Spaces:
Running
Running
Create indexer_simple.py
Browse files- helpers/indexer_simple.py +37 -0
helpers/indexer_simple.py
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from typing import List, Dict, Optional
|
| 2 |
+
|
| 3 |
+
def detect_extra_type(text: str) -> Optional[str]:
|
| 4 |
+
"""يتعرف على إن كان النص عبارة عن صورة أو رابط."""
|
| 5 |
+
lower = text.lower()
|
| 6 |
+
|
| 7 |
+
# رابط
|
| 8 |
+
if lower.startswith(("http://", "https://")):
|
| 9 |
+
# رابط صورة؟
|
| 10 |
+
if lower.endswith((".jpg", ".jpeg", ".png", ".gif", ".webp")):
|
| 11 |
+
return "image"
|
| 12 |
+
return "link"
|
| 13 |
+
|
| 14 |
+
# base64 image
|
| 15 |
+
if lower.startswith("data:image/"):
|
| 16 |
+
return "image"
|
| 17 |
+
|
| 18 |
+
# ملف صورة بدون رابط
|
| 19 |
+
if lower.endswith((".jpg", ".jpeg", ".png", ".gif", ".webp")):
|
| 20 |
+
return "image"
|
| 21 |
+
|
| 22 |
+
return None
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
def build_indexed(texts: List[Dict[str, str]]) -> List[Dict[str, str]]:
|
| 26 |
+
"""يُعيد القائمة كما هي مع تحديد النوع فقط."""
|
| 27 |
+
if not texts:
|
| 28 |
+
return []
|
| 29 |
+
|
| 30 |
+
result = []
|
| 31 |
+
for item in texts:
|
| 32 |
+
extra_type = detect_extra_type(item["text"])
|
| 33 |
+
result.append({
|
| 34 |
+
"text": item["text"].strip(),
|
| 35 |
+
"type": extra_type if extra_type else "text"
|
| 36 |
+
})
|
| 37 |
+
return result
|