feat: Nationwide Weather Discussions with AI Summarization

.gitignore

@@ -302,3 +302,4 @@ uv.lock
 .venv-test/
 src/accessiweather/_version.py
 src/accessiweather/_build_info.py
+progress.txt

CHANGELOG.md

@@ -5,41 +5,28 @@ All notable changes to this project will be documented in this file.
 ## Unreleased
 
 ### Added
-- Simplified auto-updater with one-click download and install - check for stable or nightly updates from Help menu or Settings, download with progress, and restart to apply
-- Startup update check - automatically checks for updates when the app launches (configurable in Settings > Updates)
-- "Check for Updates" in Help menu - shows your current channel (stable/nightly) and lets you check on demand
-- `--fake-version` and `--fake-nightly` CLI flags for testing the update flow without modifying code
-- AI model validation at startup - if your configured model was removed from OpenRouter, you'll be warned and offered to reset to default or open Settings
-- Auto-fallback for AI models - if your configured model fails, the app automatically tries the default model before giving up
-- AI model validation - invalid or removed models now show a clear error message directing you to Settings instead of cryptic API errors
-- Model validation methods in OpenRouter client - check if a model ID exists before using it
-- Report Issue dialog (Help menu) - quickly report bugs or request features directly to GitHub with auto-collected system info
-- Per-sound volume control in Sound Pack Manager - adjust volume for individual sounds directly in the UI, with live preview at the selected volume level
-- Model browser dialog - browse and select from 300+ OpenRouter AI models directly in Settings instead of opening a web browser. Filter by provider (OpenAI, Anthropic, Meta, Google, etc.), search by name, and toggle free-only models. Provider list updates dynamically based on your filters
-- Weather model selection for Open-Meteo - choose from 11 forecast models including ECMWF, GFS, ICON, Météo-France, and more. Find it in Settings > Data Sources when using Open-Meteo or Auto mode
-- Settings export and import - backup your preferences to a file and restore them on another machine, perfect for keeping your setup in sync across devices. Find it in Settings > Advanced. Your API keys stay secure in your system keyring and aren't included in the export file
-- Config file protection on Windows - your configuration file now has Windows-equivalent permissions (user-only access), matching the existing protection on macOS and Linux. This adds defense-in-depth for your location data and preferences
-- UV Index Dialog - dedicated view for detailed UV index information and sun protection recommendations accessible from the View menu
-- Data source attribution - now shows "Data from: National Weather Service, Open-Meteo" etc. so you always know where your weather data comes from
-- Stale data warnings - when you're seeing cached weather data, the app now tells you with messages like "Showing cached data from 2:30 PM (API timeout)"
+- Implemented Nationwide weather discussions feature, including:
+    - "Show Nationwide location" setting and dynamic filtering of the Nationwide location.
+    - Main window integration for displaying discussion summaries.
+    - AI summarization button in the nationwide dialog.
+    - Shared `NationalDiscussionService` instance with caching.
+- Bundled `prismatoid`/`prism` in PyInstaller nightly builds (PR #294).
+- Added missing tests for `national_discussion_service.py` to meet coverage gate requirements.
+- Updated Antfarm to v0.2.2 and configured feature-dev workflow for 80% diff-coverage.
+- Scheduled weekly disk cleanup cron job.
 
 ### Changed
--
+- Consolidated CPC outlooks into a single "6-10 & 8-14 Day Outlook" field.
+- Updated tab labels to show full names (e.g., "WPC (Weather Prediction Center)").
+- Tabs with no available data are now hidden, and the NHC tab is only shown during hurricane season.
+- Standardized classification of PMD/SWO discussions using WMO header codes.
 
 ### Fixed
-- Weather alerts now trigger desktop notifications - previously alerts showed in the UI but never sent notifications
-- Location switching now updates the weather display - switching locations was silently failing due to a type mismatch
-- Cleaned up Visual Crossing alert processing - removed orphaned AlertManager that was losing state between calls
-- Area Forecast Discussion AI summaries now respect your custom system prompt and instructions from Settings > AI
-- macOS PyInstaller builds on Apple Silicon now use an x86_64 toolchain and filter sound_lib binaries so the DMG build completes under Rosetta
-- AI explanation now correctly reads your configured model preference instead of always falling back to default
-- Canceling AI explanation generation no longer crashes the app
-- OpenRouter auto-router now works in Area Forecast Discussion dialog
-- Your chosen AI model won't silently fall back to other models when it returns short responses
-- Increased AI token limit for models with thinking/reasoning features (Gemini, Grok, etc.)
-
-### Removed
--
+- Fixed PyInstaller bundling for `prismatoid`/`prism` in nightly builds.
+- Enabled "Show Nationwide location" checkbox and correctly handled Nationwide location in `set_current_location()`.
+- Corrected CPC URLs to point to the actual discussion page (`fxus06.html`).
+- Ensured AI explanation logic correctly uses configured models and respects system prompt instructions.
+- Improved CI coverage gate to require >=80% on changed non-UI lines.
 
 ---
 

cliff.toml

@@ -63,7 +63,7 @@ commit_parsers = [
     # Skip noise - merge commits
     { message = "^Merge", skip = true },
     { message = "^merge", skip = true },
-    
+
     # Skip noise - conventional commit types users don't care about
     { message = "^chore\\(release\\)", skip = true },
     { message = "^chore\\(deps\\)", skip = true },
@@ -74,7 +74,7 @@ commit_parsers = [
     { message = "^chore", skip = true },
     { message = "^style", skip = true },
     { message = "^test", skip = true },
-    
+
     # Skip noise - dev/CI keywords in any commit message
     { message = "(?i)git-cliff", skip = true },
     { message = "(?i)github.pages", skip = true },
@@ -87,40 +87,40 @@ commit_parsers = [
     { message = "(?i)dependabot", skip = true },
     { message = "(?i)codecov", skip = true },
     { message = "(?i)coverage", skip = true },
-    { message = "(?i)\.github", skip = true },
+    { message = "(?i)[.]github", skip = true },
     { message = "(?i)release.notes", skip = true },
     { message = "(?i)changelog", skip = true },
     { message = "(?i)update-pages", skip = true },
     { message = "(?i)nightly", skip = true },
     { message = "(?i)build_tag", skip = true },
     { message = "(?i)app\\.version", skip = true },
-    
+
     # Added - new features
     { message = "^feat", group = "Added" },
     { message = "^[a|A]dd", group = "Added" },
     { message = "^[s|S]upport", group = "Added" },
     { message = "^.*: add", group = "Added" },
     { message = "^.*: support", group = "Added" },
-    
+
     # Fixed - bug fixes (user-facing only)
     { message = "^fix", group = "Fixed" },
     { message = "^.*: fix", group = "Fixed" },
     { message = "^hotfix", group = "Fixed" },
-    
+
     # Removed
     { message = "^[r|R]emove", group = "Removed" },
     { message = "^.*: remove", group = "Removed" },
     { message = "^.*: delete", group = "Removed" },
     { message = "^deprecate", group = "Deprecated" },
-    
+
     # Security
     { body = ".*security", group = "Security" },
     { message = "^security", group = "Security" },
-    
+
     # Changed - only conventional commits that aren't caught above
     { message = "^refactor", group = "Changed" },
     { message = "^perf", group = "Changed" },
-    
+
     # Skip everything else - if it's not conventional, it's probably dev noise
     { message = "^.*", skip = true },
 ]

src/accessiweather/config/locations.py

@@ -86,6 +86,14 @@ def set_current_location(self, name: str) -> bool:
                 self.logger.info(f"Set current location: {name}")
                 return self._manager.save_config()
 
+        # Handle Nationwide (injected dynamically, not in config.locations)
+        if name == "Nationwide":
+            config.current_location = Location(
+                name="Nationwide", latitude=39.8283, longitude=-98.5795
+            )
+            self.logger.info("Set current location: Nationwide")
+            return self._manager.save_config()
+
         self.logger.warning(f"Location {name} not found")
         return False
 
@@ -94,8 +102,27 @@ def get_current_location(self) -> Location | None:
         return self._manager.get_config().current_location
 
     def get_all_locations(self) -> list[Location]:
-        """Return a shallow copy of all configured locations."""
-        return self._manager.get_config().locations.copy()
+        """
+        Return a shallow copy of all configured locations.
+
+        If show_nationwide_location is enabled in settings, ensures the
+        Nationwide location is included. If disabled, filters it out.
+        """
+        locations = self._manager.get_config().locations.copy()
+        settings = self._manager.get_config().settings
+        show_nationwide = getattr(settings, "show_nationwide_location", True)
+        data_source = getattr(settings, "data_source", "auto")
+        # Nationwide only works with NWS-compatible sources
+        nationwide_available = show_nationwide and data_source in ("auto", "nws")
+
+        has_nationwide = any(loc.name == "Nationwide" for loc in locations)
+
+        if nationwide_available and not has_nationwide:
+            locations.insert(0, Location(name="Nationwide", latitude=39.8283, longitude=-98.5795))
+        elif not nationwide_available and has_nationwide:
+            locations = [loc for loc in locations if loc.name != "Nationwide"]
+
+        return locations
 
     def get_location_names(self) -> list[str]:
         """Return the list of configured location names."""

src/accessiweather/models/config.py

@@ -104,6 +104,8 @@ class AppSettings:
     debug_mode: bool = False
     sound_enabled: bool = True
     sound_pack: str = "default"
+    # Nationwide location visibility
+    show_nationwide_location: bool = True
     # Event-based notifications (opt-in, disabled by default)
     notify_discussion_update: bool = False
     notify_severe_risk_change: bool = False

src/accessiweather/services/__init__.py

@@ -18,11 +18,13 @@
     "EnvironmentalDataClient",
     "StartupManager",
     "sync_update_channel_to_service",
+    "NationalDiscussionService",
 ]
 
 # Type hints for static type checkers (not evaluated at runtime)
 if TYPE_CHECKING:
     from .environmental_client import EnvironmentalDataClient as EnvironmentalDataClient
+    from .national_discussion_service import NationalDiscussionService as NationalDiscussionService
     from .platform_detector import PlatformDetector as PlatformDetector
     from .startup_utils import StartupManager as StartupManager
     from .update_service import (
@@ -54,4 +56,8 @@ def __getattr__(name: str) -> type:
         from .update_service import sync_update_channel_to_service
 
         return sync_update_channel_to_service
+    if name == "NationalDiscussionService":
+        from .national_discussion_service import NationalDiscussionService
+
+        return NationalDiscussionService
     raise AttributeError(f"module {__name__!r} has no attribute {name!r}")

src/accessiweather/services/national_discussion_scraper.py

@@ -327,6 +327,135 @@ def fetch_spc_discussion(self) -> dict[str, str]:
         logger.error(f"Failed to fetch SPC discussion after {self.max_retries + 1} attempts")
         return {"summary": "No discussion found. (SPC)", "full": ""}
 
+    # CPC URLs for extended outlooks
+    CPC_6_10_URL = "https://www.cpc.ncep.noaa.gov/products/predictions/610day/"
+    CPC_8_14_URL = "https://www.cpc.ncep.noaa.gov/products/predictions/814day/"
+
+    def fetch_cpc_discussions(self) -> dict[str, str]:
+        """
+        Fetch CPC 6-10 Day and 8-14 Day extended outlooks via scraping.
+
+        CPC does not provide a public API, so we scrape the outlook pages.
+
+        Returns
+        -------
+            Dictionary with keys 'outlook_6_10' and 'outlook_8_14',
+            each containing the outlook text or an error message string.
+
+        """
+        domain = "cpc.ncep.noaa.gov"
+        result: dict[str, str] = {}
+
+        for key, url, label in [
+            ("outlook_6_10", self.CPC_6_10_URL, "6-10 Day"),
+            ("outlook_8_14", self.CPC_8_14_URL, "8-14 Day"),
+        ]:
+            text = self._fetch_cpc_outlook(url, domain, label)
+            result[key] = text
+
+        return result
+
+    def _fetch_cpc_outlook(self, url: str, domain: str, label: str) -> str:
+        """
+        Fetch and parse a single CPC outlook page.
+
+        Args:
+        ----
+            url: URL of the CPC outlook page
+            domain: Domain for rate limiting
+            label: Human-readable label for logging (e.g. '6-10 Day')
+
+        Returns:
+        -------
+            The outlook text, or a descriptive error message string.
+
+        """
+        for retry in range(self.max_retries + 1):
+            backoff_delay = (
+                0 if retry == 0 else self.request_delay * (self.retry_backoff ** (retry - 1))
+            )
+            if retry > 0:
+                logger.info(
+                    f"Retrying CPC {label} outlook fetch "
+                    f"(attempt {retry + 1}/{self.max_retries + 1})"
+                )
+                if backoff_delay > 0:
+                    time.sleep(backoff_delay)
+
+            response = self._make_request(url, domain, retry)
+
+            if response.get("success"):
+                text = self._extract_cpc_outlook_text(response["text"], label)
+                if text:
+                    logger.info(f"Successfully fetched CPC {label} outlook")
+                    return text
+
+            error_msg = response.get("error", "Unknown error")
+            logger.warning(f"CPC {label} outlook fetch attempt {retry + 1} failed: {error_msg}")
+
+        logger.error(f"Failed to fetch CPC {label} outlook after {self.max_retries + 1} attempts")
+        return f"CPC {label} Outlook is currently unavailable."
+
+    def _extract_cpc_outlook_text(self, html_content: str, label: str) -> str | None:
+        """
+        Extract outlook text from a CPC outlook HTML page.
+
+        CPC outlook pages typically contain the discussion text in a <pre> tag
+        or within the main content area.
+
+        Args:
+        ----
+            html_content: Raw HTML content
+            label: Human-readable label for logging
+
+        Returns:
+        -------
+            Extracted text, or None if extraction failed.
+
+        """
+        try:
+            soup = BeautifulSoup(html_content, "html.parser")
+
+            # CPC outlook pages typically have the discussion in a <pre> tag
+            pre = soup.find("pre")
+            if pre:
+                text = pre.get_text().strip()
+                if text:
+                    return text
+
+            # Fallback: look for div with class 'contentArea' or similar
+            for selector in [
+                {"class_": "contentArea"},
+                {"class_": "mainContent"},
+                {"id": "content"},
+            ]:
+                div = soup.find("div", **selector)
+                if div:
+                    text = div.get_text().strip()
+                    if text:
+                        return text
+
+            # Last resort: try to find any substantial text block
+            body = soup.find("body")
+            if body:
+                # Look for the longest text block
+                texts = [
+                    p.get_text().strip()
+                    for p in body.find_all(["p", "div"])
+                    if p.get_text().strip()
+                ]
+                if texts:
+                    longest = max(texts, key=len)
+                    if len(longest) > 100:
+                        return longest
+
+            logger.error(f"Could not extract CPC {label} outlook text")
+            return None
+
+        except Exception as e:
+            logger.error(f"Error parsing CPC {label} outlook HTML: {e}")
+            return None
+
     def fetch_all_discussions(self) -> dict[str, dict[str, str]]:
         """
         Fetch all national discussions with parallel processing.