Implement CLI commands: list, get, export, delete

[octo-youcef]

Problem

CLI has placeholder commands for managing analyzed packages, but they're not implemented. Users can analyze code but can't:

• List what packages are stored
• View package details/statistics
• Export graph data
• Delete packages from database

Current State

mapper analyse list          # Shows "Not implemented yet"
mapper analyse get my-pkg    # Shows "Not implemented yet"  
mapper analyse export my-pkg # Shows "Not implemented yet"
mapper analyse delete my-pkg # Shows "Not implemented yet"

Required Commands

mapper analyse list

List all analyzed packages in database.

Options:

• --detailed - Show full stats (node counts, relationships, timestamps)
• --json - Output as JSON

Output:

Analyzed Packages:
  my-app       50 modules, 200 classes, 500 functions    2026-03-26
  utils        10 modules,  30 classes, 100 functions    2026-03-25
  service      25 modules, 120 classes, 300 functions    2026-03-24

mapper analyse get <package>

Show detailed package statistics.

Options:

• --depth N - Max depth for relationship traversal (default: 3)
• --stats-only - Show only statistics, no graph details

Output:

Package: my-app
Analyzed: 2026-03-26 14:30:00

Nodes:
  Modules:     50
  Classes:    200
  Functions:  500
  Methods:   1200

Relationships:
  DEFINES:    750
  CONTAINS:  1200
  CALLS:     3500
  IMPORTS:    200
  INHERITS:    50

Most Connected:
  utils.helpers.format_data (45 incoming calls)
  models.Base (30 child classes)

mapper analyse export <package>

Export graph data in various formats.

Options:

• --format <fmt> - Export format: json, cypher, graphml, dot, csv (default: json)
• --output <path> - Output file path (default: stdout)
• --only <type> - Export only nodes or relationships
• --node-type <type> - Filter by node type (Module, Class, Function, Method)
• --relationship-type <type> - Filter by relationship type
• --pretty - Pretty-print output

Formats:

JSON:

{
  "nodes": [
    {"id": "...", "label": "Class", "properties": {...}},
    ...
  ],
  "relationships": [
    {"from": "...", "to": "...", "type": "CALLS"},
    ...
  ]
}

Cypher (for importing into another Neo4j instance):

CREATE (n1:Module {name: "main", fqn: "main", ...});
CREATE (n2:Class {name: "App", fqn: "main.App", ...});
CREATE (n1)-[:DEFINES]->(n2);
...

GraphML (standard graph format):

<graphml>
  <node id="n1"><data key="label">Class</data>...</node>
  <edge source="n1" target="n2" label="INHERITS"/>
</graphml>

DOT (Graphviz format for visualization):

digraph G {
  n1 [label="MyClass"];
  n2 [label="BaseClass"];
  n1 -> n2 [label="INHERITS"];
}

CSV (two files: nodes.csv and relationships.csv):

id,label,name,fqn,package
n1,Class,MyClass,module.MyClass,my-app

mapper analyse delete <package>

Delete a package from the database.

Options:

• --force - Skip confirmation prompt
• --dry-run - Show what would be deleted without deleting

Output:

Deleting package: my-app

Nodes to delete:
  50 Modules
  200 Classes
  500 Functions
  1200 Methods

Total: 1950 nodes, 5700 relationships

Are you sure? [y/N]: y
Deleted 1950 nodes and 5700 relationships

Implementation Notes

Query Patterns

List packages:

MATCH (n)
RETURN DISTINCT n.package, count(n) as nodes
ORDER BY n.package

Get package stats:

MATCH (n {package: $package})
RETURN labels(n)[0] as type, count(*) as count

Count relationships:

MATCH (a {package: $package})-[r]-(b)
RETURN type(r) as rel_type, count(r) as count

Delete package:

MATCH (n {package: $package})
DETACH DELETE n
RETURN count(n) as deleted

Export Implementation

• Use Neo4jConnection.driver.session() to query data
• Stream results for large exports
• Format conversion in separate modules (exporters/)
• Progress bar for large datasets

Acceptance Criteria

• mapper analyse list shows all packages
• mapper analyse get <pkg> shows package details
• mapper analyse export <pkg> exports to all formats
• mapper analyse delete <pkg> removes package safely
• All commands support --json output for scripting
• Dry-run and confirmation prompts for destructive operations
• Tests for all commands and export formats

Related

• Required for production usage
• Mentioned in ROADMAP.md lines 89-93
• Enables package management workflow
• Export enables data portability

Priority

MEDIUM - Needed for production but not blocking analysis workflow.