Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,10 +1,15 @@
|
|
| 1 |
import base64
|
|
|
|
| 2 |
from datetime import datetime
|
| 3 |
from openai import OpenAI
|
| 4 |
import gradio as gr
|
| 5 |
|
| 6 |
-
# === Initialize OpenAI Client ===
|
| 7 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
|
| 9 |
# === Prompts ===
|
| 10 |
system_prompt = (
|
|
@@ -28,37 +33,48 @@ def encode_image_to_base64(image_file):
|
|
| 28 |
return base64.b64encode(image_bytes).decode("utf-8")
|
| 29 |
|
| 30 |
# === Transcription function ===
|
| 31 |
-
def
|
| 32 |
-
if
|
| 33 |
-
return "No
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
|
| 35 |
-
|
| 36 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
{"role": "system", "content": system_prompt},
|
| 42 |
-
{"role": "user", "content": [
|
| 43 |
-
{"type": "text", "text": user_prompt_template},
|
| 44 |
-
{"type": "image_url", "image_url": {"url": image_url}}
|
| 45 |
-
]}
|
| 46 |
-
],
|
| 47 |
-
max_tokens=1500
|
| 48 |
-
)
|
| 49 |
|
| 50 |
-
|
| 51 |
-
return f"🗓️ Transcribed on: {timestamp}\n\n{response.choices[0].message.content}"
|
| 52 |
|
| 53 |
# === Gradio Interface ===
|
| 54 |
app = gr.Interface(
|
| 55 |
-
fn=
|
| 56 |
-
inputs=gr.
|
| 57 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 58 |
title="Handwritten Note Transcriber",
|
| 59 |
-
description="Upload
|
| 60 |
)
|
| 61 |
|
| 62 |
# === Launch ===
|
| 63 |
if __name__ == "__main__":
|
| 64 |
app.launch()
|
|
|
|
|
|
| 1 |
import base64
|
| 2 |
+
import os
|
| 3 |
from datetime import datetime
|
| 4 |
from openai import OpenAI
|
| 5 |
import gradio as gr
|
| 6 |
|
| 7 |
+
# === Initialize OpenAI Client using Environment Variable ===
|
| 8 |
+
openai_api_key = os.environ.get("OPENAI_API_KEY")
|
| 9 |
+
if not openai_api_key:
|
| 10 |
+
raise ValueError("OPENAI_API_KEY environment variable is not set.")
|
| 11 |
+
|
| 12 |
+
client = OpenAI(api_key=openai_api_key)
|
| 13 |
|
| 14 |
# === Prompts ===
|
| 15 |
system_prompt = (
|
|
|
|
| 33 |
return base64.b64encode(image_bytes).decode("utf-8")
|
| 34 |
|
| 35 |
# === Transcription function ===
|
| 36 |
+
def transcribe_images(images):
|
| 37 |
+
if not images:
|
| 38 |
+
return "No images uploaded."
|
| 39 |
+
|
| 40 |
+
results = []
|
| 41 |
+
for image in images:
|
| 42 |
+
encoded_image = encode_image_to_base64(image)
|
| 43 |
+
image_url = f"data:image/jpeg;base64,{encoded_image}"
|
| 44 |
|
| 45 |
+
response = client.chat.completions.create(
|
| 46 |
+
model="gpt-4-turbo",
|
| 47 |
+
messages=[
|
| 48 |
+
{"role": "system", "content": system_prompt},
|
| 49 |
+
{"role": "user", "content": [
|
| 50 |
+
{"type": "text", "text": user_prompt_template},
|
| 51 |
+
{"type": "image_url", "image_url": {"url": image_url}}
|
| 52 |
+
]}
|
| 53 |
+
],
|
| 54 |
+
max_tokens=1500
|
| 55 |
+
)
|
| 56 |
|
| 57 |
+
timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
|
| 58 |
+
result_text = f"🗓️ Transcribed on: {timestamp}\n\n{response.choices[0].message.content}"
|
| 59 |
+
results.append(result_text)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 60 |
|
| 61 |
+
return "\n\n---\n\n".join(results)
|
|
|
|
| 62 |
|
| 63 |
# === Gradio Interface ===
|
| 64 |
app = gr.Interface(
|
| 65 |
+
fn=transcribe_images,
|
| 66 |
+
inputs=gr.File(
|
| 67 |
+
type="file",
|
| 68 |
+
label="Upload handwritten note images",
|
| 69 |
+
file_types=[".jpg", ".jpeg", ".png"],
|
| 70 |
+
multiple=True
|
| 71 |
+
),
|
| 72 |
+
outputs=gr.Textbox(label="Transcribed Output", lines=30),
|
| 73 |
title="Handwritten Note Transcriber",
|
| 74 |
+
description="Upload one or more images of handwritten notes to receive clean, professional transcriptions."
|
| 75 |
)
|
| 76 |
|
| 77 |
# === Launch ===
|
| 78 |
if __name__ == "__main__":
|
| 79 |
app.launch()
|
| 80 |
+
|