Citation-system profiles

Citation-system profiles constrain locator syntax through regular expressions and, where needed, additional documented validation rules. A pull request that adds or changes a citation system MUST include the full profile record.

Required profile fields

• id: persistent system URI.
• key: flat stable key used for deterministic UUID seeds.
• type: CitationSystem.
• preferred_label: human-readable label.
• description: prose description of the citation tradition and its canonical locator form.
• locator_regex: ECMAScript regular expression for machine-checkable locator pre-validation.

See Specification §7 for the full normative field list.

Canonical locator form

Every profile defines exactly one canonical spelling for each reference point. Profiles MUST follow the flat key and locator Unicode rules in Identifier syntax. Along two ASCII axes the profile MUST take an explicit position:

1. Digit sequences. The profile MUST state whether leading zeros are permitted in numeric components. Unless the citation tradition itself uses them, leading zeros MUST be forbidden.
2. Letter case. The profile MUST declare the canonical case of letter components. Locators are matched case-sensitively against that declared casing; case variants are non-canonical spellings.

A profile MAY add further locator rules for punctuation, whitespace, allowed scripts, or non-regex-checkable constraints (such as a pinned book vocabulary). Regex-checkable constraints — including the two axes above — SHOULD be encoded in locator_regex; other constraints MUST be documented in description. The machine-actionable contract is the flat key and locator_regex.

Because the canonical locator seeds the deterministic reference UUID, alternative spellings of the same reference point (John.3.16 vs john.3.16, 514a1 vs 514a01) would mint distinct permanent identities. Validators MUST reject non-canonical spellings; they MUST NOT fold them into the canonical form. The enumerated reference data in the registry is the authority on which locators are canonical and attested; locator_regex is the machine-checkable floor beneath it.

Seed profiles

Bekker profile (Aristotelian corpus). Implements Bekker numbering, the page-and-column-and-line citation introduced by August Immanuel Bekker’s 1831 edition and used as the standard reference scheme for Aristotle:

{
  "key": "bekker",
  "description": "Bekker numbering: page, column (a or b), and line, after August Immanuel Bekker's 1831 Berlin edition. Pages span the whole corpus from 1 (Categories 1a1) to roughly 1462. Canonical form: page and line are positive integers without leading zeros; the column letter is lowercase.",
  "locator_regex": "^[1-9][0-9]{0,3}[ab][1-9][0-9]?$"
}

Stephanus profile (Platonic corpus). Implements Stephanus pagination, the page-and-section citation from Henri Estienne’s 1578 edition and used as the standard reference scheme for Plato:

{
  "key": "stephanus",
  "description": "Stephanus pagination: page, section (a–e), and optional sub-line, after Henri Estienne's 1578 edition of Plato's works. Canonical form: page and sub-line are positive integers without leading zeros; the section letter is lowercase.",
  "locator_regex": "^[1-9][0-9]{0,3}[a-e](?:[1-9][0-9]?)?$"
}

Validation rule

Every CanonicalReference MUST point to a known CitationSystem. Its locator MUST be the profile’s canonical spelling and MUST match that system’s locator_regex (see Specification §8). Regex success is necessary but not sufficient: a usable TextRefs reference must resolve to a registered CanonicalReference and satisfy any additional profile validation rules.