TPSchedule/mobile/lib/features/school/school_page_adapter.dart

import 'dart:convert';

import 'package:html/dom.dart' as dom;
import 'package:html/parser.dart' as html;

import '../../data/json_schedule_import.dart';
import '../../domain/course.dart';
import 'school_navigation_policy.dart';

class UnsupportedSchoolLayout implements Exception {
  @override
  String toString() => '未识别到完整的学校课表，当前课表保持不变。请在课表查询页面保存 HTML 后重试。';
}

/// Reads the supplied graduate-system table as data; never executes its scripts.
class SchoolPageAdapter {
  ImportDraft extract(
    String content,
    Uri origin, {
    CourseSource source = CourseSource.school,
  }) {
    try {
      if (!SchoolNavigationPolicy.isTimetable(origin) ||
          source == CourseSource.demo ||
          utf8.encode(content).length > 2 * 1024 * 1024) {
        throw UnsupportedSchoolLayout();
      }
      final document = html.parse(content);
      final table = document.querySelector('#tbl');
      final selected = document.querySelector('#drpxq option[selected]');
      final term = selected?.attributes['value'];
      if (table == null ||
          term == null ||
          !RegExp(r'^\d+$').hasMatch(term) ||
          selected!.text.trim().isEmpty) {
        throw UnsupportedSchoolLayout();
      }
      final rows = table.querySelectorAll('tr');
      const weekdays = ['节次', '星期一', '星期二', '星期三', '星期四', '星期五', '星期六', '星期日'];
      if (rows.length != 6 || rows.first.children.length != 8) {
        throw UnsupportedSchoolLayout();
      }
      for (var i = 0; i < 8; i++) {
        if (rows.first.children[i].text.trim() != weekdays[i]) {
          throw UnsupportedSchoolLayout();
        }
      }
      final courses = <String, Course>{};
      final ids = <String>{};
      const starts = [1, 3, 5, 7, 9];
      for (var row = 0; row < starts.length; row++) {
        final first = starts[row], last = first == 9 ? 11 : first + 1;
        final cells = rows[row + 1].children;
        if (cells.length != 8 ||
            cells.first.id != 't0$first' ||
            cells.first.text.trim() != '$first-$last节') {
          throw UnsupportedSchoolLayout();
        }
        for (var day = 1; day <= 7; day++) {
          final cell = cells[day];
          if (cell.id != 't$day$first') throw UnsupportedSchoolLayout();
          for (final line in _lines(cell)) {
            final record = RegExp(
              r'^(\d+)/(.+?)/(\d+(?:\.\d+)?)/(\d+(?:\.\d+)?)/([^/]+)/([^/]+)周/(\d+)[~～-](\d+)节/(.+)$',
            ).firstMatch(line);
            if (record == null ||
                int.parse(record[7]!) != first ||
                int.parse(record[8]!) != last) {
              throw UnsupportedSchoolLayout();
            }
            final code = record[1]!,
                name = record[2]!.trim(),
                teacher = record[5]!.trim(),
                room = record[9]!.trim();
            final weeks = _weeks(record[6]!);
            final identity = jsonEncode([
              term,
              code,
              name,
              teacher,
              room,
              day,
              first,
              last,
            ]);
            final previous = courses[identity];
            final id =
                previous?.id ??
                'uestc-$term-$code-$day-$first-$last-${_hash(identity)}';
            if (previous == null && !ids.add(id)) {
              throw UnsupportedSchoolLayout();
            }
            courses[identity] = Course(
              id: id,
              name: name,
              teacher: teacher,
              room: room,
              weekday: day,
              startPeriod: first,
              endPeriod: last,
              weeks: {...?previous?.weeks, ...weeks},
              source: source,
            );
          }
        }
      }
      // An empty DOM can be an expired session or an incomplete response.
      if (courses.isEmpty) throw UnsupportedSchoolLayout();
      return ImportDraft(
        courses: List.unmodifiable(courses.values),
        explicitlyEmpty: false,
        source: source,
        parserVersion: 'uestc-graduate-html-v1',
        semesterLabel: selected.text.trim(),
      );
    } on UnsupportedSchoolLayout {
      rethrow;
    } catch (_) {
      throw UnsupportedSchoolLayout();
    }
  }

  List<String> _lines(dom.Element cell) {
    final buffer = StringBuffer();
    void visit(dom.Node node) {
      if (node is dom.Text) {
        buffer.write(node.text);
      } else if (node is dom.Element) {
        if (node.localName == 'br') {
          buffer.write('\n');
        } else {
          if (node.localName == 'script' || node.localName == 'style') {
            throw UnsupportedSchoolLayout();
          }
          for (final child in node.nodes) {
            visit(child);
          }
        }
      }
    }

    for (final node in cell.nodes) {
      visit(node);
    }
    return buffer
        .toString()
        .split('\n')
        .map((s) => s.trim())
        .where((s) => s.isNotEmpty)
        .toList();
  }

  Set<int> _weeks(String text) {
    final result = <int>{};
    for (final part in text.split(RegExp('[,，、]'))) {
      final range = RegExp(r'^\s*(\d+)(?:\s*[-~～]\s*(\d+))?\s*$')
          .firstMatch(part);
      if (range == null) throw UnsupportedSchoolLayout();
      final first = int.parse(range[1]!),
          last = int.parse(range[2] ?? range[1]!);
      if (first < 1 || last > 30 || last < first) {
        throw UnsupportedSchoolLayout();
      }
      result.addAll(List.generate(last - first + 1, (i) => first + i));
    }
    return result;
  }

  String _hash(String value) {
    var hash = 2166136261;
    for (final byte in utf8.encode(value)) {
      hash = ((hash ^ byte) * 16777619) & 0xffffffff;
    }
    return hash.toRadixString(16).padLeft(8, '0');
  }
}

class SchoolReader {
  SchoolReader(this.adapter);
  final SchoolPageAdapter adapter;
  Future<ImportDraft>? _pending;
  int _generation = 0;
  Uri? _origin;
  bool get busy => _pending != null;
  void invalidate() {
    _generation++;
    _pending = null;
    _origin = null;
  }

  Future<ImportDraft> read(Uri origin, Future<String> Function() getHtml) {
    if (!SchoolNavigationPolicy.isTimetable(origin) ||
        (_pending != null && origin != _origin)) {
      return Future.error(UnsupportedSchoolLayout());
    }
    if (_pending != null) return _pending!;
    final generation = _generation;
    _origin = origin;
    _pending = Future<ImportDraft>(() async {
      try {
        if (!SchoolNavigationPolicy.isTimetable(origin)) {
          throw UnsupportedSchoolLayout();
        }
        final content = await getHtml();
        if (generation != _generation) throw UnsupportedSchoolLayout();
        return adapter.extract(content, origin);
      } finally {
        if (generation == _generation) {
          _pending = null;
          _origin = null;
        }
      }
    });
    return _pending!;
  }
}