Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -1,286 +1,3 @@
|
|
1 |
-
# import os
|
2 |
-
# import requests
|
3 |
-
# import cv2
|
4 |
-
# import re
|
5 |
-
# from flask import Flask, request, jsonify, render_template
|
6 |
-
# from deepgram import DeepgramClient, PrerecordedOptions
|
7 |
-
# from dotenv import load_dotenv
|
8 |
-
# import tempfile
|
9 |
-
# import json
|
10 |
-
# import subprocess
|
11 |
-
|
12 |
-
|
13 |
-
# import warnings
|
14 |
-
# warnings.filterwarnings("ignore", message="FP16 is not supported on CPU; using FP32 instead")
|
15 |
-
|
16 |
-
# app = Flask(__name__)
|
17 |
-
# print("APP IS RUNNING, ANIKET")
|
18 |
-
|
19 |
-
# # Load the .env file
|
20 |
-
# load_dotenv()
|
21 |
-
|
22 |
-
# print("ENV LOADED, ANIKET")
|
23 |
-
|
24 |
-
# # Fetch the API key from the .env file
|
25 |
-
# API_KEY = os.getenv("FIRST_API_KEY")
|
26 |
-
# DEEPGRAM_API_KEY = os.getenv("SECOND_API_KEY")
|
27 |
-
|
28 |
-
# # Ensure the API key is loaded correctly
|
29 |
-
# if not API_KEY:
|
30 |
-
# raise ValueError("API Key not found. Make sure it is set in the .env file.")
|
31 |
-
|
32 |
-
# if not DEEPGRAM_API_KEY:
|
33 |
-
# raise ValueError("DEEPGRAM_API_KEY not found. Make sure it is set in the .env file.")
|
34 |
-
|
35 |
-
# GEMINI_API_ENDPOINT = "https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash-latest:generateContent"
|
36 |
-
# GEMINI_API_KEY = API_KEY
|
37 |
-
|
38 |
-
# @app.route("/", methods=["GET"])
|
39 |
-
# def health_check():
|
40 |
-
# return jsonify({"status": "success", "message": "API is running successfully!"}), 200
|
41 |
-
|
42 |
-
|
43 |
-
# def transcribe_audio(wav_file_path):
|
44 |
-
# """
|
45 |
-
# Transcribe audio from a video file using Deepgram API synchronously.
|
46 |
-
|
47 |
-
# Args:
|
48 |
-
# wav_file_path (str): Path to save the converted WAV file.
|
49 |
-
# Returns:
|
50 |
-
# dict: A dictionary containing status, transcript, or error message.
|
51 |
-
# """
|
52 |
-
# print("Entered the transcribe_audio function")
|
53 |
-
# try:
|
54 |
-
# # Initialize Deepgram client
|
55 |
-
# deepgram = DeepgramClient(DEEPGRAM_API_KEY)
|
56 |
-
|
57 |
-
# # Open the converted WAV file
|
58 |
-
# with open(wav_file_path, 'rb') as buffer_data:
|
59 |
-
# payload = {'buffer': buffer_data}
|
60 |
-
|
61 |
-
# # Configure transcription options
|
62 |
-
# options = PrerecordedOptions(
|
63 |
-
# smart_format=True, model="nova-2", language="en-US"
|
64 |
-
# )
|
65 |
-
|
66 |
-
# # Transcribe the audio
|
67 |
-
# response = deepgram.listen.prerecorded.v('1').transcribe_file(payload, options)
|
68 |
-
|
69 |
-
# # Check if the response is valid
|
70 |
-
# if response:
|
71 |
-
# # print("Request successful! Processing response.")
|
72 |
-
|
73 |
-
# # Convert response to JSON string
|
74 |
-
# try:
|
75 |
-
# data_str = response.to_json(indent=4)
|
76 |
-
# except AttributeError as e:
|
77 |
-
# return {"status": "error", "message": f"Error converting response to JSON: {e}"}
|
78 |
-
|
79 |
-
# # Parse the JSON string to a Python dictionary
|
80 |
-
# try:
|
81 |
-
# data = json.loads(data_str)
|
82 |
-
# except json.JSONDecodeError as e:
|
83 |
-
# return {"status": "error", "message": f"Error parsing JSON string: {e}"}
|
84 |
-
|
85 |
-
# # Extract the transcript
|
86 |
-
# try:
|
87 |
-
# transcript = data["results"]["channels"][0]["alternatives"][0]["transcript"]
|
88 |
-
# except KeyError as e:
|
89 |
-
# return {"status": "error", "message": f"Error extracting transcript: {e}"}
|
90 |
-
|
91 |
-
# print(f"Transcript obtained: {transcript}")
|
92 |
-
# # Step: Save the transcript to a text file
|
93 |
-
# transcript_file_path = "transcript_from_transcribe_audio.txt"
|
94 |
-
# with open(transcript_file_path, "w", encoding="utf-8") as transcript_file:
|
95 |
-
# transcript_file.write(transcript)
|
96 |
-
# # print(f"Transcript saved to file: {transcript_file_path}")
|
97 |
-
|
98 |
-
# return transcript
|
99 |
-
# else:
|
100 |
-
# return {"status": "error", "message": "Invalid response from Deepgram."}
|
101 |
-
|
102 |
-
# except FileNotFoundError:
|
103 |
-
# return {"status": "error", "message": f"Video file not found: {wav_file_path}"}
|
104 |
-
# except Exception as e:
|
105 |
-
# return {"status": "error", "message": f"Unexpected error: {e}"}
|
106 |
-
# finally:
|
107 |
-
# # Clean up the temporary WAV file
|
108 |
-
# if os.path.exists(wav_file_path):
|
109 |
-
# os.remove(wav_file_path)
|
110 |
-
# print(f"Temporary WAV file deleted: {wav_file_path}")
|
111 |
-
|
112 |
-
|
113 |
-
|
114 |
-
# def download_video(url, temp_video_path):
|
115 |
-
# """Download video (MP4 format) from the given URL and save it to temp_video_path."""
|
116 |
-
# response = requests.get(url, stream=True)
|
117 |
-
# if response.status_code == 200:
|
118 |
-
# with open(temp_video_path, 'wb') as f:
|
119 |
-
# for chunk in response.iter_content(chunk_size=1024):
|
120 |
-
# f.write(chunk)
|
121 |
-
# print(f"Audio downloaded successfully to {temp_video_path}")
|
122 |
-
# else:
|
123 |
-
# raise Exception(f"Failed to download audio, status code: {response.status_code}")
|
124 |
-
|
125 |
-
|
126 |
-
# def preprocess_frame(frame):
|
127 |
-
# """Preprocess the frame for better OCR accuracy."""
|
128 |
-
# gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
|
129 |
-
# denoised = cv2.medianBlur(gray, 3)
|
130 |
-
# _, thresh = cv2.threshold(denoised, 0, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)
|
131 |
-
# return thresh
|
132 |
-
|
133 |
-
# def clean_ocr_text(text):
|
134 |
-
# """Clean the OCR output by removing noise and unwanted characters."""
|
135 |
-
# cleaned_text = re.sub(r'[^A-Za-z0-9\s,.!?-]', '', text)
|
136 |
-
# cleaned_text = '\n'.join([line.strip() for line in cleaned_text.splitlines() if len(line.strip()) > 2])
|
137 |
-
# return cleaned_text
|
138 |
-
|
139 |
-
# def get_information_from_video_using_OCR(video_path, interval=1):
|
140 |
-
# """Extract text from video frames using OCR and return the combined text content."""
|
141 |
-
# cap = cv2.VideoCapture(video_path)
|
142 |
-
# fps = int(cap.get(cv2.CAP_PROP_FPS))
|
143 |
-
# frame_interval = interval * fps
|
144 |
-
# frame_count = 0
|
145 |
-
# extracted_text = ""
|
146 |
-
|
147 |
-
# print("Starting text extraction from video...")
|
148 |
-
|
149 |
-
# while cap.isOpened():
|
150 |
-
# ret, frame = cap.read()
|
151 |
-
# if not ret:
|
152 |
-
# break
|
153 |
-
|
154 |
-
# if frame_count % frame_interval == 0:
|
155 |
-
# preprocessed_frame = preprocess_frame(frame)
|
156 |
-
# text = pytesseract.image_to_string(preprocessed_frame, lang='eng', config='--psm 6 --oem 3')
|
157 |
-
# cleaned_text = clean_ocr_text(text)
|
158 |
-
# if cleaned_text:
|
159 |
-
# extracted_text += cleaned_text + "\n\n"
|
160 |
-
# print(f"Text found at frame {frame_count}: {cleaned_text[:50]}...")
|
161 |
-
|
162 |
-
# frame_count += 1
|
163 |
-
|
164 |
-
# cap.release()
|
165 |
-
# print("Text extraction completed.")
|
166 |
-
# return extracted_text
|
167 |
-
|
168 |
-
|
169 |
-
|
170 |
-
|
171 |
-
# @app.route('/process-video', methods=['POST'])
|
172 |
-
# def process_video():
|
173 |
-
# if 'videoUrl' not in request.json:
|
174 |
-
# return jsonify({"error": "No video URL provided"}), 400
|
175 |
-
|
176 |
-
# video_url = request.json['videoUrl']
|
177 |
-
# temp_video_path = None
|
178 |
-
|
179 |
-
# try:
|
180 |
-
# # Step 1: Download the WAV file from the provided URL
|
181 |
-
# with tempfile.NamedTemporaryFile(delete=False, suffix=".mp4") as temp_video_file:
|
182 |
-
# temp_video_path = temp_video_file.name
|
183 |
-
# download_video(video_url, temp_video_path)
|
184 |
-
# interval = 1
|
185 |
-
# # Step 2: get the information from the downloaded MP4 file synchronously
|
186 |
-
# video_info = get_information_from_video_using_OCR(temp_video_path, interval)
|
187 |
-
|
188 |
-
# if not video_info:
|
189 |
-
# video_info = ""
|
190 |
-
|
191 |
-
|
192 |
-
|
193 |
-
# # Step 2: Convert the MP4 to WAV
|
194 |
-
# with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as temp_wav_file:
|
195 |
-
# temp_wav_path = temp_wav_file.name
|
196 |
-
# convert_mp4_to_wav(temp_video_path, temp_wav_path)
|
197 |
-
|
198 |
-
# audio_info = transcribe_audio(temp_wav_path)
|
199 |
-
|
200 |
-
# # If no transcription present, use an empty string
|
201 |
-
# if not audio_info:
|
202 |
-
# audio_info = ""
|
203 |
-
|
204 |
-
|
205 |
-
|
206 |
-
# # Step 3: Generate structured recipe information using Gemini API synchronously
|
207 |
-
# structured_data = query_gemini_api(video_info, audio_info)
|
208 |
-
|
209 |
-
# return jsonify(structured_data)
|
210 |
-
|
211 |
-
# except Exception as e:
|
212 |
-
# return jsonify({"error": str(e)}), 500
|
213 |
-
|
214 |
-
# finally:
|
215 |
-
# # Clean up temporary audio file
|
216 |
-
# if temp_video_path and os.path.exists(temp_video_path):
|
217 |
-
# os.remove(temp_video_path)
|
218 |
-
# print(f"Temporary audio file deleted: {temp_video_path}")
|
219 |
-
|
220 |
-
|
221 |
-
|
222 |
-
|
223 |
-
|
224 |
-
|
225 |
-
# def query_gemini_api(video_transcription, audio_transcription):
|
226 |
-
# """
|
227 |
-
# Send transcription text to Gemini API and fetch structured recipe information synchronously.
|
228 |
-
# """
|
229 |
-
# try:
|
230 |
-
# # Define the structured prompt
|
231 |
-
# prompt = (
|
232 |
-
# "Analyze the provided cooking video and audio transcription combined and based on the combined information extract the following structured information:\n"
|
233 |
-
# "1. Recipe Name: Identify the name of the dish being prepared.\n"
|
234 |
-
# "2. Ingredients List: Extract a detailed list of ingredients with their respective quantities (if mentioned).\n"
|
235 |
-
# "3. Steps for Preparation: Provide a step-by-step breakdown of the recipe's preparation process, organized and numbered sequentially.\n"
|
236 |
-
# "4. Cooking Techniques Used: Highlight the cooking techniques demonstrated in the video, such as searing, blitzing, wrapping, etc.\n"
|
237 |
-
# "5. Equipment Needed: List all tools, appliances, or utensils mentioned, e.g., blender, hot pan, cling film, etc.\n"
|
238 |
-
# "6. Nutritional Information (if inferred): Provide an approximate calorie count or nutritional breakdown based on the ingredients used.\n"
|
239 |
-
# "7. Serving size: In count of people or portion size.\n"
|
240 |
-
# "8. Special Notes or Variations: Include any specific tips, variations, or alternatives mentioned.\n"
|
241 |
-
# "9. Festive or Thematic Relevance: Note if the recipe has any special relevance to holidays, events, or seasons.\n"
|
242 |
-
# "Also, make sure not to provide anything else or any other information or warning or text apart from the above things mentioned."
|
243 |
-
# f"Text: {audio_transcription}\n"
|
244 |
-
# f"Text: {video_transcription}\n"
|
245 |
-
|
246 |
-
# )
|
247 |
-
|
248 |
-
# # Prepare the payload and headers
|
249 |
-
# payload = {
|
250 |
-
# "contents": [
|
251 |
-
# {
|
252 |
-
# "parts": [
|
253 |
-
# {"text": prompt}
|
254 |
-
# ]
|
255 |
-
# }
|
256 |
-
# ]
|
257 |
-
# }
|
258 |
-
# headers = {"Content-Type": "application/json"}
|
259 |
-
|
260 |
-
# # Send request to Gemini API synchronously
|
261 |
-
# response = requests.post(
|
262 |
-
# f"{GEMINI_API_ENDPOINT}?key={GEMINI_API_KEY}",
|
263 |
-
# json=payload,
|
264 |
-
# headers=headers,
|
265 |
-
# )
|
266 |
-
|
267 |
-
# # Raise error if response code is not 200
|
268 |
-
# response.raise_for_status()
|
269 |
-
|
270 |
-
# data = response.json()
|
271 |
-
|
272 |
-
# return data.get("candidates", [{}])[0].get("content", {}).get("parts", [{}])[0].get("text", "No result found")
|
273 |
-
|
274 |
-
# except requests.exceptions.RequestException as e:
|
275 |
-
# print(f"Error querying Gemini API: {e}")
|
276 |
-
# return {"error": str(e)}
|
277 |
-
|
278 |
-
|
279 |
-
# if __name__ == '__main__':
|
280 |
-
# app.run(debug=True)
|
281 |
-
|
282 |
-
|
283 |
-
|
284 |
import os
|
285 |
import requests
|
286 |
import cv2
|
@@ -292,9 +9,8 @@ from dotenv import load_dotenv
|
|
292 |
import tempfile
|
293 |
import json
|
294 |
import subprocess
|
295 |
-
|
296 |
-
|
297 |
import warnings
|
|
|
298 |
warnings.filterwarnings("ignore", message="FP16 is not supported on CPU; using FP32 instead")
|
299 |
|
300 |
app = Flask(__name__)
|
@@ -388,7 +104,6 @@ def transcribe_audio(wav_file_path):
|
|
388 |
if os.path.exists(wav_file_path):
|
389 |
os.remove(wav_file_path)
|
390 |
print(f"Temporary WAV file deleted: {wav_file_path}")
|
391 |
-
|
392 |
|
393 |
|
394 |
def download_video(url, temp_video_path):
|
@@ -411,7 +126,7 @@ def preprocess_frame(frame):
|
|
411 |
return thresh
|
412 |
|
413 |
def clean_ocr_text(text):
|
414 |
-
"""Clean the OCR output by removing noise and unwanted characters."""
|
415 |
cleaned_text = re.sub(r'[^A-Za-z0-9\s,.!?-]', '', text)
|
416 |
cleaned_text = '\n'.join([line.strip() for line in cleaned_text.splitlines() if len(line.strip()) > 2])
|
417 |
return cleaned_text
|
@@ -494,11 +209,15 @@ def process_video():
|
|
494 |
return jsonify({"error": str(e)}), 500
|
495 |
|
496 |
finally:
|
497 |
-
# Clean up temporary video file
|
498 |
if temp_video_path and os.path.exists(temp_video_path):
|
499 |
os.remove(temp_video_path)
|
500 |
print(f"Temporary video file deleted: {temp_video_path}")
|
501 |
|
|
|
|
|
|
|
|
|
502 |
|
503 |
def query_gemini_api(video_transcription, audio_transcription):
|
504 |
"""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import os
|
2 |
import requests
|
3 |
import cv2
|
|
|
9 |
import tempfile
|
10 |
import json
|
11 |
import subprocess
|
|
|
|
|
12 |
import warnings
|
13 |
+
|
14 |
warnings.filterwarnings("ignore", message="FP16 is not supported on CPU; using FP32 instead")
|
15 |
|
16 |
app = Flask(__name__)
|
|
|
104 |
if os.path.exists(wav_file_path):
|
105 |
os.remove(wav_file_path)
|
106 |
print(f"Temporary WAV file deleted: {wav_file_path}")
|
|
|
107 |
|
108 |
|
109 |
def download_video(url, temp_video_path):
|
|
|
126 |
return thresh
|
127 |
|
128 |
def clean_ocr_text(text):
|
129 |
+
"""Clean the OCR output by removing noise and unwanted characters."""
|
130 |
cleaned_text = re.sub(r'[^A-Za-z0-9\s,.!?-]', '', text)
|
131 |
cleaned_text = '\n'.join([line.strip() for line in cleaned_text.splitlines() if len(line.strip()) > 2])
|
132 |
return cleaned_text
|
|
|
209 |
return jsonify({"error": str(e)}), 500
|
210 |
|
211 |
finally:
|
212 |
+
# Clean up temporary video file and WAV file
|
213 |
if temp_video_path and os.path.exists(temp_video_path):
|
214 |
os.remove(temp_video_path)
|
215 |
print(f"Temporary video file deleted: {temp_video_path}")
|
216 |
|
217 |
+
if temp_wav_path and os.path.exists(temp_wav_path):
|
218 |
+
os.remove(temp_wav_path)
|
219 |
+
print(f"Temporary WAV file deleted: {temp_wav_path}")
|
220 |
+
|
221 |
|
222 |
def query_gemini_api(video_transcription, audio_transcription):
|
223 |
"""
|