dilemma 0.5.0

This is an expression language python package.

Dilemma Expression Language

An expression evaluation engine for Python applications that makes complex logical expressions readable and maintainable.

• Business rules
• Data validation
• Authorisation checks
• Data filtering
• Formulas and calculations

Why Dilemma?

Given context data like this:

---
project:
  status: review
signoffs:
- user:
    name: Alice Chen
    role: Audit
  timestamp: '2025-10-07T10:30:00Z'
- user:
    name: Bob Smith
    role: Audit
  timestamp: '2025-10-07T14:20:00Z'
- user:
    name: Carol Wong
    role: Audit
  timestamp: '2025-10-08T09:15:00Z'
- user:
    name: Dave Johnson
    role: Legal
  timestamp: '2025-10-07T16:45:00Z'
documents:
- name: contract.pdf
  verified: true
- name: financials.xlsx
  verified: true
- name: legal_review.docx
  verified: true

Instead of writing an expression like this:

// Complex JavaScript approach
audit_signoffs.filter((signoff) => {
    return signoff.user.role == 'Audit' && new Date(signoff.timestamp) < new Date()
}).length >= 3 
&& project.status == 'review' 
&& documents.filter((doc) => doc.verified).length === documents.length

Your users can express the business rules like this:

// dilemma expression language

    at least 3 of signoffs has `user.role == 'Audit' and timestamp is $past`
    and project.status == 'review'
    and all of documents has `verified == true`

For array operations, instead of complex function calls:

count_of(orders, `status == 'pending'`) >= 3 and any_of(orders, `total > 1000`)

User can enjoy a more natural language:

at least 3 of orders has `status == 'pending'` and any of orders has `total > 1000`

Features

• Secure evaluation - No arbitrary code execution, only safe expressions
• Rich data access - Navigate nested dictionaries and lists with ease
• Date/time operations - Natural language date comparisons
• Multiple resolvers - JsonPath, JQ, and basic dictionary lookup
• Performance optimized - Compile expressions once, evaluate many times
• Type safe - Built-in type checking and validation

Quick Start

pip install dilemma

from dilemma import evaluate

# Basic arithmetic and logic
result = evaluate("2 * (3 + 4)")  # Returns 14
result = evaluate("age >= 18 and status == 'active'", {"age": 25, "status": "active"})

# Natural language array operations
result = evaluate("at least 2 of users has `status == 'active'`", context)
result = evaluate("none of orders has `amount > 1000`", context)

# Date operations
result = evaluate("user.last_login upcoming within 7 days", context)
result = evaluate("subscription.end_date is $future", context)

# Complex data access
result = evaluate("user.permissions contains 'admin'", context)
result = evaluate("`[.users[] | select(.active == true) | .name] | length` > 0", context)

Language Features

All Language Features. Extensive Examples.

Data Access Patterns

# Dot notation for nested objects
"user.profile.settings.theme == 'dark'"

# Natural possessive syntax  
"user's subscription's status == 'premium'"

# Array/list access
"users[0].name == 'Alice'"

# Check membership
"'admin' in user.roles"
"user.permissions contains 'read'"

Natural Language Array Operations

Dilemma provides intuitive sugar syntax for common array operations:

# Quantity-based checks
"at least 3 of orders has `status == 'shipped'`"
"at most 2 of users has `role == 'admin'`"
"exactly 1 of servers has `status == 'maintenance'`"

# Existence checks
"any of products has `price > 100`"
"all of users has `email_verified == true`"
"none of alerts has `severity == 'critical'`"

# Combined with other operations
"at least 5 of reviews has `rating >= 4` and user.subscription is $active"
"any of files has `name like '*.pdf'` and all of files has `size < 10000000`"

Date and Time Operations

# Relative time checks
"user.created_at upcoming within 30 days"
"order.shipped_date older than 1 week"

# State comparisons
"subscription.expires is $future"
"last_backup is $past"
"meeting.date is $today"

# Date comparisons
"start_date before end_date"
"event.date same_day_as $now"

Advanced JQ Integration

For complex data manipulation, use JQ expressions in backticks:

# Filter and transform arrays - working with provided context
evaluate('`[.users[] | select(.active == true) | .name]`', context)

# Mathematical operations on arrays  
evaluate('`[.sales[].amount] | add` > 10000', context)

# Complex conditionals
evaluate('`[.products[] | select(.price > 100 and .category == "electronics")] | length` > 1', context)

Performance Optimization

For repeated evaluations, compile expressions once:

from dilemma import compile_expression

# Compile once - including array sugar syntax
eligibility_check = compile_expression(
    "user.age >= 18 and user.subscription.active and at least 1 of user.orders has `status == 'completed'`"
)

# Evaluate many times with different contexts
for user_data in users:
    if eligibility_check.evaluate(user_data):
        # send_premium_content(user_data)
        pass

Error Handling

Dilemma provides clear, actionable error messages:

try:
    result = evaluate("user.invalidfield == 'test'", context)
except VariableError as e:
    print(f"Expression error: {e}")
    # Suggests available fields and common fixes

Use Cases

• Form validation rules - "email like '*@*' and age >= 13"
• Business logic - "order.total > 100 and customer.tier == 'premium'"
• Access control - "user.roles contains 'admin' or resource.owner == user.id"
• Data filtering - "created_at upcoming within 24 hours and status == 'pending'"
• Workflow conditions - "approval.status == 'approved' and budget.remaining >= cost"
• Quality assurance - "all of tests has \status == 'passed'` and none of builds has `errors > 0`"`
• Inventory management - "at least 10 of products has \stock > 0` and any of suppliers has `delivery_time < 3`"`
• Security monitoring - "none of login_attempts has \failed_count > 5` and all of sessions has `encrypted == true`"`

Safety & Security

• ✅ No arbitrary Python code execution
• ✅ No access to imports or builtins
• ✅ Sandboxed evaluation environment
• ✅ Input validation and sanitization
• ✅ Memory and complexity limits

License

MIT License - see LICENSE file for details.