Update README and enhance app.py with Docker support, improved logging, and URL content loading
Browse files
README.md
CHANGED
@@ -3,13 +3,13 @@ title: 'Doc LS Chatbot '
|
|
3 |
emoji: 🔥
|
4 |
colorFrom: yellow
|
5 |
colorTo: yellow
|
6 |
-
sdk:
|
7 |
sdk_version: 1.42.2
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
short_description: It is a chat built with an AI model about www.Status.law
|
11 |
---
|
12 |
|
13 |
-
# LS Chatbot Log
|
14 |
|
15 |
-
It is a chat app built using
|
|
|
3 |
emoji: 🔥
|
4 |
colorFrom: yellow
|
5 |
colorTo: yellow
|
6 |
+
sdk: Docker
|
7 |
sdk_version: 1.42.2
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
short_description: It is a chat built with an AI model about www.Status.law
|
11 |
---
|
12 |
|
13 |
+
# LS DOC Chatbot Log
|
14 |
|
15 |
+
It is a chat app built using Hugging Face and Docker Space that allows users to interact with an AI model to communicate about www.Status.law
|
app.py
CHANGED
@@ -1,5 +1,14 @@
|
|
1 |
import os
|
2 |
import time
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
from dotenv import load_dotenv
|
4 |
from fastapi import FastAPI, HTTPException
|
5 |
from pydantic import BaseModel
|
@@ -7,15 +16,15 @@ from langchain_groq import ChatGroq
|
|
7 |
from langchain_huggingface import HuggingFaceEmbeddings
|
8 |
from langchain_community.vectorstores import FAISS
|
9 |
from langchain_text_splitters import RecursiveCharacterTextSplitter
|
10 |
-
from langchain_community.document_loaders import WebBaseLoader
|
11 |
from langchain_core.prompts import PromptTemplate
|
12 |
from langchain_core.output_parsers import StrOutputParser
|
13 |
-
from datetime import datetime
|
14 |
-
import json
|
15 |
-
import traceback
|
16 |
-
from typing import Optional, List, Dict
|
17 |
from langchain_core.tracers import ConsoleCallbackHandler
|
18 |
from langchain_core.callbacks import CallbackManager
|
|
|
|
|
|
|
|
|
19 |
|
20 |
# Initialize environment variables
|
21 |
load_dotenv()
|
@@ -49,6 +58,21 @@ URLS = [
|
|
49 |
"https://status.law/faq"
|
50 |
]
|
51 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
52 |
# Enhanced logging
|
53 |
class CustomCallbackHandler(ConsoleCallbackHandler):
|
54 |
def on_chain_end(self, run):
|
@@ -66,7 +90,6 @@ class CustomCallbackHandler(ConsoleCallbackHandler):
|
|
66 |
json.dump(log_entry, f, ensure_ascii=False)
|
67 |
f.write("\n")
|
68 |
|
69 |
-
# Initialize models
|
70 |
def init_models():
|
71 |
try:
|
72 |
callback_handler = CustomCallbackHandler()
|
@@ -85,50 +108,88 @@ def init_models():
|
|
85 |
except Exception as e:
|
86 |
raise Exception(f"Model initialization failed: {str(e)}")
|
87 |
|
88 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
89 |
def build_knowledge_base(embeddings):
|
90 |
try:
|
91 |
documents = []
|
92 |
os.makedirs(VECTOR_STORE_PATH, exist_ok=True)
|
93 |
|
94 |
-
print("Starting to load documents...")
|
|
|
|
|
|
|
|
|
95 |
|
96 |
-
|
|
|
97 |
try:
|
98 |
-
print(f"
|
99 |
-
|
100 |
-
docs
|
101 |
-
|
102 |
-
|
|
|
|
|
103 |
except Exception as e:
|
104 |
-
print(f"Failed to
|
105 |
-
traceback.print_exc() # Print full traceback
|
106 |
continue
|
107 |
|
108 |
if not documents:
|
109 |
-
raise Exception("No documents loaded!")
|
110 |
-
|
111 |
-
print(f"Total documents loaded: {len(documents)}") # Debug log
|
112 |
|
|
|
|
|
113 |
text_splitter = RecursiveCharacterTextSplitter(
|
114 |
chunk_size=500,
|
115 |
chunk_overlap=100
|
116 |
)
|
117 |
-
print("Splitting documents into chunks...")
|
118 |
chunks = text_splitter.split_documents(documents)
|
119 |
-
print(f"Created {len(chunks)} chunks")
|
120 |
|
121 |
-
print("Creating vector store...")
|
122 |
vector_store = FAISS.from_documents(chunks, embeddings)
|
123 |
|
124 |
-
print("Saving vector store...")
|
125 |
vector_store.save_local(folder_path=VECTOR_STORE_PATH, index_name="index")
|
126 |
|
127 |
-
print("Vector store successfully created and saved") # Debug log
|
128 |
return vector_store
|
129 |
except Exception as e:
|
130 |
-
print("Error in build_knowledge_base:")
|
131 |
-
traceback.print_exc()
|
132 |
raise Exception(f"Knowledge base creation failed: {str(e)}")
|
133 |
|
134 |
# Initialize models and knowledge base on startup
|
@@ -148,6 +209,7 @@ if os.path.exists(VECTOR_STORE_PATH):
|
|
148 |
if vector_store is None:
|
149 |
vector_store = build_knowledge_base(embeddings)
|
150 |
|
|
|
151 |
# API endpoints
|
152 |
@app.post("/chat", response_model=ChatResponse)
|
153 |
async def chat_endpoint(request: ChatRequest):
|
|
|
1 |
import os
|
2 |
import time
|
3 |
+
import sys
|
4 |
+
import json
|
5 |
+
import traceback
|
6 |
+
import warnings
|
7 |
+
from datetime import datetime
|
8 |
+
from typing import Optional, List, Dict
|
9 |
+
|
10 |
+
import requests
|
11 |
+
from bs4 import BeautifulSoup
|
12 |
from dotenv import load_dotenv
|
13 |
from fastapi import FastAPI, HTTPException
|
14 |
from pydantic import BaseModel
|
|
|
16 |
from langchain_huggingface import HuggingFaceEmbeddings
|
17 |
from langchain_community.vectorstores import FAISS
|
18 |
from langchain_text_splitters import RecursiveCharacterTextSplitter
|
19 |
+
from langchain_community.document_loaders import WebBaseLoader, BSHTMLLoader
|
20 |
from langchain_core.prompts import PromptTemplate
|
21 |
from langchain_core.output_parsers import StrOutputParser
|
|
|
|
|
|
|
|
|
22 |
from langchain_core.tracers import ConsoleCallbackHandler
|
23 |
from langchain_core.callbacks import CallbackManager
|
24 |
+
from langchain_core.documents import Document
|
25 |
+
|
26 |
+
# Ignore SSL warnings
|
27 |
+
warnings.filterwarnings('ignore')
|
28 |
|
29 |
# Initialize environment variables
|
30 |
load_dotenv()
|
|
|
58 |
"https://status.law/faq"
|
59 |
]
|
60 |
|
61 |
+
# Check write permissions
|
62 |
+
try:
|
63 |
+
if not os.path.exists(VECTOR_STORE_PATH):
|
64 |
+
os.makedirs(VECTOR_STORE_PATH)
|
65 |
+
test_file_path = os.path.join(VECTOR_STORE_PATH, 'test_write.txt')
|
66 |
+
with open(test_file_path, 'w') as f:
|
67 |
+
f.write('test')
|
68 |
+
os.remove(test_file_path)
|
69 |
+
print(f"Write permissions OK for {VECTOR_STORE_PATH}")
|
70 |
+
except Exception as e:
|
71 |
+
print(f"WARNING: No write permissions for {VECTOR_STORE_PATH}: {str(e)}")
|
72 |
+
print("Current working directory:", os.getcwd())
|
73 |
+
print("User:", os.getenv('USER'))
|
74 |
+
sys.exit(1)
|
75 |
+
|
76 |
# Enhanced logging
|
77 |
class CustomCallbackHandler(ConsoleCallbackHandler):
|
78 |
def on_chain_end(self, run):
|
|
|
90 |
json.dump(log_entry, f, ensure_ascii=False)
|
91 |
f.write("\n")
|
92 |
|
|
|
93 |
def init_models():
|
94 |
try:
|
95 |
callback_handler = CustomCallbackHandler()
|
|
|
108 |
except Exception as e:
|
109 |
raise Exception(f"Model initialization failed: {str(e)}")
|
110 |
|
111 |
+
def check_url_availability(url: str) -> bool:
|
112 |
+
try:
|
113 |
+
response = requests.get(url, verify=False, timeout=10)
|
114 |
+
return response.status_code == 200
|
115 |
+
except Exception as e:
|
116 |
+
print(f"Error checking {url}: {str(e)}")
|
117 |
+
return False
|
118 |
+
|
119 |
+
def load_url_content(url: str) -> List[Document]:
|
120 |
+
try:
|
121 |
+
response = requests.get(url, verify=False, timeout=30)
|
122 |
+
if response.status_code != 200:
|
123 |
+
print(f"Failed to load {url}, status code: {response.status_code}")
|
124 |
+
return []
|
125 |
+
|
126 |
+
soup = BeautifulSoup(response.text, 'html.parser')
|
127 |
+
|
128 |
+
# Remove script and style elements
|
129 |
+
for script in soup(["script", "style"]):
|
130 |
+
script.decompose()
|
131 |
+
|
132 |
+
# Get text content
|
133 |
+
text = soup.get_text()
|
134 |
+
|
135 |
+
# Clean up text
|
136 |
+
lines = (line.strip() for line in text.splitlines())
|
137 |
+
chunks = (phrase.strip() for line in lines for phrase in line.split(" "))
|
138 |
+
text = ' '.join(chunk for chunk in chunks if chunk)
|
139 |
+
|
140 |
+
return [Document(page_content=text, metadata={"source": url})]
|
141 |
+
except Exception as e:
|
142 |
+
print(f"Error processing {url}: {str(e)}")
|
143 |
+
return []
|
144 |
+
|
145 |
def build_knowledge_base(embeddings):
|
146 |
try:
|
147 |
documents = []
|
148 |
os.makedirs(VECTOR_STORE_PATH, exist_ok=True)
|
149 |
|
150 |
+
print("Starting to load documents...")
|
151 |
+
|
152 |
+
# First check which URLs are available
|
153 |
+
available_urls = [url for url in URLS if check_url_availability(url)]
|
154 |
+
print(f"\nAccessible URLs: {len(available_urls)} out of {len(URLS)}")
|
155 |
|
156 |
+
# Load content from available URLs
|
157 |
+
for url in available_urls:
|
158 |
try:
|
159 |
+
print(f"\nProcessing {url}")
|
160 |
+
docs = load_url_content(url)
|
161 |
+
if docs:
|
162 |
+
documents.extend(docs)
|
163 |
+
print(f"Successfully loaded content from {url}")
|
164 |
+
else:
|
165 |
+
print(f"No content extracted from {url}")
|
166 |
except Exception as e:
|
167 |
+
print(f"Failed to process {url}: {str(e)}")
|
|
|
168 |
continue
|
169 |
|
170 |
if not documents:
|
171 |
+
raise Exception("No documents were successfully loaded!")
|
|
|
|
|
172 |
|
173 |
+
print(f"\nTotal documents loaded: {len(documents)}")
|
174 |
+
|
175 |
text_splitter = RecursiveCharacterTextSplitter(
|
176 |
chunk_size=500,
|
177 |
chunk_overlap=100
|
178 |
)
|
179 |
+
print("Splitting documents into chunks...")
|
180 |
chunks = text_splitter.split_documents(documents)
|
181 |
+
print(f"Created {len(chunks)} chunks")
|
182 |
|
183 |
+
print("Creating vector store...")
|
184 |
vector_store = FAISS.from_documents(chunks, embeddings)
|
185 |
|
186 |
+
print("Saving vector store...")
|
187 |
vector_store.save_local(folder_path=VECTOR_STORE_PATH, index_name="index")
|
188 |
|
|
|
189 |
return vector_store
|
190 |
except Exception as e:
|
191 |
+
print(f"Error in build_knowledge_base: {str(e)}")
|
192 |
+
traceback.print_exc()
|
193 |
raise Exception(f"Knowledge base creation failed: {str(e)}")
|
194 |
|
195 |
# Initialize models and knowledge base on startup
|
|
|
209 |
if vector_store is None:
|
210 |
vector_store = build_knowledge_base(embeddings)
|
211 |
|
212 |
+
# API endpoints
|
213 |
# API endpoints
|
214 |
@app.post("/chat", response_model=ChatResponse)
|
215 |
async def chat_endpoint(request: ChatRequest):
|