@@ -127,6 +131,7 @@
+
diff --git a/src/vk_parser_app/text_utils.py b/src/vk_parser_app/text_utils.py
index a06128c..535ca55 100644
--- a/src/vk_parser_app/text_utils.py
+++ b/src/vk_parser_app/text_utils.py
@@ -23,8 +23,24 @@ def strip_trailing_hashtag_line(text: str) -> str:
return "\n".join(lines).rstrip()
-def publication_hashtags(category_tag: str, source_tag: str) -> str:
+def parse_hash_tags(value: object) -> list[str]:
+ if isinstance(value, list):
+ raw_items = [str(item) for item in value]
+ else:
+ raw_items = re.split(r"[\s,|]+", str(value or ""))
+ tags: list[str] = []
+ seen: set[str] = set()
+ for item in raw_items:
+ tag = normalize_hash_tag(item, "")
+ if tag and tag not in seen:
+ tags.append(tag)
+ seen.add(tag)
+ return tags
+
+
+def publication_hashtags(category_tag: str, source_tag: str, common_tags: object = "") -> str:
tags = [
+ *parse_hash_tags(common_tags),
normalize_hash_tag(category_tag, "category"),
normalize_hash_tag(source_tag, "source"),
]
@@ -38,6 +54,7 @@ def build_publication_text(
text: str,
category_tag: str,
source_tag: str,
+ common_tags: object = "",
format_title: bool = False,
parse_mode: str | None = None,
) -> str:
@@ -59,7 +76,7 @@ def build_publication_text(
break
if title_idx is None:
- return publication_hashtags(category_tag, source_tag)
+ return publication_hashtags(category_tag, source_tag, common_tags)
formatted_lines = []
for idx, line in enumerate(cleaned_lines):
@@ -84,7 +101,7 @@ def build_publication_text(
raw_body = "\n".join(formatted_lines)
# Collapse multiple consecutive blank lines to at most two newlines (\n\n)
normalized_body = re.sub(r"\n{3,}", "\n\n", raw_body).strip()
- hashtags = publication_hashtags(category_tag, source_tag)
+ hashtags = publication_hashtags(category_tag, source_tag, common_tags)
if hashtags:
return f"{normalized_body}\n\n{hashtags}" if normalized_body else hashtags
return normalized_body
diff --git a/src/vk_parser_app/workers/ai_writer.py b/src/vk_parser_app/workers/ai_writer.py
index 47f3de1..a5cf43d 100644
--- a/src/vk_parser_app/workers/ai_writer.py
+++ b/src/vk_parser_app/workers/ai_writer.py
@@ -16,7 +16,7 @@ from ..constants import WORKER_AI_WRITER
from ..db import fetch_bool_setting, fetch_float_setting, fetch_int_setting, fetch_setting, get_pool
from ..heartbeat import HeartbeatReporter
from ..jobs import is_worker_enabled
-from ..text_utils import build_publication_text, normalize_hash_tag, parse_categories
+from ..text_utils import normalize_hash_tag, parse_categories, strip_trailing_hashtag_line
from .ai_alerts import send_ai_worker_error_alert
@@ -53,7 +53,7 @@ OUTPUT SCHEMA (return array matching input order):
Rules:
- Input is a JSON object with key "posts" containing accepted posts.
-- Input has key "categories" with objects: id, name, tag.
+- Input has key "categories" with objects: id, name, tag, description.
- Each post includes producer_name and producer_tag. producer_name can be a manufacturer, shop, or publishing source.
- The "text" field for every rewrite must explicitly mention producer_name at least once. Do not satisfy this only through producer_tag, hashtags, notes, or metadata.
- Pick exactly one category from the categories list in input and return its numeric id as category_id.
@@ -208,7 +208,7 @@ async def load_writer_categories(pool) -> list[dict[str, Any]]:
try:
rows = await pool.fetch(
"""
- SELECT sort_order AS id, name, tag
+ SELECT sort_order AS id, name, tag, COALESCE(description, '') AS description
FROM content_categories
WHERE is_active=TRUE
ORDER BY sort_order, name
@@ -216,7 +216,15 @@ async def load_writer_categories(pool) -> list[dict[str, Any]]:
)
except Exception:
rows = []
- categories = [{"id": int(row["id"]), "name": str(row["name"]), "tag": str(row["tag"])} for row in rows]
+ categories = [
+ {
+ "id": int(row["id"]),
+ "name": str(row["name"]),
+ "tag": str(row["tag"]),
+ "description": str(row["description"] or ""),
+ }
+ for row in rows
+ ]
if categories:
return categories
legacy = parse_categories(await fetch_setting("ai_writer_categories", []))
@@ -399,7 +407,7 @@ class AIWriterWorker:
async def apply_rewrites(self, batch_id: int, rewrites: list[dict], model: str, prompt: str) -> None:
for item in rewrites:
- final_text = build_publication_text(item["text"], item["category_tag"], item["source_tag"])
+ final_text = strip_trailing_hashtag_line(item["text"])
editorial_status = "rejected" if item.get("reject_by_category") else "review"
editor_notes = "Отклонено AI: Не целевой контент" if item.get("reject_by_category") else None
await self.pool.execute(
diff --git a/src/vk_parser_app/workers/max_poster.py b/src/vk_parser_app/workers/max_poster.py
index 8882944..6661930 100644
--- a/src/vk_parser_app/workers/max_poster.py
+++ b/src/vk_parser_app/workers/max_poster.py
@@ -218,6 +218,7 @@ class MAXPoster:
self.auto_reaction = "👍"
self.reaction_path_template = "/messages/{message_id}/reactions"
self.dry_run = False
+ self.common_tags = ""
async def init(self) -> None:
self.pool = await get_pool()
@@ -243,6 +244,7 @@ class MAXPoster:
self.auto_reaction = str(await fetch_setting("max_poster_auto_reaction", "👍") or "👍").strip() or "👍"
self.reaction_path_template = str(await fetch_setting("max_poster_reaction_path_template", self.reaction_path_template) or "").strip()
self.dry_run = await fetch_bool_setting("max_poster_dry_run", False)
+ self.common_tags = str(await fetch_setting("publication_common_tags", "") or "").strip()
if not self.token:
raise RuntimeError("max_poster_bot_token is empty")
if not self.chat_id:
@@ -371,7 +373,12 @@ class MAXPoster:
def build_text(self, post: dict[str, Any]) -> str:
category_tag = str(post.get("final_category_tag") or post.get("rewrite_category_tag") or post.get("final_category") or post.get("rewrite_category") or "")
source_tag = str(post.get("final_source_tag") or post.get("rewrite_source_tag") or post.get("source_tag") or "")
- return build_publication_text(post.get("final_text") or post.get("rewritten_text") or "", category_tag, source_tag)
+ return build_publication_text(
+ post.get("final_text") or post.get("rewritten_text") or "",
+ category_tag,
+ source_tag,
+ self.common_tags,
+ )
async def download_media_to_temp(self, session: aiohttp.ClientSession, item: dict[str, Any]) -> Path | None:
media_type = str(item.get("media_type") or "")
diff --git a/src/vk_parser_app/workers/tg_poster.py b/src/vk_parser_app/workers/tg_poster.py
index 985c091..d83ae27 100644
--- a/src/vk_parser_app/workers/tg_poster.py
+++ b/src/vk_parser_app/workers/tg_poster.py
@@ -111,6 +111,7 @@ class TelegramPoster:
self.recent_window = 20
self.category_repeat_penalty = 3.0
self.source_repeat_penalty = 4.0
+ self.common_tags = ""
async def init(self) -> None:
self.pool = await get_pool()
@@ -137,6 +138,7 @@ class TelegramPoster:
self.recent_window = max(1, await fetch_int_setting("tg_poster_recent_window", 20))
self.category_repeat_penalty = max(0.0, await fetch_float_setting("tg_poster_category_repeat_penalty", 3.0))
self.source_repeat_penalty = max(0.0, await fetch_float_setting("tg_poster_source_repeat_penalty", 4.0))
+ self.common_tags = str(await fetch_setting("publication_common_tags", "") or "").strip()
logger.info("TG poster config: chat={} schedule={} caption_limit={}", self.chat_id, self.schedule, self.caption_limit)
async def close(self) -> None:
@@ -325,7 +327,14 @@ class TelegramPoster:
def build_text(self, post: dict[str, Any]) -> str:
category_tag = str(post.get("final_category_tag") or post.get("rewrite_category_tag") or post.get("final_category") or post.get("rewrite_category") or "")
source_tag = str(post.get("final_source_tag") or post.get("rewrite_source_tag") or post.get("source_tag") or "")
- return build_publication_text(post.get("final_text") or post.get("rewritten_text") or "", category_tag, source_tag, format_title=True, parse_mode="html")
+ return build_publication_text(
+ post.get("final_text") or post.get("rewritten_text") or "",
+ category_tag,
+ source_tag,
+ self.common_tags,
+ format_title=True,
+ parse_mode="html",
+ )
async def send_text(self, text: str) -> list[int]:
message_ids: list[int] = []
diff --git a/src/vk_parser_app/workers/vk_poster.py b/src/vk_parser_app/workers/vk_poster.py
index 9ffa960..6b5fa92 100644
--- a/src/vk_parser_app/workers/vk_poster.py
+++ b/src/vk_parser_app/workers/vk_poster.py
@@ -72,6 +72,7 @@ class VKPoster:
self.category_repeat_penalty = 3.0
self.source_repeat_penalty = 4.0
self.dry_run = False
+ self.common_tags = ""
async def init(self) -> None:
self.pool = await get_pool()
@@ -158,6 +159,7 @@ class VKPoster:
self.category_repeat_penalty = max(0.0, await fetch_float_setting("vk_poster_category_repeat_penalty", 3.0))
self.source_repeat_penalty = max(0.0, await fetch_float_setting("vk_poster_source_repeat_penalty", 4.0))
self.dry_run = await fetch_bool_setting("vk_poster_dry_run", False)
+ self.common_tags = str(await fetch_setting("publication_common_tags", "") or "").strip()
if not self.token:
raise RuntimeError("vk_poster_access_token, VK_GROUP_ACCESS_TOKEN and VK_ACCESS_TOKEN are empty")
if not self.owner_id:
@@ -292,7 +294,12 @@ class VKPoster:
def build_text(self, post: dict[str, Any]) -> str:
category_tag = str(post.get("final_category_tag") or post.get("rewrite_category_tag") or post.get("final_category") or post.get("rewrite_category") or "")
source_tag = str(post.get("final_source_tag") or post.get("rewrite_source_tag") or post.get("source_tag") or "")
- return build_publication_text(post.get("final_text") or post.get("rewritten_text") or "", category_tag, source_tag)[:VK_MESSAGE_LIMIT]
+ return build_publication_text(
+ post.get("final_text") or post.get("rewritten_text") or "",
+ category_tag,
+ source_tag,
+ self.common_tags,
+ )[:VK_MESSAGE_LIMIT]
async def media_attachment(self, client: VKAPIClient, item: dict[str, Any]) -> str | None:
attachment_id = str(item.get("original_attachment_id") or "").strip()
|