the-mews/app/lib/mews/gemtext.dart

/// Converts gemtext, the format Gemini and Spartan pages are written in, into
/// the document model (document.dart), so protocol pages render through the
/// same reader presentation as Mews pages. Gemtext declares neither structure
/// nor metadata: there is no inline markup, no head, no title, so the first
/// level-1 heading stands in for a title and the reader presentation decides
/// everything else.
///
/// A link line whose label begins with a date is a dated entry (mews-spec
/// 6.1), which keeps subscriptions working on protocol sites.
library;

import 'dart:convert';

import 'document.dart';

/// The conformance marker gemtext pages carry in place of the Mews one, so the
/// renderer applies the reader presentation instead of falling back to plain.
const gemtextSpecVersion = 'gemtext';

final _entryDatePattern = RegExp(r'^(\d{4}-\d{2}-\d{2}) (.+)$');

MewsPage parseGemtextPage({required String source, required String url}) {
  final blocks = <MewsBlock>[];
  final entries = <MewsEntry>[];
  String? title;

  var inPre = false;
  final pre = StringBuffer();
  final quote = StringBuffer();
  final linkItems = <MewsListItem>[];

  void closePre() {
    // One trailing newline rides with the last line; the fence is not content.
    blocks.add(PreBlock(pre.toString().replaceFirst(RegExp(r'\n$'), '')));
    pre.clear();
  }

  // Consecutive lines of one kind are gathered until a line of another kind
  // ends them.
  void closeQuote() {
    if (quote.isEmpty) return;
    final lines = quote.toString().split('\n');
    blocks.add(QuoteBlock([
      for (final line in lines)
        if (line.trim().isNotEmpty) ParagraphBlock([MewsText(line.trim())]),
    ]));
    quote.clear();
  }

  void closeLinks() {
    if (linkItems.isEmpty) return;
    blocks.add(ListBlock(ordered: false, start: 1, items: List.of(linkItems)));
    linkItems.clear();
  }

  for (final line in const LineSplitter().convert(source)) {
    if (inPre) {
      if (line.startsWith('```')) {
        closePre();
        inPre = false;
      } else {
        pre.writeln(line);
      }
      continue;
    }
    if (line.startsWith('```')) {
      closeQuote();
      closeLinks();
      inPre = true;
      continue;
    }
    if (line.startsWith('=>')) {
      closeQuote();
      final parts = line.substring(2).trim().split(RegExp(r'\s+'));
      if (parts.isEmpty || parts.first.isEmpty) continue;
      final href = Uri.parse(url).resolve(parts.first).toString();
      final label = parts.skip(1).join(' ').trim();
      final text = label.isEmpty ? parts.first : label;
      final match = _entryDatePattern.firstMatch(text);
      if (match != null) {
        final date = DateTime.tryParse(match.group(1)!);
        if (date != null) {
          entries.add(MewsEntry(url: href, date: date, title: match.group(2)!.trim()));
        }
      }
      linkItems.add(MewsListItem([ParagraphBlock([MewsLink(href, [MewsText(text)])])]));
      continue;
    }
    closeLinks();
    if (line.startsWith('###') && !line.startsWith('####')) {
      closeQuote();
      final text = line.substring(3).trim();
      blocks.add(HeadingBlock(3, [MewsText(text)]));
    } else if (line.startsWith('##') && !line.startsWith('###')) {
      closeQuote();
      final text = line.substring(2).trim();
      blocks.add(HeadingBlock(2, [MewsText(text)]));
    } else if (line.startsWith('#') && !line.startsWith('##')) {
      closeQuote();
      final text = line.substring(1).trim();
      final heading = HeadingBlock(1, [MewsText(text)]);
      title ??= text;
      blocks.add(heading);
    } else if (line.startsWith('* ')) {
      closeQuote();
      _listOf(blocks).items.add(MewsListItem([ParagraphBlock([MewsText(line.substring(2).trim())])]));
    } else if (line.startsWith('>')) {
      quote.writeln(line.substring(1).trim());
    } else if (line.trim().isNotEmpty) {
      closeQuote();
      blocks.add(ParagraphBlock([MewsText(line.trim())]));
    } else {
      closeQuote();
    }
  }
  if (inPre) closePre();
  closeQuote();
  closeLinks();

  entries.sort((a, b) => b.date.compareTo(a.date));

  final text = source;
  return MewsPage(
    url: url,
    specVersion: gemtextSpecVersion,
    title: title,
    lang: _langOf(text),
    direction: _isRtl(text) ? TextDirection.rtl : TextDirection.ltr,
    variant: MewsVariant.plain,
    entries: entries,
    body: blocks,
  );
}

/// Appends a list block to build on, reusing the last one while consecutive
/// items keep coming.
ListBlock _listOf(List<MewsBlock> blocks) {
  if (blocks.isNotEmpty && blocks.last is ListBlock && !(blocks.last as ListBlock).ordered) {
    return blocks.last as ListBlock;
  }
  final fresh = ListBlock(ordered: false, start: 1, items: []);
  blocks.add(fresh);
  return fresh;
}

final _rtlPattern = RegExp(r'[\u0590-\u08FF]');

bool _isRtl(String text) => _rtlPattern.hasMatch(text);

String _langOf(String text) {
  if (RegExp(r'[\u0590-\u05FF]').hasMatch(text)) return 'he';
  if (RegExp(r'[\u0600-\u06FF]').hasMatch(text)) return 'ar';
  return 'en';
}