alita-sdk 0.3.368__tar.gz → 0.3.370__tar.gz

{alita_sdk-0.3.368 → alita_sdk-0.3.370}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alita_sdk
-Version: 0.3.368
+Version: 0.3.370
 Summary: SDK for building langchain agents using resources from Alita
 Author-email: Artem Rozumenko <artyom.rozumenko@gmail.com>, Mikalai Biazruchka <mikalai_biazruchka@epam.com>, Roman Mitusov <roman_mitusov@epam.com>, Ivan Krakhmaliuk <lifedj27@gmail.com>, Artem Dubrovskiy <ad13box@gmail.com>
 License-Expression: Apache-2.0

{alita_sdk-0.3.368 → alita_sdk-0.3.370}/alita_sdk/runtime/langchain/assistant.py

@@ -111,7 +111,10 @@ class Assistant:
                 messages.extend(chat_history)
             self.prompt = Jinja2TemplatedChatMessagesTemplate(messages=messages)
             if input_variables:
-                self.prompt.input_variables = input_variables
+                if hasattr(self.prompt, 'input_variables') and self.prompt.input_variables is not None:
+                    self.prompt.input_variables.extend(input_variables)
+                else:
+                    self.prompt.input_variables = input_variables
             if variables:
                 self.prompt.partial_variables = variables
             try:

{alita_sdk-0.3.368 → alita_sdk-0.3.370}/alita_sdk/runtime/langchain/langraph_agent.py

@@ -786,13 +786,6 @@ class LangGraphAgentRunnable(CompiledStateGraph):
         else:
             result = super().invoke(input, config=config, *args, **kwargs)
         try:
-            # if self.output_variables and self.output_variables[0] != "messages":
-            #     # If output_variables are specified, use the value of first one or use the last messages as default
-            #     output = result.get(self.output_variables[0])
-            #     if not output:
-            #         output = result['messages'][-1].content
-            # else:
-            #     output = result['messages'][-1].content
             output = next((msg.content for msg in reversed(result['messages']) if not isinstance(msg, HumanMessage)), result['messages'][-1].content)
         except:
             output = list(result.values())[-1]

{alita_sdk-0.3.368 → alita_sdk-0.3.370}/alita_sdk/runtime/tools/vectorstore_base.py

@@ -6,6 +6,8 @@ from typing import Any, Optional, List, Dict, Generator
 
 from langchain_core.documents import Document
 from langchain_core.messages import HumanMessage
+from langchain_core.tools import ToolException
+from psycopg.errors import DataException
 from pydantic import BaseModel, model_validator, Field
 
 from alita_sdk.tools.elitea_base import BaseToolApiWrapper
@@ -316,6 +318,15 @@ class VectorStoreWrapperBase(BaseToolApiWrapper):
                     if doc_id not in unique_docs or score > chunk_type_scores.get(doc_id, 0):
                         unique_docs[doc_id] = doc
                         chunk_type_scores[doc_id] = score
+            except DataException as dimException:
+                exception_str = str(dimException)
+                if 'different vector dimensions' in exception_str:
+                    logger.error(f"Data exception: {exception_str}")
+                    raise ToolException(f"Global search cannot be completed since collections were indexed using "
+                                        f"different embedding models. Use search within a single collection."
+                                        f"\nDetails: {exception_str}")
+                raise ToolException(f"Data exception during search. Possibly invalid filter: {exception_str}")
+
             except Exception as e:
                 logger.warning(f"Error searching for document chunks: {str(e)}")
             

{alita_sdk-0.3.368 → alita_sdk-0.3.370}/alita_sdk/tools/base_indexer_toolkit.py

@@ -366,15 +366,15 @@ class BaseIndexerToolkit(VectorStoreWrapperBase):
             filter.update({"collection": {
                 "$eq": collection_suffix.strip()
             }})
-        filter = {
-            "$and": [
-                filter,
-                {"$or": [
-                    {"type": {"$exists": False}},
-                    {"type": {"$ne": IndexerKeywords.INDEX_META_TYPE.value}}
-                ]},
-            ]
-        }
+            filter = {
+                "$and": [
+                    filter,
+                    {"$or": [
+                        {"type": {"$exists": False}},
+                        {"type": {"$ne": IndexerKeywords.INDEX_META_TYPE.value}}
+                    ]},
+                ]
+            }
         return filter
 
     def index_meta_read(self):

{alita_sdk-0.3.368 → alita_sdk-0.3.370}/alita_sdk/tools/confluence/api_wrapper.py

@@ -815,6 +815,10 @@ class ConfluenceAPIWrapper(NonCodeIndexerToolkit):
         from .loader import AlitaConfluenceLoader
         from copy import copy
         content_format = kwargs.get('content_format', 'view').lower()
+
+        self._index_include_attachments = kwargs.get('include_attachments', False)
+        self._include_extensions = kwargs.get('include_extensions', [])
+        self._skip_extensions = kwargs.get('skip_extensions', [])
         base_params = {
             'url': self.base_url,
             'space_key': self.space,
@@ -847,65 +851,79 @@ class ConfluenceAPIWrapper(NonCodeIndexerToolkit):
 
     def _process_document(self, document: Document) -> Generator[Document, None, None]:
         try:
-            page_id = document.metadata.get('id')
-            attachments = self.client.get_attachments_from_content(page_id)
-            if not attachments or not attachments.get('results'):
-                return f"No attachments found for page ID {page_id}."
-
-            # Get attachment history for created/updated info
-            history_map = {}
-            for attachment in attachments['results']:
-                try:
-                    hist = self.client.history(attachment['id'])
-                    history_map[attachment['id']] = hist
-                except Exception as e:
-                    logger.warning(f"Failed to fetch history for attachment {attachment.get('title', '')}: {str(e)}")
-                    history_map[attachment['id']] = None
-
-            import re
-            for attachment in attachments['results']:
-                title = attachment.get('title', '')
-                file_ext = title.lower().split('.')[-1] if '.' in title else ''
-
-                media_type = attachment.get('metadata', {}).get('mediaType', '')
-                # Core metadata extraction with history
-                hist = history_map.get(attachment['id']) or {}
-                created_by = hist.get('createdBy', {}).get('displayName', '') if hist else attachment.get('creator', {}).get('displayName', '')
-                created_date = hist.get('createdDate', '') if hist else attachment.get('created', '')
-                last_updated = hist.get('lastUpdated', {}).get('when', '') if hist else ''
+            if self._index_include_attachments:
+                page_id = document.metadata.get('id')
+                attachments = self.client.get_attachments_from_content(page_id)
+                if not attachments or not attachments.get('results'):
+                    return f"No attachments found for page ID {page_id}."
+
+                # Get attachment history for created/updated info
+                history_map = {}
+                for attachment in attachments['results']:
+                    try:
+                        hist = self.client.history(attachment['id'])
+                        history_map[attachment['id']] = hist
+                    except Exception as e:
+                        logger.warning(f"Failed to fetch history for attachment {attachment.get('title', '')}: {str(e)}")
+                        history_map[attachment['id']] = None
+
+                import re
+                for attachment in attachments['results']:
+                    title = attachment.get('title', '')
+                    file_ext = title.lower().split('.')[-1] if '.' in title else ''
+
+                    # Re-verify extension filters
+                    # Check if file should be skipped based on skip_extensions
+                    if any(re.match(pattern.replace('*', '.*') + '$', title, re.IGNORECASE)
+                           for pattern in self._skip_extensions):
+                        continue
+
+                    # Check if file should be included based on include_extensions
+                    # If include_extensions is empty, process all files (that weren't skipped)
+                    if self._include_extensions and not (
+                    any(re.match(pattern.replace('*', '.*') + '$', title, re.IGNORECASE)
+                        for pattern in self._include_extensions)):
+                        continue
+
+                    media_type = attachment.get('metadata', {}).get('mediaType', '')
+                    # Core metadata extraction with history
+                    hist = history_map.get(attachment['id']) or {}
+                    created_by = hist.get('createdBy', {}).get('displayName', '') if hist else attachment.get('creator', {}).get('displayName', '')
+                    created_date = hist.get('createdDate', '') if hist else attachment.get('created', '')
+                    last_updated = hist.get('lastUpdated', {}).get('when', '') if hist else ''
+
+                    metadata = {
+                        'name': title,
+                        'size': attachment.get('extensions', {}).get('fileSize', None),
+                        'creator': created_by,
+                        'created': created_date,
+                        'updated': last_updated,
+                        'media_type': media_type,
+                        'labels': [label['name'] for label in
+                                   attachment.get('metadata', {}).get('labels', {}).get('results', [])],
+                        'download_url': self.base_url.rstrip('/') + attachment['_links']['download'] if attachment.get(
+                            '_links', {}).get('download') else None
+                    }
 
-                metadata = {
-                    'name': title,
-                    'size': attachment.get('extensions', {}).get('fileSize', None),
-                    'creator': created_by,
-                    'created': created_date,
-                    'updated': last_updated,
-                    'media_type': media_type,
-                    'labels': [label['name'] for label in
-                               attachment.get('metadata', {}).get('labels', {}).get('results', [])],
-                    'download_url': self.base_url.rstrip('/') + attachment['_links']['download'] if attachment.get(
-                        '_links', {}).get('download') else None
-                }
+                    download_url = self.base_url.rstrip('/') + attachment['_links']['download']
 
-                download_url = self.base_url.rstrip('/') + attachment['_links']['download']
+                    try:
+                        resp = self.client.request(method="GET", path=download_url[len(self.base_url):], advanced_mode=True)
+                        if resp.status_code == 200:
+                            content = resp.content
+                        else:
+                            content = f"[Failed to download {download_url}: HTTP status code {resp.status_code}]"
+                    except Exception as e:
+                        content = f"[Error downloading content: {str(e)}]"
 
-                try:
-                    resp = self.client.request(method="GET", path=download_url[len(self.base_url):], advanced_mode=True)
-                    if resp.status_code == 200:
-                        content = resp.content
+                    if isinstance(content, str):
+                        yield Document(page_content=content, metadata=metadata)
                     else:
-                        content = f"[Failed to download {download_url}: HTTP status code {resp.status_code}]"
-                except Exception as e:
-                    content = f"[Error downloading content: {str(e)}]"
-
-                if isinstance(content, str):
-                    yield Document(page_content=content, metadata=metadata)
-                else:
-                    yield Document(page_content="", metadata={
-                        **metadata,
-                        IndexerKeywords.CONTENT_FILE_NAME.value: f".{file_ext}",
-                        IndexerKeywords.CONTENT_IN_BYTES.value: content
-                    })
+                        yield Document(page_content="", metadata={
+                            **metadata,
+                            IndexerKeywords.CONTENT_FILE_NAME.value: f".{file_ext}",
+                            IndexerKeywords.CONTENT_IN_BYTES.value: content
+                        })
         except Exception as e:
             yield from ()
 
@@ -1648,6 +1666,13 @@ class ConfluenceAPIWrapper(NonCodeIndexerToolkit):
             "include_restricted_content": (Optional[bool], Field(description="Include restricted content.", default=False)),
             "include_archived_content": (Optional[bool], Field(description="Include archived content.", default=False)),
             "include_attachments": (Optional[bool], Field(description="Include attachments.", default=False)),
+            'include_extensions': (Optional[List[str]], Field(
+                description="List of file extensions to include when processing attachments: i.e. ['*.png', '*.jpg']. "
+                            "If empty, all files will be processed (except skip_extensions).",
+                default=[])),
+            'skip_extensions': (Optional[List[str]], Field(
+                description="List of file extensions to skip when processing attachments: i.e. ['*.png', '*.jpg']",
+                default=[])),
             "include_comments": (Optional[bool], Field(description="Include comments.", default=False)),
             "include_labels": (Optional[bool], Field(description="Include labels.", default=True)),
             "ocr_languages": (Optional[str], Field(description="OCR languages for processing attachments.", default='eng')),

{alita_sdk-0.3.368 → alita_sdk-0.3.370}/alita_sdk.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alita_sdk
-Version: 0.3.368
+Version: 0.3.370
 Summary: SDK for building langchain agents using resources from Alita
 Author-email: Artem Rozumenko <artyom.rozumenko@gmail.com>, Mikalai Biazruchka <mikalai_biazruchka@epam.com>, Roman Mitusov <roman_mitusov@epam.com>, Ivan Krakhmaliuk <lifedj27@gmail.com>, Artem Dubrovskiy <ad13box@gmail.com>
 License-Expression: Apache-2.0

{alita_sdk-0.3.368 → alita_sdk-0.3.370}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "alita_sdk"
-version = "0.3.368"
+version = "0.3.370"
 description = "SDK for building langchain agents using resources from Alita"
 readme = "README.md"
 requires-python = ">=3.10"