Upload folder using huggingface_hub

app/prompts.py

@@ -79,11 +79,17 @@ TASK_INSTRUCTIONS: Dict[str, str] = {
         "CRITICAL: The SURAH INFORMATION block is the ONLY authoritative source."
     ),
     "general": (
-        "The user has a general Islamic question. Steps:\n"
-        "1. Give a direct, clear answer first.\n"
-        "2. Support with evidence from the RETRIEVED RESULTS.\n"
-        "3. Quote relevant verses or hadiths from the results with references.\n"
-        "4. Conclude with a brief summary."
+        "The user may be asking a general Islamic question OR pasting text to look up. Steps:\n"
+        "1. Check ALL retrieved results from EVERY source (Quran AND Hadith).\n"
+        "2. For EACH result that matches the user's text or question, state WHERE it appears:\n"
+        "   • For Quran: Surah name (Arabic & English), Surah number, and Ayah number from the [REF] line.\n"
+        "   • For Hadith: collection name, book/chapter, hadith number, and grade.\n"
+        "3. If the same text appears in MULTIPLE places, list ALL of them explicitly.\n"
+        "   Example: ‘هذا النص ذُكِر في سورة البقرة (آية ٢٥٥) وآل عمران (آية ٢)’\n"
+        "4. Quote the Arabic text and English translation EXACTLY from the results.\n"
+        "5. Provide brief context or explanation of the text.\n"
+        "6. Answer the user's specific question if one was asked.\n"
+        "CRITICAL: Do NOT give a generic answer. Always mention the exact sources from the results."
     ),
 }
 

app/state.py

@@ -41,84 +41,93 @@ logger = logging.getLogger("qmodel.state")
 # SURAH INFO PROGRAMMATIC FALLBACK
 # ═══════════════════════════════════════════════════════════════════════
 def _results_fallback(results: list, lang: str) -> str:
-    """Generate a direct answer from search results when LLM returns empty."""
+    """Generate a direct answer from search results when LLM returns empty.
+
+    Presents ALL high-scoring results so the user sees every place
+    a text is mentioned (e.g. same verse in multiple surahs).
+    """
     if not results:
         return not_found_answer(lang)
 
-    top = results[0]
-    item_type = top.get("type", "")
-
-    if item_type == "quran":
-        ar_text = top.get("arabic", "")
-        en_text = top.get("english", "")
-        surah_ar = top.get("surah_name_ar", "")
-        surah_en = top.get("surah_name_en", "")
-        surah_num = top.get("surah_number", "")
-        verse_num = top.get("verse_number", "")
-        tafsir = top.get("tafsir_ar", "") if lang == "arabic" else top.get("tafsir_en", "")
-
-        if lang == "arabic":
-            lines = [
-                f"هذه الآية الكريمة من سورة {surah_ar} ({surah_en})، الآية {verse_num}.",
-                "",
-                f"┌─────────────────────────────────────────────┐",
-                f"│  ❝ {ar_text} ❞",
-                f"│  📝 Translation: {en_text}",
-                f"│  📖 Source: سورة {surah_ar} ({surah_en}) | رقم السورة: {surah_num} | الآية: {verse_num}",
-                f"└─────────────────────────────────────────────┘",
-            ]
-            if tafsir:
-                lines.append(f"\n{tafsir}")
-            lines.append("\nوالله أعلم.")
-            return "\n".join(lines)
-        else:
-            lines = [
-                f"This noble verse is from Surah {surah_en} ({surah_ar}), verse {verse_num}.",
-                "",
-                f"┌─────────────────────────────────────────────┐",
-                f"│  ❝ {ar_text} ❞",
-                f"│  📝 Translation: {en_text}",
-                f"│  📖 Source: Surah {surah_en} ({surah_ar}) | Surah Number: {surah_num} | Verse: {verse_num}",
-                f"└─────────────────────────────────────────────┘",
-            ]
-            if tafsir:
-                lines.append(f"\n{tafsir}")
-            lines.append("\nAnd Allah knows best.")
-            return "\n".join(lines)
-
-    elif item_type == "hadith":
-        ar_text = top.get("arabic", "")
-        en_text = top.get("english", "")
-        source = top.get("source") or top.get("reference", "")
-        grade = top.get("grade", "")
-        grade_str = f" [{grade}]" if grade else ""
-
-        if lang == "arabic":
-            lines = [
-                f"الحديث الشريف{grade_str}:",
-                "",
-                f"┌─────────────────────────────────────────────┐",
-                f"│  ❝ {ar_text} ❞",
-                f"│  📝 Translation: {en_text}",
-                f"│  📖 Source: {source}",
-                f"└─────────────────────────────────────────────┘",
-                "\nوالله أعلم.",
-            ]
-            return "\n".join(lines)
+    # Include all results whose score is at least 50% of the top score,
+    # or score >= 2.0 (strong match), whichever captures more.
+    top_score = results[0].get("_score", 0.0)
+    threshold = max(top_score * 0.5, 2.0) if top_score >= 2.0 else 0.0
+    relevant = [r for r in results if r.get("_score", 0.0) >= threshold] or results[:1]
+
+    quran_hits = [r for r in relevant if r.get("type") == "quran"]
+    hadith_hits = [r for r in relevant if r.get("type") == "hadith"]
+
+    lines: list[str] = []
+
+    # --- Quran results ---
+    if quran_hits:
+        if len(quran_hits) == 1:
+            r = quran_hits[0]
+            surah_ar = r.get("surah_name_ar", "")
+            surah_en = r.get("surah_name_en", "")
+            verse_num = r.get("verse_number", "")
+            if lang == "arabic":
+                lines.append(f"هذه الآية الكريمة من سورة {surah_ar} ({surah_en})، الآية {verse_num}.")
+            else:
+                lines.append(f"This noble verse is from Surah {surah_en} ({surah_ar}), verse {verse_num}.")
         else:
-            lines = [
-                f"The Hadith{grade_str}:",
-                "",
-                f"┌─────────────────────────────────────────────┐",
-                f"│  ❝ {ar_text} ❞",
-                f"│  📝 Translation: {en_text}",
-                f"│  📖 Source: {source}",
-                f"└─────────────────────────────────────────────┘",
-                "\nAnd Allah knows best.",
-            ]
-            return "\n".join(lines)
-
-    return not_found_answer(lang)
+            if lang == "arabic":
+                refs = [
+                    f"سورة {r.get('surah_name_ar', '')} ({r.get('surah_name_en', '')})، الآية {r.get('verse_number', '')}"
+                    for r in quran_hits
+                ]
+                lines.append(f"هذا النص الكريم ذُكِر في {len(quran_hits)} مواضع:")
+                for ref in refs:
+                    lines.append(f"  • {ref}")
+            else:
+                refs = [
+                    f"Surah {r.get('surah_name_en', '')} ({r.get('surah_name_ar', '')}), verse {r.get('verse_number', '')}"
+                    for r in quran_hits
+                ]
+                lines.append(f"This text appears in {len(quran_hits)} places:")
+                for ref in refs:
+                    lines.append(f"  • {ref}")
+
+        for r in quran_hits:
+            ar_text = r.get("arabic", "")
+            en_text = r.get("english", "")
+            surah_ar = r.get("surah_name_ar", "")
+            surah_en = r.get("surah_name_en", "")
+            surah_num = r.get("surah_number", "")
+            verse_num = r.get("verse_number", "")
+            lines.append("")
+            lines.append("┌─────────────────────────────────────────────┐")
+            lines.append(f"│  ❝ {ar_text} ❞")
+            lines.append(f"│  📝 Translation: {en_text}")
+            lines.append(f"│  📖 Source: سورة {surah_ar} ({surah_en}) | رقم السورة: {surah_num} | الآية: {verse_num}")
+            lines.append("└─────────────────────────────────────────────┘")
+
+    # --- Hadith results ---
+    if hadith_hits:
+        for r in hadith_hits:
+            ar_text = r.get("arabic", "")
+            en_text = r.get("english", "")
+            source = r.get("source") or r.get("reference", "")
+            grade = r.get("grade", "")
+            grade_str = f" [{grade}]" if grade else ""
+            if lang == "arabic":
+                lines.append(f"\nالحديث الشريف{grade_str}:")
+            else:
+                lines.append(f"\nHadith{grade_str}:")
+            lines.append("")
+            lines.append("┌─────────────────────────────────────────────┐")
+            lines.append(f"│  ❝ {ar_text} ❞")
+            lines.append(f"│  📝 Translation: {en_text}")
+            lines.append(f"│  📖 Source: {source}")
+            lines.append("└─────────────────────────────────────────────┘")
+
+    if lang == "arabic":
+        lines.append("\nوالله أعلم.")
+    else:
+        lines.append("\nAnd Allah knows best.")
+
+    return "\n".join(lines)
 
 
 def _surah_info_fallback(info: dict, lang: str) -> str: