1
+      } else if j < hi, chars[j] == "[",
+        let dclose = bracketClose(chars, from: j + 1, to: hi, char: "]") {
+        suffix = .description(Bracketed(open: j, body: (j + 1)..<dclose, close: dclose))
+        j = dclose + 1
+      }
+      return .anchor(open: i, name: name, suffix: suffix, end: j)
+    }
+
+    // Attached modifiers with recursively parsed content.
+    if let style = modifiers[c], isOpener(prev: prev, next: next),
+      let close = modifierClose(chars, from: i + 1, to: hi, char: c) {
+      return .modifier(style: style, open: i, body: (i + 1)..<close, close: close, end: close + 1)
+    }
+
+    return nil
+  }
+
+  // MARK: - Boundary helpers
+
+  private static func isSpace(_ c: Unicode.Scalar?) -> Bool {
+    guard let c else { return true }
+    return c.properties.isWhitespace
+  }
+
+  private static func isBoundary(_ c: Unicode.Scalar?) -> Bool {
+    guard let c else { return true }
+    if c.properties.isWhitespace { return true }
+    switch c.properties.generalCategory {
+    case .connectorPunctuation, .dashPunctuation, .openPunctuation,
+      .closePunctuation, .initialPunctuation, .finalPunctuation, .otherPunctuation,
+      .mathSymbol, .currencySymbol, .modifierSymbol, .otherSymbol:
+      return true
+    default:
+      return false
+    }
+  }
+
+  private static func isOpener(prev: Unicode.Scalar?, next: Unicode.Scalar?) -> Bool {
+    isBoundary(prev) && !isSpace(next)
+  }
+
+  private static func modifierClose(
+    _ chars: [Unicode.Scalar], from start: Int, to hi: Int, char: Unicode.Scalar
+  ) -> Int? {
+    var i = start
+    while i < hi {
+      if chars[i] == "\\" {
+        i += 2
+        continue
+      }
+      if chars[i] == char {
+        let prev: Unicode.Scalar? = i > 0 ? chars[i - 1] : nil
+        let next: Unicode.Scalar? = i + 1 < chars.count ? chars[i + 1] : nil
+        if !isSpace(prev) && isBoundary(next) { return i }
+      }
+      i += 1
+    }
+    return nil
+  }
+
+  private static func literalClose(
+    _ chars: [Unicode.Scalar], from start: Int, to hi: Int, char: Unicode.Scalar
+  ) -> Int? {
+    var i = start
+    while i < hi {
+      if chars[i] == char {
+        let prev: Unicode.Scalar? = i > 0 ? chars[i - 1] : nil
+        if !isSpace(prev) { return i }
+      }
+      i += 1
+    }
+    return nil
+  }
+
+  private static func bracketClose(
+    _ chars: [Unicode.Scalar], from start: Int, to hi: Int, char: Unicode.Scalar
+  ) -> Int? {
+    var i = start
+    while i < hi {
+      if chars[i] == "\\" {
+        i += 2
+        continue
+      }
+      if chars[i] == char { return i }
+      i += 1
+    }
+    return nil
+  }
+}