parser: reuse session for attachments, deduplicate news methods, remove unnecessary async

src/altrepo/parser/methods.py

@@ -17,6 +17,11 @@ class BaseParser:
             resp.raise_for_status()
             return await resp.text(encoding=encoding)
 
+    async def get_bytes(self, url: str) -> bytes:
+        async with self.session.get(url) as resp:
+            resp.raise_for_status()
+            return await resp.read()
+
 
 class NewsInfo:
     def __init__(self, client: BaseParser):
@@ -25,33 +30,28 @@ class NewsInfo:
     async def news_urls(self) -> models.NewsURL:
         return await urls_parser(self.client)
 
-    async def bugs(self) -> models.BugsModel:
-        url = (await self.news_urls()).bugs
+    async def _get_packages(self, branch: str) -> models.PackagesModel | None:
+        url = getattr(await self.news_urls(), branch, None)
         if not url:
             return None
         html = await self.client.get(url, "koi8-r")
-        return await bugs_parser(html, url)
+        return await packages_parser(html, url, self.client)
 
-    async def sisyphus(self) -> models.PackagesModel | None:
-        url = (await self.news_urls()).sisyphus
+    async def bugs(self) -> models.BugsModel | None:
+        url = (await self.news_urls()).bugs
         if not url:
             return None
         html = await self.client.get(url, "koi8-r")
-        return await packages_parser(html, url)
+        return await bugs_parser(html, url)
+
+    async def sisyphus(self) -> models.PackagesModel | None:
+        return await self._get_packages("sisyphus")
 
     async def p11(self) -> models.PackagesModel | None:
-        url = (await self.news_urls()).p11
-        if not url:
-            return None
-        html = await self.client.get(url, "koi8-r")
-        return await packages_parser(html, url)
+        return await self._get_packages("p11")
 
     async def p10(self) -> models.PackagesModel | None:
-        url = (await self.news_urls()).p10
-        if not url:
-            return None
-        html = await self.client.get(url, "koi8-r")
-        return await packages_parser(html, url)
+        return await self._get_packages("p10")
 
 
 class PackagesInfo:
@@ -61,7 +61,7 @@ class PackagesInfo:
     async def ftbfs(self) -> List[models.FTBFSModel]:
         url = "https://git.altlinux.org/beehive/stats/Sisyphus-x86_64/ftbfs-joined"
         text = await self.client.get(url)
-        return await ftbfs_parser(text)
+        return ftbfs_parser(text)
 
     async def watch_by_maintainer(
         self,
@@ -71,7 +71,7 @@ class PackagesInfo:
         url = f"https://watch.altlinux.org/pub/watch/{by_acl}/{maintainer_nickname}.txt"
         try:
             text = await self.client.get(url)
-            return await watch_parser(text)
+            return watch_parser(text)
         except:
             return []
 

src/altrepo/parser/news/bugs.py

@@ -28,7 +28,7 @@ async def bugs_parser(html: str, url: str):
                 current_bug = None
                 description_buffer = ""
 
-            section_name = await _get_bug_section_name(line)
+            section_name = _get_bug_section_name(line)
             continue
 
         bug_match = bug_pattern.match(line)
@@ -64,7 +64,7 @@ async def bugs_parser(html: str, url: str):
     return models.BugsModel(**data)
 
 
-async def _get_bug_section_name(line: str) -> str:
+def _get_bug_section_name(line: str) -> str:
     line = line.lower()
     if "new" in line and "resolved" in line:
         return "quickly_resolved"

src/altrepo/parser/news/packages.py

-import aiohttp
 from bs4 import BeautifulSoup
 import re
 import io
@@ -6,7 +5,7 @@ import gzip
 from .. import models
 
 
-async def packages_parser(html: str, url: str):
+async def packages_parser(html: str, url: str, client=None):
     soup = BeautifulSoup(html, "html.parser")
     pre_tag = soup.find("pre")
     if not pre_tag:
@@ -15,9 +14,11 @@ async def packages_parser(html: str, url: str):
     pre_text = pre_tag.get_text(strip=True)
     if "Было удалено вложение" in pre_text and "attachment" in pre_text:
         attachment_link = pre_tag.find("a", href=True)
-        if attachment_link:
+        if attachment_link and client:
             attachment_url = attachment_link["href"]
-            text = await _fetch(attachment_url)
+            compressed_data = await client.get_bytes(attachment_url)
+            with gzip.GzipFile(fileobj=io.BytesIO(compressed_data)) as gz:
+                text = gz.read().decode('utf-8')
         else:
             return models.PackagesModel(**{"added": [], "removed": [], "updated": [], "url": "none"})
     else:
@@ -71,7 +72,7 @@ async def packages_parser(html: str, url: str):
 
             current_package = {
                 "name": match.group(1),
-                "description": await _clean_description(match.group(2)),
+                "description": _clean_description(match.group(2)),
             }
             seen_changelog = False
             continue
@@ -100,19 +101,8 @@ async def packages_parser(html: str, url: str):
     return models.PackagesModel(**sections)
 
 
-async def _clean_description(desc: str):
+def _clean_description(desc: str):
     desc = desc.strip()
     desc = re.sub(r'\s+', ' ', desc)
     desc = re.sub(r'\[\d+[KMG]?\]', '', desc).strip()
     return desc
-
-async def _fetch(url: str) -> str:
-    async with aiohttp.ClientSession() as session:
-        async with session.get(url) as response:
-            response.raise_for_status()
-            compressed_data = await response.read()
-
-    with gzip.GzipFile(fileobj=io.BytesIO(compressed_data)) as gz:
-        text = gz.read().decode('utf-8')
-
-    return text

src/altrepo/parser/packages/ftbfs.py

 from .. import models
 
-async def ftbfs_parser(text: str):
+def ftbfs_parser(text: str):
     packages = []
     for line in text.strip().splitlines():
         parts = line.split('\t')

src/altrepo/parser/packages/watch.py

 from .. import models
 
 
-async def watch_parser(text: str):
+def watch_parser(text: str):
     return [
         models.WatchByMaintainerModel(
             pkg_name=parts[0],