313 lines
10 KiB
Dart
313 lines
10 KiB
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 '../../domain/school_term.dart';
|
||
import 'school_navigation_policy.dart';
|
||
|
||
class UnsupportedSchoolLayout implements Exception {
|
||
const UnsupportedSchoolLayout({
|
||
this.code = 'layout_unknown',
|
||
this.row,
|
||
this.column,
|
||
});
|
||
final String code;
|
||
final int? row, column;
|
||
String get description => switch (code) {
|
||
'origin_invalid' => '页面来源不符合学校课表要求',
|
||
'payload_too_large' => '课表页面超过读取大小限制',
|
||
'table_missing' => '未找到课表表格',
|
||
'semester_missing' => '课表学期信息缺失',
|
||
'row_count_invalid' => '课表时段行数不符合预期',
|
||
'weekday_header_invalid' => '星期表头格式不符合预期',
|
||
'row_cell_count_invalid' => '课表时段单元格不完整',
|
||
'period_header_invalid' => '课表节次表头无法识别',
|
||
'cell_id_invalid' => '课程单元格的星期或节次标识异常',
|
||
'course_record_invalid' => '课程名称、教师或课程字段格式无法识别',
|
||
'period_mismatch' => '课程节次与所在时段不一致',
|
||
'weeks_invalid' => '上课周次格式无法识别',
|
||
'unsafe_markup' => '课程单元格包含不支持的网页内容',
|
||
'empty_timetable' => '课表尚无可识别的课程安排',
|
||
_ => '学校课表格式暂不支持',
|
||
};
|
||
@override
|
||
String toString() => '$description($code),当前课表保持不变。请在反馈中附带学校HTML和日志。';
|
||
}
|
||
|
||
/// 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) {
|
||
throw const UnsupportedSchoolLayout(code: 'origin_invalid');
|
||
}
|
||
if (utf8.encode(content).length > 2 * 1024 * 1024) {
|
||
throw const UnsupportedSchoolLayout(code: 'payload_too_large');
|
||
}
|
||
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) {
|
||
throw const UnsupportedSchoolLayout(code: 'table_missing');
|
||
}
|
||
if (term == null ||
|
||
!RegExp(r'^\d+$').hasMatch(term) ||
|
||
selected!.text.trim().isEmpty) {
|
||
throw const UnsupportedSchoolLayout(code: 'semester_missing');
|
||
}
|
||
final rows = table.querySelectorAll('tr');
|
||
const weekdays = ['节次', '星期一', '星期二', '星期三', '星期四', '星期五', '星期六', '星期日'];
|
||
if (rows.length != 6) {
|
||
throw const UnsupportedSchoolLayout(code: 'row_count_invalid');
|
||
}
|
||
if (rows.first.children.length != 8) {
|
||
throw const UnsupportedSchoolLayout(
|
||
code: 'weekday_header_invalid',
|
||
row: 0,
|
||
);
|
||
}
|
||
for (var i = 0; i < 8; i++) {
|
||
if (rows.first.children[i].text.trim() != weekdays[i]) {
|
||
throw UnsupportedSchoolLayout(
|
||
code: 'weekday_header_invalid',
|
||
row: 0,
|
||
column: i,
|
||
);
|
||
}
|
||
}
|
||
final courses = <String, Course>{};
|
||
final ids = <String>{};
|
||
const starts = [1, 3, 5, 7, 9];
|
||
for (var row = 0; row < starts.length; row++) {
|
||
final first = starts[row], last = first == 9 ? 11 : first + 1;
|
||
final cells = rows[row + 1].children;
|
||
if (cells.length != 8) {
|
||
throw UnsupportedSchoolLayout(
|
||
code: 'row_cell_count_invalid',
|
||
row: row + 1,
|
||
);
|
||
}
|
||
if (cells.first.id != 't0$first' ||
|
||
cells.first.text.trim() != '$first-$last节') {
|
||
throw UnsupportedSchoolLayout(
|
||
code: 'period_header_invalid',
|
||
row: row + 1,
|
||
);
|
||
}
|
||
for (var day = 1; day <= 7; day++) {
|
||
final cell = cells[day];
|
||
if (cell.id != 't$day$first') {
|
||
throw UnsupportedSchoolLayout(
|
||
code: 'cell_id_invalid',
|
||
row: row + 1,
|
||
column: day,
|
||
);
|
||
}
|
||
final List<String> lines;
|
||
try {
|
||
lines = _lines(cell);
|
||
} on UnsupportedSchoolLayout catch (error) {
|
||
throw UnsupportedSchoolLayout(
|
||
code: error.code,
|
||
row: row + 1,
|
||
column: day,
|
||
);
|
||
}
|
||
for (final line in lines) {
|
||
final record = RegExp(
|
||
r'^([A-Za-z0-9][A-Za-z0-9_.-]*)\s*/(.+?)/\s*(\d+(?:\.\d+)?)\s*/\s*(\d+(?:\.\d+)?)\s*/([^/]+)/([^/]+)周/(\d+)[~~-](\d+)节/(.*)$',
|
||
).firstMatch(line);
|
||
if (record == null) {
|
||
throw UnsupportedSchoolLayout(
|
||
code: 'course_record_invalid',
|
||
row: row + 1,
|
||
column: day,
|
||
);
|
||
}
|
||
if (int.parse(record[7]!) != first ||
|
||
int.parse(record[8]!) != last) {
|
||
throw UnsupportedSchoolLayout(
|
||
code: 'period_mismatch',
|
||
row: row + 1,
|
||
column: day,
|
||
);
|
||
}
|
||
final code = record[1]!,
|
||
name = record[2]!.trim(),
|
||
teacher = record[5]!.trim(),
|
||
rawRoom = record[9]!.trim(),
|
||
room = rawRoom.isEmpty ? '教室待定' : rawRoom;
|
||
if (name.isEmpty || teacher.isEmpty) {
|
||
throw UnsupportedSchoolLayout(
|
||
code: 'course_record_invalid',
|
||
row: row + 1,
|
||
column: day,
|
||
);
|
||
}
|
||
final Set<int> weeks;
|
||
try {
|
||
weeks = _weeks(record[6]!);
|
||
} on UnsupportedSchoolLayout catch (error) {
|
||
throw UnsupportedSchoolLayout(
|
||
code: error.code,
|
||
row: row + 1,
|
||
column: day,
|
||
);
|
||
}
|
||
final identity = jsonEncode([
|
||
term,
|
||
code,
|
||
name,
|
||
teacher,
|
||
room,
|
||
day,
|
||
first,
|
||
last,
|
||
]);
|
||
final previous = courses[identity];
|
||
final id =
|
||
previous?.id ??
|
||
'uestc-$term-$code-$day-$first-$last-${_hash(identity)}';
|
||
if (previous == null && !ids.add(id)) {
|
||
throw UnsupportedSchoolLayout();
|
||
}
|
||
courses[identity] = Course(
|
||
id: id,
|
||
name: name,
|
||
teacher: teacher,
|
||
room: room,
|
||
weekday: day,
|
||
startPeriod: first,
|
||
endPeriod: last,
|
||
weeks: {...?previous?.weeks, ...weeks},
|
||
source: source,
|
||
notes: rawRoom.isEmpty ? '学校未提供教室,请以学校最新安排为准并在公布后刷新课表。' : '',
|
||
);
|
||
}
|
||
}
|
||
}
|
||
// An empty DOM can be an expired session or an incomplete response.
|
||
if (courses.isEmpty) {
|
||
throw const UnsupportedSchoolLayout(code: 'empty_timetable');
|
||
}
|
||
return ImportDraft(
|
||
courses: List.unmodifiable(courses.values),
|
||
explicitlyEmpty: false,
|
||
source: source,
|
||
parserVersion: 'uestc-graduate-html-v1',
|
||
semesterLabel: selected.text.trim(),
|
||
schoolTerm: SchoolTerm(int.parse(term)),
|
||
);
|
||
} on UnsupportedSchoolLayout {
|
||
rethrow;
|
||
} catch (_) {
|
||
throw UnsupportedSchoolLayout();
|
||
}
|
||
}
|
||
|
||
List<String> _lines(dom.Element cell) {
|
||
final buffer = StringBuffer();
|
||
void visit(dom.Node node) {
|
||
if (node is dom.Text) {
|
||
buffer.write(node.text);
|
||
} else if (node is dom.Element) {
|
||
if (node.localName == 'br') {
|
||
buffer.write('\n');
|
||
} else {
|
||
if (node.localName == 'script' || node.localName == 'style') {
|
||
throw const UnsupportedSchoolLayout(code: 'unsafe_markup');
|
||
}
|
||
for (final child in node.nodes) {
|
||
visit(child);
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
for (final node in cell.nodes) {
|
||
visit(node);
|
||
}
|
||
return buffer
|
||
.toString()
|
||
.split('\n')
|
||
.map((s) => s.trim())
|
||
.where((s) => s.isNotEmpty)
|
||
.toList();
|
||
}
|
||
|
||
Set<int> _weeks(String text) {
|
||
final result = <int>{};
|
||
for (final part in text.split(RegExp('[,,、]'))) {
|
||
final range = RegExp(r'^\s*(\d+)(?:\s*[-~~]\s*(\d+))?\s*$')
|
||
.firstMatch(part);
|
||
if (range == null) {
|
||
throw const UnsupportedSchoolLayout(code: 'weeks_invalid');
|
||
}
|
||
final first = int.parse(range[1]!),
|
||
last = int.parse(range[2] ?? range[1]!);
|
||
if (first < 1 || last > 30 || last < first) {
|
||
throw const UnsupportedSchoolLayout(code: 'weeks_invalid');
|
||
}
|
||
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!;
|
||
}
|
||
}
|