Spaces:
Running
Running
Daniel Amendoeira
commited on
Update tools.py
Browse files
tools.py
CHANGED
@@ -1,5 +1,10 @@
|
|
1 |
from langchain.tools import Tool
|
2 |
import datetime
|
|
|
|
|
|
|
|
|
|
|
3 |
|
4 |
def current_date(_):
|
5 |
return datetime.datetime.now().strftime("%Y-%m-%d")
|
@@ -33,4 +38,49 @@ datetime_tools = [
|
|
33 |
func=days_until,
|
34 |
description="Returns the number of days from today until a given date (input format: YYYY-MM-DD)."
|
35 |
)
|
36 |
-
]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
from langchain.tools import Tool
|
2 |
import datetime
|
3 |
+
import requests
|
4 |
+
import openai
|
5 |
+
import os
|
6 |
+
import tempfile
|
7 |
+
from urllib.parse import urlparse
|
8 |
|
9 |
def current_date(_):
|
10 |
return datetime.datetime.now().strftime("%Y-%m-%d")
|
|
|
38 |
func=days_until,
|
39 |
description="Returns the number of days from today until a given date (input format: YYYY-MM-DD)."
|
40 |
)
|
41 |
+
]
|
42 |
+
|
43 |
+
def transcribe_audio(audio_input: str) -> str:
|
44 |
+
"""
|
45 |
+
Transcribes an audio file from a local file or a URL
|
46 |
+
|
47 |
+
Args:
|
48 |
+
audio_input (str): A local file path or a direct URL to the audio file (.mp3, .m4a, etc.)
|
49 |
+
|
50 |
+
Returns:
|
51 |
+
str: The transcribed text from the audio.
|
52 |
+
"""
|
53 |
+
try:
|
54 |
+
# Detects if audio_input is a URL
|
55 |
+
is_url = audio_input.startswith("http://") or audio_input.startswith("https://")
|
56 |
+
|
57 |
+
if is_url:
|
58 |
+
parsed = urlparse(audio_input) # breaks down the URL into components (scheme, netloc, path, params, etc.)
|
59 |
+
path_extracted = os.path.splitext(parsed.path)[1] or ".mp3" # get the actual file extension from the URL path or define it to .mp3 if no extension is found
|
60 |
+
|
61 |
+
# Download to temporary file
|
62 |
+
with tempfile.NamedTemporaryFile(suffix=ext, delete=False) as tmp_file:
|
63 |
+
response = requests.get(audio_input)
|
64 |
+
response.raise_for_status()
|
65 |
+
tmp_file.write(response.content)
|
66 |
+
file_path = tmp_file.name
|
67 |
+
else:
|
68 |
+
file_path = audio_input
|
69 |
+
|
70 |
+
# Transcribe using OpenAI Whisper
|
71 |
+
client = openai.OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
|
72 |
+
with open(file_path, "rb") as audio_file:
|
73 |
+
transcription = client.audio.transcriptions.create(
|
74 |
+
model="whisper-1",
|
75 |
+
file=audio_file
|
76 |
+
)
|
77 |
+
return transcription.text
|
78 |
+
|
79 |
+
except Exception as e:
|
80 |
+
return f"Transcription failed: {e}"
|
81 |
+
|
82 |
+
transcribe_audio_tool = Tool(
|
83 |
+
name="transcribe_audio",
|
84 |
+
func=transcribe_audio,
|
85 |
+
description="Transcribes an audio file from a local file or a URL"
|
86 |
+
)
|