signalwire-agents 0.1.19__tar.gz → 0.1.20__tar.gz

{signalwire_agents-0.1.19/signalwire_agents.egg-info → signalwire_agents-0.1.20}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: signalwire_agents
-Version: 0.1.19
+Version: 0.1.20
 Summary: SignalWire AI Agents SDK
 Author-email: SignalWire Team <info@signalwire.com>
 Project-URL: Homepage, https://github.com/signalwire/signalwire-ai-agents
@@ -26,6 +26,11 @@ Requires-Dist: structlog==25.3.0
 Requires-Dist: uvicorn==0.34.2
 Requires-Dist: beautifulsoup4==4.12.3
 Requires-Dist: pytz==2023.3
+Provides-Extra: search-queryonly
+Requires-Dist: numpy>=1.24.0; extra == "search-queryonly"
+Requires-Dist: scikit-learn>=1.3.0; extra == "search-queryonly"
+Requires-Dist: sentence-transformers>=2.2.0; extra == "search-queryonly"
+Requires-Dist: nltk>=3.8; extra == "search-queryonly"
 Provides-Extra: search
 Requires-Dist: sentence-transformers>=2.2.0; extra == "search"
 Requires-Dist: scikit-learn>=1.3.0; extra == "search"
@@ -488,7 +493,10 @@ The SDK includes optional local search capabilities that can be installed separa
 #### Search Installation Options
 
 ```bash
-# Basic search (vector search + keyword search)
+# Query existing .swsearch files only (smallest footprint)
+pip install signalwire-agents[search-queryonly]
+
+# Basic search (vector search + keyword search + building indexes)
 pip install signalwire-agents[search]
 
 # Full search with document processing (PDF, DOCX, etc.)
@@ -505,11 +513,18 @@ pip install signalwire-agents[search-all]
 
 | Option | Size | Features |
 |--------|------|----------|
+| `search-queryonly` | ~400MB | Query existing .swsearch files only (no building/processing) |
 | `search` | ~500MB | Vector embeddings, keyword search, basic text processing |
 | `search-full` | ~600MB | + PDF, DOCX, Excel, PowerPoint, HTML, Markdown processing |
 | `search-nlp` | ~600MB | + Advanced spaCy NLP features |
 | `search-all` | ~700MB | All search features combined |
 
+**When to use `search-queryonly`:**
+- Production containers with pre-built `.swsearch` files
+- Lambda/serverless deployments
+- Agents that only need to query knowledge bases (not build them)
+- Smaller deployment footprint requirements
+
 #### Search Features
 
 - **Local/Offline Search**: No external API dependencies

{signalwire_agents-0.1.19 → signalwire_agents-0.1.20}/README.md

@@ -422,7 +422,10 @@ The SDK includes optional local search capabilities that can be installed separa
 #### Search Installation Options
 
 ```bash
-# Basic search (vector search + keyword search)
+# Query existing .swsearch files only (smallest footprint)
+pip install signalwire-agents[search-queryonly]
+
+# Basic search (vector search + keyword search + building indexes)
 pip install signalwire-agents[search]
 
 # Full search with document processing (PDF, DOCX, etc.)
@@ -439,11 +442,18 @@ pip install signalwire-agents[search-all]
 
 | Option | Size | Features |
 |--------|------|----------|
+| `search-queryonly` | ~400MB | Query existing .swsearch files only (no building/processing) |
 | `search` | ~500MB | Vector embeddings, keyword search, basic text processing |
 | `search-full` | ~600MB | + PDF, DOCX, Excel, PowerPoint, HTML, Markdown processing |
 | `search-nlp` | ~600MB | + Advanced spaCy NLP features |
 | `search-all` | ~700MB | All search features combined |
 
+**When to use `search-queryonly`:**
+- Production containers with pre-built `.swsearch` files
+- Lambda/serverless deployments
+- Agents that only need to query knowledge bases (not build them)
+- Smaller deployment footprint requirements
+
 #### Search Features
 
 - **Local/Offline Search**: No external API dependencies

{signalwire_agents-0.1.19 → signalwire_agents-0.1.20}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "signalwire_agents"
-version = "0.1.19"
+version = "0.1.20"
 description = "SignalWire AI Agents SDK"
 authors = [
     {name = "SignalWire Team", email = "info@signalwire.com"}
@@ -37,6 +37,15 @@ dependencies = [
 
 # Optional dependencies for search functionality
 [project.optional-dependencies]
+# Query existing .swsearch files only (no document processing/building)
+search-queryonly = [
+    "numpy>=1.24.0",
+    "scikit-learn>=1.3.0",
+    "sentence-transformers>=2.2.0",
+    "nltk>=3.8",
+]
+
+# Full search functionality (includes document processing and building)
 search = [
     "sentence-transformers>=2.2.0",
     "scikit-learn>=1.3.0",

{signalwire_agents-0.1.19 → signalwire_agents-0.1.20}/signalwire_agents/__init__.py

@@ -18,7 +18,7 @@ A package for building AI agents using SignalWire's AI and SWML capabilities.
 from .core.logging_config import configure_logging
 configure_logging()
 
-__version__ = "0.1.19"
+__version__ = "0.1.20"
 
 # Import core classes for easier access
 from .core.agent_base import AgentBase

{signalwire_agents-0.1.19 → signalwire_agents-0.1.20}/signalwire_agents/search/index_builder.py

@@ -367,16 +367,57 @@ class IndexBuilder:
                      global_tags: Optional[List[str]] = None) -> List[Dict[str, Any]]:
         """Process single file into chunks"""
         try:
-            # Try to read as text first
-            try:
-                content = file_path.read_text(encoding='utf-8')
-            except UnicodeDecodeError:
+            relative_path = str(file_path.relative_to(source_dir))
+            file_extension = file_path.suffix.lower()
+            
+            # Handle different file types appropriately
+            if file_extension == '.pdf':
+                # Use document processor for PDF extraction
+                content_result = self.doc_processor._extract_text_from_file(str(file_path))
+                if isinstance(content_result, str) and content_result.startswith('{"error"'):
+                    if self.verbose:
+                        print(f"Skipping PDF file (extraction failed): {file_path}")
+                    return []
+                content = content_result
+            elif file_extension in ['.docx', '.xlsx', '.pptx']:
+                # Use document processor for Office documents
+                content_result = self.doc_processor._extract_text_from_file(str(file_path))
+                if isinstance(content_result, str) and content_result.startswith('{"error"'):
+                    if self.verbose:
+                        print(f"Skipping office document (extraction failed): {file_path}")
+                    return []
+                content = content_result
+            elif file_extension == '.html':
+                # Use document processor for HTML
+                content_result = self.doc_processor._extract_text_from_file(str(file_path))
+                if isinstance(content_result, str) and content_result.startswith('{"error"'):
+                    if self.verbose:
+                        print(f"Skipping HTML file (extraction failed): {file_path}")
+                    return []
+                content = content_result
+            elif file_extension == '.rtf':
+                # Use document processor for RTF
+                content_result = self.doc_processor._extract_text_from_file(str(file_path))
+                if isinstance(content_result, str) and content_result.startswith('{"error"'):
+                    if self.verbose:
+                        print(f"Skipping RTF file (extraction failed): {file_path}")
+                    return []
+                content = content_result
+            else:
+                # Try to read as text file (markdown, txt, code, etc.)
+                try:
+                    content = file_path.read_text(encoding='utf-8')
+                except UnicodeDecodeError:
+                    if self.verbose:
+                        print(f"Skipping binary file: {file_path}")
+                    return []
+            
+            # Validate content
+            if not content or (isinstance(content, str) and len(content.strip()) == 0):
                 if self.verbose:
-                    print(f"Skipping binary file: {file_path}")
+                    print(f"Skipping empty file: {file_path}")
                 return []
             
-            relative_path = str(file_path.relative_to(source_dir))
-            
             # Create chunks using document processor - pass content directly, not file path
             chunks = self.doc_processor.create_chunks(
                 content=content,  # Pass the actual content, not the file path

{signalwire_agents-0.1.19 → signalwire_agents-0.1.20}/signalwire_agents/search/query_processor.py

@@ -118,15 +118,28 @@ stopwords_language_map = {
 # Function to ensure NLTK resources are downloaded
 def ensure_nltk_resources():
     """Download required NLTK resources if not already present"""
-    resources = ['punkt', 'wordnet', 'averaged_perceptron_tagger', 'stopwords']
+    resources = ['punkt', 'punkt_tab', 'wordnet', 'averaged_perceptron_tagger', 'stopwords']
     for resource in resources:
         try:
-            nltk.data.find(f'corpora/{resource}')
+            # Try different paths for different resource types
+            if resource in ['punkt', 'punkt_tab']:
+                nltk.data.find(f'tokenizers/{resource}')
+            elif resource in ['wordnet']:
+                nltk.data.find(f'corpora/{resource}')
+            elif resource in ['averaged_perceptron_tagger']:
+                nltk.data.find(f'taggers/{resource}')
+            elif resource in ['stopwords']:
+                nltk.data.find(f'corpora/{resource}')
+            else:
+                nltk.data.find(f'corpora/{resource}')
         except LookupError:
             try:
+                logger.info(f"Downloading NLTK resource '{resource}'...")
                 nltk.download(resource, quiet=True)
+                logger.info(f"Successfully downloaded NLTK resource '{resource}'")
             except Exception as e:
                 logger.warning(f"Failed to download NLTK resource '{resource}': {e}")
+                # Continue without this resource - some functionality may be degraded
 
 # Initialize NLTK resources
 ensure_nltk_resources()
@@ -246,7 +259,20 @@ def preprocess_query(query: str, language: str = 'en', pos_to_expand: Optional[L
         query_nlp_backend = 'nltk'
     
     # Tokenization and stop word removal
-    tokens = nltk.word_tokenize(query)
+    try:
+        tokens = nltk.word_tokenize(query)
+    except LookupError as e:
+        # If tokenization fails, try to download punkt resources
+        logger.warning(f"NLTK tokenization failed: {e}")
+        try:
+            nltk.download('punkt', quiet=True)
+            nltk.download('punkt_tab', quiet=True)
+            tokens = nltk.word_tokenize(query)
+        except Exception as fallback_error:
+            # If all else fails, use simple split as fallback
+            logger.warning(f"NLTK tokenization fallback failed: {fallback_error}. Using simple word splitting.")
+            tokens = query.split()
+    
     nltk_language = stopwords_language_map.get(language, 'english')
     
     try:
@@ -279,14 +305,29 @@ def preprocess_query(query: str, language: str = 'en', pos_to_expand: Optional[L
             logger.info(f"POS Tagging Results (spaCy): {pos_tags}")
     else:
         # Use NLTK (default or fallback)
-        nltk_pos_tags = nltk.pos_tag(tokens)
-        for token, pos_tag in nltk_pos_tags:
-            lemma = lemmatizer.lemmatize(token, get_wordnet_pos(pos_tag)).lower()
-            stemmed = stemmer.stem(lemma)
-            lemmas.append((token.lower(), stemmed))
-            pos_tags[token.lower()] = pos_tag
-        if debug:
-            logger.info(f"POS Tagging Results (NLTK): {pos_tags}")
+        try:
+            nltk_pos_tags = nltk.pos_tag(tokens)
+            for token, pos_tag in nltk_pos_tags:
+                try:
+                    lemma = lemmatizer.lemmatize(token, get_wordnet_pos(pos_tag)).lower()
+                except Exception:
+                    # Fallback if lemmatization fails
+                    lemma = token.lower()
+                stemmed = stemmer.stem(lemma)
+                lemmas.append((token.lower(), stemmed))
+                pos_tags[token.lower()] = pos_tag
+            if debug:
+                logger.info(f"POS Tagging Results (NLTK): {pos_tags}")
+        except Exception as pos_error:
+            # Fallback if POS tagging fails completely
+            logger.warning(f"NLTK POS tagging failed: {pos_error}. Using basic token processing.")
+            for token in tokens:
+                lemma = token.lower()
+                stemmed = stemmer.stem(lemma)
+                lemmas.append((token.lower(), stemmed))
+                pos_tags[token.lower()] = 'NN'  # Default to noun
+            if debug:
+                logger.info(f"Using fallback token processing for: {tokens}")
 
     # Expanding query with synonyms
     expanded_query_set = set()

{signalwire_agents-0.1.19 → signalwire_agents-0.1.20}/signalwire_agents/skills/native_vector_search/skill.py

@@ -41,18 +41,7 @@ class NativeVectorSearchSkill(SkillBase):
     def setup(self) -> bool:
         """Setup the native vector search skill"""
         
-        # Check if search functionality is available
-        try:
-            from signalwire_agents.search import IndexBuilder, SearchEngine
-            from signalwire_agents.search.query_processor import preprocess_query
-            self.search_available = True
-        except ImportError as e:
-            self.search_available = False
-            self.import_error = str(e)
-            self.logger.warning(f"Search dependencies not available: {e}")
-            # Don't fail setup - we'll provide helpful error messages at runtime
-        
-        # Get configuration
+        # Get configuration first
         self.tool_name = self.params.get('tool_name', 'search_knowledge')
         self.index_file = self.params.get('index_file')
         self.build_index = self.params.get('build_index', False)
@@ -74,7 +63,34 @@ class NativeVectorSearchSkill(SkillBase):
         # SWAIG fields for function fillers
         self.swaig_fields = self.params.get('swaig_fields', {})
         
-        # NLP backend configuration
+        # **EARLY REMOTE CHECK - Option 1**
+        # If remote URL is configured, skip all heavy local imports and just validate remote connectivity
+        if self.remote_url:
+            self.use_remote = True
+            self.search_engine = None  # No local search engine needed
+            self.logger.info(f"Using remote search server: {self.remote_url}")
+            
+            # Test remote connection (lightweight check)
+            try:
+                import requests
+                response = requests.get(f"{self.remote_url}/health", timeout=5)
+                if response.status_code == 200:
+                    self.logger.info("Remote search server is available")
+                    self.search_available = True
+                    return True  # Success - skip all local setup
+                else:
+                    self.logger.error(f"Remote search server returned status {response.status_code}")
+                    self.search_available = False
+                    return False
+            except Exception as e:
+                self.logger.error(f"Failed to connect to remote search server: {e}")
+                self.search_available = False
+                return False
+        
+        # **LOCAL MODE SETUP - Only when no remote URL**
+        self.use_remote = False
+        
+        # NLP backend configuration (only needed for local mode)
         self.nlp_backend = self.params.get('nlp_backend')  # Backward compatibility
         self.index_nlp_backend = self.params.get('index_nlp_backend', 'nltk')  # Default to fast NLTK for indexing
         self.query_nlp_backend = self.params.get('query_nlp_backend', 'nltk')  # Default to fast NLTK for search
@@ -95,6 +111,17 @@ class NativeVectorSearchSkill(SkillBase):
             self.logger.warning(f"Invalid query_nlp_backend '{self.query_nlp_backend}', using 'nltk'")
             self.query_nlp_backend = 'nltk'
         
+        # Check if local search functionality is available (heavy imports only for local mode)
+        try:
+            from signalwire_agents.search import IndexBuilder, SearchEngine
+            from signalwire_agents.search.query_processor import preprocess_query
+            self.search_available = True
+        except ImportError as e:
+            self.search_available = False
+            self.import_error = str(e)
+            self.logger.warning(f"Search dependencies not available: {e}")
+            # Don't fail setup - we'll provide helpful error messages at runtime
+        
         # Auto-build index if requested and search is available
         if self.build_index and self.source_dir and self.search_available:
             if not self.index_file:
@@ -124,7 +151,7 @@ class NativeVectorSearchSkill(SkillBase):
                     self.logger.error(f"Failed to build search index: {e}")
                     self.search_available = False
         
-        # Initialize search engine
+        # Initialize local search engine
         self.search_engine = None
         if self.search_available and self.index_file and os.path.exists(self.index_file):
             try:
@@ -134,24 +161,6 @@ class NativeVectorSearchSkill(SkillBase):
                 self.logger.error(f"Failed to load search index {self.index_file}: {e}")
                 self.search_available = False
         
-        # Check if we should use remote search mode
-        self.use_remote = bool(self.remote_url)
-        if self.use_remote:
-            self.logger.info(f"Using remote search server: {self.remote_url}")
-            # Test remote connection
-            try:
-                import requests
-                response = requests.get(f"{self.remote_url}/health", timeout=5)
-                if response.status_code == 200:
-                    self.logger.info("Remote search server is available")
-                    self.search_available = True
-                else:
-                    self.logger.error(f"Remote search server returned status {response.status_code}")
-                    self.search_available = False
-            except Exception as e:
-                self.logger.error(f"Failed to connect to remote search server: {e}")
-                self.search_available = False
-        
         return True
         
     def register_tools(self) -> None:
@@ -184,6 +193,11 @@ class NativeVectorSearchSkill(SkillBase):
     def _search_handler(self, args, raw_data):
         """Handle search requests"""
         
+        # Debug logging to see what arguments are being passed
+        self.logger.info(f"Search handler called with args: {args}")
+        self.logger.info(f"Args type: {type(args)}")
+        self.logger.info(f"Raw data: {raw_data}")
+        
         if not self.search_available:
             return SwaigFunctionResult(
                 f"Search functionality is not available. {getattr(self, 'import_error', '')}\n"
@@ -196,21 +210,27 @@ class NativeVectorSearchSkill(SkillBase):
                 f"{'Index file not found: ' + (self.index_file or 'not specified') if self.index_file else 'No index file configured'}"
             )
         
+        # Get arguments - the framework handles parsing correctly
         query = args.get('query', '').strip()
+        self.logger.error(f"DEBUG: Extracted query: '{query}' (length: {len(query)})")
+        self.logger.info(f"Query bool value: {bool(query)}")
+        
         if not query:
+            self.logger.error(f"Query validation failed - returning error message")
             return SwaigFunctionResult("Please provide a search query.")
         
+        self.logger.info(f"Query validation passed - proceeding with search")
         count = args.get('count', self.count)
         
         try:
-            # Preprocess the query
-            from signalwire_agents.search.query_processor import preprocess_query
-            enhanced = preprocess_query(query, language='en', vector=True, query_nlp_backend=self.query_nlp_backend)
-            
             # Perform search (local or remote)
             if self.use_remote:
-                results = self._search_remote(query, enhanced, count)
+                # For remote searches, let the server handle query preprocessing
+                results = self._search_remote(query, None, count)
             else:
+                # For local searches, preprocess the query locally
+                from signalwire_agents.search.query_processor import preprocess_query
+                enhanced = preprocess_query(query, language='en', vector=True, query_nlp_backend=self.query_nlp_backend)
                 results = self.search_engine.search(
                     query_vector=enhanced.get('vector', []),
                     enhanced_text=enhanced['enhanced_text'],
@@ -256,7 +276,24 @@ class NativeVectorSearchSkill(SkillBase):
             return SwaigFunctionResult("\n".join(response_parts))
             
         except Exception as e:
-            return SwaigFunctionResult(f"Search error: {str(e)}")
+            # Log the full error details for debugging
+            self.logger.error(f"Search error for query '{query}': {str(e)}", exc_info=True)
+            
+            # Return user-friendly error message
+            user_msg = "I'm sorry, I encountered an issue while searching. "
+            
+            # Check for specific error types and provide helpful guidance
+            error_str = str(e).lower()
+            if 'punkt' in error_str or 'nltk' in error_str:
+                user_msg += "It looks like some language processing resources are missing. Please try again in a moment."
+            elif 'vector' in error_str or 'embedding' in error_str:
+                user_msg += "There was an issue with the search indexing. Please try rephrasing your question."
+            elif 'timeout' in error_str or 'connection' in error_str:
+                user_msg += "The search service is temporarily unavailable. Please try again later."
+            else:
+                user_msg += "Please try rephrasing your question or contact support if the issue persists."
+                
+            return SwaigFunctionResult(user_msg)
     
     def _search_remote(self, query: str, enhanced: dict, count: int) -> list:
         """Perform search using remote search server"""

{signalwire_agents-0.1.19 → signalwire_agents-0.1.20/signalwire_agents.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: signalwire_agents
-Version: 0.1.19
+Version: 0.1.20
 Summary: SignalWire AI Agents SDK
 Author-email: SignalWire Team <info@signalwire.com>
 Project-URL: Homepage, https://github.com/signalwire/signalwire-ai-agents
@@ -26,6 +26,11 @@ Requires-Dist: structlog==25.3.0
 Requires-Dist: uvicorn==0.34.2
 Requires-Dist: beautifulsoup4==4.12.3
 Requires-Dist: pytz==2023.3
+Provides-Extra: search-queryonly
+Requires-Dist: numpy>=1.24.0; extra == "search-queryonly"
+Requires-Dist: scikit-learn>=1.3.0; extra == "search-queryonly"
+Requires-Dist: sentence-transformers>=2.2.0; extra == "search-queryonly"
+Requires-Dist: nltk>=3.8; extra == "search-queryonly"
 Provides-Extra: search
 Requires-Dist: sentence-transformers>=2.2.0; extra == "search"
 Requires-Dist: scikit-learn>=1.3.0; extra == "search"
@@ -488,7 +493,10 @@ The SDK includes optional local search capabilities that can be installed separa
 #### Search Installation Options
 
 ```bash
-# Basic search (vector search + keyword search)
+# Query existing .swsearch files only (smallest footprint)
+pip install signalwire-agents[search-queryonly]
+
+# Basic search (vector search + keyword search + building indexes)
 pip install signalwire-agents[search]
 
 # Full search with document processing (PDF, DOCX, etc.)
@@ -505,11 +513,18 @@ pip install signalwire-agents[search-all]
 
 | Option | Size | Features |
 |--------|------|----------|
+| `search-queryonly` | ~400MB | Query existing .swsearch files only (no building/processing) |
 | `search` | ~500MB | Vector embeddings, keyword search, basic text processing |
 | `search-full` | ~600MB | + PDF, DOCX, Excel, PowerPoint, HTML, Markdown processing |
 | `search-nlp` | ~600MB | + Advanced spaCy NLP features |
 | `search-all` | ~700MB | All search features combined |
 
+**When to use `search-queryonly`:**
+- Production containers with pre-built `.swsearch` files
+- Lambda/serverless deployments
+- Agents that only need to query knowledge bases (not build them)
+- Smaller deployment footprint requirements
+
 #### Search Features
 
 - **Local/Offline Search**: No external API dependencies

{signalwire_agents-0.1.19 → signalwire_agents-0.1.20}/signalwire_agents.egg-info/requires.txt

@@ -48,3 +48,9 @@ scikit-learn>=1.3.0
 nltk>=3.8
 numpy>=1.24.0
 spacy>=3.6.0
+
+[search-queryonly]
+numpy>=1.24.0
+scikit-learn>=1.3.0
+sentence-transformers>=2.2.0
+nltk>=3.8