feat: parse and merge undergraduate exam schedules

2026-10-14 23:35:30 +08:00
parent 17e4903399
commit ebe70e3eb2
4 changed files with 384 additions and 0 deletions
@@ -0,0 +1,179 @@
+import 'dart:convert';
+
+import 'package:html/dom.dart';
+import 'package:html/parser.dart' as html;
+
+import '../../../domain/exam_schedule.dart';
+import '../../../domain/school_identity.dart';
+import '../../../domain/school_term.dart';
+import '../../../domain/undergraduate_exam_query.dart';
+
+class UndergraduateExamAdapter {
+  static const headers = [
+    '课程序号',
+    '课程名称',
+    '考试日期',
+    '考试安排',
+    '考试地点',
+    '座位号',
+    '考试情况',
+    '考试类型',
+    '其它说明',
+  ];
+  static ExamSchedule extract(
+    String content, {
+    SchoolTerm? expectedTerm,
+    UndergraduateExamType? expectedType,
+  }) {
+    try {
+      if (utf8.encode(content).length > 2 * 1024 * 1024)
+        throw const FormatException('考试页面超过2MB');
+      final doc = html.parse(content);
+      if (doc.querySelector('input[type="password"]') != null)
+        throw const FormatException('请先完成学校登录');
+      final form = doc.querySelector('#stdExamTableForm');
+      final scope = form ?? doc;
+      final termValues = scope
+          .querySelectorAll('input[name="semester.id"]')
+          .map((e) => e.attributes['value'])
+          .whereType<String>()
+          .toSet();
+      final typeValues = scope
+          .querySelectorAll('input[name="examType.id"]')
+          .map((e) => e.attributes['value'])
+          .whereType<String>()
+          .toSet();
+      for (final params in scope.querySelectorAll('input[name="params"]')) {
+        final query = Uri.splitQueryString(
+          (params.attributes['value'] ?? '').replaceFirst(RegExp(r'^&'), ''),
+        );
+        if (query['semester.id'] != null) termValues.add(query['semester.id']!);
+        if (query['examType.id'] != null) typeValues.add(query['examType.id']!);
+      }
+      // Only a result params assignment is evidence, not calendar initialization.
+      for (final script in doc.querySelectorAll('script')) {
+        for (final m in RegExp(
+          r'''\bvar\s+params\s*=\s*["']([^"']+)["']''',
+        ).allMatches(script.text)) {
+          final query = Uri.splitQueryString(
+            m[1]!.replaceFirst(RegExp(r'^&'), ''),
+          );
+          if (query['semester.id'] != null)
+            termValues.add(query['semester.id']!);
+          if (query['examType.id'] != null)
+            typeValues.add(query['examType.id']!);
+        }
+      }
+      if (termValues.length != 1 || typeValues.length != 1)
+        throw const FormatException('排考结果学期或类型不明确');
+      final term = SchoolTerm(
+        int.parse(termValues.single),
+        school: SupportedSchool.uestcUndergraduate,
+      );
+      final type = UndergraduateExamType.fromId(int.parse(typeValues.single));
+      if (expectedTerm != null && expectedTerm != term ||
+          expectedType != null && expectedType != type)
+        throw const FormatException('排考结果与所选学期或类型不一致');
+      Element? table;
+      for (final candidate in doc.querySelectorAll('table')) {
+        final cells = candidate
+            .querySelector('tr')
+            ?.children
+            .map((c) => c.text.trim())
+            .toList();
+        if (cells != null &&
+            cells.length == headers.length &&
+            List.generate(
+              headers.length,
+              (i) => cells[i] == headers[i],
+            ).every((v) => v)) {
+          if (table != null) throw const FormatException('排考结果重复');
+          table = candidate;
+        }
+      }
+      if (table == null) throw const FormatException('排考页面尚未加载完整');
+      final records = <String, ExamRecord>{};
+      final explicitEmpty = RegExp(r'暂无考试|无考试安排|没有考试|暂无记录|没有符合条件的记录|没有数据|无数据')
+          .hasMatch(table.text);
+      for (final row in table.querySelectorAll('tr').skip(1)) {
+        final cells = row.children
+            .map((c) => c.text.replaceAll(RegExp(r'\s+'), ' ').trim())
+            .toList();
+        if (cells.length != 9) {
+          if (explicitEmpty && cells.length == 1) continue;
+          throw const FormatException('考试记录字段不完整');
+        }
+        final rowType = _type(cells[7]);
+        if (rowType != type.id) throw const FormatException('考试记录类型不一致');
+        final record = ExamRecord(
+          code: cells[0],
+          name: cells[1],
+          date: cells[2],
+          time: cells[3],
+          room: cells[4],
+          seat: cells[5],
+          status: cells[6],
+          examTypeId: rowType,
+          notes: cells[8],
+        );
+        if (record.code.isEmpty || record.name.isEmpty)
+          throw const FormatException('考试课程缺失');
+        records.putIfAbsent(record.undergraduateKey, () => record);
+      }
+      if (records.isEmpty && !explicitEmpty)
+        throw const FormatException('尚未识别到明确的排考结果');
+      return ExamSchedule.fromJson(
+        ExamSchedule(
+          semesterLabel: term.label,
+          termId: '${term.id}',
+          records: records.values.toList(),
+          school: term.school,
+          requestedExamTypes: {type.id},
+        ).toJson(),
+      );
+    } on FormatException {
+      rethrow;
+    } catch (_) {
+      throw const FormatException('本科考试页面内容无效，已有考试保持不变');
+    }
+  }
+
+  static int _type(String text) {
+    if (text.contains('期末')) return 1;
+    if (text.contains('期中')) return 2;
+    if (text.contains('补') || text.contains('缓')) return 3;
+    throw const FormatException('考试记录类型无法识别');
+  }
+
+  static ExamSchedule merge(
+    UndergraduateExamQuery query,
+    List<ExamSchedule> parts,
+  ) {
+    final expected = query.types.map((t) => t.id).toSet(),
+        seen = <int>{},
+        records = <String, ExamRecord>{};
+    for (final raw in parts) {
+      final part = ExamSchedule.fromJson(raw.toJson());
+      if (part.school != query.term.school ||
+          part.termId != '${query.term.id}' ||
+          part.requestedExamTypes.length != 1 ||
+          !expected.contains(part.requestedExamTypes.single) ||
+          !seen.add(part.requestedExamTypes.single))
+        throw const FormatException('考试查询批次不一致');
+      for (final record in part.records) {
+        records.putIfAbsent(record.undergraduateKey, () => record);
+      }
+    }
+    if (seen.length != expected.length || records.length > 500)
+      throw const FormatException('考试未全部读取完成或记录超限');
+    return ExamSchedule.fromJson(
+      ExamSchedule(
+        semesterLabel: query.term.label,
+        termId: '${query.term.id}',
+        records: records.values.toList(),
+        school: query.term.school,
+        requestedExamTypes: expected,
+      ).toJson(),
+    );
+  }
+}