Every legal document indexed in Dike is identified by a unique, deterministic Canonical ID. Canonical IDs serve as the universal join key across every endpoint in the platform (/v1/search, /v1/reason, /v1/cite, /v1/document/{id}, and /v1/graph/traverse).

[jurisdiction]-[doc_type]-[year]-[official_number]

Canonical ID Syntax Specification

Components

1. jurisdiction (2 characters, lowercase ISO code):
   • eg: Arab Republic of Egypt
   • sa: Kingdom of Saudi Arabia
   • ae: United Arab Emirates
2. doc_type (abbreviated document category):
   • leg: Primary Legislation / Statutes / Laws (قانون / نظام)
   • dec: Ministerial, Prime Ministerial, or Royal Decrees (قرار / أمر ملكي / مرسوم)
   • reg: Executive Regulations / Implementing Bylaws (لائحة تنفيذية)
   • cass: Court of Cassation / Supreme Court Judgements (أحكام محكمة النقض / المحكمة العليا)
   • fat: Official Fatawa / State Council Opinions (فتاوى مجلس الدولة / دار الإفتاء)
   • circ: Regulatory Circulars & Directives (تعاميم)
3. year (4 digits, Gregorian calendar):
   • For Saudi documents enacted by Hijri calendar, the year is normalized to Gregorian with the Hijri year preserved in metadata.
4. official_number (alphanumeric identifier):
   • The promulgated statute, decree, or case number (e.g. 131, 45-m, 1234-q).

Canonical ID Examples

Canonical ID Lifecycle Across Endpoints

Normalization & Parsing Guarantees

When calling POST /v1/cite, Dike parses raw citations into Canonical IDs deterministically:

1. Unicode NFKC Normalization: Strips decorative tatweel (ـ), normalizes whitespace and diacritics.
2. Arabic-Indic Digit Folding: Automatically converts Eastern Arabic digits (١، ٢، ٣) to standard digits (1, 2, 3).
3. Alef & Hamza Unification: Resolves variants (أ, إ, آ, ا) to ensure robust pattern matching against statutory gazettes.
4. Validation: Returns 404 citation_not_found if the parsed law number has no corresponding entry in the corpus.