feat: parse undergraduate timetable activity data

2026-10-14 23:32:27 +08:00
parent 141436f545
commit 17e4903399
5 changed files with 548 additions and 0 deletions
@@ -0,0 +1,139 @@
+class UndergraduateActivity {
+  UndergraduateActivity(
+    this.teacher,
+    this.name,
+    this.courseCode,
+    this.room,
+    this.indices,
+    this.weeks,
+  );
+  final String teacher, name, courseCode, room;
+  final Set<int> indices, weeks;
+  bool get cancelled => room == '停课';
+}
+
+List<UndergraduateActivity> parseActivityData(
+  String script, {
+  required int unitCount,
+}) {
+  if (unitCount != 12) throw const FormatException('本科节次结构无效');
+  final matches = RegExp(
+    r'new\s+TaskActivity\s*\((.*?)\)\s*;',
+    dotAll: true,
+  ).allMatches(script).toList();
+  final result = <UndergraduateActivity>[];
+  for (var i = 0; i < matches.length; i++) {
+    final m = matches[i], args = _strings(m[1]!);
+    if (args.length != 7) throw const FormatException('课程活动字段不完整');
+    final suffix = RegExp(r'\(([A-Za-z0-9_.-]+)\)$').firstMatch(args[3]);
+    if (suffix == null || args[1].trim().isEmpty || args[5].trim().isEmpty)
+      throw const FormatException('教师、教室或课程序号缺失');
+    final bitmap = args[6];
+    if (!RegExp(r'^0[01]+$').hasMatch(bitmap) ||
+        bitmap.length > 64 ||
+        (bitmap.length > 31 && bitmap.substring(31).contains('1')))
+      throw const FormatException('上课周次超出1–30周');
+    final weeks = {
+      for (var w = 1; w < bitmap.length && w <= 30; w++)
+        if (bitmap[w] == '1') w,
+    };
+    if (weeks.isEmpty) throw const FormatException('课程周次为空');
+    var end = i + 1 < matches.length ? matches[i + 1].start : script.length;
+    final finish = script.indexOf('table0.marshalTable', m.end);
+    if (finish >= 0 && finish < end) end = finish;
+    final body = script.substring(m.end, end);
+    final indices = <int>{};
+    final assignments = RegExp(r'\bindex\s*=\s*([^;]+);')
+        .allMatches(body)
+        .toList();
+    final appends = RegExp(
+      r'table\d+\.activities\[index\]\s*\[[^;]+?\]\s*=\s*activity\s*;',
+    ).allMatches(body).length;
+    if (assignments.isEmpty || assignments.length != appends)
+      throw const FormatException('课程节次赋值不完整');
+    for (final assignment in assignments) {
+      final expression = RegExp(r'^\s*(\d+)\s*\*\s*unitCount\s*\+\s*(\d+)\s*$')
+          .firstMatch(assignment[1]!);
+      if (expression == null) throw const FormatException('不支持的节次表达式');
+      final day = int.parse(expression[1]!), period = int.parse(expression[2]!);
+      if (day > 6 || period >= unitCount)
+        throw const FormatException('课程星期或节次无效');
+      indices.add(day * unitCount + period);
+    }
+    result.add(
+      UndergraduateActivity(
+        args[1].trim(),
+        args[3].substring(0, suffix.start).trim(),
+        suffix[1]!,
+        args[5].trim(),
+        indices,
+        weeks,
+      ),
+    );
+  }
+  return result;
+}
+
+/// Reads only quoted JS literals; expressions and calls are never evaluated.
+List<String> _strings(String input) {
+  final out = <String>[];
+  var p = 0;
+  void spaces() {
+    while (p < input.length && RegExp(r'\s').hasMatch(input[p])) {
+      p++;
+    }
+  }
+
+  while (true) {
+    spaces();
+    if (p >= input.length) break;
+    final quote = input[p++];
+    if (quote != '"' && quote != "'") throw const FormatException('活动参数必须为字符串');
+    final b = StringBuffer();
+    var closed = false;
+    while (p < input.length) {
+      var c = input[p++];
+      if (c == quote) {
+        closed = true;
+        break;
+      }
+      if (c == '\\') {
+        if (p >= input.length) throw const FormatException('字符串转义不完整');
+        c = input[p++];
+        const escapes = {
+          'n': '\n',
+          'r': '\r',
+          't': '\t',
+          'b': '\b',
+          'f': '\f',
+          'v': '\u000b',
+          '0': '\u0000',
+        };
+        if (c == 'u' || c == 'x') {
+          final count = c == 'u' ? 4 : 2;
+          if (p + count > input.length) throw const FormatException('字符串转义无效');
+          final hex = input.substring(p, p + count);
+          if (!RegExp(r'^[0-9a-fA-F]+$').hasMatch(hex))
+            throw const FormatException('字符串转义无效');
+          b.writeCharCode(int.parse(hex, radix: 16));
+          p += count;
+        } else if (escapes.containsKey(c)) {
+          b.write(escapes[c]);
+        } else if (const ['"', "'", '\\', '/'].contains(c)) {
+          b.write(c);
+        } else {
+          throw const FormatException('不支持的字符串转义');
+        }
+      } else {
+        if (c == '\n' || c == '\r') throw const FormatException('字符串未闭合');
+        b.write(c);
+      }
+    }
+    if (!closed) throw const FormatException('字符串未闭合');
+    out.add(b.toString());
+    spaces();
+    if (p == input.length) break;
+    if (input[p++] != ',') throw const FormatException('活动字段分隔无效');
+  }
+  return out;
+}
@@ -0,0 +1,189 @@
+import 'dart:convert';
+
+import 'package:html/parser.dart' as html;
+
+import '../../../data/json_schedule_import.dart';
+import '../../../domain/course.dart';
+import '../../../domain/school_identity.dart';
+import '../../../domain/school_term.dart';
+import 'activity_data.dart';
+
+class UndergraduateTimetableAdapter {
+  static ImportDraft extract(
+    String content,
+    Uri uri, {
+    CourseSource source = CourseSource.school,
+  }) {
+    try {
+      if (uri.scheme != 'https' ||
+          uri.host != 'eams.uestc.edu.cn' ||
+          uri.port != 443 ||
+          uri.userInfo.isNotEmpty ||
+          !uri.path.startsWith('/eams/') ||
+          source == CourseSource.demo ||
+          utf8.encode(content).length > 2 * 1024 * 1024)
+        throw const FormatException('不是本科课表页面');
+      final doc = html.parse(content),
+          table = html
+              .parse(content)
+              .querySelector('#manualArrangeCourseTable');
+      if (table == null || doc.querySelector('input[type="password"]') != null)
+        throw const FormatException('请先在学校网页打开课表');
+      final termIds = doc
+          .querySelectorAll('input[name="semester.id"]')
+          .map((e) => e.attributes['value'])
+          .toSet();
+      if (termIds.length != 1) throw const FormatException('课表学期不明确');
+      final term = SchoolTerm(
+        int.parse(termIds.single!),
+        school: SupportedSchool.uestcUndergraduate,
+      );
+      final scripts = doc
+          .querySelectorAll('script')
+          .map((e) => e.text)
+          .join('\n');
+      final units = RegExp(r'\b(?:var\s+)?unitCount\s*=\s*(\d+)\s*;')
+          .allMatches(scripts)
+          .map((m) => int.parse(m[1]!))
+          .toSet();
+      final activities = units.length == 1
+          ? parseActivityData(scripts, unitCount: units.single)
+          : <UndergraduateActivity>[];
+      final explicitEmpty =
+          activities.isEmpty && RegExp(r'暂无课程|无课程安排|没有课程').hasMatch(table.text);
+      if (activities.isEmpty && !explicitEmpty)
+        throw const FormatException('课表尚未加载完整');
+      final roster = <String, ({String code, String notes})>{};
+      for (final grid in doc.querySelectorAll('table')) {
+        final rows = grid.querySelectorAll('tr');
+        if (rows.isEmpty) continue;
+        final headers = rows.first.children.map((e) => e.text.trim()).toList();
+        if (!headers.contains('课程序号') || !headers.contains('课程代码')) continue;
+        final seq = headers.indexOf('课程序号'),
+            code = headers.indexOf('课程代码'),
+            note = headers.indexOf('选课备注');
+        for (final row in rows.skip(1)) {
+          final cells = row.children;
+          if (cells.length != headers.length)
+            throw const FormatException('课程名单不完整');
+          roster[cells[seq].text.trim()] = (
+            code: cells[code].text.trim(),
+            notes: note >= 0 ? cells[note].text.trim() : '',
+          );
+        }
+      }
+      final cancellations = <String, Set<int>>{};
+      for (final a in activities.where((a) => a.cancelled)) {
+        for (final index in a.indices) {
+          cancellations
+              .putIfAbsent('${a.courseCode}/$index', () => {})
+              .addAll(a.weeks);
+        }
+      }
+      final cells =
+          <String, ({UndergraduateActivity a, int index, Set<int> weeks})>{};
+      for (final a in activities.where((a) => !a.cancelled)) {
+        for (final index in a.indices) {
+          final weeks = a.weeks.difference(
+            cancellations['${a.courseCode}/$index'] ?? {},
+          );
+          if (weeks.isEmpty) continue;
+          final key = jsonEncode([
+            a.courseCode,
+            a.name,
+            a.teacher,
+            a.room,
+            index,
+          ]);
+          final cell = cells.putIfAbsent(
+            key,
+            () => (a: a, index: index, weeks: <int>{}),
+          );
+          cell.weeks.addAll(weeks);
+        }
+      }
+      final groups =
+          <
+            String,
+            ({UndergraduateActivity a, Set<int> weeks, List<int> indices})
+          >{};
+      for (final cell in cells.values) {
+        final key = jsonEncode([
+          cell.a.courseCode,
+          cell.a.name,
+          cell.a.teacher,
+          cell.a.room,
+          cell.index ~/ 12,
+          cell.weeks.toList()..sort(),
+        ]);
+        groups
+            .putIfAbsent(
+              key,
+              () => (a: cell.a, weeks: cell.weeks, indices: <int>[]),
+            )
+            .indices
+            .add(cell.index);
+      }
+      final courses = <Course>[];
+      for (final group in groups.values) {
+        final indices = group.indices..sort();
+        for (var p = 0; p < indices.length;) {
+          final first = indices[p];
+          var last = first;
+          p++;
+          while (p < indices.length && indices[p] == last + 1) {
+            last = indices[p++];
+          }
+          final a = group.a, info = roster[a.courseCode];
+          final identity = jsonEncode([
+            term.id,
+            a.courseCode,
+            a.name,
+            a.teacher,
+            a.room,
+            first,
+            last,
+            group.weeks.toList()..sort(),
+          ]);
+          courses.add(
+            Course(
+              id: 'uestc-undergraduate-${term.id}-${_hash(identity)}',
+              name: a.name,
+              teacher: a.teacher,
+              room: a.room,
+              weekday: first ~/ 12 + 1,
+              startPeriod: first % 12 + 1,
+              endPeriod: last % 12 + 1,
+              weeks: group.weeks,
+              source: source,
+              courseCode: info?.code ?? a.courseCode,
+              notes: info?.notes ?? '',
+            ),
+          );
+        }
+      }
+      if (courses.isEmpty && !explicitEmpty)
+        throw const FormatException('没有识别到有效上课安排');
+      return ImportDraft(
+        courses: List.unmodifiable(courses),
+        explicitlyEmpty: explicitEmpty,
+        source: source,
+        parserVersion: 'uestc-undergraduate-v1',
+        schoolTerm: term,
+        semesterLabel: term.label,
+      );
+    } on FormatException {
+      rethrow;
+    } catch (_) {
+      throw const FormatException('本科课表结构或学期无效，原课表保持不变');
+    }
+  }
+
+  static String _hash(String value) {
+    var h = 0x811c9dc5;
+    for (final b in utf8.encode(value)) {
+      h = ((h ^ b) * 0x01000193) & 0xffffffff;
+    }
+    return h.toRadixString(16).padLeft(8, '0');
+  }
+}