dtSpark 1.0.10__py3-none-any.whl → 1.1.0a1__py3-none-any.whl

dtSpark/_version.txt

@@ -1 +1 @@
-1.0.10
+1.1.0a1

dtSpark/core/application.py

@@ -779,6 +779,27 @@ class AWSBedrockCLI(AbstractApp):
                         'enabled': self.settings.get('embedded_tools.filesystem.enabled', False),
                         'allowed_path': self.settings.get('embedded_tools.filesystem.allowed_path', './'),
                         'access_mode': self.settings.get('embedded_tools.filesystem.access_mode', 'read')
+                    },
+                    'documents': {
+                        'enabled': self.settings.get('embedded_tools.documents.enabled', False),
+                        'allowed_path': self.settings.get('embedded_tools.documents.allowed_path', './'),
+                        'access_mode': self.settings.get('embedded_tools.documents.access_mode', 'read'),
+                        'max_file_size_mb': self.settings.get('embedded_tools.documents.max_file_size_mb', 50),
+                        'reading': {
+                            'max_pdf_pages': self.settings.get('embedded_tools.documents.reading.max_pdf_pages', 100),
+                            'max_excel_rows': self.settings.get('embedded_tools.documents.reading.max_excel_rows', 10000)
+                        },
+                        'creation': {
+                            'templates_path': self.settings.get('embedded_tools.documents.creation.templates_path'),
+                            'default_author': self.settings.get('embedded_tools.documents.creation.default_author')
+                        }
+                    },
+                    'archives': {
+                        'enabled': self.settings.get('embedded_tools.archives.enabled', False),
+                        'allowed_path': self.settings.get('embedded_tools.archives.allowed_path', './'),
+                        'access_mode': self.settings.get('embedded_tools.archives.access_mode', 'read'),
+                        'max_file_size_mb': self.settings.get('embedded_tools.archives.max_file_size_mb', 100),
+                        'max_files_to_list': self.settings.get('embedded_tools.archives.max_files_to_list', 1000)
                     }
                 }
             }
@@ -1655,6 +1676,122 @@ class AWSBedrockCLI(AbstractApp):
             )
             filesystem_access_mode = access_mode_choices[access_mode_choice]
 
+        # ═══════════════════════════════════════════════════════════════
+        # Embedded Document Tools (MS Office & PDF)
+        # ═══════════════════════════════════════════════════════════════
+        cli.console.print()
+        enable_document_tools = Confirm.ask("Enable embedded document tools (MS Office & PDF)?", default=False)
+
+        # Default values
+        document_allowed_path = "./running"
+        document_access_mode = "read"
+        document_max_file_size = "50"
+        document_max_pdf_pages = "100"
+        document_max_excel_rows = "10000"
+        document_templates_path = ""
+        document_default_author = ""
+
+        if enable_document_tools:
+            cli.console.print()
+            cli.console.print("[dim]Document tools allow reading and creating MS Office documents (Word, Excel, PowerPoint) and PDFs.[/dim]")
+            cli.console.print()
+
+            document_allowed_path = Prompt.ask(
+                "Allowed directory path for documents",
+                default="./running"
+            )
+
+            # Access mode
+            doc_access_mode_choices = {
+                "1": "read",
+                "2": "read_write"
+            }
+            cli.console.print()
+            cli.console.print("  [1] Read - Read documents only")
+            cli.console.print("  [2] Read/Write - Read and create documents")
+            cli.console.print()
+            doc_access_mode_choice = Prompt.ask(
+                "Select access mode",
+                choices=["1", "2"],
+                default="1"
+            )
+            document_access_mode = doc_access_mode_choices[doc_access_mode_choice]
+
+            document_max_file_size = Prompt.ask(
+                "Maximum file size in MB",
+                default="50"
+            )
+
+            document_max_pdf_pages = Prompt.ask(
+                "Maximum PDF pages to read",
+                default="100"
+            )
+
+            document_max_excel_rows = Prompt.ask(
+                "Maximum Excel rows to read",
+                default="10000"
+            )
+
+            if document_access_mode == "read_write":
+                cli.console.print()
+                cli.console.print("[dim]Templates allow creating documents with placeholder substitution.[/dim]")
+                document_templates_path = Prompt.ask(
+                    "Templates directory path (leave empty to disable)",
+                    default=""
+                )
+                document_default_author = Prompt.ask(
+                    "Default author for created documents (leave empty to disable)",
+                    default=""
+                )
+
+        # ═══════════════════════════════════════════════════════════════
+        # Embedded Archive Tools
+        # ═══════════════════════════════════════════════════════════════
+        cli.console.print()
+        enable_archive_tools = Confirm.ask("Enable embedded archive tools (ZIP, TAR)?", default=False)
+
+        # Default values
+        archive_allowed_path = "./running"
+        archive_access_mode = "read"
+        archive_max_file_size = "100"
+        archive_max_files_to_list = "1000"
+
+        if enable_archive_tools:
+            cli.console.print()
+            cli.console.print("[dim]Archive tools allow reading and extracting ZIP and TAR archives.[/dim]")
+            cli.console.print()
+
+            archive_allowed_path = Prompt.ask(
+                "Allowed directory path for archives",
+                default="./running"
+            )
+
+            # Access mode
+            archive_access_mode_choices = {
+                "1": "read",
+                "2": "read_write"
+            }
+            cli.console.print()
+            cli.console.print("  [1] Read - List contents and read files from archives")
+            cli.console.print("  [2] Read/Write - Read and extract archives to disk")
+            cli.console.print()
+            archive_access_mode_choice = Prompt.ask(
+                "Select access mode",
+                choices=["1", "2"],
+                default="1"
+            )
+            archive_access_mode = archive_access_mode_choices[archive_access_mode_choice]
+
+            archive_max_file_size = Prompt.ask(
+                "Maximum archive file size in MB",
+                default="100"
+            )
+
+            archive_max_files_to_list = Prompt.ask(
+                "Maximum files to list from archive",
+                default="1000"
+            )
+
         # ═══════════════════════════════════════════════════════════════
         # Tool Permissions
         # ═══════════════════════════════════════════════════════════════
@@ -2040,6 +2177,95 @@ class AWSBedrockCLI(AbstractApp):
                     config_content
                 )
 
+            # Embedded Document Tools
+            config_content = re.sub(
+                r'(documents:\s*\n\s+enabled:\s+)(true|false)',
+                f'\\g<1>{str(enable_document_tools).lower()}',
+                config_content
+            )
+            if enable_document_tools:
+                # Allowed path
+                escaped_doc_path = document_allowed_path.replace('\\', '/')
+                config_content = re.sub(
+                    r'(documents:\s*\n\s+enabled:\s+(?:true|false)\s*\n\s+allowed_path:\s+)[^\s#]+',
+                    f'\\g<1>{escaped_doc_path}',
+                    config_content
+                )
+                # Access mode
+                config_content = re.sub(
+                    r'(documents:\s*\n\s+enabled:\s+(?:true|false)\s*\n\s+allowed_path:\s+[^\s#]+\s*\n\s+access_mode:\s+)(read|read_write)',
+                    f'\\g<1>{document_access_mode}',
+                    config_content
+                )
+                # Max file size
+                config_content = re.sub(
+                    r'(documents:.*?max_file_size_mb:\s+)\d+',
+                    f'\\g<1>{document_max_file_size}',
+                    config_content,
+                    flags=re.DOTALL
+                )
+                # Max PDF pages
+                config_content = re.sub(
+                    r'(max_pdf_pages:\s+)\d+',
+                    f'\\g<1>{document_max_pdf_pages}',
+                    config_content
+                )
+                # Max Excel rows
+                config_content = re.sub(
+                    r'(max_excel_rows:\s+)\d+',
+                    f'\\g<1>{document_max_excel_rows}',
+                    config_content
+                )
+                # Templates path (if provided)
+                if document_templates_path:
+                    escaped_templates_path = document_templates_path.replace('\\', '/')
+                    config_content = re.sub(
+                        r'(templates_path:\s+)(null|[^\s#]+)',
+                        f'\\g<1>{escaped_templates_path}',
+                        config_content
+                    )
+                # Default author (if provided)
+                if document_default_author:
+                    config_content = re.sub(
+                        r'(default_author:\s+)(null|[^\s#]+)',
+                        f'\\g<1>{document_default_author}',
+                        config_content
+                    )
+
+            # Embedded Archive Tools
+            config_content = re.sub(
+                r'(archives:\s*\n\s+enabled:\s+)(true|false)',
+                f'\\g<1>{str(enable_archive_tools).lower()}',
+                config_content
+            )
+            if enable_archive_tools:
+                # Allowed path
+                escaped_archive_path = archive_allowed_path.replace('\\', '/')
+                config_content = re.sub(
+                    r'(archives:\s*\n\s+enabled:\s+(?:true|false)\s*\n\s+allowed_path:\s+)[^\s#]+',
+                    f'\\g<1>{escaped_archive_path}',
+                    config_content
+                )
+                # Access mode
+                config_content = re.sub(
+                    r'(archives:\s*\n\s+enabled:\s+(?:true|false)\s*\n\s+allowed_path:\s+[^\s#]+\s*\n\s+access_mode:\s+)(read|read_write)',
+                    f'\\g<1>{archive_access_mode}',
+                    config_content
+                )
+                # Max file size
+                config_content = re.sub(
+                    r'(archives:.*?max_file_size_mb:\s+)\d+',
+                    f'\\g<1>{archive_max_file_size}',
+                    config_content,
+                    flags=re.DOTALL
+                )
+                # Max files to list
+                config_content = re.sub(
+                    r'(max_files_to_list:\s+)\d+',
+                    f'\\g<1>{archive_max_files_to_list}',
+                    config_content
+                )
+
             # Tool Permissions
             config_content = re.sub(
                 r'(tool_permissions:\s*\n\s+auto_approve:\s+)(true|false)',

dtSpark/mcp_integration/tool_selector.py

@@ -21,6 +21,8 @@ class ToolSelector:
         'aws_infrastructure': ['ec2', 's3', 'lambda', 'cloudwatch', 'iam', 'vpc', 'rds', 'dynamodb', 'diagram'],
         'elasticsearch': ['elasticsearch', 'search', 'index', 'query', 'aggregation'],
         'ragstore': ['ragstore', 'rag', 'embedding', 'vector', 'semantic'],
+        'documents': ['word', 'excel', 'powerpoint', 'pdf', 'document', 'docx', 'xlsx', 'pptx', 'spreadsheet'],
+        'archives': ['archive', 'zip', 'tar', 'extract', 'compress', 'tgz'],
     }
 
     # Keywords in user messages that trigger specific categories
@@ -36,6 +38,9 @@ class ToolSelector:
                               'vpc', 'subnet', 'instance', 'bucket', 'function', 'diagram'],
         'elasticsearch': ['elasticsearch', 'search', 'query', 'index', 'log', 'aggregate'],
         'ragstore': ['ragstore', 'rag', 'embedding', 'semantic', 'vector', 'similarity'],
+        'documents': ['document', 'word', 'excel', 'powerpoint', 'pdf', 'docx', 'xlsx', 'pptx',
+                     'spreadsheet', 'presentation', 'template', 'office'],
+        'archives': ['archive', 'zip', 'tar', 'extract', 'unzip', 'compressed', 'tgz'],
     }
 
     def __init__(self, max_tools_per_request: int = 30):

dtSpark/resources/config.yaml.template

@@ -603,6 +603,48 @@ embedded_tools:
     # - Use read-only mode when write operations are not needed
     # - Consider using specific subdirectories rather than root paths
 
+  # Document Tools (MS Office and PDF)
+  # Tools for reading and creating Microsoft Office documents and PDFs
+  documents:
+    enabled: false  # Set to true to enable document reading/creation tools
+    allowed_path: ./  # Root path for document operations (files must be within this path)
+    access_mode: read  # Access mode: "read" (read-only) or "read_write" (read and create)
+                       # read: Only read operations (read_word_document, read_excel_document, etc.)
+                       # read_write: Adds create operations (create_word_document, etc.)
+    max_file_size_mb: 50  # Maximum file size for document operations (in megabytes)
+
+    # Document reading options
+    reading:
+      max_pdf_pages: 100  # Maximum pages to extract from PDF documents
+      max_excel_rows: 10000  # Maximum rows to read from Excel spreadsheets
+
+    # Document creation options (only applies when access_mode is 'read_write')
+    creation:
+      templates_path: null  # Optional: path to directory containing document templates
+                            # Templates use {{placeholder_name}} syntax for replacements
+      default_author: null  # Optional: default author name for created documents
+
+    # Supported formats:
+    # - Word: .docx (read and create)
+    # - Excel: .xlsx (read and create)
+    # - PowerPoint: .pptx (read and create)
+    # - PDF: .pdf (read only)
+
+  # Archive Tools
+  # Tools for reading and extracting compressed archive files
+  archives:
+    enabled: false  # Set to true to enable archive tools
+    allowed_path: ./  # Root path for archive operations (files must be within this path)
+    access_mode: read  # Access mode: "read" (list/read only) or "read_write" (includes extraction)
+                       # read: Only list_archive_contents and read_archive_file operations
+                       # read_write: Adds extract_archive operation
+    max_file_size_mb: 100  # Maximum archive size for operations (in megabytes)
+    max_files_to_list: 1000  # Maximum number of files to list from an archive
+
+    # Supported formats:
+    # - ZIP: .zip
+    # - TAR: .tar, .tar.gz, .tgz, .tar.bz2
+
 # Tool Permissions
 # Controls how tool usage permissions are handled
 tool_permissions: