Conjuncts and ZWJ Sequences

This section explains the two most important special grapheme patterns in Tamil and Sinhala: conjuncts and Zero-Width Joiner sequences.

Tamil Conjuncts

Some sequences of characters merge visually to form a completely new shape, known as a conjunct. Standard Unicode grapheme splitting does not know about these language-specific merges.

A common Tamil conjunct is க்ஷ (Ksha). It is composed of two separate clusters க் and ஷ, but visually it renders as one inseparable unit.

from grapheme_kit import Graphemizer

# Without grapheme-kit, this would split into ['க்', 'ஷ', ...]
g = Graphemizer("க்ஷத்ரியன்")

print(g.graphemes)
# Output: ['க்ஷ', 'த்', 'ரி', 'ய', 'ன்']

The library uses a look-ahead rule: if the current cluster is க் and the next cluster starts with ஷ, they are merged into a single grapheme.

The Shri Cluster

Another important conjunct is the honorific "Shri" in Tamil, which can be written as ஸ்ரீ or ஶ்ரீ. These consist of two separate Unicode clusters but visually render as one symbol.

from grapheme_kit import Graphemizer

print(Graphemizer("ஸ்ரீ").graphemes)
# Output: ['ஸ்ரீ']   — correctly kept as 1 grapheme

print(Graphemizer("ஸ்ரி").graphemes)
# Output: ['ஸ்', 'ரி']  — correctly split into 2 graphemes

Sinhala ZWJ Sequences

Sinhala script makes extensive use of the Zero-Width Joiner (ZWJ, U+200D). It is an invisible character used to force two adjacent characters to merge into a conjunct.

For example, ක් followed by ZWJ followed by ර renders as the single conjunct ක්‍ර.

from grapheme_kit import Graphemizer

# The string contains an invisible ZWJ between ක් and ර
g = Graphemizer("ක්‍රමය")

print(g.graphemes)
# Output: ['ක්‍ර', 'ම', 'ය']

grapheme-kit maintains an internal list of valid Sinhala ZWJ sequences and ensures they are always preserved as a single grapheme.

---

Next: Phonetic Decomposition