alita-sdk 0.3.351__tar.gz → 0.3.353__tar.gz

{alita_sdk-0.3.351 → alita_sdk-0.3.353}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alita_sdk
-Version: 0.3.351
+Version: 0.3.353
 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-0.3.351 → alita_sdk-0.3.353}/alita_sdk/runtime/langchain/document_loaders/constants.py

@@ -300,5 +300,43 @@ document_loaders_map = {
     }
 }
 
+code_extensions = [
+    # '.py',  # Python
+    '.js',  # JavaScript
+    '.ts',  # TypeScript
+    '.java',  # Java
+    '.cpp',  # C++
+    '.c',  # C
+    '.cs',  # C#
+    '.rb',  # Ruby
+    '.go',  # Go
+    '.php',  # PHP
+    '.swift',  # Swift
+    '.kt',  # Kotlin
+    '.rs',  # Rust
+    '.m',  # Objective-C
+    '.scala',  # Scala
+    '.pl',  # Perl
+    '.sh',  # Shell
+    '.bat',  # Batch
+    '.lua',  # Lua
+    '.r',  # R
+    '.pas',  # Pascal
+    '.asm',  # Assembly
+    '.dart',  # Dart
+    '.groovy',  # Groovy
+    '.sql',  # SQL
+]
+
+default_loader_config = {
+    'class': AlitaTextLoader,
+    'mime_type': 'text/plain',
+    'is_multimodal_processing': False,
+    'kwargs': {},
+    'allowed_to_override': ['max_tokens']
+}
+
+code_loaders_map = {ext: default_loader_config for ext in code_extensions}
+
 # Combined mapping for backward compatibility
-loaders_map = {**image_loaders_map, **document_loaders_map}
+loaders_map = {**image_loaders_map, **document_loaders_map, **code_loaders_map}

{alita_sdk-0.3.351 → alita_sdk-0.3.353}/alita_sdk/runtime/tools/llm.py

@@ -95,7 +95,7 @@ class LLMNode(BaseTool):
             messages = state.get("messages", []) if isinstance(state, dict) else []
             if messages:
                 # the last message has to be HumanMessage
-                if not isinstance(messages[:1][0], HumanMessage):
+                if not isinstance(messages[-1], HumanMessage):
                     raise ToolException("LLMNode requires the last message to be a HumanMessage")
             else:
                 raise ToolException("LLMNode requires 'messages' in state for chat-based interaction")

{alita_sdk-0.3.351 → alita_sdk-0.3.353}/alita_sdk/tools/ado/repos/repos_wrapper.py

@@ -644,6 +644,9 @@ class ReposApiWrapper(BaseCodeToolApiWrapper):
 
         return dumps(data)
 
+    def download_file(self, path):
+        return b"".join(self._client.get_item_content(self.repository_id, path=path, project=self.project, download=True))
+
     def get_file_content(self, commit_id, path):
         version_descriptor = GitVersionDescriptor(
             version=commit_id, version_type="commit"

{alita_sdk-0.3.351 → alita_sdk-0.3.353}/alita_sdk/tools/ado/wiki/ado_wrapper.py

@@ -1,5 +1,7 @@
 import hashlib
 import logging
+import re
+import requests
 from typing import Generator, Literal, Optional
 
 from azure.devops.connection import Connection
@@ -15,8 +17,11 @@ from pydantic import create_model, PrivateAttr, SecretStr
 from pydantic import model_validator
 from pydantic.fields import Field
 
+import alita_sdk.tools.ado.work_item
+from ..repos import ReposApiWrapper
 from ...non_code_indexer_toolkit import NonCodeIndexerToolkit
 from ...utils.available_tools_decorator import extend_with_parent_available_tools
+from ...utils.content_parser import parse_file_content
 from ....runtime.utils.utils import IndexerKeywords
 
 logger = logging.getLogger(__name__)
@@ -29,13 +34,17 @@ GetWikiInput = create_model(
 GetPageByPathInput = create_model(
     "GetPageByPathInput",
     wiki_identified=(str, Field(description="Wiki ID or wiki name")),
-    page_name=(str, Field(description="Wiki page path"))
+    page_name=(str, Field(description="Wiki page path")),
+    image_description_prompt=(Optional[str],
+                              Field(description="Prompt which is used for image description", default=None))
 )
 
 GetPageByIdInput = create_model(
     "GetPageByIdInput",
     wiki_identified=(str, Field(description="Wiki ID or wiki name")),
-    page_id=(int, Field(description="Wiki page ID"))
+    page_id=(int, Field(description="Wiki page ID")),
+    image_description_prompt=(Optional[str],
+                              Field(description="Prompt which is used for image description", default=None))
 )
 
 ModifyPageInput = create_model(
@@ -94,24 +103,75 @@ class AzureDevOpsApiWrapper(NonCodeIndexerToolkit):
             logger.error(f"Error during the attempt to extract wiki: {str(e)}")
             return ToolException(f"Error during the attempt to extract wiki: {str(e)}")
 
-    def get_wiki_page_by_path(self, wiki_identified: str, page_name: str):
+    def get_wiki_page_by_path(self, wiki_identified: str, page_name: str, image_description_prompt=None):
         """Extract ADO wiki page content."""
         try:
-            return self._client.get_page(project=self.project, wiki_identifier=wiki_identified, path=page_name,
-                                         include_content=True).page.content
+            return self._process_images(self._client.get_page(project=self.project, wiki_identifier=wiki_identified, path=page_name,
+                                         include_content=True).page.content,
+                                        image_description_prompt=image_description_prompt)
         except Exception as e:
             logger.error(f"Error during the attempt to extract wiki page: {str(e)}")
             return ToolException(f"Error during the attempt to extract wiki page: {str(e)}")
 
-    def get_wiki_page_by_id(self, wiki_identified: str, page_id: int):
+    def get_wiki_page_by_id(self, wiki_identified: str, page_id: int, image_description_prompt=None):
         """Extract ADO wiki page content."""
         try:
-            return (self._client.get_page_by_id(project=self.project, wiki_identifier=wiki_identified, id=page_id,
-                                                include_content=True).page.content)
+            return self._process_images(self._client.get_page_by_id(project=self.project, wiki_identifier=wiki_identified, id=page_id,
+                                                include_content=True).page.content,
+                                        image_description_prompt=image_description_prompt)
         except Exception as e:
             logger.error(f"Error during the attempt to extract wiki page: {str(e)}")
             return ToolException(f"Error during the attempt to extract wiki page: {str(e)}")
 
+    def _process_images(self, page_content: str, image_description_prompt=None):
+
+        image_pattern = r"!\[(.*?)\]\((.*?)\)"
+        matches = re.findall(image_pattern, page_content)
+
+        for image_name, image_url in matches:
+            if image_url.startswith("/.attachments/"):
+                try:
+                    description = self.process_attachment(attachment_url=image_url,
+                                                          attachment_name=image_name,
+                                                          image_description_prompt=image_description_prompt)
+                except Exception as e:
+                    logger.error(f"Error parsing attachment: {str(e)}")
+                    description = f"Error parsing attachment: {image_url}"
+            else:
+                try:
+                    response = requests.get(image_url)
+                    response.raise_for_status()
+                    file_content = response.content
+                    description = parse_file_content(
+                        file_content=file_content,
+                        file_name="image.png",
+                        llm=self.llm,
+                        prompt=image_description_prompt
+                    )
+                except Exception as e:
+                    logger.error(f"Error fetching external image: {str(e)}")
+                    description = f"Error fetching external image: image_url"
+
+            new_image_markdown = f"![{image_name}]({description})"
+            page_content = page_content.replace(f"![{image_name}]({image_url})", new_image_markdown)
+        return page_content
+
+    def process_attachment(self, attachment_url, attachment_name, image_description_prompt):
+        wiki_master_branch = "wikiMaster"
+        repos_wrapper = ReposApiWrapper(organization_url=self.organization_url,
+                                        project=self.project,
+                                        token=self.token.get_secret_value(),
+                                        repository_id="Test_agent.wiki",
+                                        base_branch=wiki_master_branch,
+                                        active_branch=wiki_master_branch)
+        attachment_content = repos_wrapper.download_file(path=attachment_url)
+        return parse_file_content(
+            file_content=attachment_content,
+            file_name=attachment_name,
+            llm=self.llm,
+            prompt=image_description_prompt
+        )
+
     def delete_page_by_path(self, wiki_identified: str, page_name: str):
         """Extract ADO wiki page content."""
         try:

{alita_sdk-0.3.351 → alita_sdk-0.3.353}/alita_sdk/tools/utils/content_parser.py

@@ -1,3 +1,4 @@
+from copy import deepcopy
 import os
 import re
 import tempfile
@@ -126,7 +127,7 @@ def load_file_docs(file_name=None, file_content=None, is_capture_image: bool = F
 
 def get_loader_kwargs(loader_object, file_name=None, file_content=None, is_capture_image: bool = False, page_number: int = None,
                     sheet_name: str = None, llm=None, file_path: str = None, excel_by_sheets: bool = False, prompt=None):
-    loader_kwargs = loader_object['kwargs']
+    loader_kwargs = deepcopy(loader_object['kwargs'])
     loader_kwargs.update({
         "file_path": file_path,
         "file_content": file_content,

{alita_sdk-0.3.351 → alita_sdk-0.3.353}/alita_sdk.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: alita_sdk
-Version: 0.3.351
+Version: 0.3.353
 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-0.3.351 → alita_sdk-0.3.353}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "alita_sdk"
-version = "0.3.351"
+version = "0.3.353"
 description = "SDK for building langchain agents using resources from Alita"
 readme = "README.md"
 requires-python = ">=3.10"