siddhartharya
commited on
Commit
•
6c57b19
1
Parent(s):
ba22d1b
Update utils.py
Browse files
utils.py
CHANGED
@@ -5,6 +5,8 @@ import os
|
|
5 |
import tiktoken
|
6 |
from gtts import gTTS
|
7 |
import tempfile
|
|
|
|
|
8 |
|
9 |
groq_client = Groq(api_key=os.environ["GROQ_API_KEY"])
|
10 |
tokenizer = tiktoken.get_encoding("cl100k_base")
|
@@ -35,10 +37,27 @@ def generate_script(system_prompt: str, input_text: str, tone: str):
|
|
35 |
temperature=0.7
|
36 |
)
|
37 |
|
|
|
|
|
|
|
|
|
38 |
try:
|
39 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
40 |
except ValidationError as e:
|
41 |
-
raise ValueError(f"Failed to
|
42 |
|
43 |
return dialogue
|
44 |
|
|
|
5 |
import tiktoken
|
6 |
from gtts import gTTS
|
7 |
import tempfile
|
8 |
+
import json
|
9 |
+
import re
|
10 |
|
11 |
groq_client = Groq(api_key=os.environ["GROQ_API_KEY"])
|
12 |
tokenizer = tiktoken.get_encoding("cl100k_base")
|
|
|
37 |
temperature=0.7
|
38 |
)
|
39 |
|
40 |
+
# Extract content and remove any markdown code block syntax
|
41 |
+
content = response.choices[0].message.content
|
42 |
+
content = re.sub(r'```json\s*|\s*```', '', content)
|
43 |
+
|
44 |
try:
|
45 |
+
# First, try to parse as JSON
|
46 |
+
json_data = json.loads(content)
|
47 |
+
dialogue = Dialogue.model_validate(json_data)
|
48 |
+
except json.JSONDecodeError as json_error:
|
49 |
+
# If JSON parsing fails, try to extract JSON from the text
|
50 |
+
match = re.search(r'\{.*\}', content, re.DOTALL)
|
51 |
+
if match:
|
52 |
+
try:
|
53 |
+
json_data = json.loads(match.group())
|
54 |
+
dialogue = Dialogue.model_validate(json_data)
|
55 |
+
except (json.JSONDecodeError, ValidationError) as e:
|
56 |
+
raise ValueError(f"Failed to parse dialogue JSON: {e}\nContent: {content}")
|
57 |
+
else:
|
58 |
+
raise ValueError(f"Failed to find valid JSON in the response: {content}")
|
59 |
except ValidationError as e:
|
60 |
+
raise ValueError(f"Failed to validate dialogue structure: {e}\nContent: {content}")
|
61 |
|
62 |
return dialogue
|
63 |
|