fix: import unpublished exam rooms and add opt-in school HTML feedback

2026-10-16 16:38:00 +08:00
parent defe7fd50e
commit 437ed14dbe
31 changed files with 969 additions and 39 deletions
@@ -0,0 +1,177 @@
+import 'dart:async';
+import 'dart:convert';
+import 'dart:typed_data';
+
+import 'package:html/parser.dart' as html;
+
+/// An opt-in mail attachment, separate from persistent structured logs.
+/// Keeps only course/exam evidence, never the surrounding account/login page.
+class SchoolPageDiagnostic {
+  SchoolPageDiagnostic({DateTime Function()? clock})
+    : clock = clock ?? DateTime.now;
+  static final instance = SchoolPageDiagnostic();
+  static const maxBytes = 2 * 1024 * 1024;
+  static const marker = '<!doctype html><!-- TPSchedule school diagnostic -->';
+  final DateTime Function() clock;
+  Uint8List? _bytes;
+  DateTime? _at;
+  Timer? _expiry;
+
+  void clear() {
+    _expiry?.cancel();
+    _expiry = null;
+    _bytes = null;
+    _at = null;
+  }
+
+  Uint8List? export() {
+    final at = _at;
+    if (at == null ||
+        clock().difference(at).isNegative ||
+        clock().difference(at) >= const Duration(minutes: 10)) {
+      clear();
+    }
+    return _bytes == null ? null : Uint8List.fromList(_bytes!);
+  }
+
+  void remember(String content, Uri uri) {
+    clear();
+    try {
+      if (uri.scheme != 'https' ||
+          uri.host != 'eams.uestc.edu.cn' ||
+          uri.port != 443 ||
+          uri.userInfo.isNotEmpty ||
+          !uri.path.startsWith('/eams/') ||
+          utf8.encode(content).length > maxBytes) {
+        return;
+      }
+      final doc = html.parse(content);
+      if (doc.querySelector('input[type="password"]') != null) return;
+      final tables = doc
+          .querySelectorAll('table')
+          .where(
+            (t) =>
+                t.id == 'manualArrangeCourseTable' ||
+                t.querySelector('tr')?.text.contains('课程序号') == true,
+          )
+          .toList();
+      if (tables.isEmpty) return;
+      final fields = <String, Set<String>>{};
+      void field(String key, String value) {
+        if (const ['semester.id', 'examType.id'].contains(key) &&
+            RegExp(r'^\d{1,6}$').hasMatch(value)) {
+          fields.putIfAbsent(key, () => {}).add(value);
+        }
+      }
+
+      for (final input in doc.querySelectorAll('input')) {
+        field(input.attributes['name'] ?? '', input.attributes['value'] ?? '');
+      }
+      for (final node in doc.querySelectorAll('script,input[name="params"]')) {
+        final value = node.localName == 'input'
+            ? node.attributes['value'] ?? ''
+            : node.text;
+        for (final m in RegExp(
+          r'(semester\.id|examType\.id)=(\d{1,6})(?=[&"\x27\s]|$)',
+        ).allMatches(value)) {
+          field(m[1]!, m[2]!);
+        }
+      }
+      final script = doc
+          .querySelectorAll('script')
+          .map((s) => s.text)
+          .join('\n');
+      // Only retain the literal course statements our parser reads. All other
+      // scripts, URL-bearing attributes and form/account fields are omitted.
+      final statements = _courseStatements(script);
+      for (final table in tables) {
+        for (final node in table.querySelectorAll(
+          'script,style,link,iframe,object,embed,form,input,textarea,select,button,img,svg,video,audio,source,meta,base',
+        )) {
+          node.remove();
+        }
+        for (final node in [table, ...table.querySelectorAll('*')]) {
+          final attrs = Map<String, String>.from(node.attributes);
+          node.attributes.clear();
+          if (attrs['id'] == 'manualArrangeCourseTable') {
+            node.id = 'manualArrangeCourseTable';
+          }
+          for (final key in ['rowspan', 'colspan']) {
+            final value = attrs[key];
+            if (value != null && RegExp(r'^\d{1,2}$').hasMatch(value)) {
+              node.attributes[key] = value;
+            }
+          }
+        }
+      }
+      final safe =
+          '$marker<html><head><meta charset="utf-8">'
+          '<meta http-equiv="Content-Security-Policy" content="default-src \'none\'; base-uri \'none\'; form-action \'none\'">'
+          '<title>TPSchedule school diagnostic</title></head><body>'
+          '${fields.entries.expand((e) => e.value.map((v) => '<input type="hidden" name="${e.key}" value="$v">')).join()}'
+          '${tables.map((t) => t.outerHtml).join()}'
+          '<script type="application/x-tpschedule-course-data">${statements.replaceAll('<', r'\u003c')}</script></body></html>';
+      final bytes = Uint8List.fromList(utf8.encode(safe));
+      if (bytes.length > maxBytes) return;
+      _bytes = bytes;
+      _at = clock();
+      _expiry = Timer(const Duration(minutes: 10), clear);
+    } catch (_) {
+      clear(); // Diagnostic collection must not interfere with importing.
+    }
+  }
+
+  // Split outside quoted strings without validating the course arguments: a
+  // malformed argument/index is precisely what this attachment must preserve.
+  static String _courseStatements(String script) {
+    final out = <String>[];
+    var start = 0;
+    String? quote;
+    var lineComment = false, blockComment = false;
+    final allowed = RegExp(
+      r'^(?:var\s+)?(?:unitCount\s*=|activity\s*=\s*new\s+TaskActivity\b|new\s+TaskActivity\b|index\s*=|table\d+\.activities\s*\[)',
+    );
+    void keep(int end) {
+      final statement = script
+          .substring(start, end)
+          .replaceFirst(
+            RegExp(r'^(?:\s|//[^\r\n]*(?:\r?\n|$)|/\*[\s\S]*?\*/)*'),
+            '',
+          )
+          .trim();
+      if (allowed.hasMatch(statement)) out.add(statement);
+      start = end;
+    }
+
+    for (var i = 0; i < script.length; i++) {
+      final c = script[i];
+      final next = i + 1 < script.length ? script[i + 1] : '';
+      if (lineComment) {
+        if (c == '\n' || c == '\r') lineComment = false;
+      } else if (blockComment) {
+        if (c == '*' && next == '/') {
+          blockComment = false;
+          i++;
+        }
+      } else if (quote != null) {
+        if (c == '\\') {
+          i++;
+        } else if (c == quote) {
+          quote = null;
+        }
+      } else if (c == '/' && next == '/') {
+        lineComment = true;
+        i++;
+      } else if (c == '/' && next == '*') {
+        blockComment = true;
+        i++;
+      } else if (c == '"' || c == "'" || c == '`') {
+        quote = c;
+      } else if (c == ';') {
+        keep(i + 1);
+      }
+    }
+    if (start < script.length) keep(script.length);
+    return out.join('\n');
+  }
+}