Update app.py
Browse files
app.py
CHANGED
|
@@ -3,15 +3,7 @@ import gradio as gr
|
|
| 3 |
from transformers import pipeline
|
| 4 |
from sentence_transformers import SentenceTransformer, util
|
| 5 |
import PyPDF2
|
| 6 |
-
|
| 7 |
-
# Set up logging with immediate writing
|
| 8 |
-
logging.basicConfig(
|
| 9 |
-
filename='support_bot_log.txt',
|
| 10 |
-
level=logging.INFO,
|
| 11 |
-
format='%(asctime)s - %(message)s',
|
| 12 |
-
force=True # Ensures any existing handlers are replaced and logging starts fresh
|
| 13 |
-
)
|
| 14 |
-
logger = logging.getLogger()
|
| 15 |
|
| 16 |
# Load models
|
| 17 |
qa_model = pipeline("question-answering", model="distilbert-base-uncased-distilled-squad")
|
|
@@ -39,39 +31,38 @@ def find_relevant_section(query, sections, section_embeddings):
|
|
| 39 |
|
| 40 |
SIMILARITY_THRESHOLD = 0.4
|
| 41 |
if similarity_score >= SIMILARITY_THRESHOLD:
|
| 42 |
-
|
| 43 |
return best_section
|
| 44 |
|
| 45 |
-
|
| 46 |
|
| 47 |
# Keyword-based fallback search with stopword filtering
|
| 48 |
-
query_words = {word for word in
|
| 49 |
-
for section in sections:
|
| 50 |
section_words = {word for word in section.lower().split() if word not in stopwords}
|
| 51 |
common_words = query_words.intersection(section_words)
|
| 52 |
if len(common_words) >= 2:
|
| 53 |
-
|
| 54 |
return section
|
| 55 |
|
| 56 |
-
|
| 57 |
return "I don’t have enough information to answer that."
|
| 58 |
|
| 59 |
# Process the uploaded file with detailed logging
|
| 60 |
def process_file(file, state):
|
| 61 |
if file is None:
|
| 62 |
-
|
| 63 |
return [("Bot", "Please upload a file.")], state
|
| 64 |
|
| 65 |
file_path = file.name
|
| 66 |
if file_path.lower().endswith(".pdf"):
|
| 67 |
-
|
| 68 |
text = extract_text_from_pdf(file_path)
|
| 69 |
elif file_path.lower().endswith(".txt"):
|
| 70 |
-
|
| 71 |
with open(file_path, 'r', encoding='utf-8') as f:
|
| 72 |
text = f.read()
|
| 73 |
else:
|
| 74 |
-
|
| 75 |
return [("Bot", "Unsupported file format. Please upload a PDF or TXT file.")], state
|
| 76 |
|
| 77 |
sections = text.split('\n\n')
|
|
@@ -83,18 +74,18 @@ def process_file(file, state):
|
|
| 83 |
state['feedback_count'] = 0
|
| 84 |
state['mode'] = 'waiting_for_query'
|
| 85 |
state['chat_history'] = [("Bot", "File processed. You can now ask questions.")]
|
| 86 |
-
|
| 87 |
return state['chat_history'], state
|
| 88 |
|
| 89 |
# Handle user input (queries and feedback)
|
| 90 |
def handle_input(user_input, state):
|
| 91 |
if state['mode'] == 'waiting_for_upload':
|
| 92 |
state['chat_history'].append(("Bot", "Please upload a file first."))
|
| 93 |
-
|
| 94 |
return state['chat_history'], state
|
| 95 |
elif state['mode'] == 'waiting_for_query':
|
| 96 |
if user_input.lower() == "exit":
|
| 97 |
-
|
| 98 |
state['mode'] = 'exited'
|
| 99 |
state['chat_history'].append(("User", "exit"))
|
| 100 |
state['chat_history'].append(("Bot", "Session ended. You can download the log file."))
|
|
@@ -114,10 +105,10 @@ def handle_input(user_input, state):
|
|
| 114 |
state['chat_history'].append(("User", query))
|
| 115 |
state['chat_history'].append(("Bot", f"Answer: {answer}\nPlease provide feedback: good, too vague, not helpful."))
|
| 116 |
# Log the query and initial answer here:
|
| 117 |
-
|
| 118 |
elif state['mode'] == 'waiting_for_feedback':
|
| 119 |
if user_input.lower() == "exit":
|
| 120 |
-
|
| 121 |
state['mode'] = 'exited'
|
| 122 |
state['chat_history'].append(("User", "exit"))
|
| 123 |
state['chat_history'].append(("Bot", "Session ended. You can download the log file."))
|
|
@@ -125,15 +116,15 @@ def handle_input(user_input, state):
|
|
| 125 |
|
| 126 |
feedback = user_input.lower()
|
| 127 |
state['chat_history'].append(("User", feedback))
|
| 128 |
-
|
| 129 |
if feedback == "good" or state['feedback_count'] >= 2:
|
| 130 |
state['mode'] = 'waiting_for_query'
|
| 131 |
if feedback == "good":
|
| 132 |
state['chat_history'].append(("Bot", "Thank you for your feedback. You can ask another question."))
|
| 133 |
-
|
| 134 |
else:
|
| 135 |
state['chat_history'].append(("Bot", "Maximum feedback iterations reached. You can ask another question."))
|
| 136 |
-
|
| 137 |
else:
|
| 138 |
query = state['current_query']
|
| 139 |
context = find_relevant_section(query, state['sections'], state['section_embeddings'])
|
|
@@ -143,15 +134,15 @@ def handle_input(user_input, state):
|
|
| 143 |
adjusted_answer = qa_model(question=query + " Please provide more detailed information with examples.", context=context)['answer']
|
| 144 |
else:
|
| 145 |
state['chat_history'].append(("Bot", "Please provide valid feedback: good, too vague, not helpful."))
|
| 146 |
-
|
| 147 |
return state['chat_history'], state
|
| 148 |
state['last_answer'] = adjusted_answer
|
| 149 |
state['feedback_count'] += 1
|
| 150 |
state['chat_history'].append(("Bot", f"Updated answer: {adjusted_answer}\nPlease provide feedback: good, too vague, not helpful."))
|
| 151 |
-
|
| 152 |
elif state['mode'] == 'exited':
|
| 153 |
state['chat_history'].append(("Bot", "Session is over. Please download the log."))
|
| 154 |
-
|
| 155 |
return state['chat_history'], state
|
| 156 |
|
| 157 |
# Initial state
|
|
@@ -163,9 +154,22 @@ initial_state = {
|
|
| 163 |
'feedback_count': 0,
|
| 164 |
'mode': 'waiting_for_upload',
|
| 165 |
'chat_history': [("Bot", "Please upload a PDF or TXT file to start.")],
|
| 166 |
-
'last_answer': None
|
|
|
|
| 167 |
}
|
| 168 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 169 |
# Gradio interface
|
| 170 |
with gr.Blocks() as demo:
|
| 171 |
state = gr.State(initial_state)
|
|
@@ -173,7 +177,7 @@ with gr.Blocks() as demo:
|
|
| 173 |
chat = gr.Chatbot()
|
| 174 |
user_input = gr.Textbox(label="Your query or feedback")
|
| 175 |
submit_btn = gr.Button("Submit")
|
| 176 |
-
log_file = gr.File(label="Download Log File"
|
| 177 |
|
| 178 |
# Process file upload
|
| 179 |
file_upload.upload(process_file, inputs=[file_upload, state], outputs=[chat, state])
|
|
@@ -181,4 +185,17 @@ with gr.Blocks() as demo:
|
|
| 181 |
# Handle user input and clear the textbox
|
| 182 |
submit_btn.click(handle_input, inputs=[user_input, state], outputs=[chat, state]).then(lambda: "", None, user_input)
|
| 183 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 184 |
demo.launch(share=True)
|
|
|
|
| 3 |
from transformers import pipeline
|
| 4 |
from sentence_transformers import SentenceTransformer, util
|
| 5 |
import PyPDF2
|
| 6 |
+
import os
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
|
| 8 |
# Load models
|
| 9 |
qa_model = pipeline("question-answering", model="distilbert-base-uncased-distilled-squad")
|
|
|
|
| 31 |
|
| 32 |
SIMILARITY_THRESHOLD = 0.4
|
| 33 |
if similarity_score >= SIMILARITY_THRESHOLD:
|
| 34 |
+
log_message(f"Found relevant section using embeddings for query: {query}")
|
| 35 |
return best_section
|
| 36 |
|
| 37 |
+
log_message(f"Low similarity ({similarity_score}). Falling back to keyword search.")
|
| 38 |
|
| 39 |
# Keyword-based fallback search with stopword filtering
|
| 40 |
+
query_words = {word for word in sections:
|
|
|
|
| 41 |
section_words = {word for word in section.lower().split() if word not in stopwords}
|
| 42 |
common_words = query_words.intersection(section_words)
|
| 43 |
if len(common_words) >= 2:
|
| 44 |
+
log_message(f"Keyword match found for query: {query} with common words: {common_words}")
|
| 45 |
return section
|
| 46 |
|
| 47 |
+
log_message(f"No good keyword match found. Returning default fallback response.")
|
| 48 |
return "I don’t have enough information to answer that."
|
| 49 |
|
| 50 |
# Process the uploaded file with detailed logging
|
| 51 |
def process_file(file, state):
|
| 52 |
if file is None:
|
| 53 |
+
log_message("No file uploaded.")
|
| 54 |
return [("Bot", "Please upload a file.")], state
|
| 55 |
|
| 56 |
file_path = file.name
|
| 57 |
if file_path.lower().endswith(".pdf"):
|
| 58 |
+
log_message(f"Uploaded PDF file: {file_path}")
|
| 59 |
text = extract_text_from_pdf(file_path)
|
| 60 |
elif file_path.lower().endswith(".txt"):
|
| 61 |
+
log_message(f"Uploaded TXT file: {file_path}")
|
| 62 |
with open(file_path, 'r', encoding='utf-8') as f:
|
| 63 |
text = f.read()
|
| 64 |
else:
|
| 65 |
+
log_message(f"Unsupported file format: {file_path}")
|
| 66 |
return [("Bot", "Unsupported file format. Please upload a PDF or TXT file.")], state
|
| 67 |
|
| 68 |
sections = text.split('\n\n')
|
|
|
|
| 74 |
state['feedback_count'] = 0
|
| 75 |
state['mode'] = 'waiting_for_query'
|
| 76 |
state['chat_history'] = [("Bot", "File processed. You can now ask questions.")]
|
| 77 |
+
log_message(f"Processed file: {file_path}")
|
| 78 |
return state['chat_history'], state
|
| 79 |
|
| 80 |
# Handle user input (queries and feedback)
|
| 81 |
def handle_input(user_input, state):
|
| 82 |
if state['mode'] == 'waiting_for_upload':
|
| 83 |
state['chat_history'].append(("Bot", "Please upload a file first."))
|
| 84 |
+
log_message("User attempted to interact without uploading a file.")
|
| 85 |
return state['chat_history'], state
|
| 86 |
elif state['mode'] == 'waiting_for_query':
|
| 87 |
if user_input.lower() == "exit":
|
| 88 |
+
log_message("User entered 'exit'. Ending session.")
|
| 89 |
state['mode'] = 'exited'
|
| 90 |
state['chat_history'].append(("User", "exit"))
|
| 91 |
state['chat_history'].append(("Bot", "Session ended. You can download the log file."))
|
|
|
|
| 105 |
state['chat_history'].append(("User", query))
|
| 106 |
state['chat_history'].append(("Bot", f"Answer: {answer}\nPlease provide feedback: good, too vague, not helpful."))
|
| 107 |
# Log the query and initial answer here:
|
| 108 |
+
log_message(f"Query: {query}, Answer: {answer}")
|
| 109 |
elif state['mode'] == 'waiting_for_feedback':
|
| 110 |
if user_input.lower() == "exit":
|
| 111 |
+
log_message("User entered 'exit'. Ending session.")
|
| 112 |
state['mode'] = 'exited'
|
| 113 |
state['chat_history'].append(("User", "exit"))
|
| 114 |
state['chat_history'].append(("Bot", "Session ended. You can download the log file."))
|
|
|
|
| 116 |
|
| 117 |
feedback = user_input.lower()
|
| 118 |
state['chat_history'].append(("User", feedback))
|
| 119 |
+
log_message(f"Feedback: {feedback}")
|
| 120 |
if feedback == "good" or state['feedback_count'] >= 2:
|
| 121 |
state['mode'] = 'waiting_for_query'
|
| 122 |
if feedback == "good":
|
| 123 |
state['chat_history'].append(("Bot", "Thank you for your feedback. You can ask another question."))
|
| 124 |
+
log_message("Feedback accepted as 'good'. Waiting for next query.")
|
| 125 |
else:
|
| 126 |
state['chat_history'].append(("Bot", "Maximum feedback iterations reached. You can ask another question."))
|
| 127 |
+
log_message("Max feedback iterations reached. Waiting for next query.")
|
| 128 |
else:
|
| 129 |
query = state['current_query']
|
| 130 |
context = find_relevant_section(query, state['sections'], state['section_embeddings'])
|
|
|
|
| 134 |
adjusted_answer = qa_model(question=query + " Please provide more detailed information with examples.", context=context)['answer']
|
| 135 |
else:
|
| 136 |
state['chat_history'].append(("Bot", "Please provide valid feedback: good, too vague, not helpful."))
|
| 137 |
+
log_message(f"Invalid feedback received: {feedback}")
|
| 138 |
return state['chat_history'], state
|
| 139 |
state['last_answer'] = adjusted_answer
|
| 140 |
state['feedback_count'] += 1
|
| 141 |
state['chat_history'].append(("Bot", f"Updated answer: {adjusted_answer}\nPlease provide feedback: good, too vague, not helpful."))
|
| 142 |
+
log_message(f"Adjusted answer: {adjusted_answer}")
|
| 143 |
elif state['mode'] == 'exited':
|
| 144 |
state['chat_history'].append(("Bot", "Session is over. Please download the log."))
|
| 145 |
+
log_message("User interacted after exiting.")
|
| 146 |
return state['chat_history'], state
|
| 147 |
|
| 148 |
# Initial state
|
|
|
|
| 154 |
'feedback_count': 0,
|
| 155 |
'mode': 'waiting_for_upload',
|
| 156 |
'chat_history': [("Bot", "Please upload a PDF or TXT file to start.")],
|
| 157 |
+
'last_answer': None,
|
| 158 |
+
'log_messages':# Initialize an empty list for log messages
|
| 159 |
}
|
| 160 |
|
| 161 |
+
# Logging function to store messages in memory
|
| 162 |
+
def log_message(message):
|
| 163 |
+
timestamp = datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")
|
| 164 |
+
log_entry = f"{timestamp} - {message}"
|
| 165 |
+
initial_state['log_messages'].append(log_entry)
|
| 166 |
+
|
| 167 |
+
# Function to save logs to file
|
| 168 |
+
def save_logs_to_file():
|
| 169 |
+
with open("support_bot_log.txt", "w") as log_file:
|
| 170 |
+
for log_message in initial_state['log_messages']:
|
| 171 |
+
log_file.write(log_message + "\n")
|
| 172 |
+
|
| 173 |
# Gradio interface
|
| 174 |
with gr.Blocks() as demo:
|
| 175 |
state = gr.State(initial_state)
|
|
|
|
| 177 |
chat = gr.Chatbot()
|
| 178 |
user_input = gr.Textbox(label="Your query or feedback")
|
| 179 |
submit_btn = gr.Button("Submit")
|
| 180 |
+
log_file = gr.File(label="Download Log File") # Changed: No initial value
|
| 181 |
|
| 182 |
# Process file upload
|
| 183 |
file_upload.upload(process_file, inputs=[file_upload, state], outputs=[chat, state])
|
|
|
|
| 185 |
# Handle user input and clear the textbox
|
| 186 |
submit_btn.click(handle_input, inputs=[user_input, state], outputs=[chat, state]).then(lambda: "", None, user_input)
|
| 187 |
|
| 188 |
+
# Update the log file just before download
|
| 189 |
+
log_file.click(save_logs_to_file,, [log_file]) # Trigger save on click
|
| 190 |
+
|
| 191 |
+
# Also save logs when user exits
|
| 192 |
+
user_input.submit(
|
| 193 |
+
lambda user_input, state: (
|
| 194 |
+
save_logs_to_file() if user_input.lower() == "exit" else None,
|
| 195 |
+
state
|
| 196 |
+
),
|
| 197 |
+
[user_input, state],
|
| 198 |
+
[log_file, state]
|
| 199 |
+
)
|
| 200 |
+
|
| 201 |
demo.launch(share=True)
|