django-datalog 0.4.0

Django Datalog - Logic programming and inference engine for Django applications

django-datalog

A high-performance logic programming and inference engine for Django applications with advanced query optimization.

✨ Key Features

• 🧠 Logic Programming: Define facts and rules using intuitive Python syntax
• 🚀 Advanced Query Optimization: AST-based analysis with up to 75% query reduction
• 🔗 Cross-Variable Constraints: Complex relational queries with automatic optimization
• 🛡️ Security First: 100% Django ORM - eliminates SQL injection vulnerabilities
• ⚡ Zero Configuration: Transparent optimization - no code changes required
• 🔧 Developer Tools: CLI tools for query analysis and optimization insights

Installation

pip install django-datalog

# settings.py
INSTALLED_APPS = ['django_datalog']

python manage.py migrate

📈 Performance Note: All existing code automatically benefits from the new advanced query optimization system. No changes required - just upgrade and get better performance!

Core Concepts

Facts

Define facts as Python classes with Django model integration. Use Term[X] (a shorthand for X | Var[X]) for each slot — it reads as "an X, or a variable standing for an X":

from django_datalog.models import Fact, Term, Var

class WorksFor(Fact):
    subject: Term[Employee]  # Employee
    object: Term[Company]    # Company

class ColleaguesOf(Fact, inferred=True):  # Inferred facts can't be stored directly
    subject: Term[Employee]
    object: Term[Employee]

Typed variables

Var is parametric: Var[Employee]("emp") records that the variable stands for an Employee, so it only fits into Employee-typed slots. Create each variable once and reuse it across a rule/query — the type is carried along and a type checker rejects a variable used in the wrong position:

emp = Var[Employee]("emp")
company = Var[Company]("company")

query(WorksFor(emp, company))    # ✅ ok
query(WorksFor(company, emp))    # ✗ type error: Var[Company] can't be an Employee slot

Bare Var("emp") is still valid (inferred from the slot it fills), so the type parameter is fully opt-in and backward compatible.

Rules

Define inference logic with tuples (AND) and lists (OR):

from django_datalog.rules import rule

# Simple rule: Colleagues work at same company
emp1, emp2 = Var[Employee]("emp1"), Var[Employee]("emp2")
company = Var[Company]("company")
rule(
    ColleaguesOf(emp1, emp2),
    WorksFor(emp1, company) & WorksFor(emp2, company)
)

# Disjunctive rule: HasAccess via admin OR manager
user, resource = Var[User]("user"), Var[Resource]("resource")
rule(
    HasAccess(user, resource),
    IsAdmin(user) | IsManager(user, resource)
)

# Mixed rule: Complex access control
user, doc, folder = Var[User]("user"), Var[Document]("doc"), Var[Folder]("folder")
rule(
    CanEdit(user, doc),
    IsOwner(user, doc) |
    (IsManager(user, folder) & Contains(folder, doc))
)

Fact Operators

Use | (OR) and & (AND) operators:

# Modern operator syntax (recommended):
rule(head, fact1 | fact2)           # OR: fact1 OR fact2
rule(head, fact1 & fact2)           # AND: fact1 AND fact2

# Combining operators:
rule(head, (fact1 & fact2) | fact3)  # (fact1 AND fact2) OR fact3
rule(head, fact1 & fact2 & fact3)    # fact1 AND fact2 AND fact3
rule(head, fact1 | fact2 | fact3)    # fact1 OR fact2 OR fact3

# Legacy syntax (still supported):
rule(head, [fact1, fact2])           # OR (list syntax)
rule(head, (fact1, fact2))          # AND (tuple syntax)

Storing Facts

from django_datalog.models import store_facts

store_facts(
    WorksFor(subject=alice, object=tech_corp),
    WorksFor(subject=bob, object=tech_corp),
)

Querying

from django_datalog.models import query

# Find Alice's colleagues
colleagues = list(query(ColleaguesOf(alice, Var("colleague"))))

# With Django Q constraints
managers = list(query(WorksFor(Var("emp", where=Q(is_manager=True)), tech_corp)))

# Complex cross-variable constraints (automatically optimized)
results = list(query(
    WorksFor(Var("emp"), Var("company")),
    WorksOn(Var("emp"), Var("project", where=Q(company=Var("company"))))
))
# ↑ Automatically converts to optimized Django ORM with EXISTS subqueries

# Complex queries
results = list(query(
    ColleaguesOf(Var("emp1"), Var("emp2")),
    WorksFor(Var("emp1"), Var("company", where=Q(is_active=True)))
))

Rule Context

Isolate rules for testing or temporary logic:

from django_datalog.models import rule_context

# As context manager
with rule_context():
    rule(TestFact(Var("x")), LocalFact(Var("x")))
    results = query(TestFact(Var("x")))  # Rules active here

# As decorator
@rule_context
def test_something(self):
    rule(TestFact(Var("x")), LocalFact(Var("x")))
    assert len(query(TestFact(Var("x")))) > 0

Variables & Constraints

# Basic variable (typed — only fits Employee slots)
emp = Var[Employee]("employee")

# With Django Q constraints
senior_emp = Var[Employee]("employee", where=Q(years_experience__gte=5))

# Multiple constraints
constrained = Var[Employee]("emp", where=Q(is_active=True) & Q(department="Engineering"))

# Cross-variable constraints (reference other variables)
query(
    WorksFor(Var("emp"), Var("company")),
    WorksOn(Var("emp"), Var("project", where=Q(company=Var("company"))))
)
# Finds employees working on projects from their own company

# Complex cross-variable relationships
query(
    MemberOf(Var("emp"), Var("dept")),
    WorksFor(Var("emp"), Var("company", where=Q(is_active=True, department__in=[Var("dept")])))
)
# Finds employees in departments that belong to active companies

Performance Features

Advanced Query Analysis System

The engine features a sophisticated AST-based optimization system that works transparently:

• Query AST Parser: Automatically parses queries into abstract syntax trees
• Dependency Analysis: Maps variable relationships and constraint dependencies
• Execution Planning: Creates optimal execution plans based on query structure
• Recursive ORM Construction: Builds complex Django ORM queries automatically
• Cross-Variable Constraint Resolution: Transforms complex constraints into optimized EXISTS subqueries

Automatic Optimization

The engine automatically:

• Converts complex patterns to optimized Django ORM queries (up to 75% query reduction)
• Propagates constraints across same-named variables
• Orders execution by selectivity (most selective first)
• Learns from execution times for better planning
• Pushes constraints to the database
• Eliminates SQL injection by using 100% Django ORM

# You write natural cross-variable queries:
query(
    WorksFor(Var("emp"), Var("company")),
    WorksOn(Var("emp"), Var("project", where=Q(company=Var("company"))))
)

# Engine automatically generates optimized SQL like:
# SELECT ... FROM worksforstorage 
# INNER JOIN employee ON (...) 
# INNER JOIN company ON (...)
# WHERE EXISTS(
#     SELECT 1 FROM worksonstorage U0 
#     INNER JOIN project U2 ON (...) 
#     WHERE (...) AND U2.company_id = worksforstorage.object_id
# )
# Result: 16 queries → 4 queries (75% improvement)

Performance Analysis Tools

# Analyze query patterns and get optimization recommendations
python manage.py convert_to_orm --analyze

# Interactive query analysis
python manage.py convert_to_orm --interactive

# Process file with django-datalog queries
python manage.py convert_to_orm --file my_queries.py

Example: Complete Employee System

# models.py
class Employee(models.Model):
    name = models.CharField(max_length=100)
    is_manager = models.BooleanField(default=False)
    company = models.ForeignKey(Company, on_delete=models.CASCADE)

class Project(models.Model):
    name = models.CharField(max_length=100)
    company = models.ForeignKey(Company, on_delete=models.CASCADE)

class WorksFor(Fact):
    subject: Term[Employee]
    object: Term[Company]

class WorksOn(Fact):
    subject: Term[Employee]
    object: Term[Project]

class ColleaguesOf(Fact, inferred=True):
    subject: Term[Employee]
    object: Term[Employee]

# rules.py
emp1, emp2 = Var[Employee]("emp1"), Var[Employee]("emp2")
company = Var[Company]("company")
rule(
    ColleaguesOf(emp1, emp2),
    WorksFor(emp1, company) & WorksFor(emp2, company)
)

# usage.py
store_facts(
    WorksFor(subject=alice, object=tech_corp),
    WorksFor(subject=bob, object=tech_corp),
    WorksOn(subject=alice, object=tech_project),
    WorksOn(subject=bob, object=other_project),
)

# Simple queries (automatically optimized)
colleagues = query(ColleaguesOf(alice, Var[Employee]("colleague")))

# Complex cross-variable constraints (75% query reduction!)
emp, company = Var[Employee]("emp"), Var[Company]("company")
same_company_projects = query(
    WorksFor(emp, company),
    WorksOn(emp, Var[Project]("project", where=Q(company=company)))
)
# ↑ Finds employees working on projects from their own company
# Automatically converts to optimized Django ORM with EXISTS subqueries

Testing

class MyTest(TestCase):
    @rule_context  # Isolate rules per test
    def test_access_control(self):
        rule(CanAccess(Var("user")), IsAdmin(Var("user")))
        
        results = query(CanAccess(admin_user))
        self.assertEqual(len(results), 1)

Documentation

• docs/ - Technical documentation and reference materials
• docs/django_orm_equivalents.md - Django ORM equivalent queries and conversion patterns

Requirements

• Python 3.12+
• Django 5.2 (LTS)

License

MIT License