@robertying/thu-learn-lib 5.0.0 → 6.0.0

package/src/utils.ts

@@ -1,21 +1,61 @@
+import * as cheerio from 'cheerio';
 import { decodeHTML as _decodeHTML } from 'entities';
 
-import type { DetailSelector } from './types.ts';
+import type { ApiError, DetailSelector, Fetch } from './types.ts';
 import { ContentType, FailReason, HomeworkGradeLevel, QuestionnaireType, SemesterType } from './types.ts';
 
-export function parseSemesterType(n: number): SemesterType {
-  if (n === 1) {
-    return SemesterType.FALL;
-  } else if (n === 2) {
-    return SemesterType.SPRING;
-  } else if (n === 3) {
-    return SemesterType.SUMMER;
-  } else {
-    return SemesterType.UNKNOWN;
-  }
+const CHEERIO_CONFIG: cheerio.CheerioOptions = { xml: true };
+
+/**
+ * The error every API rejects with: a plain `{ reason, extra }` object (not an `Error`), so callers can compare
+ * `reason` with `FailReason`. `extra` is left out when not given.
+ */
+export const apiError = (reason: FailReason, extra?: unknown): ApiError =>
+  extra === undefined ? { reason } : { reason, extra };
+
+/** throw `apiError(reason, json)` unless the server's JSON answer says `result: 'success'` */
+export function expectSuccess(json: any, reason: FailReason): void {
+  if (json?.result !== 'success') throw apiError(reason, json);
 }
 
-const CONTENT_TYPE_MK_MAP = {
+/** a DOM element as cheerio's selections hold it (domhandler's `Element`, taken from cheerio's types so it always matches) */
+export type Element = ReturnType<ReturnType<cheerio.CheerioAPI['root']>['children']>[number];
+
+/** load an HTML page or fragment for querying */
+export const $ = (html: Parameters<typeof cheerio.load>[0]): cheerio.CheerioAPI => cheerio.load(html, CHEERIO_CONFIG);
+
+/** what the server writes for "yes" in its flags (and 否 for "no") */
+export const YES = '是';
+
+/** the URL of a `fetch` input */
+export const urlOf = (input: Parameters<Fetch>[0]): string =>
+  typeof input === 'string' ? input : input instanceof URL ? input.href : input.url;
+
+/** pages longer than this are taken as real pages, never as the short login / redirect page of an expired session */
+const MAX_LOGIN_PAGE_LENGTH = 4096;
+
+/**
+ * Whether a 200 page is the short login or redirect page of an expired session rather than the page asked for: short,
+ * and either Web Learning's "login timeout" page (what the server sends for an HTML page of an expired session, in
+ * either website language: title 登录超时, image `log_fail.png`, text 您未登录或登录失效) or a page pointing at the
+ * login timeout page or the SSO login.
+ */
+export const looksLikeLoginPage = (text: string) =>
+  text.length < MAX_LOGIN_PAGE_LENGTH &&
+  /登录超时|您未登录或登录失效|log_fail\.png|login_timeout|id\.tsinghua\.edu\.cn\/do\/off\/ui\/auth\/login|j_spring_security_thauth_roaming_entry/.test(
+    text,
+  );
+
+const SEMESTER_TYPES: Record<number, SemesterType> = {
+  1: SemesterType.FALL,
+  2: SemesterType.SPRING,
+  3: SemesterType.SUMMER,
+};
+
+/** the semester type from the last digit of a semester id (`2024-2025-1`) */
+export const parseSemesterType = (n: number): SemesterType => SEMESTER_TYPES[n] ?? SemesterType.UNKNOWN;
+
+const CONTENT_TYPE_MK_MAP: Record<ContentType, string> = {
   [ContentType.NOTIFICATION]: 'kcgg',
   [ContentType.FILE]: 'kcwj',
   [ContentType.HOMEWORK]: 'kczy',
@@ -25,7 +65,7 @@ const CONTENT_TYPE_MK_MAP = {
 };
 
 export function getMkFromType(type: ContentType): string {
-  return 'mk_' + (CONTENT_TYPE_MK_MAP[type] ?? 'UNKNOWN');
+  return 'mk_' + CONTENT_TYPE_MK_MAP[type];
 }
 
 export function decodeHTML(html: string): string {
@@ -40,6 +80,34 @@ export function decodeHTML(html: string): string {
         : text;
 }
 
+const BEIJING_OFFSET_MS = 8 * 60 * 60 * 1000;
+
+/**
+ * A server time as an ISO 8601 string in Beijing time with its offset, e.g. `2026-11-07T18:28:00+08:00`. The server
+ * sends either a Unix timestamp (a number or a string of digits; milliseconds, or seconds for a 10-digit value such as
+ * a favorite's pin time) or a Beijing wall-clock time without a zone (`2026-11-07 18:28`, `2026-11-07 18:28:00`).
+ * `undefined` for a missing or empty value, or a timestamp that is 0, negative or out of range; a string in any other
+ * shape is returned as it is, so nothing the server sent is lost.
+ */
+export function toServerTime(value: unknown): string | undefined {
+  if (value === null || value === undefined || value === '') return undefined;
+  if (typeof value === 'number' || (typeof value === 'string' && /^\d{10,}$/.test(value))) {
+    const number = Number(value);
+    // 0 (or less) is the server's "no time", e.g. a homework without a late submission deadline
+    if (!Number.isFinite(number) || number <= 0) return undefined;
+    // seconds until the year 5138, milliseconds after 1973
+    const date = new Date((number < 1e11 ? number * 1000 : number) + BEIJING_OFFSET_MS);
+    if (Number.isNaN(date.getTime())) return undefined;
+    return date.toISOString().replace(/(\.\d{3})?Z$/, '+08:00');
+  }
+  if (typeof value !== 'string') return undefined;
+  const wallClock = /^(\d{4}-\d{2}-\d{2})[ T](\d{2}:\d{2})(:\d{2})?(?:\.\d+)?$/.exec(value.trim());
+  return wallClock ? `${wallClock[1]}T${wallClock[2]}${wallClock[3] ?? ':00'}+08:00` : value;
+}
+
+/** `toServerTime` for a time every item has; '' when the server left it out */
+export const toRequiredServerTime = (value: unknown): string => toServerTime(value) ?? '';
+
 export function trimAndDefine(text: string | undefined | null): string | undefined {
   if (text === undefined || text === null) {
     return undefined;
@@ -64,7 +132,7 @@ export const GRADE_LEVEL_MAP = new Map([
   [-66, HomeworkGradeLevel.D_PLUS],
   [-64, HomeworkGradeLevel.D],
   [-65, HomeworkGradeLevel.EXEMPTED_COURSE],
-  [-63, HomeworkGradeLevel.PASS],
+  [-63, HomeworkGradeLevel.P],
   [-62, HomeworkGradeLevel.EX],
   [-61, HomeworkGradeLevel.EXEMPTION],
   [-60, HomeworkGradeLevel.PASS],
@@ -78,14 +146,21 @@ export const GRADE_LEVEL_MAP = new Map([
 
 export const JSONP_EXTRACTOR_NAME = 'thu_learn_lib_jsonp_extractor';
 
-export function extractJSONPResult(jsonp: string): any {
-  // check jsonp format
-  if (!jsonp.startsWith(JSONP_EXTRACTOR_NAME)) {
-    throw FailReason.INVALID_RESPONSE;
+/**
+ * The argument of a `thu_learn_lib_jsonp_extractor(...)` JSONP body, parsed as JSON. The body is never run as code,
+ * so whatever the registrar (or anything between) sends can't execute in the app.
+ */
+export function extractJSONPResult(jsonp: string): unknown {
+  const prefix = `${JSONP_EXTRACTOR_NAME}(`;
+  const body = jsonp.trim().replace(/;$/, '').trimEnd();
+  if (!body.startsWith(prefix) || !body.endsWith(')')) {
+    throw apiError(FailReason.INVALID_RESPONSE, jsonp.slice(0, 200));
+  }
+  try {
+    return JSON.parse(body.slice(prefix.length, -1));
+  } catch (err) {
+    throw apiError(FailReason.INVALID_RESPONSE, { error: String(err), text: jsonp.slice(0, 200) });
   }
-  // evaluate the result
-  // oxlint-disable-next-line typescript/no-implied-eval -- JSONP from the registrar is a JS expression, not JSON
-  return Function(`"use strict";const ${JSONP_EXTRACTOR_NAME}=(s)=>s;return ${jsonp};`)();
 }
 
 export function formatFileSize(size: number): string {
@@ -105,14 +180,7 @@ export const CONTENT_TYPE_MAP = new Map([
   [ContentType.QUESTIONNAIRE, 'KCWJ'],
   // omitted: 课表(KCKB)
 ]);
-export const CONTENT_TYPE_MAP_REVERSE = new Map([
-  [CONTENT_TYPE_MAP.get(ContentType.NOTIFICATION)!, ContentType.NOTIFICATION],
-  [CONTENT_TYPE_MAP.get(ContentType.FILE)!, ContentType.FILE],
-  [CONTENT_TYPE_MAP.get(ContentType.HOMEWORK)!, ContentType.HOMEWORK],
-  [CONTENT_TYPE_MAP.get(ContentType.DISCUSSION)!, ContentType.DISCUSSION],
-  [CONTENT_TYPE_MAP.get(ContentType.QUESTION)!, ContentType.QUESTION],
-  [CONTENT_TYPE_MAP.get(ContentType.QUESTIONNAIRE)!, ContentType.QUESTIONNAIRE],
-]);
+export const CONTENT_TYPE_MAP_REVERSE = new Map([...CONTENT_TYPE_MAP].map(([type, code]) => [code, type]));
 
 export const QNR_TYPE_MAP = new Map([
   ['投票', QuestionnaireType.VOTE],