From 8ae283fe418a030114f0d7a9df23916f36a40c31 Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Sat, 16 May 2026 16:45:31 +0000
Subject: [PATCH] v0.4 Phase 3: asset validation in mdcms build

- Add validate_assets(): scans config.yml and theme.yml recursively for
  string values starting with assets/, and all markdown files in pages/
  and posts/ via regex; deduplicates by (path, source) before checking
- Add _collect_yaml_assets() helper for recursive YAML traversal
- Call validate_assets() in run_build() after writing nav.yml and
  search.json; prints yellow warnings for each missing asset, build
  continues on all warnings

https://claude.ai/code/session_015XtsgTMi8UtmgxEgb5Qt2c
---
 mdcms.py | 64 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 64 insertions(+)

diff --git a/mdcms.py b/mdcms.py
index c092823..8b298cd 100644
--- a/mdcms.py
+++ b/mdcms.py
@@ -339,6 +339,66 @@ def generate_search_json(
     return json.dumps(out, indent=2, ensure_ascii=False)
 
 
+# ─── Asset validation ─────────────────────────────────────────
+
+_ASSET_RE = re.compile(r'assets/[\w.\-/]+')
+
+
+def _collect_yaml_assets(val, source: str, out: list):
+    if isinstance(val, str):
+        if val.startswith("assets/"):
+            out.append((val, source))
+    elif isinstance(val, dict):
+        for v in val.values():
+            _collect_yaml_assets(v, source, out)
+    elif isinstance(val, list):
+        for item in val:
+            _collect_yaml_assets(item, source, out)
+
+
+def validate_assets(site_path: Path, cfg: dict) -> list:
+    """Return list of warning strings for assets/ references that don't exist on disk."""
+    refs: list = []
+
+    _collect_yaml_assets(cfg, "config.yml", refs)
+
+    theme_file = cfg.get("theme")
+    if theme_file:
+        theme_path = site_path / theme_file
+        if theme_path.exists():
+            try:
+                theme_data = yaml.safe_load(theme_path.read_text(encoding="utf-8")) or {}
+                _collect_yaml_assets(theme_data, theme_file, refs)
+            except (OSError, yaml.YAMLError):
+                pass
+
+    for folder in ("pages", "posts"):
+        d = site_path / folder
+        if not d.is_dir():
+            continue
+        for md_file in sorted(d.rglob("*.md")):
+            try:
+                content = md_file.read_text(encoding="utf-8")
+                rel = str(md_file.relative_to(site_path)).replace("\\", "/")
+                for m in _ASSET_RE.finditer(content):
+                    refs.append((m.group(), rel))
+            except OSError:
+                pass
+
+    warnings = []
+    seen: set = set()
+    for asset_path, source in refs:
+        key = (asset_path, source)
+        if key in seen:
+            continue
+        seen.add(key)
+        if not (site_path / asset_path).exists():
+            warnings.append(
+                f"Warning: asset not found: {asset_path}\n  Referenced in: {source}"
+            )
+    return warnings
+
+
 # ─── Core build logic ─────────────────────────────────────────
 
 def run_build(site_path: Path):
@@ -420,6 +480,10 @@ def run_build(site_path: Path):
     )
     click.echo(f"  Wrote search.json ({len(live_pages) + len(post_records)} entries)")
 
+    asset_warnings = validate_assets(site_path, cfg)
+    for w in asset_warnings:
+        click.echo(click.style(w, fg="yellow"))
+
     if auto_created:
         click.echo(click.style(
             f"\nNotice: {len(auto_created)} section(s) auto-created: {', '.join(auto_created)}\n"