Spaces:
Paused
Paused
Update app.py
Browse files
app.py
CHANGED
|
@@ -10,22 +10,14 @@ AA_HOST = "youtube-transcriptor.p.rapidapi.com"
|
|
| 10 |
|
| 11 |
# ์ ํ๋ธ URL์์ ๋น๋์ค ID๋ฅผ ์ถ์ถํ๋ ํจ์
|
| 12 |
def get_video_id(youtube_url):
|
| 13 |
-
#
|
| 14 |
-
|
| 15 |
-
if match:
|
| 16 |
-
return match.group(1)
|
| 17 |
|
| 18 |
-
# YouTube Shorts URL
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
return match.group(1)
|
| 22 |
|
| 23 |
-
|
| 24 |
-
match = re.search(r"\/shorts\/([0-9A-Za-z_-]{11})", youtube_url)
|
| 25 |
-
if match:
|
| 26 |
-
return match.group(1)
|
| 27 |
-
|
| 28 |
-
return None
|
| 29 |
|
| 30 |
# ์๋ง ์ธ์ด ์ฐ์ ์์ ๋ฆฌ์คํธ
|
| 31 |
LANGUAGE_PRIORITY = ['ko', 'en', 'ja', 'zh']
|
|
@@ -49,20 +41,16 @@ def get_youtube_transcript(youtube_url):
|
|
| 49 |
querystring = {"video_id": video_id, "lang": lang}
|
| 50 |
response = requests.get(url, headers=headers, params=querystring)
|
| 51 |
|
|
|
|
| 52 |
if response.status_code == 200:
|
| 53 |
try:
|
| 54 |
data = response.json()
|
| 55 |
-
|
| 56 |
-
#
|
| 57 |
-
|
| 58 |
-
return {"language": lang, "data": data}
|
| 59 |
-
elif isinstance(data, list) and len(data) > 0 and "text" in data[0]:
|
| 60 |
-
return {"language": lang, "data": {"transcript": data}}
|
| 61 |
-
else:
|
| 62 |
-
continue # ํ์ฌ ์ธ์ด์ ์๋ง์ด ์์ผ๋ฉด ๋ค์ ์ธ์ด๋ก ์๋
|
| 63 |
|
| 64 |
except json.JSONDecodeError as e:
|
| 65 |
-
|
| 66 |
|
| 67 |
# ๋ชจ๋ ์ธ์ด์์ ์๋ง์ ์ฐพ์ง ๋ชปํ ๊ฒฝ์ฐ
|
| 68 |
return {"error": "์ฐ์ ์์ ์ธ์ด๋ก ์๋ง์ ์ฐพ์ ์ ์์ต๋๋ค."}
|
|
|
|
| 10 |
|
| 11 |
# ์ ํ๋ธ URL์์ ๋น๋์ค ID๋ฅผ ์ถ์ถํ๋ ํจ์
|
| 12 |
def get_video_id(youtube_url):
|
| 13 |
+
# ์ ํ๋ธ URL ๋๋ youtu.be ๋จ์ถ URL์์ video_id ์ถ์ถ
|
| 14 |
+
video_id_match = re.search(r"(?<=v=)[^#&?]*", youtube_url) or re.search(r"(?<=youtu.be/)[^#&?]*", youtube_url)
|
|
|
|
|
|
|
| 15 |
|
| 16 |
+
# YouTube Shorts URL ์ฒ๋ฆฌ
|
| 17 |
+
if not video_id_match:
|
| 18 |
+
video_id_match = re.search(r"(?<=shorts/)[^#&?]*", youtube_url)
|
|
|
|
| 19 |
|
| 20 |
+
return video_id_match.group(0) if video_id_match else None
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
|
| 22 |
# ์๋ง ์ธ์ด ์ฐ์ ์์ ๋ฆฌ์คํธ
|
| 23 |
LANGUAGE_PRIORITY = ['ko', 'en', 'ja', 'zh']
|
|
|
|
| 41 |
querystring = {"video_id": video_id, "lang": lang}
|
| 42 |
response = requests.get(url, headers=headers, params=querystring)
|
| 43 |
|
| 44 |
+
# ์ํ ์ฝ๋ ํ์ธ ๋ฐ ์ ์ฒด ์๋ต ๋ฐํ
|
| 45 |
if response.status_code == 200:
|
| 46 |
try:
|
| 47 |
data = response.json()
|
| 48 |
+
|
| 49 |
+
# ์ ์ฒด ์๋ต ๋ฐ์ดํฐ๋ฅผ ๊ทธ๋๋ก ๋ฐํ
|
| 50 |
+
return {"language": lang, "data": data}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 51 |
|
| 52 |
except json.JSONDecodeError as e:
|
| 53 |
+
return {"error": f"JSON ๋์ฝ๋ฉ ์ค๋ฅ ๋ฐ์: {str(e)}"}
|
| 54 |
|
| 55 |
# ๋ชจ๋ ์ธ์ด์์ ์๋ง์ ์ฐพ์ง ๋ชปํ ๊ฒฝ์ฐ
|
| 56 |
return {"error": "์ฐ์ ์์ ์ธ์ด๋ก ์๋ง์ ์ฐพ์ ์ ์์ต๋๋ค."}
|