Merge pull request 'feat: adds oauth server to docker' (#13) from feature/add-oauth-server into main

Reviewed-on: #13

.env.example

@@ -15,6 +15,12 @@ ADMIN_USERNAME=admin
 ADMIN_PASSWORD=change-me
 SESSION_SECRET=change-me-session-secret
 MCP_TOKEN=change-me-mcp-token
+MCP_AUTH_MODE=oauth
+MCP_RESOURCE_URL=http://localhost:8001/mcp
+MCP_OAUTH_ISSUER=
+MCP_OAUTH_AUDIENCE=
+MCP_OAUTH_JWKS_URL=
+MCP_OAUTH_REQUIRED_SCOPE=mcp:tools
 
 API_PORT=8000
 MCP_PORT=8001

.gitignore

@@ -4,6 +4,7 @@ __pycache__/
 *.py[cod]
 *.db
 .pytest_cache/
+pytest-cache-files-*/
 .coverage
 htmlcov/
 postgres_data/

README.md

@@ -6,7 +6,7 @@
 
 - `api`: FastAPI, REST API, HTML-админка, healthcheck.
 - `worker`: weekly scheduler, который запускает парсинг по `CRAWL_CRON`.
-- `mcp`: HTTP MCP endpoint с bearer token.
+- `mcp`: HTTP MCP endpoint с OAuth/OIDC access token для внешних агентов или legacy static token для локального режима.
 - `postgres`: основная БД.
 
 Парсер использует фиксированный источник сотрудников, по умолчанию `https://miem.hse.ru/persons`. Для каждой карточки сохраняются ФИО, должности, год начала работы, контакты, идентификаторы, вкладки профиля, секции, публикации, курсы, ВКР, JSON-снапшот и сжатый HTML-снапшот. Ссылки обходятся только из меню профиля самого сотрудника (`person-menu`), например `#sci`, `#teaching`, `#main`.
@@ -27,7 +27,13 @@ cp .env.example .env
 - `CRAWL_LIMIT`: опциональный лимит профилей для тестового запуска.
 - `ADMIN_USERNAME`, `ADMIN_PASSWORD`: логин и пароль админки.
 - `SESSION_SECRET`: секрет подписи cookie.
-- `MCP_TOKEN`: bearer token для `/mcp`.
+- `MCP_TOKEN`: статический bearer token для legacy/local режима `MCP_AUTH_MODE=token`.
+- `MCP_AUTH_MODE`: режим авторизации MCP: `oauth` для внешних агентов или `token` для локальной отладки.
+- `MCP_RESOURCE_URL`: публичный URL MCP endpoint, например `https://example.com/mcp`.
+- `MCP_OAUTH_ISSUER`: issuer внешнего OIDC-провайдера.
+- `MCP_OAUTH_AUDIENCE`: ожидаемый `aud` в OAuth access token.
+- `MCP_OAUTH_JWKS_URL`: JWKS endpoint; если не задан, используется `<issuer>/.well-known/jwks.json`.
+- `MCP_OAUTH_REQUIRED_SCOPE`: scope для доступа к MCP tools, по умолчанию `mcp:tools`.
 - `PARSER_USE_PLAYWRIGHT`: включение Playwright-рендера динамических вкладок.
 
 ## Локальный запуск
@@ -82,7 +88,9 @@ curl -X POST http://localhost:8000/api/crawl-runs --cookie "miem_admin_session=.
 
 ## MCP
 
-Endpoint: `POST /mcp`, авторизация `Authorization: Bearer <MCP_TOKEN>`.
+Endpoint: `POST /mcp`, авторизация `Authorization: Bearer <token>`.
+
+Для внешних ИИ-агентов используйте `MCP_AUTH_MODE=oauth`. В этом режиме статический `MCP_TOKEN` не принимается: клиент должен передать OAuth/OIDC access token с нужным scope.
 
 Поддерживаемые tools:
 
@@ -92,7 +100,7 @@ Endpoint: `POST /mcp`, авторизация `Authorization: Bearer <MCP_TOKEN>
 - `list_employee_courses(profile_id_or_url)`
 - `get_crawl_status()`
 
-Пример:
+Пример локального legacy-режима со статическим токеном:
 
 ```bash
 curl http://localhost:8001/mcp \
@@ -101,6 +109,19 @@ curl http://localhost:8001/mcp \
   -d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'
 ```
 
+Для production OAuth/OIDC настройте внешний authorization server и включите режим `oauth`:
+
+```env
+MCP_AUTH_MODE=oauth
+MCP_RESOURCE_URL=https://example.com/mcp
+MCP_OAUTH_ISSUER=https://auth.example.com
+MCP_OAUTH_AUDIENCE=miem-mcp
+MCP_OAUTH_JWKS_URL=https://auth.example.com/.well-known/jwks.json
+MCP_OAUTH_REQUIRED_SCOPE=mcp:tools
+```
+
+MCP server работает как OAuth protected resource: он не выдает токены, а проверяет JWT access token по JWKS, `issuer`, `audience`, сроку действия и scope. Metadata для MCP-клиентов доступна по `GET /.well-known/oauth-protected-resource`.
+
 ## Обслуживание
 
 ```bash
@@ -110,4 +131,4 @@ docker compose exec postgres pg_dump -U miem miem_workers > backup.sql
 docker compose down
 ```
 
-Версия сервиса: `0.2.7`. Админка всегда показывает версии backend и frontend в footer.
+Версия сервиса: `0.3.0`. Админка всегда показывает версии backend и frontend в footer.

app/admin.py

@@ -108,7 +108,7 @@ def directory(
                 "has_email": has_email or "",
                 "sort": sort,
                 "direction": direction,
-                "limit": limit,
+                "limit": page["limit"],
                 "offset": offset,
             },
         },

app/config.py

@@ -1,4 +1,6 @@
 from functools import lru_cache
+from typing import Literal
+
 from pydantic import Field, field_validator
 from pydantic_settings import BaseSettings, SettingsConfigDict
 
@@ -18,6 +20,12 @@ class Settings(BaseSettings):
     admin_password: str = "admin"
     session_secret: str = Field(default="dev-session-secret", min_length=8)
     mcp_token: str = "dev-mcp-token"
+    mcp_auth_mode: Literal["token", "oauth"] = "oauth"
+    mcp_resource_url: str = "http://localhost:8001/mcp"
+    mcp_oauth_issuer: str = ""
+    mcp_oauth_audience: str = ""
+    mcp_oauth_jwks_url: str = ""
+    mcp_oauth_required_scope: str = "mcp:tools"
 
     @field_validator("crawl_limit", mode="before")
     @classmethod
@@ -26,6 +34,14 @@ class Settings(BaseSettings):
             return None
         return value
 
+    def oauth_jwks_url(self) -> str:
+        if self.mcp_oauth_jwks_url:
+            return self.mcp_oauth_jwks_url
+        issuer = self.mcp_oauth_issuer.rstrip("/")
+        if not issuer:
+            return ""
+        return f"{issuer}/.well-known/jwks.json"
+
 
 @lru_cache
 def get_settings() -> Settings:

app/main.py

@@ -4,6 +4,7 @@ from fastapi.staticfiles import StaticFiles
 from app.admin import router as admin_router
 from app.api import router as api_router
 from app.db import init_db
+from app.mcp import metadata_router as mcp_metadata_router
 from app.mcp import router as mcp_router
 from app.version import BACKEND_VERSION
 
@@ -12,6 +13,7 @@ app.mount("/static", StaticFiles(directory="app/static"), name="static")
 app.include_router(api_router)
 app.include_router(admin_router)
 app.include_router(mcp_router)
+app.include_router(mcp_metadata_router)
 
 
 @app.on_event("startup")

app/mcp.py

@@ -7,9 +7,10 @@ from sqlalchemy.orm import Session
 from app.config import Settings, get_settings
 from app.db import get_db
 from app.models import CrawlRun, Employee
-from app.security import require_mcp_token
+from app.security import mcp_protected_resource_metadata, require_mcp_auth
 
 router = APIRouter(prefix="/mcp")
+metadata_router = APIRouter()
 
 
 TOOLS = [
@@ -55,7 +56,7 @@ async def mcp_http(
     db: Session = Depends(get_db),
     settings: Settings = Depends(get_settings),
 ) -> dict:
-    require_mcp_token(request, settings)
+    require_mcp_auth(request, settings)
     payload = await request.json()
     method = payload.get("method")
     request_id = payload.get("id")
@@ -168,3 +169,8 @@ def _run_payload(run: CrawlRun) -> dict:
 
 def _tool_response(data: object) -> dict:
     return {"content": [{"type": "text", "text": json.dumps(data, ensure_ascii=False, default=str)}]}
+
+
+@metadata_router.get("/.well-known/oauth-protected-resource")
+def oauth_protected_resource(settings: Settings = Depends(get_settings)) -> dict:
+    return mcp_protected_resource_metadata(settings)

app/parser/profile.py

@@ -164,6 +164,7 @@ def parse_person_profile(
     header = extract_person_header(soup, normalized_url)
     tabs = extract_person_tabs(soup, normalized_url)
     sections = extract_sections(soup, normalized_url)
+    sections = enrich_sections_from_hse_widgets(session, soup, normalized_url, headers, timeout, sections)
     internal_links = [tab["href"] for tab in tabs if tab.get("href")]
 
     return {
@@ -183,6 +184,25 @@ def parse_person_profile(
     }
 
 
+def enrich_sections_from_hse_widgets(
+    session: Session,
+    soup: BeautifulSoup,
+    source_url: str,
+    headers: dict[str, str],
+    timeout: int,
+    sections: list[dict],
+) -> list[dict]:
+    enriched = list(sections)
+    publications = _load_widget_publications(session, soup, headers, timeout)
+    if publications:
+        enriched = _upsert_publications_section(enriched, publications)
+
+    theses = _load_widget_graduation_theses(session, soup, source_url, headers, timeout)
+    if theses:
+        enriched = _upsert_graduation_theses_section(enriched, theses)
+    return enriched
+
+
 def _render_with_playwright(source_url: str, fallback_html: str) -> str:
     try:
         from playwright.sync_api import sync_playwright
@@ -206,6 +226,89 @@ def _render_with_playwright(source_url: str, fallback_html: str) -> str:
         return fallback_html
 
 
+def _load_widget_publications(session: Session, soup: BeautifulSoup, headers: dict[str, str], timeout: int) -> list[dict]:
+    script = soup.select_one('script[data-widget-name="AuthorSearch"][data-author]')
+    if not script:
+        return []
+    author_id = normalize_ws(script.get("data-author"))
+    if not author_id:
+        return []
+
+    publications = []
+    page_id = 1
+    per_page = 100
+    while page_id <= 20:
+        payload = {
+            "type": "ANY",
+            "filterParams": (
+                f'"acceptLanguage":"ru"|"fullTextPublicEnabled": 1|'
+                f'"pubsAuthor": {author_id}|"widgetName": "AuthorSearch"'
+            ),
+            "paginationParams": {
+                "publsSort": ["TITLE_ASC"],
+                "publsCount": per_page,
+                "pageId": page_id,
+            },
+        }
+        try:
+            response = session.post(
+                "https://publications.hse.ru/api/searchPubs",
+                json=payload,
+                headers=headers,
+                timeout=timeout,
+            )
+            response.raise_for_status()
+            data = response.json()
+        except Exception:
+            return publications
+
+        result = data.get("result") if isinstance(data, dict) else {}
+        items = result.get("items") if isinstance(result, dict) else []
+        if not isinstance(items, list) or not items:
+            break
+        publications.extend(_normalize_publication_item(item) for item in items if isinstance(item, dict))
+
+        total = int(result.get("total") or 0)
+        if not result.get("more") and len(publications) >= total:
+            break
+        page_id += 1
+    return _dedupe_publications(publications)
+
+
+def _load_widget_graduation_theses(
+    session: Session,
+    soup: BeautifulSoup,
+    source_url: str,
+    headers: dict[str, str],
+    timeout: int,
+) -> list[dict]:
+    script = soup.select_one('script[src*="/n/stat/vkr/app.js"][data-person-id]')
+    if not script:
+        return []
+    person_id = normalize_ws(script.get("data-person-id"))
+    api_url = normalize_ws(script.get("data-api-url")) or "/n/vkr/api/"
+    if not person_id:
+        return []
+
+    request_headers = {**headers, "x-portal-language": "ru"}
+    try:
+        response = session.get(
+            urljoin(source_url, api_url),
+            params={"supervisorId": person_id},
+            headers=request_headers,
+            timeout=timeout,
+        )
+        response.raise_for_status()
+        data = response.json()
+    except Exception:
+        return []
+
+    items = data.get("data") if isinstance(data, dict) else []
+    if not isinstance(items, list):
+        return []
+    return [_normalize_vkr_item(item, source_url) for item in items if isinstance(item, dict)]
+
+
 def _collect_between_h2(start_h2: Tag) -> list[Tag | NavigableString | str]:
     nodes = []
     for sibling in start_h2.next_siblings:
@@ -353,6 +456,122 @@ def _parse_vkr_items(nodes: list) -> list[str]:
     return [item for item in dict.fromkeys(items) if item]
 
 
+def _normalize_publication_item(item: dict) -> dict:
+    publication_id = str(item.get("id") or "").strip()
+    title = _html_to_text(item.get("title"))
+    year = item.get("year")
+    publication_type = str(item.get("type") or "").strip() or None
+    description = item.get("description") if isinstance(item.get("description"), dict) else {}
+    short_description = _localized_value(description.get("short")) or _localized_value(description.get("shortLeft"))
+    text = normalize_ws(" ".join(part for part in [title, str(year or ""), short_description] if part))
+    return {
+        "id": publication_id or None,
+        "title": title or publication_id,
+        "year": year,
+        "type": publication_type,
+        "url": f"https://publications.hse.ru/view/{publication_id}" if publication_id else None,
+        "text": text or title or publication_id,
+    }
+
+
+def _normalize_vkr_item(item: dict, source_url: str) -> dict:
+    thesis_id = item.get("id")
+    program = item.get("learnProgram") if isinstance(item.get("learnProgram"), dict) else {}
+    org_unit = item.get("orgUnit") if isinstance(item.get("orgUnit"), dict) else {}
+    supervisors = []
+    for supervisor in item.get("supervisors") or []:
+        if not isinstance(supervisor, dict):
+            continue
+        name = normalize_ws(supervisor.get("name"))
+        url = normalize_ws(supervisor.get("url"))
+        if name or url:
+            supervisors.append({"name": name or url, "url": url or None})
+    return {
+        "id": thesis_id,
+        "student": normalize_ws(item.get("student")),
+        "title": normalize_ws(item.get("title")),
+        "defense_year": item.get("year"),
+        "level": normalize_ws(item.get("level")),
+        "rating": item.get("rating"),
+        "project_url": urljoin(source_url, f"/edu/vkr/{thesis_id}") if thesis_id else None,
+        "program": normalize_ws(program.get("title")),
+        "program_url": urljoin(source_url, program.get("url")) if program.get("url") else None,
+        "org_unit": normalize_ws(org_unit.get("title")),
+        "org_unit_url": urljoin(source_url, org_unit.get("url")) if org_unit.get("url") else None,
+        "supervisors": supervisors,
+        "text": normalize_ws(" ".join(str(part) for part in [item.get("student"), item.get("title"), item.get("year")] if part)),
+    }
+
+
+def _upsert_publications_section(sections: list[dict], publications: list[dict]) -> list[dict]:
+    merged = []
+    inserted = False
+    for section in sections:
+        if section.get("type") != "publications":
+            merged.append(section)
+            continue
+        existing = section.get("publications") or []
+        section = {
+            **section,
+            "publications_count": max(section.get("publications_count") or 0, len(publications)),
+            "publications": _dedupe_publications([*existing, *publications]),
+        }
+        section["items"] = [item["text"] for item in section["publications"] if item.get("text")]
+        merged.append(section)
+        inserted = True
+    if not inserted:
+        merged.append(
+            {
+                "title": "Публикации и исследования",
+                "slug": "publikacii_i_issledovaniya",
+                "type": "publications",
+                "raw_text": "",
+                "paragraphs": [],
+                "items": [item["text"] for item in publications if item.get("text")],
+                "links": [],
+                "publications_count": len(publications),
+                "publications": publications,
+            }
+        )
+    return merged
+
+
+def _upsert_graduation_theses_section(sections: list[dict], theses: list[dict]) -> list[dict]:
+    section = {
+        "title": "Выпускные квалификационные работы студентов НИУ ВШЭ",
+        "slug": "vypusknye_kvalifikacionnye_raboty_studentov_niu_vshe",
+        "type": "graduation_theses",
+        "raw_text": "",
+        "paragraphs": [],
+        "items": [item["text"] for item in theses if item.get("text")],
+        "links": [{"text": item["title"], "url": item["project_url"]} for item in theses if item.get("title") and item.get("project_url")],
+        "theses_count": len(theses),
+        "theses": theses,
+    }
+    return [item for item in sections if item.get("type") != "graduation_theses"] + [section]
+
+
+def _dedupe_publications(items: list[dict]) -> list[dict]:
+    seen = set()
+    unique = []
+    for item in items:
+        key = item.get("id") or item.get("url") or item.get("title")
+        if key and key not in seen:
+            seen.add(key)
+            unique.append(item)
+    return unique
+
+
+def _html_to_text(value: object) -> str:
+    return normalize_ws(BeautifulSoup(str(value or ""), "html.parser").get_text(" ", strip=True))
+
+
+def _localized_value(value: object) -> str:
+    if isinstance(value, dict):
+        return normalize_ws(value.get("ru") or value.get("publ") or value.get("en"))
+    return normalize_ws(str(value or ""))
+
+
 def _slugify(value: str) -> str:
     cleaned = re.sub(r"[^\w\s-]", "", value.lower(), flags=re.UNICODE)
     return re.sub(r"[-\s]+", "_", cleaned).strip("_") or "section"

app/security.py

@@ -3,7 +3,10 @@ import hashlib
 import hmac
 import json
 import time
+from functools import lru_cache
 
+import jwt
+from jwt import PyJWKClient, PyJWTError
 from fastapi import HTTPException, Request, status
 
 from app.config import Settings
@@ -46,7 +49,91 @@ def require_admin(request: Request, settings: Settings) -> str:
     return username
 
 
-def require_mcp_token(request: Request, settings: Settings) -> None:
+def require_mcp_auth(request: Request, settings: Settings) -> None:
     auth = request.headers.get("authorization", "")
-    if not auth.startswith("Bearer ") or not hmac.compare_digest(auth.removeprefix("Bearer ").strip(), settings.mcp_token):
-        raise HTTPException(status_code=status.HTTP_401_UNAUTHORIZED, detail="Invalid MCP token")
+    if not auth.startswith("Bearer "):
+        raise _mcp_unauthorized(settings, "Missing bearer token")
+
+    token = auth.removeprefix("Bearer ").strip()
+    if _mcp_static_token_allowed(settings) and hmac.compare_digest(token, settings.mcp_token):
+        return
+    if _mcp_oauth_allowed(settings):
+        _validate_mcp_oauth_token(token, settings)
+        return
+    raise _mcp_unauthorized(settings, "Invalid MCP token")
+
+
+def require_mcp_token(request: Request, settings: Settings) -> None:
+    require_mcp_auth(request, settings)
+
+
+def mcp_protected_resource_metadata(settings: Settings) -> dict:
+    authorization_servers = [settings.mcp_oauth_issuer.rstrip("/")] if settings.mcp_oauth_issuer else []
+    return {
+        "resource": settings.mcp_resource_url,
+        "authorization_servers": authorization_servers,
+        "bearer_methods_supported": ["header"],
+        "scopes_supported": [settings.mcp_oauth_required_scope],
+        "resource_documentation": settings.mcp_resource_url,
+    }
+
+
+def _mcp_static_token_allowed(settings: Settings) -> bool:
+    return settings.mcp_auth_mode == "token"
+
+
+def _mcp_oauth_allowed(settings: Settings) -> bool:
+    return settings.mcp_auth_mode == "oauth"
+
+
+def _validate_mcp_oauth_token(token: str, settings: Settings) -> None:
+    if not settings.mcp_oauth_issuer or not settings.mcp_oauth_audience or not settings.oauth_jwks_url():
+        raise _mcp_unauthorized(settings, "MCP OAuth is not configured")
+    try:
+        signing_key = _get_mcp_oauth_signing_key(token, settings).key
+        claims = jwt.decode(
+            token,
+            signing_key,
+            algorithms=["RS256", "RS384", "RS512", "ES256", "ES384", "ES512"],
+            audience=settings.mcp_oauth_audience,
+            issuer=settings.mcp_oauth_issuer.rstrip("/"),
+        )
+    except PyJWTError as exc:
+        raise _mcp_unauthorized(settings, "Invalid OAuth access token") from exc
+    if not _claims_have_scope(claims, settings.mcp_oauth_required_scope):
+        raise HTTPException(status_code=status.HTTP_403_FORBIDDEN, detail="Missing required MCP OAuth scope")
+
+
+def _claims_have_scope(claims: dict, required_scope: str) -> bool:
+    scopes: set[str] = set()
+    scope = claims.get("scope")
+    if isinstance(scope, str):
+        scopes.update(scope.split())
+    scp = claims.get("scp")
+    if isinstance(scp, str):
+        scopes.update(scp.split())
+    elif isinstance(scp, list):
+        scopes.update(str(item) for item in scp)
+    return required_scope in scopes
+
+
+@lru_cache(maxsize=16)
+def _get_jwk_client(jwks_url: str) -> PyJWKClient:
+    return PyJWKClient(jwks_url)
+
+
+def _get_mcp_oauth_signing_key(token: str, settings: Settings):
+    return _get_jwk_client(settings.oauth_jwks_url()).get_signing_key_from_jwt(token)
+
+
+def _mcp_unauthorized(settings: Settings, detail: str) -> HTTPException:
+    headers = {}
+    if _mcp_oauth_allowed(settings):
+        headers["WWW-Authenticate"] = f'Bearer resource_metadata="{_mcp_metadata_url(settings)}"'
+    return HTTPException(status_code=status.HTTP_401_UNAUTHORIZED, detail=detail, headers=headers)
+
+
+def _mcp_metadata_url(settings: Settings) -> str:
+    resource_url = settings.mcp_resource_url.rstrip("/")
+    base_url = resource_url[: -len("/mcp")] if resource_url.endswith("/mcp") else resource_url
+    return f"{base_url}/.well-known/oauth-protected-resource"

app/services/admin_data.py

@@ -112,7 +112,7 @@ def list_employees_page(
     limit: int = 50,
     offset: int = 0,
 ) -> dict[str, Any]:
-    limit = max(1, min(limit, 200))
+    limit = limit if limit in {25, 50, 100} else 50
     offset = max(0, offset)
     base_stmt = build_employee_query(
         status=status,
@@ -281,6 +281,8 @@ def _normalize_section(section: Any) -> dict[str, Any]:
         "year_entries": _normalize_year_entries(section.get("year_entries")),
         "publications": _normalize_publications(section.get("publications")),
         "publications_count": section.get("publications_count"),
+        "theses": _normalize_theses(section.get("theses")),
+        "theses_count": section.get("theses_count"),
         "academic_year": section.get("academic_year"),
         "courses": _normalize_courses(section.get("courses")),
         "table": _normalize_table(section.get("table")),
@@ -349,6 +351,35 @@ def _normalize_courses(items: Any) -> list[dict[str, str | None]]:
     return normalized
 
 
+def _normalize_theses(items: Any) -> list[dict[str, Any]]:
+    normalized = []
+    if not isinstance(items, list):
+        return normalized
+    for item in items:
+        if not isinstance(item, dict):
+            continue
+        title = str(item.get("title") or "").strip()
+        student = str(item.get("student") or "").strip()
+        if not title and not student:
+            continue
+        normalized.append(
+            {
+                "id": item.get("id"),
+                "student": student,
+                "title": title,
+                "defense_year": item.get("defense_year") or item.get("year"),
+                "level": str(item.get("level") or "").strip(),
+                "rating": item.get("rating"),
+                "project_url": str(item.get("project_url") or "").strip() or None,
+                "program": str(item.get("program") or "").strip(),
+                "program_url": str(item.get("program_url") or "").strip() or None,
+                "org_unit": str(item.get("org_unit") or "").strip(),
+                "org_unit_url": str(item.get("org_unit_url") or "").strip() or None,
+            }
+        )
+    return normalized
+
+
 def _normalize_table(table: Any) -> dict[str, Any] | None:
     if not isinstance(table, dict):
         return None

app/static/admin.css

@@ -270,6 +270,18 @@
   line-height: 1.55;
 }
 
+.employee-section__meta {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 8px 12px;
+  color: #4b5563;
+  font-size: 13px;
+}
+
+.employee-section__meta-item {
+  line-height: 1.4;
+}
+
 .employee-section__table-wrap {
   overflow-x: auto;
 }

app/templates/directory.html

@@ -33,6 +33,12 @@
       <option value="asc" {% if filters.direction == "asc" %}selected{% endif %}>По возрастанию</option>
       <option value="desc" {% if filters.direction == "desc" %}selected{% endif %}>По убыванию</option>
     </select>
+    <select class="directory__input" name="limit" onchange="this.form.offset.value = 0; this.form.submit()">
+      {% for value in [25, 50, 100] %}
+      <option value="{{ value }}" {% if filters.limit == value %}selected{% endif %}>На странице: {{ value }}</option>
+      {% endfor %}
+    </select>
+    <input type="hidden" name="offset" value="{{ filters.offset }}">
     <button class="button" type="submit">Применить</button>
   </form>
 

app/templates/employee_detail.html

@@ -138,6 +138,34 @@
           </li>
           {% endfor %}
         </ul>
+        {% elif section.type == "graduation_theses" and section.theses %}
+        {% if section.theses_count %}<p class="employee-section__note">Всего: {{ section.theses_count }}</p>{% endif %}
+        <ul class="employee-card__list">
+          {% for thesis in section.theses %}
+          <li class="employee-card__list-item">
+            {% if thesis.student %}<strong>{{ thesis.student }}</strong>{% endif %}
+            {% if thesis.title %}
+            <div class="employee-section__text">
+              {% if thesis.project_url %}
+              <a class="admin__link" href="{{ thesis.project_url }}">{{ thesis.title }}</a>
+              {% else %}
+              {{ thesis.title }}
+              {% endif %}
+            </div>
+            {% endif %}
+            <div class="employee-section__meta">
+              {% if thesis.defense_year %}<span class="employee-section__meta-item">Год защиты: {{ thesis.defense_year }}</span>{% endif %}
+              {% if thesis.level %}<span class="employee-section__meta-item">{{ thesis.level }}</span>{% endif %}
+              {% if thesis.rating is not none %}<span class="employee-section__meta-item">Оценка: {{ thesis.rating }}</span>{% endif %}
+              {% if thesis.program %}
+              <span class="employee-section__meta-item">
+                {% if thesis.program_url %}<a class="admin__link" href="{{ thesis.program_url }}">{{ thesis.program }}</a>{% else %}{{ thesis.program }}{% endif %}
+              </span>
+              {% endif %}
+            </div>
+          </li>
+          {% endfor %}
+        </ul>
         {% elif section.type == "table" and section.table %}
         <div class="employee-section__table-wrap">
           <table class="employee-section__table">
@@ -170,7 +198,7 @@
         {% endif %}
         {% endif %}
 
-        {% if section.links and section.type not in ["courses_by_year"] %}
+        {% if section.links and section.type not in ["courses_by_year", "graduation_theses"] %}
         <div class="employee-section__links">
           {% for link in section.links %}
           <a class="employee-section__link" href="{{ link.url }}">{{ link.text }}</a>

app/version.py

@@ -1,3 +1,3 @@
-APP_VERSION = "0.2.7"
-FRONTEND_VERSION = "0.2.7"
-BACKEND_VERSION = "0.2.7"
+APP_VERSION = "0.3.0"
+FRONTEND_VERSION = "0.3.0"
+BACKEND_VERSION = "0.3.0"

docker-compose.yml

@@ -47,5 +47,31 @@ services:
       postgres:
         condition: service_healthy
 
+  keycloak:
+    image: quay.io/keycloak/keycloak:latest
+    container_name: keycloak
+    restart: unless-stopped
+    environment:
+      KC_DB: postgres
+      KC_DB_URL: jdbc:postgresql://postgres:5432/keycloak
+      KC_DB_USERNAME: keycloak
+      KC_DB_PASSWORD: CHANGE_ME_KEYCLOAK_DB_PASSWORD
+
+      KEYCLOAK_ADMIN: admin
+      KEYCLOAK_ADMIN_PASSWORD: admin-KEYCLOAK-p45sw0rd
+
+      KC_HTTP_ENABLED: true
+      KC_PROXY_HEADERS: xforwarded
+      KC_HOSTNAME: https://oauth.miem.space
+
+      KC_HEALTH_ENABLED: true
+      KC_METRICS_ENABLED: true
+    command: start
+    ports:
+      - "127.0.0.1:8080:8080"
+    depends_on:
+      postgres:
+        condition: service_healthy
+
 volumes:
   postgres_data:

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "miem-workers"
-version = "0.2.6"
+version = "0.3.0"
 description = "MIEM employees parser, admin API, and MCP server"
 requires-python = ">=3.11"
 dependencies = [
@@ -12,6 +12,7 @@ dependencies = [
   "lxml>=5.2.0",
   "psycopg[binary]>=3.2.0",
   "pydantic-settings>=2.4.0",
+  "PyJWT[crypto]>=2.9.0",
   "python-multipart>=0.0.9",
   "requests>=2.32.0",
   "sqlalchemy>=2.0.32",

requirements.txt

@@ -6,6 +6,7 @@ jinja2>=3.1.4
 lxml>=5.2.0
 psycopg[binary]>=3.2.0
 pydantic-settings>=2.4.0
+PyJWT[crypto]>=2.9.0
 python-multipart>=0.0.9
 requests>=2.32.0
 sqlalchemy>=2.0.32

tests/test_admin_data.py

@@ -85,6 +85,19 @@ def test_employee_detail_payload_normalizes_human_readable_sections(db_session):
                     "academic_year": "2025/2026",
                     "courses": [{"title": "Course", "url": "https://example.test/course"}],
                 },
+                {
+                    "title": "ВКР",
+                    "type": "graduation_theses",
+                    "theses_count": 1,
+                    "theses": [
+                        {
+                            "student": "Student Name",
+                            "title": "Thesis title",
+                            "defense_year": 2025,
+                            "project_url": "https://www.hse.ru/edu/vkr/1",
+                        }
+                    ],
+                },
                 {
                     "title": "Fallback",
                     "type": "generic",
@@ -102,7 +115,8 @@ def test_employee_detail_payload_normalizes_human_readable_sections(db_session):
     assert payload["sections"][0]["year_entries"][0]["text"] == "Master degree"
     assert payload["sections"][1]["publications"][0]["title"] == "Paper"
     assert payload["sections"][2]["courses"][0]["title"] == "Course"
-    assert payload["sections"][3]["paragraphs"] == ["Fallback text"]
+    assert payload["sections"][3]["theses"][0]["student"] == "Student Name"
+    assert payload["sections"][4]["paragraphs"] == ["Fallback text"]
 
 
 def test_employee_payloads_tolerate_malformed_current_data(db_session):
@@ -155,6 +169,7 @@ def test_list_employees_page_filters_sorts_and_paginates(db_session):
 
     assert page["total"] == 1
     assert page["employees"][0]["full_name"] == "Alpha"
+    assert page["limit"] == 50
 
 
 def test_stats_payload_uses_latest_run_new_count(db_session):

tests/test_admin_templates.py

@@ -18,6 +18,8 @@ def test_directory_template_is_russian_and_uses_display_dates():
     assert "Сотрудники" in template
     assert "Колонки" in template
     assert "Применить" in template
+    assert "На странице: {{ value }}" in template
+    assert "{% for value in [25, 50, 100] %}" in template
     assert "Найдено:" in template
     assert "employee.first_seen_display" in template
     assert "employee.last_seen_display" in template

tests/test_api_mcp.py

@@ -1,10 +1,15 @@
+import time
 from datetime import datetime, timezone
+from types import SimpleNamespace
 
+import jwt
 from fastapi.testclient import TestClient
+from cryptography.hazmat.primitives.asymmetric import rsa
 from sqlalchemy import create_engine
 from sqlalchemy.orm import sessionmaker
 from sqlalchemy.pool import StaticPool
 
+import app.security as security
 from app.config import Settings, get_settings
 from app.db import Base, get_db
 from app.main import app
@@ -18,7 +23,7 @@ def test_health_returns_versions():
     response = client.get("/api/health")
 
     assert response.status_code == 200
-    assert response.json()["backend_version"] == "0.2.7"
+    assert response.json()["backend_version"] == "0.3.0"
 
 
 def test_mcp_requires_token_and_lists_tools():
@@ -38,7 +43,9 @@ def test_mcp_requires_token_and_lists_tools():
             session.close()
 
     app.dependency_overrides[get_db] = override_db
-    app.dependency_overrides[get_settings] = lambda: Settings(mcp_token="secret", session_secret="session-secret")
+    app.dependency_overrides[get_settings] = lambda: Settings(
+        mcp_auth_mode="token", mcp_token="secret", session_secret="session-secret"
+    )
     client = TestClient(app)
 
     unauthorized = client.post("/mcp", json={"jsonrpc": "2.0", "id": 1, "method": "tools/list", "params": {}})
@@ -88,7 +95,9 @@ def test_mcp_search_employees_returns_matching_employee():
             db.close()
 
     app.dependency_overrides[get_db] = override_db
-    app.dependency_overrides[get_settings] = lambda: Settings(mcp_token="secret", session_secret="session-secret")
+    app.dependency_overrides[get_settings] = lambda: Settings(
+        mcp_auth_mode="token", mcp_token="secret", session_secret="session-secret"
+    )
     client = TestClient(app)
 
     response = client.post(
@@ -108,6 +117,148 @@ def test_mcp_search_employees_returns_matching_employee():
     app.dependency_overrides.clear()
 
 
+def test_mcp_oauth_rejects_static_token():
+    engine = create_engine(
+        "sqlite:///:memory:",
+        connect_args={"check_same_thread": False},
+        poolclass=StaticPool,
+    )
+    Base.metadata.create_all(engine)
+    Session = sessionmaker(bind=engine)
+
+    def override_db():
+        session = Session()
+        try:
+            yield session
+        finally:
+            session.close()
+
+    settings = Settings(
+        mcp_auth_mode="oauth",
+        mcp_token="secret",
+        session_secret="session-secret",
+        mcp_oauth_issuer="https://auth.example.com",
+        mcp_oauth_audience="miem-mcp",
+        mcp_oauth_jwks_url="https://auth.example.com/.well-known/jwks.json",
+    )
+    app.dependency_overrides[get_db] = override_db
+    app.dependency_overrides[get_settings] = lambda: settings
+    client = TestClient(app)
+
+    response = client.post(
+        "/mcp",
+        headers={"Authorization": "Bearer secret"},
+        json={"jsonrpc": "2.0", "id": 1, "method": "tools/list", "params": {}},
+    )
+
+    assert response.status_code == 401
+    assert response.headers["www-authenticate"] == (
+        'Bearer resource_metadata="http://localhost:8001/.well-known/oauth-protected-resource"'
+    )
+
+    app.dependency_overrides.clear()
+
+
+def test_mcp_oauth_missing_auth_returns_metadata_challenge():
+    settings = Settings(
+        mcp_auth_mode="oauth",
+        mcp_resource_url="https://api.example.com/mcp",
+        mcp_oauth_issuer="https://auth.example.com",
+        mcp_oauth_audience="miem-mcp",
+        mcp_oauth_jwks_url="https://auth.example.com/.well-known/jwks.json",
+    )
+    app.dependency_overrides[get_settings] = lambda: settings
+    client = TestClient(app)
+
+    response = client.post("/mcp", json={"jsonrpc": "2.0", "id": 1, "method": "tools/list", "params": {}})
+
+    assert response.status_code == 401
+    assert response.headers["www-authenticate"] == (
+        'Bearer resource_metadata="https://api.example.com/.well-known/oauth-protected-resource"'
+    )
+
+    app.dependency_overrides.clear()
+
+
+def test_mcp_accepts_valid_oauth_jwt(monkeypatch):
+    public_key, token = _oauth_key_and_token()
+    monkeypatch.setattr(security, "_get_mcp_oauth_signing_key", lambda _token, _settings: SimpleNamespace(key=public_key))
+    app.dependency_overrides[get_settings] = lambda: _oauth_settings()
+    client = TestClient(app)
+
+    response = client.post(
+        "/mcp",
+        headers={"Authorization": f"Bearer {token}"},
+        json={"jsonrpc": "2.0", "id": 1, "method": "tools/list", "params": {}},
+    )
+
+    assert response.status_code == 200
+    assert response.json()["result"]["tools"][0]["name"] == "search_employees"
+
+    app.dependency_overrides.clear()
+
+
+def test_mcp_rejects_invalid_oauth_jwts(monkeypatch):
+    public_key, expired_token = _oauth_key_and_token(exp=int(time.time()) - 60)
+    _, wrong_issuer_token = _oauth_key_and_token(issuer="https://other.example.com")
+    _, wrong_audience_token = _oauth_key_and_token(audience="other-audience")
+    _, bad_signature_token = _oauth_key_and_token(public_key=public_key)
+    monkeypatch.setattr(security, "_get_mcp_oauth_signing_key", lambda _token, _settings: SimpleNamespace(key=public_key))
+    app.dependency_overrides[get_settings] = lambda: _oauth_settings()
+    client = TestClient(app)
+
+    for token in [expired_token, wrong_issuer_token, wrong_audience_token, bad_signature_token]:
+        response = client.post(
+            "/mcp",
+            headers={"Authorization": f"Bearer {token}"},
+            json={"jsonrpc": "2.0", "id": 1, "method": "tools/list", "params": {}},
+        )
+
+        assert response.status_code == 401
+
+    app.dependency_overrides.clear()
+
+
+def test_mcp_rejects_oauth_jwt_without_required_scope(monkeypatch):
+    public_key, token = _oauth_key_and_token(scope="profile")
+    monkeypatch.setattr(security, "_get_mcp_oauth_signing_key", lambda _token, _settings: SimpleNamespace(key=public_key))
+    app.dependency_overrides[get_settings] = lambda: _oauth_settings()
+    client = TestClient(app)
+
+    response = client.post(
+        "/mcp",
+        headers={"Authorization": f"Bearer {token}"},
+        json={"jsonrpc": "2.0", "id": 1, "method": "tools/list", "params": {}},
+    )
+
+    assert response.status_code == 403
+
+    app.dependency_overrides.clear()
+
+
+def test_mcp_protected_resource_metadata_uses_settings():
+    settings = Settings(
+        mcp_resource_url="https://api.example.com/mcp",
+        mcp_oauth_issuer="https://auth.example.com/",
+        mcp_oauth_required_scope="mcp:tools",
+    )
+    app.dependency_overrides[get_settings] = lambda: settings
+    client = TestClient(app)
+
+    response = client.get("/.well-known/oauth-protected-resource")
+
+    assert response.status_code == 200
+    assert response.json() == {
+        "resource": "https://api.example.com/mcp",
+        "authorization_servers": ["https://auth.example.com"],
+        "bearer_methods_supported": ["header"],
+        "scopes_supported": ["mcp:tools"],
+        "resource_documentation": "https://api.example.com/mcp",
+    }
+
+    app.dependency_overrides.clear()
+
+
 def test_api_employees_and_stats_require_admin_session():
     engine = create_engine(
         "sqlite:///:memory:",
@@ -157,3 +308,35 @@ def test_api_employees_and_stats_require_admin_session():
     assert stats.json()["new_in_last_run"] == 1
 
     app.dependency_overrides.clear()
+
+
+def _oauth_settings() -> Settings:
+    return Settings(
+        mcp_auth_mode="oauth",
+        mcp_resource_url="https://api.example.com/mcp",
+        mcp_oauth_issuer="https://auth.example.com",
+        mcp_oauth_audience="miem-mcp",
+        mcp_oauth_jwks_url="https://auth.example.com/.well-known/jwks.json",
+        session_secret="session-secret",
+    )
+
+
+def _oauth_key_and_token(
+    *,
+    issuer: str = "https://auth.example.com",
+    audience: str = "miem-mcp",
+    scope: str = "mcp:tools",
+    exp: int | None = None,
+    public_key=None,
+):
+    private_key = rsa.generate_private_key(public_exponent=65537, key_size=2048)
+    claims = {
+        "iss": issuer,
+        "aud": audience,
+        "scope": scope,
+        "sub": "mcp-client",
+        "iat": int(time.time()),
+        "exp": exp or int(time.time()) + 300,
+    }
+    token = jwt.encode(claims, private_key, algorithm="RS256", headers={"kid": "test-key"})
+    return public_key or private_key.public_key(), token

tests/test_config.py

@@ -1,3 +1,6 @@
+import pytest
+from pydantic import ValidationError
+
 from app.config import Settings
 
 
@@ -11,3 +14,8 @@ def test_numeric_crawl_limit_is_parsed():
     settings = Settings(crawl_limit="25")
 
     assert settings.crawl_limit == 25
+
+
+def test_mcp_auth_mode_rejects_oauth_or_token_fallback():
+    with pytest.raises(ValidationError):
+        Settings(mcp_auth_mode="oauth_or_token")

tests/test_employee_detail_template.py

@@ -14,6 +14,8 @@ def test_employee_detail_template_is_human_readable():
     assert "Основная информация" in template
     assert "Контакты" in template
     assert "Разделы профиля" in template
+    assert "graduation_theses" in template
+    assert "Год защиты" in template
     assert "Parser version" not in template
     assert "First seen" not in template
     assert "Last seen" not in template

tests/test_parser.py

@@ -1,9 +1,69 @@
 from bs4 import BeautifulSoup
 
-from app.parser.profile import extract_person_tabs
+from app.parser.profile import enrich_sections_from_hse_widgets, extract_person_tabs
 from app.parser.profile_url import normalize_profile_url, parse_profile_identity
 
 
+class FakeResponse:
+    def __init__(self, payload):
+        self.payload = payload
+
+    def raise_for_status(self):
+        return None
+
+    def json(self):
+        return self.payload
+
+
+class FakeSession:
+    def __init__(self):
+        self.posts = []
+        self.gets = []
+
+    def post(self, url, **kwargs):
+        self.posts.append((url, kwargs))
+        return FakeResponse(
+            {
+                "status": "ok",
+                "result": {
+                    "more": False,
+                    "total": 1,
+                    "items": [
+                        {
+                            "id": "888959076",
+                            "type": "ARTICLE",
+                            "title": "Дублирование пакетов",
+                            "year": 2023,
+                            "description": {"short": {"ru": "Информационные процессы. 2023."}},
+                        }
+                    ],
+                },
+            }
+        )
+
+    def get(self, url, **kwargs):
+        self.gets.append((url, kwargs))
+        return FakeResponse(
+            {
+                "lang": "ru",
+                "success": True,
+                "data": [
+                    {
+                        "id": 1045750164,
+                        "year": 2025,
+                        "level": "Бакалавриат",
+                        "title": "Аппаратно-программный комплекс защиты сети",
+                        "rating": 8,
+                        "student": "Лесняк Владислав Евгеньевич",
+                        "learnProgram": {"title": "Информатика и вычислительная техника", "url": "https://hse.ru/ba/isct/"},
+                        "orgUnit": {"title": "МИЭМ", "url": "https://www.hse.ru/org/url/59315150"},
+                        "supervisors": [{"url": "https://www.hse.ru/org/persons/803294906", "name": "Борисов Сергей Петрович"}],
+                    }
+                ],
+            }
+        )
+
+
 def test_normalize_profile_url_supports_staff_and_org_persons():
     assert normalize_profile_url("/staff/avsergeev#sci") == "https://www.hse.ru/staff/avsergeev"
     assert normalize_profile_url("https://www.hse.ru/org/persons/123/") == "https://www.hse.ru/org/persons/123"
@@ -26,3 +86,34 @@ def test_extract_person_tabs_prefers_person_menu_addition():
 
     assert [tab["title"] for tab in tabs] == ["Домашняя страница", "Публикации"]
     assert tabs[1]["href"] == "https://www.hse.ru/staff/avsergeev#sci"
+
+
+def test_enrich_sections_from_hse_widgets_loads_publications_and_vkr():
+    soup = BeautifulSoup(
+        """
+        <script src="/n/stat/publications/dist-w/publs.js" data-author="568398853" data-widget-name="AuthorSearch"></script>
+        <script src="/n/stat/vkr/app.js" data-api-url="/n/vkr/api/" data-person-id="803294906"></script>
+        """,
+        "html.parser",
+    )
+    session = FakeSession()
+
+    sections = enrich_sections_from_hse_widgets(
+        session,
+        soup,
+        "https://www.hse.ru/org/persons/803294906",
+        {"User-Agent": "test"},
+        10,
+        [],
+    )
+
+    publications = next(section for section in sections if section["type"] == "publications")
+    theses = next(section for section in sections if section["type"] == "graduation_theses")
+
+    assert publications["publications_count"] == 1
+    assert publications["publications"][0]["url"] == "https://publications.hse.ru/view/888959076"
+    assert theses["theses_count"] == 1
+    assert theses["theses"][0]["student"] == "Лесняк Владислав Евгеньевич"
+    assert theses["theses"][0]["project_url"] == "https://www.hse.ru/edu/vkr/1045750164"
+    assert session.posts[0][0] == "https://publications.hse.ru/api/searchPubs"
+    assert session.gets[0][1]["params"] == {"supervisorId": "803294906"}