Update app.py
Browse files
app.py
CHANGED
@@ -2,7 +2,9 @@ import os
|
|
2 |
import json
|
3 |
import logging
|
4 |
import requests
|
|
|
5 |
from uuid import uuid4
|
|
|
6 |
from flask import Flask, request, Response, jsonify, send_from_directory
|
7 |
from webscout.Provider.Deepinfra import DeepInfra
|
8 |
|
@@ -28,39 +30,114 @@ BASE_MODEL = DeepInfra(is_conversation=False, update_file=False, system_prompt=S
|
|
28 |
|
29 |
|
30 |
# -------------------- TTS Generator --------------------
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
42 |
params = {
|
43 |
-
'input': text,
|
44 |
'prompt': VOICE,
|
45 |
'voice': 'alloy',
|
46 |
-
'generation': str(uuid4())
|
47 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
48 |
|
49 |
-
response = requests.get('https://www.openai.fm/api/generate', params=params, headers=headers)
|
50 |
-
if response.status_code == 200:
|
51 |
filename = f"{uuid4().hex}.mp3"
|
52 |
filepath = os.path.join(AUDIO_DIR, filename)
|
53 |
-
with open(
|
54 |
-
|
55 |
-
|
|
|
|
|
|
|
56 |
return f"static/audio/{filename}"
|
57 |
-
|
58 |
-
logger.warning(f"TTS failed with status {response.status_code}: {response.text}")
|
59 |
except Exception as e:
|
60 |
-
logger.exception("
|
61 |
return None
|
62 |
|
63 |
-
|
64 |
# -------------------- Chat Route --------------------
|
65 |
@app.route("/chat", methods=["POST"])
|
66 |
def chat():
|
|
|
2 |
import json
|
3 |
import logging
|
4 |
import requests
|
5 |
+
from typing import Dict
|
6 |
from uuid import uuid4
|
7 |
+
from concurrent.futures import ThreadPoolExecutor, as_completed
|
8 |
from flask import Flask, request, Response, jsonify, send_from_directory
|
9 |
from webscout.Provider.Deepinfra import DeepInfra
|
10 |
|
|
|
30 |
|
31 |
|
32 |
# -------------------- TTS Generator --------------------
|
33 |
+
|
34 |
+
def generate_tts(text:str, verbose:bool = True):
|
35 |
+
PROVIDER_URL: str = "https://www.openai.fm/api/generate"
|
36 |
+
headers = {
|
37 |
+
'sec-ch-ua-platform': '"Windows"',
|
38 |
+
'Referer': 'https://www.openai.fm/',
|
39 |
+
'sec-ch-ua': '"Microsoft Edge";v="137", "Chromium";v="137", "Not/A)Brand";v="24"',
|
40 |
+
'sec-ch-ua-mobile': '?0',
|
41 |
+
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/137.0.0.0 Safari/537.36 Edg/137.0.0.0',
|
42 |
+
'DNT': '1',
|
43 |
+
'Range': 'bytes=0-',
|
44 |
+
}
|
45 |
+
|
46 |
+
def _generate_audio_chunk(text: str, chunk_number: int) -> tuple[int, bytes]:
|
47 |
+
"""
|
48 |
+
Generate audio for a single text chunk.
|
49 |
+
|
50 |
+
Args:
|
51 |
+
text (str): The text chunk to convert.
|
52 |
+
chunk_number (int): The sequence number of the chunk.
|
53 |
+
|
54 |
+
Returns:
|
55 |
+
tuple[int, bytes]: Chunk number and audio data.
|
56 |
+
|
57 |
+
Raises:
|
58 |
+
TTSRequestError: If the request fails.
|
59 |
+
"""
|
60 |
params = {
|
61 |
+
'input': text.strip(),
|
62 |
'prompt': VOICE,
|
63 |
'voice': 'alloy',
|
64 |
+
'generation': str(uuid4()),
|
65 |
}
|
66 |
+
if verbose:logger.info(f"Processing Text: {params['input'][:70]}...")
|
67 |
+
while True:
|
68 |
+
try:
|
69 |
+
response = requests.get(
|
70 |
+
PROVIDER_URL,
|
71 |
+
params=params,
|
72 |
+
headers=headers
|
73 |
+
)
|
74 |
+
response.raise_for_status()
|
75 |
+
|
76 |
+
if response.content:
|
77 |
+
if verbose:
|
78 |
+
logger.info(f"Chunk {chunk_number} processed successfully.")
|
79 |
+
return chunk_number, response.content
|
80 |
+
|
81 |
+
if verbose:
|
82 |
+
logger.info(f"No data received for chunk {chunk_number}.")
|
83 |
+
|
84 |
+
except Exception as e:
|
85 |
+
if verbose:
|
86 |
+
logger.info(f"Error processing chunk {chunk_number}: {e}")
|
87 |
+
time.sleep(1)
|
88 |
+
|
89 |
+
def _split_text_by_fullstop(text:str, max_length:int = 800):
|
90 |
+
parts = []
|
91 |
+
while len(text) > max_length:
|
92 |
+
# Find the last period (.) within the first max_length characters
|
93 |
+
split_index = text.rfind('.', 0, max_length)
|
94 |
+
if split_index == -1:
|
95 |
+
# If there's no full stop, force split at max_length
|
96 |
+
split_index = max_length
|
97 |
+
else:
|
98 |
+
split_index += 1 # Include the period in the current part
|
99 |
+
|
100 |
+
part = text[:split_index].strip()
|
101 |
+
parts.append(part)
|
102 |
+
text = text[split_index:].strip()
|
103 |
+
if text:
|
104 |
+
parts.append(text)
|
105 |
+
return parts
|
106 |
+
|
107 |
+
sentences = _split_text_by_fullstop(text)
|
108 |
+
|
109 |
+
try:
|
110 |
+
with ThreadPoolExecutor() as executor:
|
111 |
+
futures = {
|
112 |
+
executor.submit(_generate_audio_chunk, sentence.strip(), i): i
|
113 |
+
for i, sentence in enumerate(sentences, start=1)
|
114 |
+
}
|
115 |
+
|
116 |
+
audio_chunks: Dict[int, bytes] = {}
|
117 |
+
|
118 |
+
for future in as_completed(futures):
|
119 |
+
chunk_num = futures[future]
|
120 |
+
try:
|
121 |
+
part_number, audio_data = future.result()
|
122 |
+
audio_chunks[part_number] = audio_data
|
123 |
+
except Exception as e:
|
124 |
+
if verbose:
|
125 |
+
logger.info(f"Failed to generate audio for chunk {chunk_num}: {e}")
|
126 |
|
|
|
|
|
127 |
filename = f"{uuid4().hex}.mp3"
|
128 |
filepath = os.path.join(AUDIO_DIR, filename)
|
129 |
+
with filepath.open('wb') as f:
|
130 |
+
for chunk_num in sorted(audio_chunks.keys()):
|
131 |
+
f.write(audio_chunks[chunk_num])
|
132 |
+
|
133 |
+
if verbose:
|
134 |
+
logger.info(f"TTS audio generated: {filename}")
|
135 |
return f"static/audio/{filename}"
|
136 |
+
|
|
|
137 |
except Exception as e:
|
138 |
+
logger.exception(f"Failed to generate audio: {e}")
|
139 |
return None
|
140 |
|
|
|
141 |
# -------------------- Chat Route --------------------
|
142 |
@app.route("/chat", methods=["POST"])
|
143 |
def chat():
|