Sinhala Scripts¶

A practical guide to using graphemes++ with Sinhala (සිංහල) text.

Why Grapheme-Aware Processing Matters for Sinhala¶

Sinhala uses Zero Width Joiner (ZWJ) characters to form conjunct consonants. These invisible characters connect multiple consonants into a single visual glyph:

ක්‍රි 1 grapheme · Multiple Unicode code points (with ZWJ)

Sinhala Grapheme Segmentation¶

Basic Examples¶

from graphemes_plusplus import Graphemizer

# Simple Sinhala text
g = Graphemizer("සිංහල")
print(g.graphemes)
# ['සිං', 'හ', 'ල']

# ZWJ conjuncts
g = Graphemizer("ක්‍රිකට්")
print(g.graphemes)
# ['ක්‍රි', 'ක', 'ට්']

ZWJ Handling¶

The standard grapheme library may split Sinhala ZWJ conjuncts incorrectly. graphemes++ detects ZWJ (U+200D) sequences and merges them:

# ZWJ conjuncts are kept as single graphemes
g = Graphemizer("ක්‍රීඩා")
print(g.graphemes)
# ['ක්‍රී', 'ඩා'] — correctly merged

Zero Width Joiner (ZWJ) The ZWJ character (U+200D) is invisible but linguistically significant in Sinhala. It tells the rendering engine to form a conjunct ligature. graphemes++ respects this and keeps ZWJ-connected characters together.

Sinhala Decomposition¶

Decompose Sinhala graphemes into consonant base + vowel:

from graphemes_plusplus import decompose, compose

# Decompose
print(decompose("ක"))    # ක් + අ
print(decompose("කා"))   # ක් + ආ
print(decompose("කැ"))   # ක් + ඇ

# Compose back
print(compose(decompose("කා")))  # කා

The Sinhala Phonetic System¶

graph TD
    A["Sinhala Letters"] --> B["ස්වර (Vowels)"]
    A --> C["ව්‍යඤ්ජන (Consonants)"]
    A --> D["Combined Forms"]

    B --> E["අ, ආ, ඇ, ඈ, ඉ, ඊ, උ, ඌ, ..."]
    C --> F["ක්, ග්, ච්, ජ්, ට්, ඩ්, ..."]
    D --> G["ක් + අ = ක, ක් + ආ = කා, ..."]

    style A fill:#7c4dff,color:#fff
    style D fill:#00bfa5,color:#fff

Sinhala Normalization¶

The Normalizer fixes common Sinhala encoding confusables:

Incorrect Sequence	Corrected	Description
`ේා`	`ෝ`	Kombuwa + aela-pilla → Kombu deka-pilla
`්ො`	`ෝ`	Hal + kombuva-with-aela-pilla
`ෟෙ`	`ෞ`	Gayanukitta confusion
`ෙ‌ෙ`	`ෛ`	Double kombuva
`්ෙ`	`ේ`	Hal + kombuva → Kombu deka-pilla

from graphemes_plusplus.utils import Normalizer

n = Normalizer()
# All variant encodings produce the correct form
n.normalize("ේා")   # → ෝ
n.normalize("්ො")   # → ෝ

Sinhala Distance Metrics¶

from graphemes_plusplus import levenshtein, hamming

print(levenshtein("ක්‍රම", "කම"))       # Grapheme-aware edit distance
print(hamming("ස්වාගතයි", "ස්වාගතයි"))  # 0 (identical)