Spaces:
Running
Running
Update helpers/indexer.py
Browse files- helpers/indexer.py +25 -48
helpers/indexer.py
CHANGED
|
@@ -1,11 +1,12 @@
|
|
| 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(
|
| 9 |
# رابط صورة؟
|
| 10 |
if lower.endswith((".jpg", ".jpeg", ".png", ".gif", ".webp")):
|
| 11 |
return "image"
|
|
@@ -24,8 +25,8 @@ def detect_extra_type(text: str) -> Optional[str]:
|
|
| 24 |
|
| 25 |
def build_indexed_response(
|
| 26 |
texts: List[Dict[str, str]],
|
| 27 |
-
title_index:
|
| 28 |
-
preamble_end:
|
| 29 |
body_end: Optional[int] = None,
|
| 30 |
*,
|
| 31 |
preamble_start: Optional[int] = None,
|
|
@@ -36,6 +37,18 @@ def build_indexed_response(
|
|
| 36 |
return []
|
| 37 |
|
| 38 |
last_idx = len(texts) - 1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 39 |
result = []
|
| 40 |
|
| 41 |
def append_item(text: str, fallback_type: str):
|
|
@@ -45,51 +58,15 @@ def build_indexed_response(
|
|
| 45 |
"type": extra_type if extra_type else fallback_type
|
| 46 |
})
|
| 47 |
|
| 48 |
-
#
|
| 49 |
-
|
| 50 |
-
# ================================================================
|
| 51 |
-
if title_index is None and preamble_end is None:
|
| 52 |
-
for t in texts:
|
| 53 |
-
append_item(t["text"], "text")
|
| 54 |
-
return result
|
| 55 |
-
|
| 56 |
-
# ================================================================
|
| 57 |
-
# (2) العنوان
|
| 58 |
-
# ================================================================
|
| 59 |
-
if title_index is not None:
|
| 60 |
-
title_index = min(title_index, last_idx)
|
| 61 |
-
append_item(texts[title_index]["text"], "title")
|
| 62 |
-
|
| 63 |
-
# ================================================================
|
| 64 |
-
# (3) المقدمة
|
| 65 |
-
# ================================================================
|
| 66 |
-
if preamble_end is not None:
|
| 67 |
-
preamble_end = min(preamble_end, last_idx)
|
| 68 |
-
|
| 69 |
-
# استنتاج البداية تلقائياً إذا لم يحددها المستخدم
|
| 70 |
-
if preamble_start is None:
|
| 71 |
-
preamble_start = (title_index + 1) if title_index is not None else 0
|
| 72 |
-
|
| 73 |
-
preamble_start = min(preamble_start, last_idx)
|
| 74 |
-
|
| 75 |
-
for i in range(preamble_start, preamble_end + 1):
|
| 76 |
-
append_item(texts[i]["text"], "preamble")
|
| 77 |
-
|
| 78 |
-
# ================================================================
|
| 79 |
-
# (4) النص الأساسي (body)
|
| 80 |
-
# ================================================================
|
| 81 |
-
if preamble_end is not None and body_start is None:
|
| 82 |
-
body_start = preamble_end + 1
|
| 83 |
-
|
| 84 |
-
if body_start is not None:
|
| 85 |
-
body_start = min(body_start, last_idx)
|
| 86 |
-
|
| 87 |
-
if body_end is None:
|
| 88 |
-
body_end = last_idx
|
| 89 |
|
| 90 |
-
|
|
|
|
|
|
|
| 91 |
|
| 92 |
-
|
| 93 |
-
|
|
|
|
| 94 |
|
| 95 |
return result
|
|
|
|
| 1 |
+
from typing import List, Dict, Any, Optional
|
| 2 |
+
|
| 3 |
|
| 4 |
def detect_extra_type(text: str) -> Optional[str]:
|
| 5 |
"""يتعرف على إن كان النص عبارة عن صورة أو رابط."""
|
| 6 |
lower = text.lower()
|
| 7 |
|
| 8 |
# رابط
|
| 9 |
+
if lower.startswith("http://") or lower.startswith("https://"):
|
| 10 |
# رابط صورة؟
|
| 11 |
if lower.endswith((".jpg", ".jpeg", ".png", ".gif", ".webp")):
|
| 12 |
return "image"
|
|
|
|
| 25 |
|
| 26 |
def build_indexed_response(
|
| 27 |
texts: List[Dict[str, str]],
|
| 28 |
+
title_index: int,
|
| 29 |
+
preamble_end: int,
|
| 30 |
body_end: Optional[int] = None,
|
| 31 |
*,
|
| 32 |
preamble_start: Optional[int] = None,
|
|
|
|
| 37 |
return []
|
| 38 |
|
| 39 |
last_idx = len(texts) - 1
|
| 40 |
+
title_index = min(title_index, last_idx)
|
| 41 |
+
preamble_end = min(preamble_end, last_idx)
|
| 42 |
+
body_end = min(body_end if body_end is not None else last_idx, last_idx)
|
| 43 |
+
|
| 44 |
+
if preamble_start is None:
|
| 45 |
+
preamble_start = title_index + 1
|
| 46 |
+
if body_start is None:
|
| 47 |
+
body_start = preamble_end + 1
|
| 48 |
+
|
| 49 |
+
preamble_start = min(preamble_start, last_idx)
|
| 50 |
+
body_start = min(body_start, last_idx)
|
| 51 |
+
|
| 52 |
result = []
|
| 53 |
|
| 54 |
def append_item(text: str, fallback_type: str):
|
|
|
|
| 58 |
"type": extra_type if extra_type else fallback_type
|
| 59 |
})
|
| 60 |
|
| 61 |
+
# العنوان
|
| 62 |
+
append_item(texts[title_index]["text"], "title")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 63 |
|
| 64 |
+
# المقدمة
|
| 65 |
+
for i in range(preamble_start, preamble_end + 1):
|
| 66 |
+
append_item(texts[i]["text"], "preamble")
|
| 67 |
|
| 68 |
+
# النص الأساسي (body)
|
| 69 |
+
for i in range(body_start, body_end + 1):
|
| 70 |
+
append_item(texts[i]["text"], "body")
|
| 71 |
|
| 72 |
return result
|