altrepo: add appstream module

.gitignore

@@ -2,5 +2,6 @@
 .venv
 
 *.db
+/appstream/
 
 __pycache__/
\ No newline at end of file

pyproject.toml

@@ -13,7 +13,8 @@ dependencies = [
     "pydantic (>=2.11.7,<3.0.0)",
     "pydantic-settings (>=2.10.1,<3.0.0)",
     "peewee (>=3.18.2,<4.0.0)",
-    "bs4 (>=0.0.2,<0.0.3)"
+    "bs4 (>=0.0.2,<0.0.3)",
+    "lxml (>=6.0.0,<7.0.0)"
 ]
 
 [project.urls]

src/altrepo/__init__.py

 import aiohttp
 
 from .api import ALTRepoAPI
+from .appstream import ALTRepoAppStream
 from .parser import ALTRepoParser
 
 class ALTRepo:
@@ -8,17 +9,21 @@ class ALTRepo:
     def __init__(self):
         self._session: aiohttp.ClientSession | None = None
         self.api: ALTRepoAPI | None = None
+        self.appstream: ALTRepoAppStream | None = None
         self.parser: ALTRepoParser | None = None
 
-    async def init(self):
+    async def init(
+            self,
+            appstream_dir: str = "appstream"
+    ):
         if self._session is None:
             self._session = aiohttp.ClientSession()
             self.api = ALTRepoAPI(self._session)
+            self.appstream = ALTRepoAppStream(self._session, appstream_dir)
             self.parser = ALTRepoParser(self._session)
     
     async def close(self):
         await self._session.close()
-        
 
 
 altrepo = ALTRepo()

src/altrepo/appstream/__init__.py

+from .methods import ALTRepoAppStream

src/altrepo/appstream/methods.py

+import os
+
+import aiohttp
+from lxml import etree
+
+
+class AppStreamClient:
+    BASE_URL = (
+        "https://git.altlinux.org/gears/a/appstream-data-desktop.git?"
+        "a=blob_plain;f=xmls/altlinux.xml;hb=refs/heads/{}"
+    )
+
+    def __init__(self, session: aiohttp.ClientSession):
+        self.session = session
+
+    async def _handle_response(self, resp: aiohttp.ClientResponse):
+
+        match resp.status:
+            case 200 | 201:
+                return await resp.text(encoding="utf-8")
+            case _:
+                return None
+
+    async def get(self, branch: str) -> str | None:
+        url = self.BASE_URL.format(branch)
+        async with self.session.get(url) as resp:
+            return await self._handle_response(resp)
+
+
+class DataInfo:
+    def __init__(self, client: AppStreamClient, appstream_dir: str):
+        self.client = client
+        self.dir = appstream_dir
+        os.makedirs(self.dir, exist_ok=True)
+
+    async def load_by_branch(self, branch: str, version: str):
+        text = await self.client.get(branch)
+        if not text:
+            return
+
+        old_file = self.get_file_path(branch)
+        if old_file:
+            os.remove(old_file)
+
+        filename = f"{branch}-{version}.xml"
+        filepath = os.path.join(self.dir, filename)
+        with open(filepath, "w", encoding="utf-8") as f:
+            f.write(text)
+
+    def get_file_path(self, branch: str) -> str | None:
+        for fname in os.listdir(self.dir):
+            if fname.startswith(f"{branch}-") and fname.endswith(".xml"):
+                return os.path.join(self.dir, fname)
+        return None
+
+    def get_current_version(self, branch: str) -> str | None:
+        path = self.get_file_path(branch)
+        if path:
+            fname = os.path.basename(path)
+            return fname[len(branch) + 1 : -4]
+        return None
+
+    def has_file(self, branch: str) -> bool:
+        return self.get_file_path(branch) is not None
+
+
+class PackageInfo:
+    def __init__(self, client: AppStreamClient, data: DataInfo):
+        self.client = client
+        self._data = data
+
+    def id_by_pkgname(self, pkgname: str, branch: str) -> str | None:
+        path = self._data.get_file_path(branch)
+        if not path:
+            return None
+
+        context = etree.iterparse(path, events=("end",), recover=True)
+        for _, elem in context:
+            if elem.tag == "component":
+                if elem.findtext("pkgname") == pkgname:
+                    return elem.findtext("id")
+            elem.clear()
+            while elem.getprevious() is not None:
+                del elem.getparent()[0]
+        return None
+
+
+class ALTRepoAppStream:
+    def __init__(self, session: aiohttp.ClientSession, appstream_dir: str):
+        self._client = AppStreamClient(session)
+        self.data = DataInfo(self._client, appstream_dir)
+        self.package = PackageInfo(self._client, self.data)

src/handlers/help.py

@@ -32,6 +32,10 @@ async def info_handler(m: Message) -> None:
         f"    Сопровождающих в базе: {DB.maintainer.count()}\n"
         f"    Пакетов переведено: {DB.package.count()}\n\n"
 
+        f"{_bold("ALTRepo Bot AppStream Data:")}\n"
+        f"    Sisyphus: {altrepo.appstream.data.get_current_version("sisyphus")}\n"
+        f"    P11: {altrepo.appstream.data.get_current_version("p11")}\n\n"
+
         f"{_bold("Исходный код:")}\n"
         f"{repos_formatted}\n\n"
 

src/main.py

@@ -11,12 +11,14 @@ from middlewares import UserMiddleware
 
 from services.test_api_version import test_api_version
 from services.update_maintainers import update_maintainers
+from services.appstream.update_appstream_data import update_appstream_data
 
 bot = Telegrinder(tg_api)
 bot.dispatch.load_from_dir("src/handlers")
 
 bot.on.message.register_middleware(UserMiddleware)
 
+
 @bot.loop_wrapper.lifespan.on_startup
 async def startup():
     db.create_tables([Maintainer, User, Package])
@@ -24,31 +26,43 @@ async def startup():
     await altrepo.init()
     await test_api_version()
     await update_maintainers()
-    await bot.api.set_my_commands(commands=[
-        BotCommand("watch", "Отслеживание по пакетам"),
-        BotCommand("bugs", "Отслеживание по ошибкам"),
-        BotCommand("ftbfs", "Ошибки пересборки"),
-        BotCommand("statistics", "Статистика репозитория"),
-        BotCommand("altrepo_info", "Информация о боте"),
-        BotCommand("help", "Справка")
-    ])
+    await update_appstream_data()
+    await bot.api.set_my_commands(
+        commands=[
+            BotCommand("watch", "Отслеживание по пакетам"),
+            BotCommand("bugs", "Отслеживание по ошибкам"),
+            BotCommand("ftbfs", "Ошибки пересборки"),
+            BotCommand("statistics", "Статистика репозитория"),
+            BotCommand("altrepo_info", "Информация о боте"),
+            BotCommand("help", "Справка"),
+        ]
+    )
+
 
 @bot.loop_wrapper.lifespan.on_shutdown
 async def shutdown():
     logger.info("stopping ALTRepo")
     await altrepo.close()
 
+
 @bot.loop_wrapper.interval(days=1)
 async def api_test():
     await test_api_version()
-    
+
+
 @bot.loop_wrapper.interval(days=1)
 async def update_maintainers_db():
     await update_maintainers()
 
 
+@bot.loop_wrapper.interval(days=1)
+async def update_appstream():
+    await update_appstream_data()
+
+
 @bot.on.error()
 async def error_handler(err: Error[TooManyRequests], m: Message):
     await m.answer(f"Слишком много запросов!")
 
+
 bot.run_forever(skip_updates=True)

src/services/appstream/update_appstream_data.py

+from telegrinder.modules import logger
+
+from altrepo import altrepo
+
+
+async def update_appstream_data():
+
+    for branch in ["sisyphus", "p11"]:
+
+        package = (
+            await altrepo.api.package.package_info(
+                "appstream-data-desktop", branch=branch, source=True
+            )
+        ).packages[0]
+        current_version = altrepo.appstream.data.get_current_version(branch)
+        package_version = f"{package.version}-{package.release}"
+        if (
+            not altrepo.appstream.data.has_file(branch)
+            or package_version != current_version
+        ):
+            await altrepo.appstream.data.load_by_branch(branch, package_version)
+            logger.info(f"Update AppStream data for {branch}")