Composition Exclusions

Characters with the property Full_Composition_Exclusion can never appear in normalised text because the normalisation algorithm will not combine their constituent parts into a precomposed character. They can be categorised as follows:

Script Specifics

Characters with regular decomposition mappings that have been exceptionally excluded from composition because their decomposed form is preferred in actual usage

Post Composition Version Precomposed Characters

Characters with regular decomposition mappings that have been exceptionally excluded from composition because they were encoded after or close to Unicode 3.1, the “composition version”, and fulfill certain other conditions, such as decomposing into characters from earlier Unicode versions

Singleton Decompositions

Characters that decompose into a single code point instead of a pair of code points as required for composition

Non‐Starter Decompositions

Characters that decompose into a pair of code points where the first code point has a Canonical_Combining_Class value other than 0 (Not_Reordered), which the normalisation algorithm does not recognise as the start of a composable sequence

See Unicode Standard Annex #15: Unicode Normalization Forms for more information.

Sources: CompositionExclusions.txt, DerivedNormalizationProps.txt