fix: add cross-day dedupe

ai_daily_report/runner.py

@@ -6,10 +6,11 @@ from pathlib import Path
 from typing import Any
 
 from .clients import BlogApiClient, OpenAICompatibleClient, fetch_text as default_fetch_text
-from .config import load_source_configs
+from .config import load_pipeline_config, load_source_configs
 from .env import load_env, resolve_blog_token, resolve_llm_config
 from .models import SourceConfig
 from .pipeline import run_stage0_to_stage8
+from .publish import load_published_urls, update_published_urls
 from .sources.registry import get_source_fetcher
 
 
@@ -89,6 +90,8 @@ def run_daily_report(
     out_dir: Path,
     base_url: str,
     sources_path: Path | None = None,
+    pipeline_path: Path | None = None,
+    history_path: Path | None = None,
     fetch_text=None,
     env: dict[str, str] | None = None,
     llm_client_factory=OpenAICompatibleClient,
@@ -96,6 +99,15 @@ def run_daily_report(
 ) -> dict[str, Any]:
     fetch_text = fetch_text or default_fetch_text
     env = env if env is not None else load_env()
+    pipeline_config_path = pipeline_path or Path("config") / "pipeline.json"
+    pipeline_config = load_pipeline_config(pipeline_config_path)
+    cross_day_config = pipeline_config.get("cross_day_dedup", {}) or {}
+    cross_day_enabled = bool(cross_day_config.get("enabled", True))
+    cross_day_max_age_days = int(cross_day_config.get("max_age_days", 7))
+    configured_history_path = history_path or Path(
+        str(cross_day_config.get("history_path") or "~/.hermes/scripts/ai_morning_out/published_urls.json")
+    ).expanduser()
+    published_urls = load_published_urls(configured_history_path) if cross_day_enabled else None
 
     if source_mode == "mock":
         source_configs = _mock_source_configs()
@@ -141,8 +153,19 @@ def run_daily_report(
         mode=mode,
         base_url=base_url,
         client=blog_client,
+        published_urls=published_urls,
+        cross_day_dedup_enabled=cross_day_enabled,
+        cross_day_dedup_max_age_days=cross_day_max_age_days,
     )
 
+    if cross_day_enabled and result["publish"].mode == "publish" and result["publish"].status == "ok":
+        update_published_urls(
+            configured_history_path,
+            result["items"],
+            run_date=run_date,
+            max_age_days=cross_day_max_age_days,
+        )
+
     run_dir = out_dir / run_date
     run_dir.mkdir(parents=True, exist_ok=True)
     (run_dir / "blog_markdown.md").write_text(result["markdown"], encoding="utf-8")