From c4b5b7f2c2c614417117c2cf06e9aeae5059608c Mon Sep 17 00:00:00 2001
From: Aleksandr Meshchriakov <avm@sh-inc.ru>
Date: Fri, 20 Mar 2026 11:43:12 +0100
Subject: [PATCH] fix(parsers): support proxy fallback from env for external
 sources

---
 .env.prod.example         |  3 +++
 src/apps/parsers/tasks.py | 48 +++++++++++++++++++++++----------------
 src/settings/base.py      |  4 +++-
 3 files changed, 34 insertions(+), 21 deletions(-)

diff --git a/.env.prod.example b/.env.prod.example
index be3eecf..fd06fa3 100644
--- a/.env.prod.example
+++ b/.env.prod.example
@@ -25,6 +25,9 @@ CELERY_WORKER_CONCURRENCY=4
 # Parsers API keys
 CHECKO_API_KEY=CHANGE_ME_CHECKO_API_KEY
 ZAKUPKI_TOKEN=CHANGE_ME_ZAKUPKI_TOKEN
+# Optional: comma-separated HTTP(S) proxies for parser tasks
+# Example: PARSER_PROXIES=http://user:pass@proxy1:8080,http://user:pass@proxy2:8080
+PARSER_PROXIES=
 
 # 1 to collect static files during migrate service, 0 to skip
 COLLECTSTATIC_ON_MIGRATE=1
diff --git a/src/apps/parsers/tasks.py b/src/apps/parsers/tasks.py
index 39f7432..436d49b 100644
--- a/src/apps/parsers/tasks.py
+++ b/src/apps/parsers/tasks.py
@@ -32,6 +32,7 @@ from apps.parsers.services import (
     ProxyService,
 )
 from celery import shared_task
+from django.conf import settings
 from requests.adapters import BaseAdapter
 
 logger = logging.getLogger(__name__)
@@ -41,6 +42,26 @@ DEFAULT_START_YEAR = 2025
 DEFAULT_START_MONTH = 1
 
 
+def _resolve_proxies(proxies: list[str] | None) -> list[str] | None:
+    """
+    Разрешить итоговый список прокси.
+
+    Приоритет:
+    1. Явно переданные в задачу `proxies`
+    2. Активные прокси из БД
+    3. `settings.PARSER_PROXIES` (например, из ENV)
+    """
+    if proxies is not None:
+        return proxies
+
+    db_proxies = ProxyService.get_active_proxies_or_none()
+    if db_proxies:
+        return db_proxies
+
+    configured_proxies = getattr(settings, "PARSER_PROXIES", []) or []
+    return configured_proxies or None
+
+
 def _get_or_create_background_job(
     *,
     task_id: str,
@@ -301,9 +322,7 @@ def parse_industrial_production(
     )
     task_id = self.request.id or str(uuid.uuid4())
 
-    # Если прокси не переданы, берём из БД
-    if proxies is None:
-        proxies = ProxyService.get_active_proxies_or_none()
+    proxies = _resolve_proxies(proxies)
 
     logger.info(
         "Starting industrial production parsing (task_id=%s, batch_id=%d, proxies=%d)",
@@ -395,9 +414,7 @@ def parse_manufactures(
     )
     task_id = self.request.id or str(uuid.uuid4())
 
-    # Если прокси не переданы, берём из БД
-    if proxies is None:
-        proxies = ProxyService.get_active_proxies_or_none()
+    proxies = _resolve_proxies(proxies)
 
     logger.info(
         "Starting manufactures parsing (task_id=%s, batch_id=%d, proxies=%d)",
@@ -488,8 +505,7 @@ def parse_industrial_products(
     )
     task_id = self.request.id or str(uuid.uuid4())
 
-    if proxies is None:
-        proxies = ProxyService.get_active_proxies_or_none()
+    proxies = _resolve_proxies(proxies)
 
     logger.info(
         "Starting industrial products parsing (task_id=%s, batch_id=%d, proxies=%d)",
@@ -634,9 +650,7 @@ def parse_inspections(
     )
     task_id = self.request.id or str(uuid.uuid4())
 
-    # Если прокси не переданы, берём из БД
-    if proxies is None:
-        proxies = ProxyService.get_active_proxies_or_none()
+    proxies = _resolve_proxies(proxies)
 
     logger.info(
         "Starting inspections parsing (task_id=%s, batch_id=%d, year=%s, month=%s, proxies=%d)",
@@ -828,9 +842,7 @@ def sync_inspections(  # noqa: C901
     )
     task_id = self.request.id or str(uuid.uuid4())
 
-    # Если прокси не переданы, берём из БД
-    if proxies is None:
-        proxies = ProxyService.get_active_proxies_or_none()
+    proxies = _resolve_proxies(proxies)
 
     logger.info(
         "Starting inspections sync (task_id=%s, batch_id=%d)", task_id, batch_id
@@ -1039,9 +1051,7 @@ def parse_procurements(
     )
     task_id = self.request.id or str(uuid.uuid4())
 
-    # Если прокси не переданы, берём из БД
-    if proxies is None:
-        proxies = ProxyService.get_active_proxies_or_none()
+    proxies = _resolve_proxies(proxies)
 
     logger.info(
         "Starting procurements parsing "
@@ -1174,9 +1184,7 @@ def sync_procurements(  # noqa: C901
     )
     task_id = self.request.id or str(uuid.uuid4())
 
-    # Если прокси не переданы, берём из БД
-    if proxies is None:
-        proxies = ProxyService.get_active_proxies_or_none()
+    proxies = _resolve_proxies(proxies)
 
     logger.info(
         "Starting procurements sync (task_id=%s, batch_id=%d, region=%s, law=%s-FZ)",
diff --git a/src/settings/base.py b/src/settings/base.py
index 7da16a3..a2cf6c9 100644
--- a/src/settings/base.py
+++ b/src/settings/base.py
@@ -197,7 +197,9 @@ WSGI_APPLICATION = "core.wsgi.application"
 
 ZAKUPKI_TOKEN = os.getenv("ZAKUPKI_TOKEN", "")
 FNS_LOCK_TTL_SECONDS = 3600
-PARSER_PROXIES = []
+PARSER_PROXIES = [
+    item.strip() for item in os.getenv("PARSER_PROXIES", "").split(",") if item.strip()
+]
 BACKUP_ENCRYPTION_KEY = os.getenv("BACKUP_ENCRYPTION_KEY", "")
 BACKUP_KEY_ID = os.getenv("BACKUP_KEY_ID", "default")
 BACKUP_EXPORT_DIRECTORY = os.getenv(