Improve Smoke Signal profile/OCR diagnostics for rights/status gating

smoke_signal_tab.py

@@ -602,6 +602,8 @@ def run_profile() -> tuple:
     except ImportError:
         return _profile_status_html(), "PyMuPDF not installed. Run: pip install pymupdf"
 
+    skipped_rights = 0
+
     for _, row in pending.iterrows():
         book_id  = row["book_id"]
         filename = row["filename"]
@@ -612,6 +614,7 @@ def run_profile() -> tuple:
             continue
         if row.get("rights_class") in ("unknown", "excluded"):
             log.append(log_line(f"↩ {book_id}: skipped — rights={row['rights_class']}"))
+            skipped_rights += 1
             continue
 
         try:
@@ -663,6 +666,14 @@ def run_profile() -> tuple:
         except Exception as e:
             log.append(log_line(f"✗ {book_id}: {e}"))
 
+    if skipped_rights:
+        log.append(
+            log_line(
+                f"⚠ {skipped_rights} source(s) skipped due rights_class=unknown/excluded. "
+                "Set rights in Step 1 and rerun Profile."
+            )
+        )
+
     save_manifest_df(df)
     return _profile_status_html(), "\n".join(log)
 
@@ -711,6 +722,15 @@ def run_ocr() -> tuple:
 
     eligible = df[df["status"].isin(["profiled", "ocred", "rendered"])]
     if eligible.empty:
+        unknown_or_excluded = df[df["rights_class"].isin(["unknown", "excluded"])] if "rights_class" in df.columns else df.iloc[0:0]
+        pending = df[df["status"] == "pending"] if "status" in df.columns else df.iloc[0:0]
+        debug += (
+            f"Eligible rows={len(eligible)}\n"
+            f"Pending rows={len(pending)}\n"
+            f"Unknown/excluded rights={len(unknown_or_excluded)}\n"
+            "Tip: In Step 1, update rights_class away from unknown/excluded, "
+            "then rerun Step 2 Profile.\n"
+        )
         return _ocr_status_html(), debug + "No profiled PDFs. Complete Step 2 first."
 
     log = []