Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -2,90 +2,61 @@ import gradio as gr
|
|
| 2 |
from huggingface_hub import InferenceClient
|
| 3 |
|
| 4 |
import nltk
|
| 5 |
-
import json
|
| 6 |
-
import io
|
| 7 |
-
import base64
|
| 8 |
-
from fpdf import FPDF
|
| 9 |
-
from textblob import TextBlob
|
| 10 |
import PyPDF2
|
| 11 |
-
import tempfile
|
| 12 |
|
| 13 |
-
# Download the NLTK punkt tokenizer if not already present
|
| 14 |
nltk.download("punkt", quiet=True)
|
| 15 |
|
| 16 |
###############################################################################
|
| 17 |
# Hugging Face Chat Code #
|
| 18 |
###############################################################################
|
| 19 |
-
"""
|
| 20 |
-
For more information on Hugging Face Inference API support, please check:
|
| 21 |
-
https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
|
| 22 |
-
"""
|
| 23 |
|
| 24 |
-
# Initialize
|
| 25 |
client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
|
| 26 |
|
| 27 |
-
def respond(
|
| 28 |
-
message,
|
| 29 |
-
history: list[tuple[str, str]],
|
| 30 |
-
system_message,
|
| 31 |
-
max_tokens,
|
| 32 |
-
temperature,
|
| 33 |
-
top_p,
|
| 34 |
-
file_content
|
| 35 |
-
):
|
| 36 |
"""
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
Yields tokens as they arrive.
|
| 40 |
"""
|
| 41 |
-
# Append file content to
|
| 42 |
if file_content and file_content.strip():
|
| 43 |
-
system_message
|
| 44 |
-
|
| 45 |
-
#
|
| 46 |
messages = [{"role": "system", "content": system_message}]
|
| 47 |
-
for
|
| 48 |
-
if
|
| 49 |
-
messages.append({"role": "user", "content":
|
| 50 |
-
if
|
| 51 |
-
messages.append({"role": "assistant", "content":
|
| 52 |
messages.append({"role": "user", "content": message})
|
| 53 |
-
|
| 54 |
-
|
| 55 |
try:
|
| 56 |
-
|
| 57 |
messages,
|
| 58 |
max_tokens=max_tokens,
|
| 59 |
-
stream=True,
|
| 60 |
temperature=temperature,
|
| 61 |
top_p=top_p,
|
| 62 |
-
)
|
| 63 |
-
|
| 64 |
-
response += token
|
| 65 |
-
yield response
|
| 66 |
except Exception as e:
|
| 67 |
-
|
| 68 |
|
| 69 |
###############################################################################
|
| 70 |
-
#
|
| 71 |
###############################################################################
|
| 72 |
|
| 73 |
def parse_file(file_obj):
|
| 74 |
"""
|
| 75 |
-
Parses
|
| 76 |
-
|
| 77 |
-
For other file types, it attempts to decode as UTF-8 text.
|
| 78 |
"""
|
| 79 |
-
# Expect file_obj to be a file-like object with a 'name' attribute.
|
| 80 |
file_extension = file_obj.name.split('.')[-1].lower()
|
| 81 |
if file_extension == "pdf":
|
| 82 |
try:
|
| 83 |
reader = PyPDF2.PdfReader(file_obj)
|
| 84 |
-
|
| 85 |
-
for page in reader.pages:
|
| 86 |
-
extracted = page.extract_text() or ""
|
| 87 |
-
text += extracted + "\n"
|
| 88 |
-
return text
|
| 89 |
except Exception as e:
|
| 90 |
return f"Error reading PDF: {e}"
|
| 91 |
else:
|
|
@@ -96,83 +67,66 @@ def parse_file(file_obj):
|
|
| 96 |
|
| 97 |
def load_files(files):
|
| 98 |
"""
|
| 99 |
-
|
| 100 |
-
Opens each file, extracts its text, and concatenates the results.
|
| 101 |
"""
|
| 102 |
-
|
| 103 |
-
for
|
| 104 |
try:
|
| 105 |
-
with open(
|
| 106 |
content = parse_file(f)
|
| 107 |
-
|
| 108 |
except Exception as e:
|
| 109 |
-
|
| 110 |
-
return
|
| 111 |
|
| 112 |
###############################################################################
|
| 113 |
-
#
|
| 114 |
###############################################################################
|
| 115 |
|
| 116 |
with gr.Blocks() as demo:
|
| 117 |
-
gr.Markdown("# **
|
| 118 |
gr.Markdown(
|
| 119 |
"""
|
| 120 |
-
This app
|
| 121 |
-
|
| 122 |
-
|
| 123 |
-
- **Upload File(s):** (PDF or TXT files) The content is automatically parsed.
|
| 124 |
-
- **Chat:** Your messages are sent to the AI along with the uploaded file's context.
|
| 125 |
-
"""
|
| 126 |
)
|
| 127 |
|
| 128 |
-
#
|
| 129 |
file_content_state = gr.State("")
|
| 130 |
-
chat_history_state = gr.State([])
|
| 131 |
-
|
| 132 |
-
# File
|
| 133 |
file_input = gr.File(label="Upload File(s)", file_count="multiple", type="filepath")
|
| 134 |
-
# When the file(s) are uploaded, automatically load them and store the combined text.
|
| 135 |
file_input.change(fn=load_files, inputs=file_input, outputs=file_content_state)
|
| 136 |
-
|
| 137 |
-
|
| 138 |
-
|
| 139 |
-
|
| 140 |
-
|
| 141 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 142 |
if not user_message.strip():
|
| 143 |
-
return "",
|
| 144 |
-
# Append
|
| 145 |
-
|
| 146 |
-
#
|
| 147 |
-
|
| 148 |
-
|
| 149 |
-
|
| 150 |
-
system_message="You are a helpful AI assistant.",
|
| 151 |
-
max_tokens=512,
|
| 152 |
-
temperature=0.7,
|
| 153 |
-
top_p=0.95,
|
| 154 |
-
file_content=file_content
|
| 155 |
-
)
|
| 156 |
-
response = ""
|
| 157 |
-
try:
|
| 158 |
-
for partial in response_gen:
|
| 159 |
-
response = partial
|
| 160 |
-
# Update the last chat history entry with the current response.
|
| 161 |
-
chat_history[-1] = (user_message, response)
|
| 162 |
-
yield "", chat_history
|
| 163 |
-
except Exception as e:
|
| 164 |
-
chat_history[-1] = (user_message, f"Error: {e}")
|
| 165 |
-
yield "", chat_history
|
| 166 |
|
| 167 |
-
#
|
| 168 |
user_input.submit(
|
| 169 |
-
|
| 170 |
-
inputs=[user_input, chat_history_state, file_content_state],
|
| 171 |
-
outputs=[user_input, chatbot]
|
| 172 |
-
queue=True
|
| 173 |
)
|
| 174 |
-
|
| 175 |
demo.launch(server_name="0.0.0.0", server_port=7860)
|
| 176 |
-
|
| 177 |
-
if __name__ == "__main__":
|
| 178 |
-
demo.launch()
|
|
|
|
| 2 |
from huggingface_hub import InferenceClient
|
| 3 |
|
| 4 |
import nltk
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
import PyPDF2
|
|
|
|
| 6 |
|
|
|
|
| 7 |
nltk.download("punkt", quiet=True)
|
| 8 |
|
| 9 |
###############################################################################
|
| 10 |
# Hugging Face Chat Code #
|
| 11 |
###############################################################################
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
|
| 13 |
+
# Initialize the Hugging Face model client
|
| 14 |
client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
|
| 15 |
|
| 16 |
+
def respond(message, history, system_message, max_tokens, temperature, top_p, file_content):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
"""
|
| 18 |
+
Calls the Hugging Face model for a response.
|
| 19 |
+
Includes file content in the system message for context.
|
|
|
|
| 20 |
"""
|
| 21 |
+
# Append file content to system message if available
|
| 22 |
if file_content and file_content.strip():
|
| 23 |
+
system_message += f"\n\nFile content:\n{file_content}"
|
| 24 |
+
|
| 25 |
+
# Prepare the message payload
|
| 26 |
messages = [{"role": "system", "content": system_message}]
|
| 27 |
+
for user, assistant in history:
|
| 28 |
+
if user:
|
| 29 |
+
messages.append({"role": "user", "content": user})
|
| 30 |
+
if assistant:
|
| 31 |
+
messages.append({"role": "assistant", "content": assistant})
|
| 32 |
messages.append({"role": "user", "content": message})
|
| 33 |
+
|
| 34 |
+
# Get the model response
|
| 35 |
try:
|
| 36 |
+
completion = client.chat_completion(
|
| 37 |
messages,
|
| 38 |
max_tokens=max_tokens,
|
|
|
|
| 39 |
temperature=temperature,
|
| 40 |
top_p=top_p,
|
| 41 |
+
)
|
| 42 |
+
return completion.choices[0].message["content"]
|
|
|
|
|
|
|
| 43 |
except Exception as e:
|
| 44 |
+
return f"Error: {e}"
|
| 45 |
|
| 46 |
###############################################################################
|
| 47 |
+
# File Upload & Parsing Functions #
|
| 48 |
###############################################################################
|
| 49 |
|
| 50 |
def parse_file(file_obj):
|
| 51 |
"""
|
| 52 |
+
Parses uploaded files and extracts content.
|
| 53 |
+
Supports PDFs and plain text.
|
|
|
|
| 54 |
"""
|
|
|
|
| 55 |
file_extension = file_obj.name.split('.')[-1].lower()
|
| 56 |
if file_extension == "pdf":
|
| 57 |
try:
|
| 58 |
reader = PyPDF2.PdfReader(file_obj)
|
| 59 |
+
return "\n".join(page.extract_text() or "" for page in reader.pages)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 60 |
except Exception as e:
|
| 61 |
return f"Error reading PDF: {e}"
|
| 62 |
else:
|
|
|
|
| 67 |
|
| 68 |
def load_files(files):
|
| 69 |
"""
|
| 70 |
+
Loads multiple files, parses their content, and concatenates the text.
|
|
|
|
| 71 |
"""
|
| 72 |
+
combined_text = ""
|
| 73 |
+
for file in files:
|
| 74 |
try:
|
| 75 |
+
with open(file, "rb") as f:
|
| 76 |
content = parse_file(f)
|
| 77 |
+
combined_text += content + "\n"
|
| 78 |
except Exception as e:
|
| 79 |
+
combined_text += f"Error processing file {file}: {e}\n"
|
| 80 |
+
return combined_text
|
| 81 |
|
| 82 |
###############################################################################
|
| 83 |
+
# Gradio UI Layout #
|
| 84 |
###############################################################################
|
| 85 |
|
| 86 |
with gr.Blocks() as demo:
|
| 87 |
+
gr.Markdown("# **Chat with File Context**")
|
| 88 |
gr.Markdown(
|
| 89 |
"""
|
| 90 |
+
This app lets you upload file(s) and chat with an AI assistant.
|
| 91 |
+
Uploaded file content will provide context for the conversation.
|
| 92 |
+
"""
|
|
|
|
|
|
|
|
|
|
| 93 |
)
|
| 94 |
|
| 95 |
+
# States to store file content and chat history
|
| 96 |
file_content_state = gr.State("")
|
| 97 |
+
chat_history_state = gr.State([])
|
| 98 |
+
|
| 99 |
+
# File Upload Section
|
| 100 |
file_input = gr.File(label="Upload File(s)", file_count="multiple", type="filepath")
|
|
|
|
| 101 |
file_input.change(fn=load_files, inputs=file_input, outputs=file_content_state)
|
| 102 |
+
|
| 103 |
+
# Chat Section
|
| 104 |
+
gr.Markdown("## Chat")
|
| 105 |
+
chatbot = gr.Chatbot(label="Conversation")
|
| 106 |
+
user_input = gr.Textbox(label="Your Message", placeholder="Ask something...", lines=2)
|
| 107 |
+
|
| 108 |
+
# Model Configuration Sliders
|
| 109 |
+
system_prompt = gr.Textbox(label="System Prompt", value="You are a helpful AI assistant.", interactive=True)
|
| 110 |
+
max_tokens = gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max Tokens")
|
| 111 |
+
temperature = gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature")
|
| 112 |
+
top_p = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p")
|
| 113 |
+
|
| 114 |
+
# Chat Function
|
| 115 |
+
def chat_function(user_message, history, file_content, system_prompt, max_tokens, temperature, top_p):
|
| 116 |
if not user_message.strip():
|
| 117 |
+
return "", history
|
| 118 |
+
# Append user's message to the chat history
|
| 119 |
+
history.append((user_message, ""))
|
| 120 |
+
# Get response from the model
|
| 121 |
+
assistant_response = respond(user_message, history, system_prompt, max_tokens, temperature, top_p, file_content)
|
| 122 |
+
history[-1] = (user_message, assistant_response)
|
| 123 |
+
return "", history
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 124 |
|
| 125 |
+
# Submit Chat Input
|
| 126 |
user_input.submit(
|
| 127 |
+
fn=chat_function,
|
| 128 |
+
inputs=[user_input, chat_history_state, file_content_state, system_prompt, max_tokens, temperature, top_p],
|
| 129 |
+
outputs=[user_input, chatbot]
|
|
|
|
| 130 |
)
|
| 131 |
+
|
| 132 |
demo.launch(server_name="0.0.0.0", server_port=7860)
|
|
|
|
|
|
|
|
|