Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -18,9 +18,10 @@ import pytz
|
|
| 18 |
import re
|
| 19 |
from PIL import Image
|
| 20 |
import zipfile
|
|
|
|
| 21 |
import time
|
| 22 |
|
| 23 |
-
# App Configuration
|
| 24 |
Site_Name = 'πGitπCosmosπ« - Azure Cosmos DB and Github Agent'
|
| 25 |
title = "πGitπCosmosπ« - Azure Cosmos DB and Github Agent"
|
| 26 |
helpURL = 'https://huggingface.co/awacke1'
|
|
@@ -39,23 +40,25 @@ st.set_page_config(
|
|
| 39 |
}
|
| 40 |
)
|
| 41 |
|
| 42 |
-
# Cosmos DB configuration
|
| 43 |
ENDPOINT = "https://acae-afd.documents.azure.com:443/"
|
| 44 |
DATABASE_NAME = os.environ.get("COSMOS_DATABASE_NAME")
|
| 45 |
CONTAINER_NAME = os.environ.get("COSMOS_CONTAINER_NAME")
|
| 46 |
Key = os.environ.get("Key")
|
| 47 |
|
| 48 |
-
# Your local app URL
|
| 49 |
LOCAL_APP_URL = "https://huggingface.co/spaces/awacke1/AzureCosmosDBUI"
|
| 50 |
|
| 51 |
-
# Anthropic configuration
|
| 52 |
client = anthropic.Anthropic(api_key=os.environ.get("ANTHROPIC_API_KEY"))
|
| 53 |
|
| 54 |
-
# Initialize session state
|
| 55 |
if "chat_history" not in st.session_state:
|
| 56 |
st.session_state.chat_history = []
|
| 57 |
|
| 58 |
-
# Helper Functions
|
|
|
|
|
|
|
| 59 |
def get_download_link(file_path):
|
| 60 |
with open(file_path, "rb") as file:
|
| 61 |
contents = file.read()
|
|
@@ -63,30 +66,33 @@ def get_download_link(file_path):
|
|
| 63 |
file_name = os.path.basename(file_path)
|
| 64 |
return f'<a href="data:file/txt;base64,{b64}" download="{file_name}">Download {file_name}π</a>'
|
| 65 |
|
|
|
|
| 66 |
def generate_unique_id():
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
unique_id = f"{date_time}{ms:03d}"
|
| 71 |
-
return unique_id
|
| 72 |
|
|
|
|
| 73 |
def generate_filename(prompt, file_type):
|
| 74 |
central = pytz.timezone('US/Central')
|
| 75 |
safe_date_time = datetime.now(central).strftime("%m%d_%H%M")
|
| 76 |
safe_prompt = re.sub(r'\W+', '', prompt)[:90]
|
| 77 |
return f"{safe_date_time}{safe_prompt}.{file_type}"
|
| 78 |
|
|
|
|
| 79 |
def create_file(filename, prompt, response, should_save=True):
|
| 80 |
if not should_save:
|
| 81 |
return
|
| 82 |
with open(filename, 'w', encoding='utf-8') as file:
|
| 83 |
file.write(prompt + "\n\n" + response)
|
| 84 |
|
|
|
|
| 85 |
def load_file(file_name):
|
| 86 |
with open(file_name, "r", encoding='utf-8') as file:
|
| 87 |
content = file.read()
|
| 88 |
return content
|
| 89 |
|
|
|
|
| 90 |
def display_glossary_entity(k):
|
| 91 |
search_urls = {
|
| 92 |
"ππArXiv": lambda k: f"/?q={quote(k)}",
|
|
@@ -97,6 +103,7 @@ def display_glossary_entity(k):
|
|
| 97 |
links_md = ' '.join([f"<a href='{url(k)}' target='_blank'>{emoji}</a>" for emoji, url in search_urls.items()])
|
| 98 |
st.markdown(f"{k} {links_md}", unsafe_allow_html=True)
|
| 99 |
|
|
|
|
| 100 |
def create_zip_of_files(files):
|
| 101 |
zip_name = "all_files.zip"
|
| 102 |
with zipfile.ZipFile(zip_name, 'w') as zipf:
|
|
@@ -104,6 +111,7 @@ def create_zip_of_files(files):
|
|
| 104 |
zipf.write(file)
|
| 105 |
return zip_name
|
| 106 |
|
|
|
|
| 107 |
def get_video_html(video_path, width="100%"):
|
| 108 |
video_url = f"data:video/mp4;base64,{base64.b64encode(open(video_path, 'rb').read()).decode()}"
|
| 109 |
return f'''
|
|
@@ -113,6 +121,7 @@ def get_video_html(video_path, width="100%"):
|
|
| 113 |
</video>
|
| 114 |
'''
|
| 115 |
|
|
|
|
| 116 |
def get_audio_html(audio_path, width="100%"):
|
| 117 |
audio_url = f"data:audio/mpeg;base64,{base64.b64encode(open(audio_path, 'rb').read()).decode()}"
|
| 118 |
return f'''
|
|
@@ -122,18 +131,23 @@ def get_audio_html(audio_path, width="100%"):
|
|
| 122 |
</audio>
|
| 123 |
'''
|
| 124 |
|
| 125 |
-
# Cosmos DB functions
|
|
|
|
|
|
|
| 126 |
def get_databases(client):
|
| 127 |
return [db['id'] for db in client.list_databases()]
|
| 128 |
|
|
|
|
| 129 |
def get_containers(database):
|
| 130 |
return [container['id'] for container in database.list_containers()]
|
| 131 |
|
|
|
|
| 132 |
def get_documents(container, limit=None):
|
| 133 |
query = "SELECT * FROM c ORDER BY c._ts DESC"
|
| 134 |
items = list(container.query_items(query=query, enable_cross_partition_query=True, max_item_count=limit))
|
| 135 |
return items
|
| 136 |
|
|
|
|
| 137 |
def insert_record(container, record):
|
| 138 |
try:
|
| 139 |
container.create_item(body=record)
|
|
@@ -143,6 +157,7 @@ def insert_record(container, record):
|
|
| 143 |
except Exception as e:
|
| 144 |
return False, f"An unexpected error occurred: {str(e)} π±"
|
| 145 |
|
|
|
|
| 146 |
def update_record(container, updated_record):
|
| 147 |
try:
|
| 148 |
container.upsert_item(body=updated_record)
|
|
@@ -152,6 +167,7 @@ def update_record(container, updated_record):
|
|
| 152 |
except Exception as e:
|
| 153 |
return False, f"An unexpected error occurred: {traceback.format_exc()} π±"
|
| 154 |
|
|
|
|
| 155 |
def delete_record(container, name, id):
|
| 156 |
try:
|
| 157 |
container.delete_item(item=id, partition_key=id)
|
|
@@ -163,11 +179,71 @@ def delete_record(container, name, id):
|
|
| 163 |
except Exception as e:
|
| 164 |
return False, f"An unexpected error occurred: {traceback.format_exc()} π±"
|
| 165 |
|
| 166 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 167 |
|
| 168 |
-
|
| 169 |
-
|
| 170 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 171 |
def save_or_clone_to_cosmos_db(container, query=None, response=None, clone_id=None):
|
| 172 |
try:
|
| 173 |
if not container:
|
|
@@ -216,34 +292,52 @@ def save_or_clone_to_cosmos_db(container, query=None, response=None, clone_id=No
|
|
| 216 |
except Exception as e:
|
| 217 |
st.error(f"An unexpected error occurred: {str(e)}")
|
| 218 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 219 |
|
| 220 |
-
#
|
| 221 |
def search_glossary(query):
|
| 222 |
-
st.markdown(f"### π
|
| 223 |
-
|
| 224 |
# Dropdown for model selection
|
| 225 |
model_options = ['mistralai/Mixtral-8x7B-Instruct-v0.1', 'mistralai/Mistral-7B-Instruct-v0.2', 'google/gemma-7b-it', 'None']
|
| 226 |
model_choice = st.selectbox('π§ Select LLM Model', options=model_options, index=1)
|
| 227 |
-
|
| 228 |
# Dropdown for database selection
|
| 229 |
database_options = ['Semantic Search', 'Arxiv Search - Latest - (EXPERIMENTAL)']
|
| 230 |
database_choice = st.selectbox('π Select Database', options=database_options, index=0)
|
| 231 |
|
| 232 |
-
|
| 233 |
-
|
| 234 |
-
# Run Button with Emoji
|
| 235 |
-
#if st.button("π Run"):
|
| 236 |
-
|
| 237 |
# π΅οΈββοΈ Searching the glossary for: query
|
| 238 |
all_results = ""
|
| 239 |
st.markdown(f"- {query}")
|
| 240 |
|
| 241 |
# π ArXiv RAG researcher expert ~-<>-~ Paper Summary & Ask LLM
|
| 242 |
-
#database_choice Literal['Semantic Search', 'Arxiv Search - Latest - (EXPERIMENTAL)'] Default: "Semantic Search"
|
| 243 |
-
#llm_model_picked Literal['mistralai/Mixtral-8x7B-Instruct-v0.1', 'mistralai/Mistral-7B-Instruct-v0.2', 'google/gemma-7b-it', 'None'] Default: "mistralai/Mistral-7B-Instruct-v0.2"
|
| 244 |
client = Client("awacke1/Arxiv-Paper-Search-And-QA-RAG-Pattern")
|
| 245 |
|
| 246 |
-
|
| 247 |
# π ArXiv RAG researcher expert ~-<>-~ Paper Summary & Ask LLM - api_name: /ask_llm
|
| 248 |
result = client.predict(
|
| 249 |
prompt=query,
|
|
@@ -253,7 +347,7 @@ def search_glossary(query):
|
|
| 253 |
)
|
| 254 |
st.markdown(result)
|
| 255 |
st.code(result, language="python", line_numbers=True)
|
| 256 |
-
|
| 257 |
# π ArXiv RAG researcher expert ~-<>-~ Paper Summary & Ask LLM - api_name: /ask_llm
|
| 258 |
result2 = client.predict(
|
| 259 |
prompt=query,
|
|
@@ -274,7 +368,6 @@ def search_glossary(query):
|
|
| 274 |
st.markdown(result3)
|
| 275 |
st.code(result3, language="python", line_numbers=True)
|
| 276 |
|
| 277 |
-
|
| 278 |
# π ArXiv RAG researcher expert ~-<>-~ Paper Summary & Ask LLM - api_name: /update_with_rag_md
|
| 279 |
response2 = client.predict(
|
| 280 |
message=query, # str in 'parameter_13' Textbox component
|
|
@@ -282,173 +375,25 @@ def search_glossary(query):
|
|
| 282 |
database_choice="Semantic Search",
|
| 283 |
llm_model_picked="mistralai/Mistral-7B-Instruct-v0.2",
|
| 284 |
api_name="/update_with_rag_md"
|
| 285 |
-
)
|
| 286 |
|
| 287 |
st.markdown(response2[0])
|
| 288 |
st.code(response2[0], language="python", line_numbers=True, wrap_lines=True)
|
| 289 |
-
|
| 290 |
st.markdown(response2[1])
|
| 291 |
st.code(response2[1], language="python", line_numbers=True, wrap_lines=True)
|
| 292 |
|
| 293 |
-
|
| 294 |
-
try:
|
| 295 |
-
#save_to_cosmos_db(st.session_state.cosmos_container, query, result, result)
|
| 296 |
-
#save_to_cosmos_db(st.session_state.cosmos_container, query, result2, result2)
|
| 297 |
-
#save_to_cosmos_db(st.session_state.cosmos_container, query, result3, result3)
|
| 298 |
-
#save_to_cosmos_db(st.session_state.cosmos_container, query, response2[0], response2[0])
|
| 299 |
-
#save_to_cosmos_db(st.session_state.cosmos_container, query, response2[1], response2[1])
|
| 300 |
-
|
| 301 |
-
save_or_clone_to_cosmos_db(container=st.session_state.cosmos_container, query=query, response=result, clone_id=None)
|
| 302 |
-
save_or_clone_to_cosmos_db(container=st.session_state.cosmos_container, query=query, response=result2, clone_id=None)
|
| 303 |
-
save_or_clone_to_cosmos_db(container=st.session_state.cosmos_container, query=query, response=result3, clone_id=None)
|
| 304 |
-
save_or_clone_to_cosmos_db(container=st.session_state.cosmos_container, query=query, response=response2[0], clone_id=None)
|
| 305 |
-
save_or_clone_to_cosmos_db(container=st.session_state.cosmos_container, query=query, response=response2[1], clone_id=None)
|
| 306 |
-
|
| 307 |
-
|
| 308 |
-
|
| 309 |
-
except exceptions.CosmosHttpResponseError as e:
|
| 310 |
-
return False, f"HTTP error occurred: {str(e)} π¨"
|
| 311 |
-
except Exception as e:
|
| 312 |
-
return False, f"An unexpected error occurred: {str(e)} π±"
|
| 313 |
-
|
| 314 |
-
|
| 315 |
-
try:
|
| 316 |
-
# Aggregate hyperlinks and show with emojis
|
| 317 |
-
hyperlinks = extract_hyperlinks([response1, response2])
|
| 318 |
-
st.markdown("### π Aggregated Hyperlinks")
|
| 319 |
-
for link in hyperlinks:
|
| 320 |
-
st.markdown(f"π [{link}]({link})")
|
| 321 |
-
|
| 322 |
-
# Show responses in a code format with line numbers
|
| 323 |
-
st.markdown("### π Response Outputs with Line Numbers")
|
| 324 |
-
st.code(f"Response 1: \n{format_with_line_numbers(response1)}\n\nResponse 2: \n{format_with_line_numbers(response2)}", language="json")
|
| 325 |
-
except exceptions.CosmosHttpResponseError as e:
|
| 326 |
-
return False, f"HTTP error occurred: {str(e)} π¨"
|
| 327 |
-
except Exception as e:
|
| 328 |
-
return False, f"An unexpected error occurred: {str(e)} π±"
|
| 329 |
-
|
| 330 |
-
|
| 331 |
-
|
| 332 |
-
# π€ Function to process text input
|
| 333 |
-
def process_text(text_input):
|
| 334 |
-
# π€ Processing text inputβtranslating human words into cosmic signals! π‘
|
| 335 |
-
if text_input:
|
| 336 |
-
if 'messages' not in st.session_state:
|
| 337 |
-
st.session_state.messages = []
|
| 338 |
-
|
| 339 |
-
st.session_state.messages.append({"role": "user", "content": text_input})
|
| 340 |
-
|
| 341 |
-
with st.chat_message("user"):
|
| 342 |
-
st.markdown(text_input)
|
| 343 |
-
|
| 344 |
-
with st.chat_message("assistant"):
|
| 345 |
-
search_glossary(text_input)
|
| 346 |
-
|
| 347 |
-
# π Function to generate a filename
|
| 348 |
-
def generate_filename(text, file_type):
|
| 349 |
-
# π Generate a filename based on the text input
|
| 350 |
-
safe_text = "".join(c if c.isalnum() or c in (' ', '.', '_') else '_' for c in text)
|
| 351 |
-
safe_text = "_".join(safe_text.strip().split())
|
| 352 |
-
filename = f"{safe_text}.{file_type}"
|
| 353 |
-
return filename
|
| 354 |
-
|
| 355 |
-
# π΅οΈββοΈ Function to extract markdown title
|
| 356 |
-
def extract_markdown_title(content):
|
| 357 |
-
# π΅οΈββοΈ Extracting markdown titleβfinding the headline in the cosmic news! π°
|
| 358 |
-
lines = content.splitlines()
|
| 359 |
-
for line in lines:
|
| 360 |
-
if line.startswith('#'):
|
| 361 |
-
return line.lstrip('#').strip()
|
| 362 |
-
return None
|
| 363 |
-
|
| 364 |
-
# πΎ Function to create and save a file
|
| 365 |
-
def create_and_save_file(content, file_type="md", prompt=None, is_image=False, should_save=True):
|
| 366 |
-
# πΎ Creating and saving a fileβcapturing cosmic wisdom! π
|
| 367 |
-
if not should_save:
|
| 368 |
-
return None
|
| 369 |
-
|
| 370 |
-
# Step 1: Generate filename based on the prompt or content
|
| 371 |
-
filename = generate_filename(prompt if prompt else content, file_type)
|
| 372 |
|
| 373 |
-
|
| 374 |
-
if file_type == "md":
|
| 375 |
-
title_from_content = extract_markdown_title(content)
|
| 376 |
-
if title_from_content:
|
| 377 |
-
filename = generate_filename(title_from_content, file_type)
|
| 378 |
-
|
| 379 |
-
# Step 3: Save the file
|
| 380 |
-
with open(filename, "w", encoding="utf-8") as f:
|
| 381 |
-
if is_image:
|
| 382 |
-
f.write(content)
|
| 383 |
-
else:
|
| 384 |
-
f.write(prompt + "\n\n" + content)
|
| 385 |
-
|
| 386 |
-
return filename
|
| 387 |
-
|
| 388 |
-
# π€ Function to insert an auto-generated record
|
| 389 |
-
def insert_auto_generated_record(container):
|
| 390 |
-
# π€ Automatically generating a record and inserting it into Cosmos DB!
|
| 391 |
-
try:
|
| 392 |
-
# Generate a unique id
|
| 393 |
-
new_id = generate_unique_id()
|
| 394 |
-
# Create a sample JSON document
|
| 395 |
-
new_doc = {
|
| 396 |
-
'id': new_id,
|
| 397 |
-
'name': f'Sample Name {new_id[:8]}',
|
| 398 |
-
'description': 'This is a sample auto-generated description.',
|
| 399 |
-
'timestamp': datetime.utcnow().isoformat()
|
| 400 |
-
}
|
| 401 |
-
# Insert the document
|
| 402 |
-
container.create_item(body=new_doc)
|
| 403 |
-
return True, f"Record inserted successfully with id: {new_id} π"
|
| 404 |
-
except exceptions.CosmosHttpResponseError as e:
|
| 405 |
-
return False, f"HTTP error occurred: {str(e)} π¨"
|
| 406 |
-
except Exception as e:
|
| 407 |
-
return False, f"An unexpected error occurred: {str(e)} π±"
|
| 408 |
-
|
| 409 |
-
|
| 410 |
-
|
| 411 |
-
# GitHub functions
|
| 412 |
-
def download_github_repo(url, local_path):
|
| 413 |
-
if os.path.exists(local_path):
|
| 414 |
-
shutil.rmtree(local_path)
|
| 415 |
-
Repo.clone_from(url, local_path)
|
| 416 |
-
|
| 417 |
-
def create_zip_file(source_dir, output_filename):
|
| 418 |
-
shutil.make_archive(output_filename, 'zip', source_dir)
|
| 419 |
-
|
| 420 |
-
def create_repo(g, repo_name):
|
| 421 |
-
user = g.get_user()
|
| 422 |
-
return user.create_repo(repo_name)
|
| 423 |
-
|
| 424 |
-
def push_to_github(local_path, repo, github_token):
|
| 425 |
-
repo_url = f"https://{github_token}@github.com/{repo.full_name}.git"
|
| 426 |
-
local_repo = Repo(local_path)
|
| 427 |
-
|
| 428 |
-
if 'origin' in [remote.name for remote in local_repo.remotes]:
|
| 429 |
-
origin = local_repo.remote('origin')
|
| 430 |
-
origin.set_url(repo_url)
|
| 431 |
-
else:
|
| 432 |
-
origin = local_repo.create_remote('origin', repo_url)
|
| 433 |
-
|
| 434 |
-
if not local_repo.heads:
|
| 435 |
-
local_repo.git.checkout('-b', 'main')
|
| 436 |
-
current_branch = 'main'
|
| 437 |
-
else:
|
| 438 |
-
current_branch = local_repo.active_branch.name
|
| 439 |
-
|
| 440 |
-
local_repo.git.add(A=True)
|
| 441 |
-
|
| 442 |
-
if local_repo.is_dirty():
|
| 443 |
-
local_repo.git.commit('-m', 'Initial commit')
|
| 444 |
-
|
| 445 |
-
origin.push(refspec=f'{current_branch}:{current_branch}')
|
| 446 |
-
|
| 447 |
-
# Main function
|
| 448 |
def main():
|
| 449 |
st.title("πGitπCosmosπ« - Azure Cosmos DB and Github Agent")
|
| 450 |
|
| 451 |
-
# Initialize session state
|
|
|
|
|
|
|
|
|
|
|
|
|
| 452 |
if 'client' not in st.session_state:
|
| 453 |
st.session_state.client = None
|
| 454 |
if 'selected_database' not in st.session_state:
|
|
@@ -459,27 +404,47 @@ def main():
|
|
| 459 |
st.session_state.selected_document_id = None
|
| 460 |
if 'current_index' not in st.session_state:
|
| 461 |
st.session_state.current_index = 0
|
| 462 |
-
if '
|
| 463 |
-
st.session_state.
|
| 464 |
-
|
| 465 |
-
|
| 466 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 467 |
|
| 468 |
-
# Automatic Login
|
| 469 |
if Key:
|
| 470 |
st.session_state.primary_key = Key
|
| 471 |
st.session_state.logged_in = True
|
| 472 |
else:
|
| 473 |
st.error("Cosmos DB Key is not set in environment variables. πβ")
|
| 474 |
-
return
|
| 475 |
|
| 476 |
if st.session_state.logged_in:
|
| 477 |
-
# Initialize Cosmos DB client
|
| 478 |
try:
|
| 479 |
if st.session_state.client is None:
|
| 480 |
st.session_state.client = CosmosClient(ENDPOINT, credential=st.session_state.primary_key)
|
| 481 |
|
| 482 |
-
# Sidebar for database, container, and document selection
|
| 483 |
st.sidebar.title("πGitπCosmosπ«ποΈNavigator")
|
| 484 |
|
| 485 |
databases = get_databases(st.session_state.client)
|
|
@@ -505,24 +470,8 @@ def main():
|
|
| 505 |
|
| 506 |
if st.session_state.selected_container:
|
| 507 |
container = database.get_container_client(st.session_state.selected_container)
|
| 508 |
-
|
| 509 |
-
if selected_db != st.session_state.selected_database:
|
| 510 |
-
st.session_state.selected_database = selected_db
|
| 511 |
-
st.session_state.selected_container = None
|
| 512 |
-
st.session_state.selected_document_id = None
|
| 513 |
-
st.session_state.current_index = 0
|
| 514 |
-
st.rerun()
|
| 515 |
-
|
| 516 |
-
|
| 517 |
-
|
| 518 |
-
|
| 519 |
-
|
| 520 |
-
|
| 521 |
-
|
| 522 |
-
|
| 523 |
-
|
| 524 |
|
| 525 |
-
# Add Export button
|
| 526 |
if st.button("π¦ Export Container Data"):
|
| 527 |
download_link = archive_current_container(st.session_state.selected_database, st.session_state.selected_container, st.session_state.client)
|
| 528 |
if download_link.startswith('<a'):
|
|
@@ -533,28 +482,6 @@ def main():
|
|
| 533 |
# Fetch documents
|
| 534 |
documents = get_documents(container)
|
| 535 |
total_docs = len(documents)
|
| 536 |
-
|
| 537 |
-
# Add buttons for saving new document and cloning
|
| 538 |
-
if st.button("πΎ Save New Document"):
|
| 539 |
-
user_query = st.session_state.get('user_query', '')
|
| 540 |
-
ai_response = st.session_state.get('ai_response', '')
|
| 541 |
-
save_or_clone_to_cosmos_db(container, query=user_query, response=ai_response)
|
| 542 |
-
|
| 543 |
-
if st.button("π Clone Document"):
|
| 544 |
-
doc_to_clone = st.selectbox("Select document to clone", [doc['id'] for doc in st.session_state.documents])
|
| 545 |
-
if doc_to_clone:
|
| 546 |
-
save_or_clone_to_cosmos_db(container, clone_id=doc_to_clone)
|
| 547 |
-
|
| 548 |
-
# Display documents
|
| 549 |
-
if st.session_state.documents:
|
| 550 |
-
for doc in st.session_state.documents:
|
| 551 |
-
st.write(f"ID: {doc['id']}")
|
| 552 |
-
st.write(f"Query: {doc.get('query', 'N/A')}")
|
| 553 |
-
st.write(f"Response: {doc.get('response', 'N/A')}")
|
| 554 |
-
st.write("---")
|
| 555 |
-
|
| 556 |
-
|
| 557 |
-
|
| 558 |
|
| 559 |
if total_docs > 5:
|
| 560 |
documents_to_display = documents[:5]
|
|
@@ -564,17 +491,17 @@ def main():
|
|
| 564 |
st.info(f"Showing all {len(documents_to_display)} documents.")
|
| 565 |
|
| 566 |
if documents_to_display:
|
| 567 |
-
# Add Viewer/Editor selection
|
| 568 |
view_options = ['Show as Markdown', 'Show as Code Editor', 'Show as Edit and Save', 'Clone Document', 'New Record']
|
| 569 |
selected_view = st.selectbox("Select Viewer/Editor", view_options, index=2)
|
| 570 |
|
| 571 |
if selected_view == 'Show as Markdown':
|
| 572 |
-
# Show each record as Markdown with navigation
|
| 573 |
total_docs = len(documents)
|
| 574 |
doc = documents[st.session_state.current_index]
|
| 575 |
st.markdown(f"#### Document ID: {doc.get('id', '')}")
|
| 576 |
|
| 577 |
-
#
|
| 578 |
values_with_space = []
|
| 579 |
def extract_values(obj):
|
| 580 |
if isinstance(obj, dict):
|
|
@@ -589,7 +516,7 @@ def main():
|
|
| 589 |
|
| 590 |
extract_values(doc)
|
| 591 |
|
| 592 |
-
#
|
| 593 |
st.markdown("#### π Links for Extracted Texts")
|
| 594 |
for term in values_with_space:
|
| 595 |
display_glossary_entity(term)
|
|
@@ -612,7 +539,7 @@ def main():
|
|
| 612 |
st.rerun()
|
| 613 |
|
| 614 |
elif selected_view == 'Show as Code Editor':
|
| 615 |
-
# Show each record in a code editor with navigation
|
| 616 |
total_docs = len(documents)
|
| 617 |
doc = documents[st.session_state.current_index]
|
| 618 |
st.markdown(f"#### Document ID: {doc.get('id', '')}")
|
|
@@ -642,7 +569,7 @@ def main():
|
|
| 642 |
st.error(f"Invalid JSON: {str(e)} π«")
|
| 643 |
|
| 644 |
elif selected_view == 'Show as Edit and Save':
|
| 645 |
-
# Show as Edit and Save in columns
|
| 646 |
st.markdown("#### Edit the document fields below:")
|
| 647 |
|
| 648 |
# Create columns for each document
|
|
@@ -680,8 +607,8 @@ def main():
|
|
| 680 |
search_glossary(json.dumps(editable_doc, indent=2))
|
| 681 |
|
| 682 |
elif selected_view == 'Clone Document':
|
| 683 |
-
# Clone Document per record
|
| 684 |
-
st.markdown("#### Clone a document:")
|
| 685 |
for idx, doc in enumerate(documents_to_display):
|
| 686 |
st.markdown(f"##### Document ID: {doc.get('id', '')}")
|
| 687 |
if st.button("π Clone Document", key=f'clone_button_{idx}'):
|
|
@@ -712,10 +639,10 @@ def main():
|
|
| 712 |
st.error(f"Invalid JSON: {str(e)} π«")
|
| 713 |
|
| 714 |
elif selected_view == 'New Record':
|
| 715 |
-
# New Record
|
| 716 |
st.markdown("#### Create a new document:")
|
| 717 |
if st.button("π€ Insert Auto-Generated Record"):
|
| 718 |
-
success, message =
|
| 719 |
if success:
|
| 720 |
st.success(message)
|
| 721 |
st.rerun()
|
|
@@ -742,7 +669,7 @@ def main():
|
|
| 742 |
else:
|
| 743 |
st.sidebar.info("No documents found in this container. π")
|
| 744 |
|
| 745 |
-
# Main content area
|
| 746 |
st.subheader(f"π Container: {st.session_state.selected_container}")
|
| 747 |
if st.session_state.selected_container:
|
| 748 |
if documents_to_display:
|
|
@@ -751,7 +678,7 @@ def main():
|
|
| 751 |
else:
|
| 752 |
st.info("No documents to display. π§")
|
| 753 |
|
| 754 |
-
# GitHub section
|
| 755 |
st.subheader("π GitHub Operations")
|
| 756 |
github_token = os.environ.get("GITHUB") # Read GitHub token from environment variable
|
| 757 |
source_repo = st.text_input("Source GitHub Repository URL", value="https://github.com/AaronCWacker/AIExamples-8-24-Streamlit")
|
|
@@ -777,7 +704,7 @@ def main():
|
|
| 777 |
os.remove(zip_filename)
|
| 778 |
else:
|
| 779 |
st.error("Please ensure GitHub token is set in environment variables and source repository URL is provided. πβ")
|
| 780 |
-
|
| 781 |
with col2:
|
| 782 |
if st.button("π€ Push to New Repository"):
|
| 783 |
if github_token and source_repo:
|
|
@@ -796,7 +723,7 @@ def main():
|
|
| 796 |
else:
|
| 797 |
st.error("Please ensure GitHub token is set in environment variables and source repository URL is provided. πβ")
|
| 798 |
|
| 799 |
-
# Chat with Claude
|
| 800 |
st.subheader("π¬ Chat with Claude")
|
| 801 |
user_input = st.text_area("Message π¨:", height=100)
|
| 802 |
|
|
@@ -818,7 +745,7 @@ def main():
|
|
| 818 |
st.session_state.chat_history.append({"user": user_input, "claude": response.content[0].text})
|
| 819 |
|
| 820 |
# Save to Cosmos DB
|
| 821 |
-
|
| 822 |
|
| 823 |
# Display Chat History
|
| 824 |
st.subheader("Past Conversations π")
|
|
@@ -873,20 +800,7 @@ def main():
|
|
| 873 |
except Exception as e:
|
| 874 |
st.error(f"An unexpected error occurred: {str(e)} π±")
|
| 875 |
|
| 876 |
-
|
| 877 |
-
# βοΈ q= Run ArXiv search from query parameters
|
| 878 |
-
try:
|
| 879 |
-
query_params = st.query_params
|
| 880 |
-
query = query_params.get('q') or query_params.get('query') or ''
|
| 881 |
-
if query:
|
| 882 |
-
# π΅οΈββοΈ We have a query! Let's process it!
|
| 883 |
-
process_text(query)
|
| 884 |
-
#st.stop() # Stop further execution
|
| 885 |
-
except Exception as e:
|
| 886 |
-
st.markdown(' ')
|
| 887 |
-
|
| 888 |
-
|
| 889 |
-
# Logout button
|
| 890 |
if st.session_state.logged_in and st.sidebar.button("πͺ Logout"):
|
| 891 |
st.session_state.logged_in = False
|
| 892 |
st.session_state.selected_records.clear()
|
|
|
|
| 18 |
import re
|
| 19 |
from PIL import Image
|
| 20 |
import zipfile
|
| 21 |
+
import hashlib
|
| 22 |
import time
|
| 23 |
|
| 24 |
+
# π App Configuration - Because every app needs a good costume!
|
| 25 |
Site_Name = 'πGitπCosmosπ« - Azure Cosmos DB and Github Agent'
|
| 26 |
title = "πGitπCosmosπ« - Azure Cosmos DB and Github Agent"
|
| 27 |
helpURL = 'https://huggingface.co/awacke1'
|
|
|
|
| 40 |
}
|
| 41 |
)
|
| 42 |
|
| 43 |
+
# π Cosmos DB configuration - Where data goes to party!
|
| 44 |
ENDPOINT = "https://acae-afd.documents.azure.com:443/"
|
| 45 |
DATABASE_NAME = os.environ.get("COSMOS_DATABASE_NAME")
|
| 46 |
CONTAINER_NAME = os.environ.get("COSMOS_CONTAINER_NAME")
|
| 47 |
Key = os.environ.get("Key")
|
| 48 |
|
| 49 |
+
# π Your local app URL - Home sweet home
|
| 50 |
LOCAL_APP_URL = "https://huggingface.co/spaces/awacke1/AzureCosmosDBUI"
|
| 51 |
|
| 52 |
+
# π€ Anthropic configuration - Teaching machines to be more human (and funnier)
|
| 53 |
client = anthropic.Anthropic(api_key=os.environ.get("ANTHROPIC_API_KEY"))
|
| 54 |
|
| 55 |
+
# π§ Initialize session state - Because even apps need a good memory
|
| 56 |
if "chat_history" not in st.session_state:
|
| 57 |
st.session_state.chat_history = []
|
| 58 |
|
| 59 |
+
# π οΈ Helper Functions - The unsung heroes of our code
|
| 60 |
+
|
| 61 |
+
# π Get a file download link - Making file sharing as easy as stealing candy from a baby
|
| 62 |
def get_download_link(file_path):
|
| 63 |
with open(file_path, "rb") as file:
|
| 64 |
contents = file.read()
|
|
|
|
| 66 |
file_name = os.path.basename(file_path)
|
| 67 |
return f'<a href="data:file/txt;base64,{b64}" download="{file_name}">Download {file_name}π</a>'
|
| 68 |
|
| 69 |
+
# π² Generate a unique ID - Because being unique is important (just ask your mother)
|
| 70 |
def generate_unique_id():
|
| 71 |
+
timestamp = datetime.utcnow().strftime('%Y%m%d%H%M%S%f')
|
| 72 |
+
unique_uuid = str(uuid.uuid4())
|
| 73 |
+
return f"{timestamp}-{unique_uuid}"
|
|
|
|
|
|
|
| 74 |
|
| 75 |
+
# π Generate a filename - Naming files like a pro (or a very confused librarian)
|
| 76 |
def generate_filename(prompt, file_type):
|
| 77 |
central = pytz.timezone('US/Central')
|
| 78 |
safe_date_time = datetime.now(central).strftime("%m%d_%H%M")
|
| 79 |
safe_prompt = re.sub(r'\W+', '', prompt)[:90]
|
| 80 |
return f"{safe_date_time}{safe_prompt}.{file_type}"
|
| 81 |
|
| 82 |
+
# πΎ Create and save a file - Because data hoarding is a legitimate hobby
|
| 83 |
def create_file(filename, prompt, response, should_save=True):
|
| 84 |
if not should_save:
|
| 85 |
return
|
| 86 |
with open(filename, 'w', encoding='utf-8') as file:
|
| 87 |
file.write(prompt + "\n\n" + response)
|
| 88 |
|
| 89 |
+
# π Load file content - Bringing words back from the digital grave
|
| 90 |
def load_file(file_name):
|
| 91 |
with open(file_name, "r", encoding='utf-8') as file:
|
| 92 |
content = file.read()
|
| 93 |
return content
|
| 94 |
|
| 95 |
+
# π Display glossary entity - Making search fun again (as if it ever was)
|
| 96 |
def display_glossary_entity(k):
|
| 97 |
search_urls = {
|
| 98 |
"ππArXiv": lambda k: f"/?q={quote(k)}",
|
|
|
|
| 103 |
links_md = ' '.join([f"<a href='{url(k)}' target='_blank'>{emoji}</a>" for emoji, url in search_urls.items()])
|
| 104 |
st.markdown(f"{k} {links_md}", unsafe_allow_html=True)
|
| 105 |
|
| 106 |
+
# ποΈ Create zip of files - Squeezing files together like sardines in a can
|
| 107 |
def create_zip_of_files(files):
|
| 108 |
zip_name = "all_files.zip"
|
| 109 |
with zipfile.ZipFile(zip_name, 'w') as zipf:
|
|
|
|
| 111 |
zipf.write(file)
|
| 112 |
return zip_name
|
| 113 |
|
| 114 |
+
# π¬ Get video HTML - Making videos play nice (or at least trying to)
|
| 115 |
def get_video_html(video_path, width="100%"):
|
| 116 |
video_url = f"data:video/mp4;base64,{base64.b64encode(open(video_path, 'rb').read()).decode()}"
|
| 117 |
return f'''
|
|
|
|
| 121 |
</video>
|
| 122 |
'''
|
| 123 |
|
| 124 |
+
# π΅ Get audio HTML - Let the music play (and hope it's not Baby Shark)
|
| 125 |
def get_audio_html(audio_path, width="100%"):
|
| 126 |
audio_url = f"data:audio/mpeg;base64,{base64.b64encode(open(audio_path, 'rb').read()).decode()}"
|
| 127 |
return f'''
|
|
|
|
| 131 |
</audio>
|
| 132 |
'''
|
| 133 |
|
| 134 |
+
# π Cosmos DB functions - Where data goes to live its best life
|
| 135 |
+
|
| 136 |
+
# π Get databases - Collecting databases like Pokemon cards
|
| 137 |
def get_databases(client):
|
| 138 |
return [db['id'] for db in client.list_databases()]
|
| 139 |
|
| 140 |
+
# π¦ Get containers - Finding where all the good stuff is hidden
|
| 141 |
def get_containers(database):
|
| 142 |
return [container['id'] for container in database.list_containers()]
|
| 143 |
|
| 144 |
+
# π Get documents - Retrieving the sacred texts (or just some JSON)
|
| 145 |
def get_documents(container, limit=None):
|
| 146 |
query = "SELECT * FROM c ORDER BY c._ts DESC"
|
| 147 |
items = list(container.query_items(query=query, enable_cross_partition_query=True, max_item_count=limit))
|
| 148 |
return items
|
| 149 |
|
| 150 |
+
# π₯ Insert record - Adding new data (and crossing fingers it doesn't break anything)
|
| 151 |
def insert_record(container, record):
|
| 152 |
try:
|
| 153 |
container.create_item(body=record)
|
|
|
|
| 157 |
except Exception as e:
|
| 158 |
return False, f"An unexpected error occurred: {str(e)} π±"
|
| 159 |
|
| 160 |
+
# π Update record - Giving data a makeover
|
| 161 |
def update_record(container, updated_record):
|
| 162 |
try:
|
| 163 |
container.upsert_item(body=updated_record)
|
|
|
|
| 167 |
except Exception as e:
|
| 168 |
return False, f"An unexpected error occurred: {traceback.format_exc()} π±"
|
| 169 |
|
| 170 |
+
# ποΈ Delete record - Saying goodbye to data (it's not you, it's me)
|
| 171 |
def delete_record(container, name, id):
|
| 172 |
try:
|
| 173 |
container.delete_item(item=id, partition_key=id)
|
|
|
|
| 179 |
except Exception as e:
|
| 180 |
return False, f"An unexpected error occurred: {traceback.format_exc()} π±"
|
| 181 |
|
| 182 |
+
# πΎ Save to Cosmos DB - Preserving data for future generations (or just until the next update)
|
| 183 |
+
def save_to_cosmos_db(container, query, response1, response2):
|
| 184 |
+
try:
|
| 185 |
+
if container:
|
| 186 |
+
record = {
|
| 187 |
+
"id": generate_unique_id(),
|
| 188 |
+
"query": query,
|
| 189 |
+
"response1": response1,
|
| 190 |
+
"response2": response2,
|
| 191 |
+
"timestamp": datetime.utcnow().isoformat()
|
| 192 |
+
}
|
| 193 |
+
try:
|
| 194 |
+
container.create_item(body=record)
|
| 195 |
+
st.success(f"Record saved successfully with ID: {record['id']}")
|
| 196 |
+
# Refresh the documents display
|
| 197 |
+
st.session_state.documents = get_documents(container)
|
| 198 |
+
except exceptions.CosmosHttpResponseError as e:
|
| 199 |
+
st.error(f"Error saving record to Cosmos DB: {e}")
|
| 200 |
+
else:
|
| 201 |
+
st.error("Cosmos DB container is not initialized.")
|
| 202 |
+
except Exception as e:
|
| 203 |
+
st.error(f"An unexpected error occurred: {str(e)}")
|
| 204 |
|
| 205 |
+
# π GitHub functions - Where code goes to socialize
|
| 206 |
+
|
| 207 |
+
# π₯ Download GitHub repo - Cloning repos like it's going out of style
|
| 208 |
+
def download_github_repo(url, local_path):
|
| 209 |
+
if os.path.exists(local_path):
|
| 210 |
+
shutil.rmtree(local_path)
|
| 211 |
+
Repo.clone_from(url, local_path)
|
| 212 |
+
|
| 213 |
+
# ποΈ Create zip file - Squeezing files tighter than your budget
|
| 214 |
+
def create_zip_file(source_dir, output_filename):
|
| 215 |
+
shutil.make_archive(output_filename, 'zip', source_dir)
|
| 216 |
+
|
| 217 |
+
# ποΈ Create repo - Building digital homes for lonely code
|
| 218 |
+
def create_repo(g, repo_name):
|
| 219 |
+
user = g.get_user()
|
| 220 |
+
return user.create_repo(repo_name)
|
| 221 |
+
|
| 222 |
+
# π Push to GitHub - Sending code to the cloud (hopefully not the rainy kind)
|
| 223 |
+
def push_to_github(local_path, repo, github_token):
|
| 224 |
+
repo_url = f"https://{github_token}@github.com/{repo.full_name}.git"
|
| 225 |
+
local_repo = Repo(local_path)
|
| 226 |
+
|
| 227 |
+
if 'origin' in [remote.name for remote in local_repo.remotes]:
|
| 228 |
+
origin = local_repo.remote('origin')
|
| 229 |
+
origin.set_url(repo_url)
|
| 230 |
+
else:
|
| 231 |
+
origin = local_repo.create_remote('origin', repo_url)
|
| 232 |
+
|
| 233 |
+
if not local_repo.heads:
|
| 234 |
+
local_repo.git.checkout('-b', 'main')
|
| 235 |
+
current_branch = 'main'
|
| 236 |
+
else:
|
| 237 |
+
current_branch = local_repo.active_branch.name
|
| 238 |
+
|
| 239 |
+
local_repo.git.add(A=True)
|
| 240 |
+
|
| 241 |
+
if local_repo.is_dirty():
|
| 242 |
+
local_repo.git.commit('-m', 'Initial commit')
|
| 243 |
+
|
| 244 |
+
origin.push(refspec=f'{current_branch}:{current_branch}')
|
| 245 |
+
|
| 246 |
+
# πΎ Save or clone to Cosmos DB - Because every document deserves a twin
|
| 247 |
def save_or_clone_to_cosmos_db(container, query=None, response=None, clone_id=None):
|
| 248 |
try:
|
| 249 |
if not container:
|
|
|
|
| 292 |
except Exception as e:
|
| 293 |
st.error(f"An unexpected error occurred: {str(e)}")
|
| 294 |
|
| 295 |
+
# π¦ Archive current container - Packing up data like you're moving to a new digital house
|
| 296 |
+
def archive_current_container(database_name, container_name, client):
|
| 297 |
+
try:
|
| 298 |
+
base_dir = "./cosmos_archive_current_container"
|
| 299 |
+
if os.path.exists(base_dir):
|
| 300 |
+
shutil.rmtree(base_dir)
|
| 301 |
+
os.makedirs(base_dir)
|
| 302 |
+
|
| 303 |
+
db_client = client.get_database_client(database_name)
|
| 304 |
+
container_client = db_client.get_container_client(container_name)
|
| 305 |
+
items = list(container_client.read_all_items())
|
| 306 |
+
|
| 307 |
+
container_dir = os.path.join(base_dir, container_name)
|
| 308 |
+
os.makedirs(container_dir)
|
| 309 |
+
|
| 310 |
+
for item in items:
|
| 311 |
+
item_id = item.get('id', f"unknown_{datetime.now().strftime('%Y%m%d%H%M%S')}")
|
| 312 |
+
with open(os.path.join(container_dir, f"{item_id}.json"), 'w') as f:
|
| 313 |
+
json.dump(item, f, indent=2)
|
| 314 |
+
|
| 315 |
+
archive_name = f"{container_name}_archive_{datetime.now().strftime('%Y%m%d%H%M%S')}"
|
| 316 |
+
shutil.make_archive(archive_name, 'zip', base_dir)
|
| 317 |
+
|
| 318 |
+
return get_download_link(f"{archive_name}.zip")
|
| 319 |
+
except Exception as e:
|
| 320 |
+
return f"An error occurred while archiving data: {str(e)} π’"
|
| 321 |
|
| 322 |
+
# π Search glossary - Finding needles in digital haystacks
|
| 323 |
def search_glossary(query):
|
| 324 |
+
st.markdown(f"### π SearchGlossary for: {query}")
|
| 325 |
+
|
| 326 |
# Dropdown for model selection
|
| 327 |
model_options = ['mistralai/Mixtral-8x7B-Instruct-v0.1', 'mistralai/Mistral-7B-Instruct-v0.2', 'google/gemma-7b-it', 'None']
|
| 328 |
model_choice = st.selectbox('π§ Select LLM Model', options=model_options, index=1)
|
| 329 |
+
|
| 330 |
# Dropdown for database selection
|
| 331 |
database_options = ['Semantic Search', 'Arxiv Search - Latest - (EXPERIMENTAL)']
|
| 332 |
database_choice = st.selectbox('π Select Database', options=database_options, index=0)
|
| 333 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 334 |
# π΅οΈββοΈ Searching the glossary for: query
|
| 335 |
all_results = ""
|
| 336 |
st.markdown(f"- {query}")
|
| 337 |
|
| 338 |
# π ArXiv RAG researcher expert ~-<>-~ Paper Summary & Ask LLM
|
|
|
|
|
|
|
| 339 |
client = Client("awacke1/Arxiv-Paper-Search-And-QA-RAG-Pattern")
|
| 340 |
|
|
|
|
| 341 |
# π ArXiv RAG researcher expert ~-<>-~ Paper Summary & Ask LLM - api_name: /ask_llm
|
| 342 |
result = client.predict(
|
| 343 |
prompt=query,
|
|
|
|
| 347 |
)
|
| 348 |
st.markdown(result)
|
| 349 |
st.code(result, language="python", line_numbers=True)
|
| 350 |
+
|
| 351 |
# π ArXiv RAG researcher expert ~-<>-~ Paper Summary & Ask LLM - api_name: /ask_llm
|
| 352 |
result2 = client.predict(
|
| 353 |
prompt=query,
|
|
|
|
| 368 |
st.markdown(result3)
|
| 369 |
st.code(result3, language="python", line_numbers=True)
|
| 370 |
|
|
|
|
| 371 |
# π ArXiv RAG researcher expert ~-<>-~ Paper Summary & Ask LLM - api_name: /update_with_rag_md
|
| 372 |
response2 = client.predict(
|
| 373 |
message=query, # str in 'parameter_13' Textbox component
|
|
|
|
| 375 |
database_choice="Semantic Search",
|
| 376 |
llm_model_picked="mistralai/Mistral-7B-Instruct-v0.2",
|
| 377 |
api_name="/update_with_rag_md"
|
| 378 |
+
)
|
| 379 |
|
| 380 |
st.markdown(response2[0])
|
| 381 |
st.code(response2[0], language="python", line_numbers=True, wrap_lines=True)
|
| 382 |
+
|
| 383 |
st.markdown(response2[1])
|
| 384 |
st.code(response2[1], language="python", line_numbers=True, wrap_lines=True)
|
| 385 |
|
| 386 |
+
return result, result2, result3, response2
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 387 |
|
| 388 |
+
# π Main function - Where the magic happens (and occasionally breaks)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 389 |
def main():
|
| 390 |
st.title("πGitπCosmosπ« - Azure Cosmos DB and Github Agent")
|
| 391 |
|
| 392 |
+
# π¦ Initialize session state
|
| 393 |
+
if 'logged_in' not in st.session_state:
|
| 394 |
+
st.session_state.logged_in = False
|
| 395 |
+
if 'selected_records' not in st.session_state:
|
| 396 |
+
st.session_state.selected_records = []
|
| 397 |
if 'client' not in st.session_state:
|
| 398 |
st.session_state.client = None
|
| 399 |
if 'selected_database' not in st.session_state:
|
|
|
|
| 404 |
st.session_state.selected_document_id = None
|
| 405 |
if 'current_index' not in st.session_state:
|
| 406 |
st.session_state.current_index = 0
|
| 407 |
+
if 'cloned_doc' not in st.session_state:
|
| 408 |
+
st.session_state.cloned_doc = None
|
|
|
|
|
|
|
| 409 |
|
| 410 |
+
# βοΈ q= Run ArXiv search from query parameters
|
| 411 |
+
try:
|
| 412 |
+
query_params = st.query_params
|
| 413 |
+
query = query_params.get('q') or query_params.get('query') or ''
|
| 414 |
+
if query:
|
| 415 |
+
# π΅οΈββοΈ We have a query! Let's process it!
|
| 416 |
+
result, result2, result3, response2 = search_glossary(query)
|
| 417 |
+
# When saving results, pass the container
|
| 418 |
+
try:
|
| 419 |
+
save_to_cosmos_db(st.session_state.cosmos_container, query, result, result)
|
| 420 |
+
save_to_cosmos_db(st.session_state.cosmos_container, query, result2, result2)
|
| 421 |
+
save_to_cosmos_db(st.session_state.cosmos_container, query, result3, result3)
|
| 422 |
+
save_to_cosmos_db(st.session_state.cosmos_container, query, response2[0], response2[0])
|
| 423 |
+
save_to_cosmos_db(st.session_state.cosmos_container, query, response2[1], response2[1])
|
| 424 |
+
except exceptions.CosmosHttpResponseError as e:
|
| 425 |
+
st.error(f"HTTP error occurred: {str(e)} π¨")
|
| 426 |
+
except Exception as e:
|
| 427 |
+
st.error(f"An unexpected error occurred: {str(e)} π±")
|
| 428 |
+
|
| 429 |
+
st.stop() # Stop further execution
|
| 430 |
+
except Exception as e:
|
| 431 |
+
st.markdown(' ')
|
| 432 |
|
| 433 |
+
# π Automatic Login
|
| 434 |
if Key:
|
| 435 |
st.session_state.primary_key = Key
|
| 436 |
st.session_state.logged_in = True
|
| 437 |
else:
|
| 438 |
st.error("Cosmos DB Key is not set in environment variables. πβ")
|
| 439 |
+
return # Can't proceed without a key
|
| 440 |
|
| 441 |
if st.session_state.logged_in:
|
| 442 |
+
# π Initialize Cosmos DB client
|
| 443 |
try:
|
| 444 |
if st.session_state.client is None:
|
| 445 |
st.session_state.client = CosmosClient(ENDPOINT, credential=st.session_state.primary_key)
|
| 446 |
|
| 447 |
+
# ποΈ Sidebar for database, container, and document selection
|
| 448 |
st.sidebar.title("πGitπCosmosπ«ποΈNavigator")
|
| 449 |
|
| 450 |
databases = get_databases(st.session_state.client)
|
|
|
|
| 470 |
|
| 471 |
if st.session_state.selected_container:
|
| 472 |
container = database.get_container_client(st.session_state.selected_container)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 473 |
|
| 474 |
+
# π¦ Add Export button
|
| 475 |
if st.button("π¦ Export Container Data"):
|
| 476 |
download_link = archive_current_container(st.session_state.selected_database, st.session_state.selected_container, st.session_state.client)
|
| 477 |
if download_link.startswith('<a'):
|
|
|
|
| 482 |
# Fetch documents
|
| 483 |
documents = get_documents(container)
|
| 484 |
total_docs = len(documents)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 485 |
|
| 486 |
if total_docs > 5:
|
| 487 |
documents_to_display = documents[:5]
|
|
|
|
| 491 |
st.info(f"Showing all {len(documents_to_display)} documents.")
|
| 492 |
|
| 493 |
if documents_to_display:
|
| 494 |
+
# π¨ Add Viewer/Editor selection
|
| 495 |
view_options = ['Show as Markdown', 'Show as Code Editor', 'Show as Edit and Save', 'Clone Document', 'New Record']
|
| 496 |
selected_view = st.selectbox("Select Viewer/Editor", view_options, index=2)
|
| 497 |
|
| 498 |
if selected_view == 'Show as Markdown':
|
| 499 |
+
# ποΈ Show each record as Markdown with navigation
|
| 500 |
total_docs = len(documents)
|
| 501 |
doc = documents[st.session_state.current_index]
|
| 502 |
st.markdown(f"#### Document ID: {doc.get('id', '')}")
|
| 503 |
|
| 504 |
+
# π΅οΈββοΈ Let's extract values from the JSON that have at least one space
|
| 505 |
values_with_space = []
|
| 506 |
def extract_values(obj):
|
| 507 |
if isinstance(obj, dict):
|
|
|
|
| 516 |
|
| 517 |
extract_values(doc)
|
| 518 |
|
| 519 |
+
# π Let's create a list of links for these values
|
| 520 |
st.markdown("#### π Links for Extracted Texts")
|
| 521 |
for term in values_with_space:
|
| 522 |
display_glossary_entity(term)
|
|
|
|
| 539 |
st.rerun()
|
| 540 |
|
| 541 |
elif selected_view == 'Show as Code Editor':
|
| 542 |
+
# π» Show each record in a code editor with navigation
|
| 543 |
total_docs = len(documents)
|
| 544 |
doc = documents[st.session_state.current_index]
|
| 545 |
st.markdown(f"#### Document ID: {doc.get('id', '')}")
|
|
|
|
| 569 |
st.error(f"Invalid JSON: {str(e)} π«")
|
| 570 |
|
| 571 |
elif selected_view == 'Show as Edit and Save':
|
| 572 |
+
# βοΈ Show as Edit and Save in columns
|
| 573 |
st.markdown("#### Edit the document fields below:")
|
| 574 |
|
| 575 |
# Create columns for each document
|
|
|
|
| 607 |
search_glossary(json.dumps(editable_doc, indent=2))
|
| 608 |
|
| 609 |
elif selected_view == 'Clone Document':
|
| 610 |
+
# 𧬠Clone Document per record
|
| 611 |
+
st.markdownst.markdown("#### Clone a document:")
|
| 612 |
for idx, doc in enumerate(documents_to_display):
|
| 613 |
st.markdown(f"##### Document ID: {doc.get('id', '')}")
|
| 614 |
if st.button("π Clone Document", key=f'clone_button_{idx}'):
|
|
|
|
| 639 |
st.error(f"Invalid JSON: {str(e)} π«")
|
| 640 |
|
| 641 |
elif selected_view == 'New Record':
|
| 642 |
+
# π New Record
|
| 643 |
st.markdown("#### Create a new document:")
|
| 644 |
if st.button("π€ Insert Auto-Generated Record"):
|
| 645 |
+
success, message = save_or_clone_to_cosmos_db(container, query="Auto-generated", response="This is an auto-generated record.")
|
| 646 |
if success:
|
| 647 |
st.success(message)
|
| 648 |
st.rerun()
|
|
|
|
| 669 |
else:
|
| 670 |
st.sidebar.info("No documents found in this container. π")
|
| 671 |
|
| 672 |
+
# π Main content area
|
| 673 |
st.subheader(f"π Container: {st.session_state.selected_container}")
|
| 674 |
if st.session_state.selected_container:
|
| 675 |
if documents_to_display:
|
|
|
|
| 678 |
else:
|
| 679 |
st.info("No documents to display. π§")
|
| 680 |
|
| 681 |
+
# π GitHub section
|
| 682 |
st.subheader("π GitHub Operations")
|
| 683 |
github_token = os.environ.get("GITHUB") # Read GitHub token from environment variable
|
| 684 |
source_repo = st.text_input("Source GitHub Repository URL", value="https://github.com/AaronCWacker/AIExamples-8-24-Streamlit")
|
|
|
|
| 704 |
os.remove(zip_filename)
|
| 705 |
else:
|
| 706 |
st.error("Please ensure GitHub token is set in environment variables and source repository URL is provided. πβ")
|
| 707 |
+
|
| 708 |
with col2:
|
| 709 |
if st.button("π€ Push to New Repository"):
|
| 710 |
if github_token and source_repo:
|
|
|
|
| 723 |
else:
|
| 724 |
st.error("Please ensure GitHub token is set in environment variables and source repository URL is provided. πβ")
|
| 725 |
|
| 726 |
+
# π¬ Chat with Claude
|
| 727 |
st.subheader("π¬ Chat with Claude")
|
| 728 |
user_input = st.text_area("Message π¨:", height=100)
|
| 729 |
|
|
|
|
| 745 |
st.session_state.chat_history.append({"user": user_input, "claude": response.content[0].text})
|
| 746 |
|
| 747 |
# Save to Cosmos DB
|
| 748 |
+
save_to_cosmos_db(container, user_input, response.content[0].text, "")
|
| 749 |
|
| 750 |
# Display Chat History
|
| 751 |
st.subheader("Past Conversations π")
|
|
|
|
| 800 |
except Exception as e:
|
| 801 |
st.error(f"An unexpected error occurred: {str(e)} π±")
|
| 802 |
|
| 803 |
+
# πͺ Logout button
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 804 |
if st.session_state.logged_in and st.sidebar.button("πͺ Logout"):
|
| 805 |
st.session_state.logged_in = False
|
| 806 |
st.session_state.selected_records.clear()
|