Spaces:
Sleeping
Sleeping
File size: 21,368 Bytes
8c85b97 b812a47 0cfaa67 8c85b97 b812a47 8c85b97 b812a47 8c85b97 b812a47 8c85b97 b812a47 8c85b97 0cfaa67 8c85b97 0cfaa67 8c85b97 b812a47 8c85b97 0cfaa67 8c85b97 b812a47 0cfaa67 b812a47 8c85b97 b812a47 8c85b97 b812a47 0cfaa67 b812a47 8c85b97 b812a47 8c85b97 b812a47 8c85b97 b812a47 8c85b97 b812a47 8c85b97 b812a47 8c85b97 b812a47 8c85b97 b812a47 8c85b97 b812a47 8c85b97 b812a47 8c85b97 b812a47 8c85b97 b812a47 8c85b97 b812a47 8c85b97 0cfaa67 8c85b97 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 | """
LangChain Tools for Meeting Intelligence Agent
This module defines tools that can be used by LangChain agents to interact
with meeting transcripts stored in Pinecone.
Tools follow the official @tool decorator pattern from LangChain.
Reference: https://docs.langchain.com/oss/python/langchain/tools#create-tools
"""
from typing import List, Dict, Any, Optional
from datetime import datetime
import uuid
import requests
import traceback
from langchain.tools import tool
from langchain_core.documents import Document
from src.retrievers.pipeline import process_transcript_to_documents
from src.processing.metadata_extractor import MetadataExtractor
from src.config.settings import Config
# Global reference to PineconeManager (will be set during initialization)
_pinecone_manager = None
def initialize_tools(pinecone_manager):
"""
Initialize tools with a PineconeManager instance.
Args:
pinecone_manager: Instance of PineconeManager for database access
"""
global _pinecone_manager
_pinecone_manager = pinecone_manager
@tool
def search_meetings(query: str, max_results: int = 5, meeting_id: Optional[str] = None) -> str:
"""
Search meeting transcripts for relevant information using semantic search.
Use this tool when you need to find specific information across meeting transcripts.
The search uses AI-powered semantic matching to find the most relevant segments.
Args:
query: The search query or question to find relevant meeting content
max_results: Maximum number of results to return (default: 5)
meeting_id: Optional meeting ID to search within a specific meeting (e.g., "meeting_abc12345"). DO NOT use indices like "1" or "2".
Returns:
A formatted string containing the most relevant meeting transcript segments
Example:
search_meetings("What were the action items?", max_results=3)
search_meetings("budget discussion", meeting_id="meeting_abc12345")
"""
if not _pinecone_manager:
return "Error: Pinecone service is not initialized. Cannot search meetings."
try:
# Build search kwargs
search_kwargs = {"k": max_results}
# Add meeting_id filter if provided
if meeting_id:
search_kwargs["filter"] = {"meeting_id": {"$eq": meeting_id}}
# Get retriever and perform search
retriever = _pinecone_manager.get_retriever(
namespace=Config.PINECONE_NAMESPACE,
search_kwargs=search_kwargs
)
docs = retriever.invoke(query)
if not docs:
return "No relevant meeting segments found for your query."
# Format results
result_parts = [f"Found {len(docs)} relevant meeting segments:\n"]
for i, doc in enumerate(docs, 1):
metadata = doc.metadata
meeting_id = metadata.get("meeting_id", "unknown")
meeting_date = metadata.get("meeting_date", "N/A") # Fixed missing variable
meeting_title = metadata.get("meeting_title", "Untitled") # Added title
chunk_index = metadata.get("chunk_index", "?")
summary = metadata.get("summary", "N/A")
speakers = metadata.get("speaker_mapping", "N/A")
result_parts.append(
f"\n--- Segment {i} ---\n"
f"Meeting: {meeting_title} (ID: {meeting_id})\n"
f"Date: {meeting_date}\n"
f"Summary: {summary}\n"
f"Speakers: {speakers}\n"
f"Chunk: {chunk_index}\n"
f"Content:\n{doc.page_content}\n"
)
return "".join(result_parts)
except Exception as e:
traceback.print_exc()
return f"Error searching meetings: {str(e)}"
@tool
def get_meeting_metadata(meeting_id: str) -> str:
"""
Retrieve metadata and summary information for a specific meeting.
Use this tool when you need to get details about a specific meeting,
such as date, title, participants, or other metadata.
Args:
meeting_id: The unique identifier for the meeting (e.g., "meeting_abc12345")
Returns:
A formatted string containing the meeting's metadata
Example:
get_meeting_metadata("meeting_abc12345")
"""
if not _pinecone_manager:
return "Error: Pinecone service is not initialized. Cannot retrieve metadata."
try:
# Search for any document from this meeting to get metadata
retriever = _pinecone_manager.get_retriever(
namespace=Config.PINECONE_NAMESPACE,
search_kwargs={
"k": 1,
"filter": {"meeting_id": {"$eq": meeting_id}}
}
)
# Use a generic query to get any chunk from this meeting
docs = retriever.invoke("meeting content")
if not docs:
return f"No meeting found with ID: {meeting_id}"
# Extract metadata from the first document
metadata = docs[0].metadata
result_parts = [
f"Meeting Information for {meeting_id}:\n",
f"- Date: {metadata.get('meeting_date', 'N/A')}", # β
Fixed: was 'date'
f"- Title: {metadata.get('meeting_title', 'N/A')}", # β
Fixed: was 'title'
f"- Summary: {metadata.get('summary', 'N/A')}", # β
Added summary
f"- Source: {metadata.get('source', 'N/A')}",
f"- Source File: {metadata.get('source_file', 'N/A')}",
f"- Language: {metadata.get('language', 'N/A')}",
f"- Transcription Model: {metadata.get('transcription_model', 'N/A')}",
f"- Duration: {metadata.get('meeting_duration', 'N/A')}", # β
Added duration
]
return "\n".join(result_parts)
except Exception as e:
traceback.print_exc()
return f"Error retrieving meeting metadata: {str(e)}"
@tool
def list_recent_meetings(limit: int = 10) -> str:
"""
Get a list of recent meetings stored in the system.
Use this tool when you need to see what meetings are available,
or to help the user understand what they can ask about.
Args:
limit: Maximum number of meetings to return (default: 10)
Returns:
A formatted string listing recent meetings with their IDs and dates
Example:
list_recent_meetings(limit=5)
"""
if not _pinecone_manager:
return "Error: Pinecone service is not initialized. Cannot list meetings."
try:
# Get retriever with high k to fetch many documents
retriever = _pinecone_manager.get_retriever(
namespace=Config.PINECONE_NAMESPACE,
search_kwargs={"k": 500} # Fetch many to find unique meetings
)
# Use a generic query to get documents
docs = retriever.invoke("meeting")
if not docs:
return "No meetings found in the system."
# Extract unique meetings
meetings_dict = {}
for doc in docs:
metadata = doc.metadata
meeting_id = metadata.get("meeting_id")
if meeting_id and meeting_id not in meetings_dict:
meetings_dict[meeting_id] = {
"date": metadata.get("meeting_date", "N/A"), # β
Fixed: was "date"
"title": metadata.get("meeting_title", "N/A"), # β
Fixed: was "title"
"source_file": metadata.get("source_file", "N/A")
}
# Stop if we've found enough unique meetings
if len(meetings_dict) >= limit:
break
if not meetings_dict:
return "No meetings found in the system."
# Format results
result_parts = [f"Found {len(meetings_dict)} recent meetings:\n"]
for i, (meeting_id, info) in enumerate(meetings_dict.items(), 1):
result_parts.append(
f"\n{i}. {meeting_id}\n"
f" Date: {info['date']}\n"
f" Title: {info['title']}\n"
f" Source: {info['source_file']}"
)
return "\n".join(result_parts)
except Exception as e:
traceback.print_exc()
return f"Error listing meetings: {str(e)}"
@tool
def get_current_time() -> str:
"""
Get the current date and time.
Use this tool when you need to answer questions about relative time
(e.g., "what happened yesterday?", "meetings from last week?").
Returns:
Current date and time in YYYY-MM-DD HH:MM format
"""
return datetime.now().strftime("%Y-%m-%d %H:%M")
@tool
def import_notion_to_pinecone(query: str) -> str:
"""
Directly import a Notion page to Pinecone by name.
Fetch a Notion page and save it TO Pinecone.
Use this tool ONLY when the user wants to *Import* or *Sync* a page FROM Notion INTO the database.
Do NOT use this tool to write content TO Notion. Use `API-post-page` or `API-append-block-children` for that.
This tool handles the entire process (Search -> Fetch Content -> Upsert) automatically.
Args:
query: The name of the Notion page to find (e.g., "Meeting 1").
Returns:
Status message indicating success or failure.
"""
if not Config.NOTION_TOKEN:
return "β Error: NOTION_TOKEN not set in configuration."
headers = {
"Authorization": f"Bearer {Config.NOTION_TOKEN}",
"Notion-Version": "2022-06-28",
"Content-Type": "application/json"
}
def fetch_blocks_recursive(block_id: str, depth: int = 0) -> List[str]:
"""Recursive helper to fetch blocks and their children."""
if depth > 5: # Safety limit for recursion depth
return []
collected_text = []
cursor = None
has_more = True
while has_more:
blocks_url = f"https://api.notion.com/v1/blocks/{block_id}/children"
params = {"page_size": 100}
if cursor:
params["start_cursor"] = cursor
resp = requests.get(blocks_url, headers=headers, params=params)
if resp.status_code != 200:
print(f"β οΈ Error fetching sub-blocks for {block_id}: {resp.text}")
return []
data = resp.json()
blocks = data.get("results", [])
for block in blocks:
# 1. Extract text from this block
b_type = block.get("type")
plain_text = ""
if b_type and block.get(b_type) and "rich_text" in block[b_type]:
rich_text = block[b_type]["rich_text"]
plain_text = "".join([t.get("plain_text", "") for t in rich_text])
# Append text if present
if plain_text.strip():
collected_text.append(plain_text)
# 2. Check for children (Recursion)
if block.get("has_children", False):
# Fetch children text and append
children_text = fetch_blocks_recursive(block["id"], depth + 1)
collected_text.extend(children_text)
has_more = data.get("has_more", False)
cursor = data.get("next_cursor")
return collected_text
try:
# 1. Search for the page
print(f"π Searching Notion for: {query}...")
search_url = "https://api.notion.com/v1/search"
search_payload = {
"query": query,
"filter": {"value": "page", "property": "object"},
"sort": {"direction": "descending", "timestamp": "last_edited_time"},
"page_size": 25
}
response = requests.post(search_url, headers=headers, json=search_payload)
if response.status_code != 200:
return f"β Notion Search Error: {response.text}"
results = response.json().get("results", [])
if not results:
return f"β No Notion page found matching '{query}'."
# Select best match
best_page = None
exact_match = None
substring_match = None
for p in results:
# Extract title for this page
p_props = p.get("properties", {})
p_title_prop = next((v for k, v in p_props.items() if v["id"] == "title"), None)
p_title = ""
if p_title_prop and p_title_prop.get("title"):
p_title = "".join([t.get("plain_text", "") for t in p_title_prop.get("title", [])])
p_title_clean = p_title.lower().strip()
query_clean = query.lower().strip()
# Check 1: Exact Match
if p_title_clean == query_clean:
exact_match = p
print(f"β
Exact match found: '{p_title}'")
break # Found the perfect match
# Check 2: Substring Match (save the first one found)
# Check 2: Substring Match (save the first one found)
if query_clean in p_title_clean and substring_match is None:
substring_match = p
print(f"π Substring match candidate: '{p_title}'")
# Print for debugging
print(f" - Found result: '{p_title}'")
# Decide which page to use
if exact_match:
best_page = exact_match
elif substring_match:
best_page = substring_match
print("β οΈ Using substring match.")
else:
# Generate list of titles found to guide the user
titles_found = []
for p in results:
p_props = p.get("properties", {})
p_title_prop = next((v for k, v in p_props.items() if v["id"] == "title"), None)
if p_title_prop and p_title_prop.get("title"):
titles_found.append("".join([t.get("plain_text", "") for t in p_title_prop.get("title", [])]))
return f"β Could not find a specific match for '{query}'. Found these pages instead: {', '.join(titles_found)}. Please try again with the exact name."
page = best_page
page_id = page["id"]
# Re-extract title for the selected page for final usage
props = page.get("properties", {})
title_prop = next((v for k, v in props.items() if v["id"] == "title"), None)
title = "Untitled"
if title_prop and title_prop.get("title"):
title = "".join([t.get("plain_text", "") for t in title_prop.get("title", [])])
print(f"π Found Page: '{title}' ({page_id})")
# 2. Recursive Fetch of All Content
all_text_lines = fetch_blocks_recursive(page_id)
if not all_text_lines:
return f"β οΈ Page '{title}' found but appears empty or has no text blocks."
full_content = "\n\n".join(all_text_lines)
# 3. Upsert to Pinecone
return upsert_text_to_pinecone.invoke({"text": full_content, "title": title, "source": "Notion"})
except Exception as e:
traceback.print_exc()
return f"β Import failed: {str(e)}"
# Export all tools for easy import
__all__ = [
"initialize_tools",
"search_meetings",
"get_meeting_metadata",
"list_recent_meetings",
"upsert_text_to_pinecone",
"import_notion_to_pinecone",
"create_notion_page",
"get_current_time"
]
@tool
def create_notion_page(title: str, content: str) -> str:
"""
Create a new page in Notion with a Title and Text Content.
Use this tool for ANY request to "Write to Notion", "Save to Notion", "Create a page", "Draft an email in Notion".
This tool handles all the formatting automatically.
Args:
title: The title of the new page.
content: The text content of the page.
Returns:
Status message with link to the new page.
"""
if not Config.NOTION_TOKEN:
return "β Error: NOTION_TOKEN not set."
headers = {
"Authorization": f"Bearer {Config.NOTION_TOKEN}",
"Notion-Version": "2022-06-28",
"Content-Type": "application/json"
}
# Split content into chunks of 2000 chars (Notion block limit)
chunks = [content[i:i+2000] for i in range(0, len(content), 2000)]
children_blocks = []
for chunk in chunks:
children_blocks.append({
"object": "block",
"type": "paragraph",
"paragraph": {
"rich_text": [{"type": "text", "text": {"content": chunk}}]
}
})
# Default parent page: Meetings Summary Test
parent_page_id = "2bc5a424-5cbb-80ec-8aa9-c4fd989e67bc"
payload = {
"parent": {"page_id": parent_page_id},
"properties": {
"title": [
{
"text": {
"content": title
}
}
]
},
"children": children_blocks
}
try:
url = "https://api.notion.com/v1/pages"
resp = requests.post(url, headers=headers, json=payload)
if resp.status_code == 200:
data = resp.json()
url = data.get('url', 'URL not found')
return f"β
Successfully created Notion page: '{title}'.\nLink: {url}"
else:
return f"β Failed to create Notion page: {resp.status_code} - {resp.text}"
except Exception as e:
traceback.print_exc()
return f"β Error creating page: {str(e)}"
@tool
def upsert_text_to_pinecone(text: str, title: str, source: str = "Manual Entry", date: str = None) -> str:
"""
Upsert any text content (e.g., Notion pages, manual notes) to Pinecone.
Automatically extracts metadata (summary, date, speakers) from the text.
Use this tool when retrieving full content from Notion or other sources.
CRITICAL: Do NOT use this tool if the user wants to "Save to Notion" or "Create a Page".
Use the Notion tools (`API-post-page`) for that. Use this ONLY for saving to Pinecone/Database.
Args:
text: The FULL content to save (do not summarize!)
title: Title of the document/meeting
source: Source of the content (e.g., "Notion", "Manual Entry")
date: Optional date override (YYYY-MM-DD). If not provided, AI extracts it from text or uses today.
Returns:
Success message with the generated meeting_id
"""
if not _pinecone_manager:
return "Error: Pinecone service is not initialized."
try:
# 1. Extract intelligent metadata
print(f"π Extracting metadata for '{title}'...")
extractor = MetadataExtractor()
extracted = extractor.extract_metadata(text)
# 2. Resolve final metadata values
final_summary = extracted.get("summary") or f"Imported from {source}"
# Date logic: Argument > Extracted > Today
if date:
final_date = date
elif extracted.get("meeting_date"):
final_date = extracted.get("meeting_date")
else:
final_date = datetime.now().strftime("%Y-%m-%d")
speaker_mapping = extracted.get("speaker_mapping", {})
# 3. Apply speaker mapping to text (improves searchability)
# Replaces "SPEAKER_00" -> "Name" directly in the text content
processed_text = extractor.apply_speaker_mapping(text, speaker_mapping)
# 4. Generate ID and prepare metadata
meeting_id = f"doc_{uuid.uuid4().hex[:8]}"
meeting_metadata = {
"meeting_id": meeting_id,
"meeting_date": final_date,
"date_transcribed": datetime.now().strftime("%Y-%m-%d"),
"source": source,
"meeting_title": title,
"summary": final_summary,
"source_file": f"{source.lower()}_upload",
"transcription_model": "text_import",
"language": "en",
"speaker_mapping": speaker_mapping
}
# 5. Process text into documents
docs = process_transcript_to_documents(
transcript_text=processed_text,
speaker_data=None, # Uses fallback chunking
meeting_id=meeting_id,
meeting_metadata=meeting_metadata
)
# 6. Upsert to Pinecone
_pinecone_manager.upsert_documents(docs, namespace=Config.PINECONE_NAMESPACE)
return (f"β
Successfully saved '{title}' to Pinecone (ID: {meeting_id})\n"
f" - Date: {final_date}\n"
f" - Extracted Speakers: {', '.join(speaker_mapping.values()) if speaker_mapping else 'None'}")
except Exception as e:
traceback.print_exc()
return f"β Error saving to Pinecone: {str(e)}"
|