Update helpers/indexer.py

helpers/indexer.py

@@ -1,70 +1,51 @@
 from typing import List, Dict, Any, Optional
 
-def detect_extra_type(text: str) -> Optional[str]:
-"""يتعرف على إن كان النص عبارة عن صورة أو رابط."""
-lower = text.lower()
-
-# رابط  
-if lower.startswith("http://") or lower.startswith("https://"):  
-    # رابط صورة؟  
-    if lower.endswith((".jpg", ".jpeg", ".png", ".gif", ".webp")):  
-        return "image"  
-    return "link"  
-
-# base64 image  
-if lower.startswith("data:image/"):  
-    return "image"  
-
-# ملف صورة بدون رابط  
-if lower.endswith((".jpg", ".jpeg", ".png", ".gif", ".webp")):  
-    return "image"  
-
-return None
-
 def build_indexed_response(
-texts: List[Dict[str, str]],
-title_index: int,
-preamble_end: int,
-body_end: Optional[int] = None,
-*,
-preamble_start: Optional[int] = None,
-body_start: Optional[int] = None
+    texts: List[Dict[str, str]],
+    title_index: int,
+    preamble_end: int,
+    body_end: Optional[int] = None,
+    *,
+    preamble_start: Optional[int] = None,
+    body_start: Optional[int] = None
 ) -> List[Dict[str, str]]:
 
-if not texts:  
-    return []  
-
-last_idx = len(texts) - 1  
-title_index = min(title_index, last_idx)  
-preamble_end = min(preamble_end, last_idx)  
-body_end = min(body_end if body_end is not None else last_idx, last_idx)  
-
-if preamble_start is None:  
-    preamble_start = title_index + 1  
-if body_start is None:  
-    body_start = preamble_end + 1  
-
-preamble_start = min(preamble_start, last_idx)  
-body_start = min(body_start, last_idx)  
-
-result = []  
-
-def append_item(text: str, fallback_type: str):  
-    extra_type = detect_extra_type(text)  
-    result.append({  
-        "text": text.strip(),  
-        "type": extra_type if extra_type else fallback_type  
-    })  
-
-# العنوان  
-append_item(texts[title_index]["text"], "title")  
-
-# المقدمة  
-for i in range(preamble_start, preamble_end + 1):  
-    append_item(texts[i]["text"], "preamble")  
-
-# النص الأساسي (body)  
-for i in range(body_start, body_end + 1):  
-    append_item(texts[i]["text"], "body")  
-
-return result
+    if not texts:
+        return []
+
+    last_idx = len(texts) - 1
+    title_index = min(title_index, last_idx)
+    preamble_end = min(preamble_end, last_idx)
+    body_end = min(body_end if body_end is not None else last_idx, last_idx)
+
+    if preamble_start is None:
+        preamble_start = title_index + 1
+    if body_start is None:
+        body_start = preamble_end + 1
+
+    preamble_start = min(preamble_start, last_idx)
+    body_start = min(body_start, last_idx)
+
+    result = []
+
+    # العنوان
+    result.append({
+        "text": texts[title_index]["text"].strip(),
+        "type": "title"
+    })
+
+    # المقدمة
+    for i in range(preamble_start, preamble_end + 1):
+        result.append({
+            "text": texts[i]["text"].strip(),
+            "type": "preamble"
+        })
+
+    # بقية النص (body)
+    for i in range(body_start, body_end + 1):
+        result.append({
+            "text": texts[i]["text"].strip(),
+            "type": "body"
+        })
+
+    return result