chuvsu-js 4.1.4 → 4.2.0

package/dist/common/parse.js

@@ -28,7 +28,7 @@ export function parseWeekParity(html) {
     return match[1] === "**" ? "even" : "odd";
 }
 export function parseTeacher(s) {
-    const trimmed = s.trim();
+    const trimmed = s.replace(/\s*\(\s*ДОТ\s*\)\s*$/iu, "").trim();
     if (!trimmed)
         return { name: "" };
     const posMatch = trimmed.match(/^(доц\.|проф\.|ст\.преп\.|ст\. преп\.|преп\.|асс\.|зав\.каф\.)\s*/);

package/dist/tt/parse/audience.js

@@ -3,6 +3,7 @@ import { parseSemesterScheduleWith } from "./full-schedule.js";
 import { parseGroupsString } from "./groups.js";
 import { parseSubstituteForDiv, parseSubstitutionDiv, parseTransferDiv, } from "./overlays.js";
 import { LESSON_TYPE_GLOBAL_RE, LESSON_TYPE_RE, SUBGROUP_ANNOTATION_RE, SUBGROUP_RE, WEEKS_GLOBAL_RE, WEEKS_RE, } from "./patterns.js";
+import { linesAfterSubject, stripDistanceMarker } from "./entry-parts.js";
 const DISTANCE_RE = /дистанционно|ДОТ/i;
 export function parseAudienceInfo(html) {
     const doc = parseHtml(html);
@@ -111,28 +112,19 @@ function parseAudienceSemesterEntry(el) {
     const weeksMatch = cleanText.match(WEEKS_RE);
     const subgroupMatch = cleanText.match(SUBGROUP_RE);
     const weekParity = parseWeekParity(cleanHtml);
-    // Audience entries layout:
-    //   <span blue>SUBJ</span> (TYPE) (WEEKS) <br>TEACHER<br>GROUPS
-    // Teacher is the first line after </span>...<br>, groups is the next line.
-    const afterSubject = cleanHtml.split(/<\/span>/i).slice(1).join("</span>");
-    const parts = afterSubject
-        .split(/<br\s*\/?>/i)
-        .map((p) => p.replace(/<[^>]*>/g, "").trim())
-        .filter((p) => p.length > 0);
-    // parts[0] = " (лк) (1 - 16 нед.) " — trailing metadata; drop tokens that
-    // look like (type)/(weeks)/(N подгруппа). First real text line = teacher.
-    const textLines = [];
-    for (const p of parts) {
-        const cleaned = p
-            .replace(LESSON_TYPE_GLOBAL_RE, "")
-            .replace(WEEKS_GLOBAL_RE, "")
-            .replace(SUBGROUP_ANNOTATION_RE, "")
-            .trim();
-        if (cleaned)
-            textLines.push(cleaned);
-    }
-    const teacherLine = textLines[0] ?? "";
-    const groupsLine = textLines.slice(1).join(" ").trim();
+    // Audience entries: subject line, then teacher line, then group line(s).
+    const parts = linesAfterSubject(cleanHtml, subject);
+    const teacherLine = stripDistanceMarker(parts[0] ?? "");
+    const groupsLine = parts
+        .slice(1)
+        .map(stripDistanceMarker)
+        .map((line) => line
+        .replace(LESSON_TYPE_GLOBAL_RE, "")
+        .replace(WEEKS_GLOBAL_RE, "")
+        .replace(SUBGROUP_ANNOTATION_RE, "")
+        .trim())
+        .filter(Boolean)
+        .join(" ");
     return {
         room: "",
         subject,

package/dist/tt/parse/entry-parts.d.ts

@@ -0,0 +1,7 @@
+export declare function entryHtmlLines(html: string): string[];
+export declare function entryTextLines(html: string): string[];
+export declare function stripHtml(html: string): string;
+export declare function linesAfterSubject(html: string, subject: string): string[];
+export declare function parseEntryRoom(html: string, subject: string): string;
+export declare function stripDistanceMarker(value: string): string;
+export declare function containsGroupCode(value: string): boolean;

package/dist/tt/parse/entry-parts.js

@@ -0,0 +1,49 @@
+const BR_RE = /<br\s*\/?>/i;
+const TAG_RE = /<[^>]*>/g;
+const DISTANCE_RE = /дистанционно|ДОТ/i;
+const GROUP_TOKEN_RE = /^[A-ZА-ЯЁ]{1,}(?:-[A-ZА-ЯЁa-zа-яё0-9]+)+$/u;
+export function entryHtmlLines(html) {
+    return html.split(BR_RE).map((line) => line.trim());
+}
+export function entryTextLines(html) {
+    return entryHtmlLines(html)
+        .map(stripHtml)
+        .filter((line) => line.length > 0);
+}
+export function stripHtml(html) {
+    return html.replace(TAG_RE, " ").replace(/\s+/g, " ").trim();
+}
+export function linesAfterSubject(html, subject) {
+    const lines = entryTextLines(html);
+    const index = lines.findIndex((line) => line.includes(subject));
+    return index < 0 ? [] : lines.slice(index + 1);
+}
+export function parseEntryRoom(html, subject) {
+    const lines = entryHtmlLines(html);
+    const line = lines.find((part) => stripHtml(part).includes(subject));
+    if (!line)
+        return "";
+    const lineText = stripHtml(line);
+    const subjectIndex = lineText.indexOf(subject);
+    if (subjectIndex < 0)
+        return "";
+    const beforeSubject = lineText
+        .slice(0, subjectIndex)
+        .replace(/^\*+\s*/, "")
+        .trim();
+    if (DISTANCE_RE.test(stripHtml(html))) {
+        return "Дистанционно (ДОТ)";
+    }
+    if (!beforeSubject)
+        return "";
+    return beforeSubject;
+}
+export function stripDistanceMarker(value) {
+    return value
+        .replace(/\(\s*ДОТ\s*\)/giu, " ")
+        .replace(/\s+/g, " ")
+        .trim();
+}
+export function containsGroupCode(value) {
+    return value.split(/\s+/).some((token) => GROUP_TOKEN_RE.test(token));
+}

package/dist/tt/parse/full-schedule.js

@@ -2,6 +2,7 @@ import { parseHtml, parseTeacher, parseTime, parseWeekParity, parseWeeks, text,
 import { getLessonNumber } from "../utils/index.js";
 import { parseSubstitutionDiv, parseTransferDiv, } from "./overlays.js";
 import { FLEXIBLE_LESSON_TYPE_RE_I, LESSON_TYPE_RE, SUBGROUP_RE, WEEKS_RE, } from "./patterns.js";
+import { containsGroupCode, linesAfterSubject, parseEntryRoom, stripDistanceMarker, } from "./entry-parts.js";
 const DISTANCE_RE = /дистанционно|ДОТ/i;
 export function parseFullSchedule(html, educationType) {
     const doc = parseHtml(html);
@@ -30,6 +31,13 @@ export function parseSemesterScheduleWith(doc, entryParser) {
         }
         if (!currentDay)
             continue;
+        const selfStudyMarker = row.querySelector('span[style*="color: blue"]');
+        if (selfStudyMarker &&
+            /^День самостоятельной работы$/i.test(text(selfStudyMarker)) &&
+            selfStudyMarker.closest("tr") === row) {
+            currentDay.isSelfStudyDay = true;
+            continue;
+        }
         const timeCell = row.querySelector("td.trf");
         const dataCell = row.querySelector("td.trdata:not(.trf)");
         if (!timeCell || !dataCell)
@@ -98,20 +106,23 @@ function parseSemesterEntry(el) {
         return null;
     const typeMatch = cleanText.match(LESSON_TYPE_RE);
     const weeksMatch = cleanText.match(WEEKS_RE);
-    const roomMatch = cleanHtml.match(/(?:<sup>[^<]*<\/sup>)?([А-Яа-яA-Za-z]-\d+)/);
-    const teacherMatch = cleanHtml.match(/<br\s*\/?>\s*([^<]+?)(?:<br|<\/td|<div|<i|$)/);
+    const room = parseEntryRoom(cleanHtml, subject);
+    const teacherLine = linesAfterSubject(cleanHtml, subject).find((line) => {
+        const candidate = stripDistanceMarker(line);
+        return candidate.length > 0 && !SUBGROUP_RE.test(candidate);
+    }) ?? "";
     const subgroupMatch = cleanText.match(SUBGROUP_RE);
     const weekParity = parseWeekParity(cleanHtml);
     return {
-        room: roomMatch?.[1] ?? "",
+        room,
         subject,
         type: typeMatch?.[1] ?? "",
         weeks: parseWeeks(weeksMatch?.[1] ?? ""),
-        teacher: parseTeacher(teacherMatch?.[1] ?? ""),
+        teacher: parseTeacher(teacherLine),
         groups: [],
         subgroup: subgroupMatch ? parseInt(subgroupMatch[1]) : undefined,
         weekParity,
-        isDistance: DISTANCE_RE.test(cleanText) || DISTANCE_RE.test(roomMatch?.[1] ?? ""),
+        isDistance: DISTANCE_RE.test(cleanText) || DISTANCE_RE.test(room),
         substitutions: substitutions.length > 0 ? substitutions : undefined,
         possibleChanges,
     };
@@ -168,20 +179,16 @@ function parseSessionEntry(td) {
     const subject = subjectEl ? text(subjectEl) : "";
     if (!subject)
         return null;
-    // Room: text before the first <span
-    const roomMatch = fullHtml.match(/^([^<]*?)\s*<span/);
-    const room = roomMatch ? roomMatch[1].trim() : "";
+    const room = parseEntryRoom(fullHtml, subject);
     // Type: parenthesized text after </span>, case-insensitive
     const typeMatch = plainText.match(FLEXIBLE_LESSON_TYPE_RE_I);
     const type = typeMatch ? typeMatch[1].replace(/\.$/, "").toLowerCase() : "";
     const subgroupMatch = plainText.match(SUBGROUP_RE);
-    const parts = fullHtml
-        .split(/<br\s*\/?>/i)
-        .map((part) => part.replace(/<[^>]*>/g, "").trim())
-        .filter((part) => part.length > 0);
-    const teacherPart = parts.find((part) => !part.includes(subject) &&
+    const parts = linesAfterSubject(fullHtml, subject);
+    const teacherPart = parts.find((part) => stripDistanceMarker(part).length > 0 &&
         !/^\d{2}:\d{2}\s*-\s*\d{2}:\d{2}$/.test(part) &&
-        !SUBGROUP_RE.test(part)) ?? "";
+        !SUBGROUP_RE.test(part) &&
+        !containsGroupCode(stripDistanceMarker(part))) ?? "";
     // Time: after <br>, format HH:MM - HH:MM
     const timeMatch = fullHtml.match(/<br\s*\/?>\s*(\d{2}:\d{2})\s*-\s*(\d{2}:\d{2})/);
     if (!timeMatch)

package/dist/tt/parse/patterns.d.ts

@@ -1,5 +1,5 @@
-export declare const LESSON_TYPE_PATTERN = "\u043B\u043A|\u043F\u0440|\u043B\u0431|\u0437\u0430\u0447\u043E|\u0437\u0430\u0447|\u044D\u043A\u0437|\u043A\u043E\u043D\u0441|\u043A\u043F";
-export declare const FLEXIBLE_LESSON_TYPE_PATTERN = "(?:\u043B\u043A|\u043F\u0440|\u043B\u0431|\u0437\u0430\u0447\u043E|\u0437\u0430\u0447|\u044D\u043A\u0437|\u043A\u043E\u043D\u0441|\u043A\u043F)\\.?|\u042D\u043A\u0437";
+export declare const LESSON_TYPE_PATTERN = "\u043B\u043A|\u043F\u0440|\u043B\u0431|\u0437\u0430\u0447\u043E|\u0437\u0430\u0447|\u044D\u043A\u0437|\u043A\u043E\u043D\u0441|\u043A\u043F|\u0438\u0437|\u0433\u0437|\u043A\u0440\u043F";
+export declare const FLEXIBLE_LESSON_TYPE_PATTERN = "(?:\u043B\u043A|\u043F\u0440|\u043B\u0431|\u0437\u0430\u0447\u043E|\u0437\u0430\u0447|\u044D\u043A\u0437|\u043A\u043E\u043D\u0441|\u043A\u043F|\u0438\u0437|\u0433\u0437|\u043A\u0440\u043F)\\.?|\u042D\u043A\u0437";
 export declare const LESSON_TYPE_RE: RegExp;
 export declare const LESSON_TYPE_RE_I: RegExp;
 export declare const FLEXIBLE_LESSON_TYPE_RE_I: RegExp;

package/dist/tt/parse/patterns.js

@@ -1,9 +1,9 @@
-export const LESSON_TYPE_PATTERN = "лк|пр|лб|зачо|зач|экз|конс|кп";
+export const LESSON_TYPE_PATTERN = "лк|пр|лб|зачо|зач|экз|конс|кп|из|гз|крп";
 export const FLEXIBLE_LESSON_TYPE_PATTERN = `(?:${LESSON_TYPE_PATTERN})\\.?|Экз`;
-export const LESSON_TYPE_RE = new RegExp(`\\((${LESSON_TYPE_PATTERN})\\)`);
+export const LESSON_TYPE_RE = new RegExp(`\\((${LESSON_TYPE_PATTERN})\\)`, "i");
 export const LESSON_TYPE_RE_I = new RegExp(`\\((${LESSON_TYPE_PATTERN})\\)`, "i");
 export const FLEXIBLE_LESSON_TYPE_RE_I = new RegExp(`\\((${FLEXIBLE_LESSON_TYPE_PATTERN})\\)`, "i");
-export const LESSON_TYPE_GLOBAL_RE = new RegExp(`\\((${LESSON_TYPE_PATTERN})\\)`, "g");
+export const LESSON_TYPE_GLOBAL_RE = new RegExp(`\\((${LESSON_TYPE_PATTERN})\\)`, "gi");
 export const WEEKS_RE = /\(([^)]*нед\.?[^)]*)\)/;
 export const WEEKS_GLOBAL_RE = /\([^)]*нед\.?[^)]*\)/g;
 export const SUBGROUP_RE = /(\d+)\s*подгруппа/;

package/dist/tt/parse/teacher.js

@@ -4,6 +4,7 @@ import { parseSemesterScheduleWith } from "./full-schedule.js";
 import { parseGroupsString } from "./groups.js";
 import { parseSubstituteForDiv, parseSubstitutionDiv, parseTransferDiv, } from "./overlays.js";
 import { FLEXIBLE_LESSON_TYPE_RE_I, LESSON_TYPE_RE, SUBGROUP_RE, WEEKS_RE, } from "./patterns.js";
+import { containsGroupCode, linesAfterSubject, parseEntryRoom, stripDistanceMarker, } from "./entry-parts.js";
 const DISTANCE_RE = /дистанционно|ДОТ/i;
 export function parseTeacherFullSchedule(html, educationType) {
     const doc = parseHtml(html);
@@ -56,20 +57,20 @@ function parseTeacherSemesterEntry(el) {
         return null;
     const typeMatch = cleanText.match(LESSON_TYPE_RE);
     const weeksMatch = cleanText.match(WEEKS_RE);
-    const roomMatch = cleanHtml.match(/(?:<sup>[^<]*<\/sup>)?([А-Яа-яA-Za-z]-\d+)/);
-    const groupsMatch = cleanHtml.match(/<br\s*\/?>\s*([^<]+?)(?:<br|<\/td|<div|<i|$)/);
+    const room = parseEntryRoom(cleanHtml, subject);
+    const groupsLine = linesAfterSubject(cleanHtml, subject).find((line) => containsGroupCode(stripDistanceMarker(line))) ?? "";
     const subgroupMatch = cleanText.match(SUBGROUP_RE);
     const weekParity = parseWeekParity(cleanHtml);
     return {
-        room: roomMatch?.[1] ?? "",
+        room,
         subject,
         type: typeMatch?.[1] ?? "",
         weeks: parseWeeks(weeksMatch?.[1] ?? ""),
         teacher: { name: "" },
-        groups: parseGroupsString(groupsMatch?.[1]),
+        groups: parseGroupsString(stripDistanceMarker(groupsLine)),
         subgroup: subgroupMatch ? parseInt(subgroupMatch[1]) : undefined,
         weekParity,
-        isDistance: DISTANCE_RE.test(cleanText) || DISTANCE_RE.test(roomMatch?.[1] ?? ""),
+        isDistance: DISTANCE_RE.test(cleanText) || DISTANCE_RE.test(room),
         substitutions: substitutions.length > 0 ? substitutions : undefined,
         possibleChanges,
     };
@@ -121,20 +122,17 @@ function parseTeacherSessionEntry(td) {
     const subject = subjectEl ? text(subjectEl) : "";
     if (!subject)
         return null;
-    const roomMatch = fullHtml.match(/^([^<]*?)\s*<span/);
-    const room = roomMatch ? roomMatch[1].trim() : "";
+    const room = parseEntryRoom(fullHtml, subject);
     const typeMatch = plainText.match(FLEXIBLE_LESSON_TYPE_RE_I);
     const type = typeMatch ? typeMatch[1].replace(/\.$/, "").toLowerCase() : "";
     const subgroupMatch = plainText.match(SUBGROUP_RE);
     const timeMatch = fullHtml.match(/<br\s*\/?>\s*(\d{2}:\d{2})\s*-\s*(\d{2}:\d{2})/);
     if (!timeMatch)
         return null;
-    const parts = fullHtml
-        .split(/<br\s*\/?>/i)
-        .map((part) => part.replace(/<[^>]*>/g, "").trim())
-        .filter((part) => part.length > 0);
-    const groupsPart = parts.find((part) => !part.includes(subject) &&
-        !/^\d{2}:\d{2}\s*-\s*\d{2}:\d{2}$/.test(part)) ?? "";
+    const parts = linesAfterSubject(fullHtml, subject);
+    const groupsPart = parts.find((part) => stripDistanceMarker(part).length > 0 &&
+        !/^\d{2}:\d{2}\s*-\s*\d{2}:\d{2}$/.test(part) &&
+        containsGroupCode(stripDistanceMarker(part))) ?? "";
     return {
         entry: {
             room,
@@ -142,7 +140,7 @@ function parseTeacherSessionEntry(td) {
             type,
             weeks: { from: 0, to: 0 },
             teacher: { name: "" },
-            groups: parseGroupsString(groupsPart),
+            groups: parseGroupsString(stripDistanceMarker(groupsPart)),
             subgroup: subgroupMatch ? parseInt(subgroupMatch[1]) : undefined,
             isDistance: DISTANCE_RE.test(plainText) || DISTANCE_RE.test(room),
             possibleChanges,

package/dist/tt/schedule.js

@@ -135,7 +135,7 @@ export class Schedule {
                 year: this.getSemesterYear(semesterPeriod),
                 date,
             });
-            if (week >= 0 && week <= 17) {
+            if (week >= 1 && week <= 17) {
                 const weekday = date.getDay();
                 const slots = this.getSlotsForWeekday(weekday, semesterDays, {
                     subgroup: opts?.subgroup,

package/dist/tt/types.d.ts

@@ -100,6 +100,8 @@ export interface FullScheduleSlot {
 export interface FullScheduleDay {
     weekday: string;
     date?: Date;
+    /** True when portal marks this weekday as a self-study day. */
+    isSelfStudyDay?: boolean;
     slots: FullScheduleSlot[];
 }
 export interface LessonTimeSlot {

package/dist/tt/utils/semester.d.ts

@@ -14,7 +14,7 @@ export declare function getSemesterStart(opts: {
 }): Date;
 /**
  * All weeks in a semester with their start/end dates.
- * Week 0 starts from the semester start date.
+ * Week 1 is the calendar week containing the semester start date.
  */
 export declare function getSemesterWeeks(opts: {
     period: Period;

package/dist/tt/utils/semester.js

@@ -36,16 +36,16 @@ export function getSemesterStart(opts) {
 }
 /**
  * All weeks in a semester with their start/end dates.
- * Week 0 starts from the semester start date.
+ * Week 1 is the calendar week containing the semester start date.
  */
 export function getSemesterWeeks(opts) {
     const weekCount = opts.weekCount ?? 17;
     const semesterStart = getSemesterStart(opts);
     const startMonday = getMonday(semesterStart);
     const weeks = [];
-    for (let i = 0; i <= weekCount; i++) {
+    for (let i = 1; i <= weekCount; i++) {
         const start = new Date(startMonday);
-        start.setDate(startMonday.getDate() + i * 7);
+        start.setDate(startMonday.getDate() + (i - 1) * 7);
         const end = new Date(start);
         end.setDate(start.getDate() + 6);
         end.setHours(23, 59, 59, 999);
@@ -60,5 +60,5 @@ export function getWeekNumber(opts) {
     const startMonday = getMonday(semesterStart);
     const targetMonday = getMonday(date);
     const diff = targetMonday.getTime() - startMonday.getTime();
-    return Math.floor(diff / (7 * 24 * 60 * 60 * 1000));
+    return Math.floor(diff / (7 * 24 * 60 * 60 * 1000)) + 1;
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "chuvsu-js",
-  "version": "4.1.4",
+  "version": "4.2.0",
   "description": "Node.js library for ChuvSU student portal (lk.chuvsu.ru) and schedule (tt.chuvsu.ru)",
   "type": "module",
   "main": "dist/index.js",
@@ -43,6 +43,9 @@
   "scripts": {
     "clean": "node -e \"require('fs').rmSync('dist',{recursive:true,force:true})\"",
     "build": "pnpm clean && tsc",
-    "test": "pnpm build && node --test"
+    "test": "pnpm build && node --test",
+    "test:live:schedules": "pnpm build && node --env-file=.env utils/testSchedules.mjs",
+    "fixtures:audit": "pnpm build && node utils/auditScheduleExpectations.mjs",
+    "fixtures:collect": "pnpm build && node --env-file=.env utils/collectScheduleFixtures.mjs"
   }
 }