pvw-cli 1.2.4__tar.gz → 1.2.5__tar.gz

{pvw_cli-1.2.4 → pvw_cli-1.2.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pvw-cli
-Version: 1.2.4
+Version: 1.2.5
 Summary: Microsoft Purview CLI with comprehensive automation capabilities
 Author-email: AYOUB KEBAILI <keayoub@msn.com>
 Maintainer-email: AYOUB KEBAILI <keayoub@msn.com>
@@ -56,9 +56,9 @@ Requires-Dist: pytest-asyncio>=0.20.0; extra == "test"
 Requires-Dist: pytest-cov>=4.0.0; extra == "test"
 Requires-Dist: requests-mock>=1.9.0; extra == "test"
 
-# PURVIEW CLI v1.2.4 - Microsoft Purview Automation & Data Governance
+# PURVIEW CLI v1.2.5 - Microsoft Purview Automation & Data Governance
 
-> **LATEST UPDATE v1.2.4 (October 2025):**
+> **LATEST UPDATE v1.2.5 (October 2025):**
 >
 > - **[NEW]** Lineage CSV Import - Bulk import lineage relationships from CSV files with validation and sample generation
 > - **[FIXED]** Search API Integration - Fixed `suggest` and `autocomplete` API payload format (HTTP 400 errors resolved)
@@ -77,7 +77,7 @@ Requires-Dist: requests-mock>=1.9.0; extra == "test"
 
 ## What is PVW CLI?
 
-**PVW CLI v1.2.4** is a modern, full-featured command-line interface and Python library for Microsoft Purview. It enables automation and management of *all major Purview APIs* including:
+**PVW CLI v1.2.5** is a modern, full-featured command-line interface and Python library for Microsoft Purview. It enables automation and management of *all major Purview APIs* including:
 
 - **Unified Catalog (UC) Management** - Complete governance domains, glossary terms, data products, OKRs, CDEs
 - **Bulk Operations** - Import/export terms from CSV/JSON, bulk delete scripts with progress tracking
@@ -92,6 +92,15 @@ Requires-Dist: requests-mock>=1.9.0; extra == "test"
 
 The CLI is designed for data engineers, stewards, architects, and platform teams to automate, scale, and enhance their Microsoft Purview experience.
 
+### NEW: MCP Server for AI Assistants
+
+**[NEW]** Model Context Protocol (MCP) server enables LLM-powered data governance workflows! 
+
+- Natural language interface to Purview catalog
+- 20+ tools for AI assistants (Claude, Cline, etc.)
+- Automate complex multi-step operations
+- See `mcp/README.md` for setup instructions
+
 ---
 
 ## Getting Started
@@ -169,8 +178,9 @@ For more advanced usage, see the documentation in `doc/` or the project docs: <h
 
 ## Overview
 
-**PVW CLI v1.2.4** is a modern command-line interface and Python library for Microsoft Purview, enabling:
+**PVW CLI v1.2.5** is a modern command-line interface and Python library for Microsoft Purview, enabling:
 
+- **MCP Server** - Natural language interface for AI assistants (Claude, Cline)
 - Advanced data catalog search and discovery
 - Bulk import/export of entities, glossary terms, and lineage
 - Real-time monitoring and analytics
@@ -533,7 +543,7 @@ The PVW CLI provides advanced search using the latest Microsoft Purview Discover
 - Use autocomplete and suggestion endpoints
 - Perform faceted, time-based, and entity-type-specific queries
 
-**v1.2.4 Improvements:**
+**v1.2.5 Improvements:**
 
 - Fixed `suggest` and `autocomplete` API payload format (removed empty filter causing HTTP 400 errors)
 - Enhanced collection display with robust type checking and fallback logic
@@ -1297,7 +1307,7 @@ PVW CLI includes comprehensive sample files and scripts for bulk operations:
 - Success/failure tracking per term
 - Rate limiting (200ms delay)
 
-### Critical Fixes (v1.2.4)
+### Critical Fixes (v1.2.5)
 
 - **Search API Suggest/Autocomplete:** Fixed HTTP 400 errors by removing empty filter objects from payload
 - **Collection Display:** Enhanced collection name detection with proper fallback logic (isinstance checks)
@@ -1360,9 +1370,9 @@ See [LICENSE](LICENSE) file for details.
 
 ---
 
-**PVW CLI v1.2.4 empowers data engineers, stewards, and architects to automate, scale, and enhance their Microsoft Purview experience with powerful command-line and programmatic capabilities.**
+**PVW CLI v1.2.5 empowers data engineers, stewards, and architects to automate, scale, and enhance their Microsoft Purview experience with powerful command-line and programmatic capabilities.**
 
-**Latest in v1.2.4:**
+**Latest in v1.2.5:**
 
 - Fixed Search API suggest/autocomplete (HTTP 400 errors resolved)
 - Enhanced collection display with robust fallback logic

{pvw_cli-1.2.4 → pvw_cli-1.2.5}/README.md

@@ -1,6 +1,6 @@
-# PURVIEW CLI v1.2.4 - Microsoft Purview Automation & Data Governance
+# PURVIEW CLI v1.2.5 - Microsoft Purview Automation & Data Governance
 
-> **LATEST UPDATE v1.2.4 (October 2025):**
+> **LATEST UPDATE v1.2.5 (October 2025):**
 >
 > - **[NEW]** Lineage CSV Import - Bulk import lineage relationships from CSV files with validation and sample generation
 > - **[FIXED]** Search API Integration - Fixed `suggest` and `autocomplete` API payload format (HTTP 400 errors resolved)
@@ -19,7 +19,7 @@
 
 ## What is PVW CLI?
 
-**PVW CLI v1.2.4** is a modern, full-featured command-line interface and Python library for Microsoft Purview. It enables automation and management of *all major Purview APIs* including:
+**PVW CLI v1.2.5** is a modern, full-featured command-line interface and Python library for Microsoft Purview. It enables automation and management of *all major Purview APIs* including:
 
 - **Unified Catalog (UC) Management** - Complete governance domains, glossary terms, data products, OKRs, CDEs
 - **Bulk Operations** - Import/export terms from CSV/JSON, bulk delete scripts with progress tracking
@@ -34,6 +34,15 @@
 
 The CLI is designed for data engineers, stewards, architects, and platform teams to automate, scale, and enhance their Microsoft Purview experience.
 
+### NEW: MCP Server for AI Assistants
+
+**[NEW]** Model Context Protocol (MCP) server enables LLM-powered data governance workflows! 
+
+- Natural language interface to Purview catalog
+- 20+ tools for AI assistants (Claude, Cline, etc.)
+- Automate complex multi-step operations
+- See `mcp/README.md` for setup instructions
+
 ---
 
 ## Getting Started
@@ -111,8 +120,9 @@ For more advanced usage, see the documentation in `doc/` or the project docs: <h
 
 ## Overview
 
-**PVW CLI v1.2.4** is a modern command-line interface and Python library for Microsoft Purview, enabling:
+**PVW CLI v1.2.5** is a modern command-line interface and Python library for Microsoft Purview, enabling:
 
+- **MCP Server** - Natural language interface for AI assistants (Claude, Cline)
 - Advanced data catalog search and discovery
 - Bulk import/export of entities, glossary terms, and lineage
 - Real-time monitoring and analytics
@@ -475,7 +485,7 @@ The PVW CLI provides advanced search using the latest Microsoft Purview Discover
 - Use autocomplete and suggestion endpoints
 - Perform faceted, time-based, and entity-type-specific queries
 
-**v1.2.4 Improvements:**
+**v1.2.5 Improvements:**
 
 - Fixed `suggest` and `autocomplete` API payload format (removed empty filter causing HTTP 400 errors)
 - Enhanced collection display with robust type checking and fallback logic
@@ -1239,7 +1249,7 @@ PVW CLI includes comprehensive sample files and scripts for bulk operations:
 - Success/failure tracking per term
 - Rate limiting (200ms delay)
 
-### Critical Fixes (v1.2.4)
+### Critical Fixes (v1.2.5)
 
 - **Search API Suggest/Autocomplete:** Fixed HTTP 400 errors by removing empty filter objects from payload
 - **Collection Display:** Enhanced collection name detection with proper fallback logic (isinstance checks)
@@ -1302,9 +1312,9 @@ See [LICENSE](LICENSE) file for details.
 
 ---
 
-**PVW CLI v1.2.4 empowers data engineers, stewards, and architects to automate, scale, and enhance their Microsoft Purview experience with powerful command-line and programmatic capabilities.**
+**PVW CLI v1.2.5 empowers data engineers, stewards, and architects to automate, scale, and enhance their Microsoft Purview experience with powerful command-line and programmatic capabilities.**
 
-**Latest in v1.2.4:**
+**Latest in v1.2.5:**
 
 - Fixed Search API suggest/autocomplete (HTTP 400 errors resolved)
 - Enhanced collection display with robust fallback logic

{pvw_cli-1.2.4 → pvw_cli-1.2.5}/purviewcli/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "1.2.4"
+__version__ = "1.2.5"
 
 # Import main client modules
 from .client import *

{pvw_cli-1.2.4 → pvw_cli-1.2.5}/purviewcli/cli/types.py

@@ -495,4 +495,337 @@ def update_enum_def(payload_file, dry_run, validate):
     except Exception as e:
         click.echo(f"Error: {e}")
 
+
+@types.command(name="list-business-attributes")
+@click.option('--output', type=click.Choice(['table', 'json']), default='table', help='Output format')
+@click.option('--show-empty-groups/--hide-empty-groups', default=True, help='Show groups with no attributes')
+def list_business_attributes(output, show_empty_groups):
+    """List all business metadata attributes (Custom metadata in Purview UI)
+    
+    This command displays individual attributes organized by their parent groups,
+    matching what you see in the Purview "Custom metadata (preview)" interface
+    under "Business concept attributes" tab.
+    
+    Examples:
+        pvw types list-business-attributes
+        pvw types list-business-attributes --output json
+        pvw types list-business-attributes --hide-empty-groups
+    """
+    from rich.console import Console
+    from rich.table import Table
+    
+    console = Console()
+    
+    try:
+        client = Types()
+        result = client.typesRead({})
+        
+        if not result:
+            console.print("[red]ERROR:[/red] Failed to retrieve type definitions")
+            return
+        
+        biz = result.get('businessMetadataDefs', [])
+        
+        if not biz:
+            console.print("[yellow]No business metadata found[/yellow]")
+            return
+        
+        if output == 'json':
+            # JSON output: list of attributes with their group info
+            attributes_list = []
+            for group in biz:
+                group_name = group.get('name', 'N/A')
+                group_guid = group.get('guid', 'N/A')
+                
+                attributes = group.get('attributeDefs', [])
+                if not attributes and not show_empty_groups:
+                    continue
+                
+                for attr in attributes:
+                    attributes_list.append({
+                        'attributeName': attr.get('name'),
+                        'group': group_name,
+                        'groupGuid': group_guid,
+                        'type': attr.get('typeName'),
+                        'description': attr.get('description', ''),
+                        'isOptional': attr.get('isOptional', True),
+                        'isIndexable': attr.get('isIndexable', False)
+                    })
+            
+            click.echo(json.dumps(attributes_list, indent=2))
+            return
+        
+        # Table output
+        table = Table(title="[bold cyan]Business Concept Attributes (Custom Metadata)[/bold cyan]")
+        table.add_column("Attribute Name", style="green", no_wrap=True)
+        table.add_column("Group", style="cyan")
+        table.add_column("Type", style="yellow")
+        table.add_column("Description", style="white", max_width=40)
+        table.add_column("Scope", style="magenta")
+        
+        total_attributes = 0
+        
+        for group in biz:
+            group_name = group.get('name', 'N/A')
+            
+            # Parse data governance options for scope
+            scope_info = "N/A"
+            options = group.get('options', {})
+            if 'dataGovernanceOptions' in options:
+                try:
+                    dg_opts_str = options.get('dataGovernanceOptions', '{}')
+                    dg_opts = json.loads(dg_opts_str) if isinstance(dg_opts_str, str) else dg_opts_str
+                    applicable = dg_opts.get('applicableConstructs', [])
+                    if applicable:
+                        # Extract construct types (domain, businessConcept, etc.)
+                        scope_parts = []
+                        for construct in applicable[:2]:  # Show first 2
+                            if ':' in construct:
+                                scope_parts.append(construct.split(':')[0])
+                            else:
+                                scope_parts.append(construct)
+                        scope_info = ', '.join(scope_parts)
+                        if len(applicable) > 2:
+                            scope_info += f", +{len(applicable)-2} more"
+                except:
+                    pass
+            
+            # List all attributes in this group
+            attributes = group.get('attributeDefs', [])
+            
+            if attributes:
+                for attr in attributes:
+                    total_attributes += 1
+                    attr_name = attr.get('name', 'N/A')
+                    attr_type = attr.get('typeName', 'N/A')
+                    
+                    # Simplify enum types for display
+                    if 'ATTRIBUTE_ENUM_' in attr_type:
+                        attr_type = 'Enum (Single choice)'
+                    
+                    attr_desc = attr.get('description', '')
+                    
+                    # Get scope from attribute if it overrides group
+                    attr_scope = scope_info
+                    attr_opts = attr.get('options', {})
+                    if 'dataGovernanceOptions' in attr_opts:
+                        try:
+                            attr_dg_str = attr_opts.get('dataGovernanceOptions', '{}')
+                            attr_dg = json.loads(attr_dg_str) if isinstance(attr_dg_str, str) else attr_dg_str
+                            inherit = attr_dg.get('inheritApplicableConstructsFromGroup', True)
+                            if not inherit:
+                                attr_applicable = attr_dg.get('applicableConstructs', [])
+                                if attr_applicable:
+                                    attr_scope = f"{len(attr_applicable)} custom scope(s)"
+                        except:
+                            pass
+                    
+                    table.add_row(
+                        attr_name,
+                        group_name,
+                        attr_type,
+                        attr_desc[:40] + "..." if len(attr_desc) > 40 else attr_desc,
+                        attr_scope
+                    )
+            elif show_empty_groups:
+                # Show group with no attributes
+                table.add_row(
+                    f"[dim](no attributes)[/dim]",
+                    group_name,
+                    "-",
+                    f"[dim]Empty group[/dim]",
+                    scope_info
+                )
+        
+        console.print(table)
+        console.print(f"\n[cyan]Total:[/cyan] {total_attributes} attribute(s) in {len(biz)} group(s)")
+        
+        if total_attributes > 0:
+            console.print("\n[dim]Tip: Use 'pvw types read-business-metadata-def --name <GroupName>' for details[/dim]")
+            console.print("[dim]Tip: Use 'pvw types list-business-metadata-groups' to see group-level summary[/dim]")
+    
+    except Exception as e:
+        console.print(f"[red]ERROR:[/red] {str(e)}")
+
+
+@types.command(name="list-business-metadata-groups")
+@click.option('--output', type=click.Choice(['table', 'json']), default='table', help='Output format')
+def list_business_metadata_groups(output):
+    """List business metadata groups with their scope (Business Concept vs Data Asset).
+    
+    Shows a summary view of metadata groups to distinguish which apply to:
+    - Business Concepts (Terms, Domains, Business Rules)
+    - Data Assets (Tables, Files, Databases)
+    - Universal (Both)
+    
+    Examples:
+        pvw types list-business-metadata-groups
+        pvw types list-business-metadata-groups --output json
+    """
+    from rich.console import Console
+    from rich.table import Table
+    import json
+    
+    console = Console()
+    
+    try:
+        client = Types()
+        result = client.typesRead({})
+        
+        if not result:
+            console.print("[red]ERROR:[/red] Failed to retrieve type definitions")
+            return
+        
+        biz = result.get('businessMetadataDefs', [])
+        
+        if not biz:
+            console.print("[yellow]No business metadata groups found[/yellow]")
+            return
+        
+        if output == 'json':
+            # JSON output
+            groups_list = []
+            for group in biz:
+                group_name = group.get('name', 'N/A')
+                group_guid = group.get('guid', 'N/A')
+                attr_count = len(group.get('attributeDefs', []))
+                
+                # Determine scope
+                scope = "N/A"
+                scope_type = "unknown"
+                options = group.get('options', {})
+                
+                if 'dataGovernanceOptions' in options:
+                    try:
+                        dg_opts_str = options.get('dataGovernanceOptions', '{}')
+                        dg_opts = json.loads(dg_opts_str) if isinstance(dg_opts_str, str) else dg_opts_str
+                        applicable = dg_opts.get('applicableConstructs', [])
+                        
+                        if applicable:
+                            has_business_concept = any('businessConcept' in c or 'domain' in c for c in applicable)
+                            has_dataset = any('dataset' in c.lower() for c in applicable)
+                            
+                            if has_business_concept and has_dataset:
+                                scope = "Universal (Concept + Dataset)"
+                                scope_type = "universal"
+                            elif has_business_concept:
+                                scope = "Business Concept"
+                                scope_type = "business_concept"
+                            elif has_dataset:
+                                scope = "Data Asset"
+                                scope_type = "data_asset"
+                            else:
+                                scope = ', '.join([c.split(':')[0] if ':' in c else c for c in applicable[:3]])
+                                scope_type = "custom"
+                    except:
+                        pass
+                
+                # Check for legacy applicableEntityTypes in attributes
+                if scope == "N/A":
+                    for attr in group.get('attributeDefs', []):
+                        attr_opts = attr.get('options', {})
+                        if 'applicableEntityTypes' in attr_opts:
+                            try:
+                                entity_types_str = attr_opts.get('applicableEntityTypes', '[]')
+                                entity_types = json.loads(entity_types_str) if isinstance(entity_types_str, str) else entity_types_str
+                                if entity_types and isinstance(entity_types, list):
+                                    if any('table' in et.lower() or 'database' in et.lower() for et in entity_types):
+                                        scope = "Data Asset (Legacy)"
+                                        scope_type = "data_asset_legacy"
+                                        break
+                            except:
+                                pass
+                
+                groups_list.append({
+                    'groupName': group_name,
+                    'groupGuid': group_guid,
+                    'scope': scope,
+                    'scopeType': scope_type,
+                    'attributeCount': attr_count,
+                    'description': group.get('description', '')
+                })
+            
+            click.echo(json.dumps(groups_list, indent=2))
+            return
+        
+        # Table output
+        table = Table(title="[bold cyan]Business Metadata Groups[/bold cyan]", show_header=True)
+        table.add_column("Group Name", style="cyan", no_wrap=True)
+        table.add_column("Scope", style="magenta", max_width=30)
+        table.add_column("Attributes", style="yellow", justify="center")
+        table.add_column("Description", style="white", max_width=40)
+        
+        for group in biz:
+            group_name = group.get('name', 'N/A')
+            attr_count = len(group.get('attributeDefs', []))
+            group_desc = group.get('description', '')
+            
+            # Determine scope
+            scope = "N/A"
+            scope_style = "white"
+            options = group.get('options', {})
+            
+            if 'dataGovernanceOptions' in options:
+                try:
+                    dg_opts_str = options.get('dataGovernanceOptions', '{}')
+                    dg_opts = json.loads(dg_opts_str) if isinstance(dg_opts_str, str) else dg_opts_str
+                    applicable = dg_opts.get('applicableConstructs', [])
+                    
+                    if applicable:
+                        has_business_concept = any('businessConcept' in c or 'domain' in c for c in applicable)
+                        has_dataset = any('dataset' in c.lower() for c in applicable)
+                        
+                        if has_business_concept and has_dataset:
+                            scope = "Universal"
+                            scope_style = "magenta bold"
+                        elif has_business_concept:
+                            scope = "Business Concept"
+                            scope_style = "green"
+                        elif has_dataset:
+                            scope = "Data Asset"
+                            scope_style = "blue"
+                        else:
+                            scope = ', '.join([c.split(':')[0] if ':' in c else c for c in applicable[:2]])
+                            scope_style = "yellow"
+                except:
+                    pass
+            
+            # Check for legacy applicableEntityTypes
+            if scope == "N/A":
+                for attr in group.get('attributeDefs', []):
+                    attr_opts = attr.get('options', {})
+                    if 'applicableEntityTypes' in attr_opts:
+                        try:
+                            entity_types_str = attr_opts.get('applicableEntityTypes', '[]')
+                            entity_types = json.loads(entity_types_str) if isinstance(entity_types_str, str) else entity_types_str
+                            if entity_types and isinstance(entity_types, list):
+                                if any('table' in et.lower() or 'database' in et.lower() for et in entity_types):
+                                    scope = "Data Asset (Legacy)"
+                                    scope_style = "blue dim"
+                                    break
+                        except:
+                            pass
+            
+            table.add_row(
+                group_name,
+                f"[{scope_style}]{scope}[/{scope_style}]",
+                str(attr_count),
+                group_desc[:40] + "..." if len(group_desc) > 40 else group_desc
+            )
+        
+        console.print(table)
+        console.print(f"\n[cyan]Total:[/cyan] {len(biz)} group(s)")
+        
+        console.print("\n[bold]Legend:[/bold]")
+        console.print("  [green]Business Concept[/green] = Applies to Glossary Terms, Domains, Business Rules")
+        console.print("  [blue]Data Asset[/blue] = Applies to Tables, Files, Databases, etc.")
+        console.print("  [magenta bold]Universal[/magenta bold] = Applies to both Concepts and Assets")
+        
+        console.print("\n[dim]Tip: Use 'pvw types list-business-attributes' to see individual attributes[/dim]")
+        console.print("[dim]Tip: Use 'pvw types read-business-metadata-def --name <GroupName>' for full details[/dim]")
+    
+    except Exception as e:
+        console.print(f"[red]ERROR:[/red] {str(e)}")
+
+
 __all__ = ['types']