Normalizer API¶

Overview¶

Unicode normalization, ZWNJ/ZWJ stripping, and script detection for Nepali text.

Functions¶

`normalize(text, form="NFC")`¶

Normalize text with all options enabled.

from nepalikit.normalizer import normalize

normalize("म  नाम   राम हो")  # "म नाम राम हो"
normalize("म\u200Cनाम")       # "मनाम"

`detect_script(text)`¶

Detect the script of the text.

from nepalikit.normalizer import detect_script

detect_script("मेरो नाम राम हो")  # "devanagari"
detect_script("mero naam ram ho")  # "latin"
detect_script("मेरो name ram")     # "mixed"
detect_script("12345")              # "other"

`is_devanagari(text)`¶

Check if text is purely Devanagari (ignoring whitespace/punctuation).

from nepalikit.normalizer import is_devanagari

is_devanagari("नेपाल")      # True
is_devanagari("नेपाल a")    # False

`contains_devanagari(text)`¶

Check if text contains at least one Devanagari character.

from nepalikit.normalizer import contains_devanagari

contains_devanagari("Hello नमस्ते")  # True
contains_devanagari("Hello")          # False

`contains_latin(text)`¶

Check if text contains at least one Latin character.

from nepalikit.normalizer import contains_latin

contains_latin("Hello नमस्ते")  # True
contains_latin("नेपाल")          # False

`mixed_script_ratio(text)`¶

Compute Devanagari fraction over Devanagari + Latin characters.

from nepalikit.normalizer import mixed_script_ratio

mixed_script_ratio("नेपाल Hello")  # 0.5
mixed_script_ratio("नेपाल")          # 1.0
mixed_script_ratio("Hello")          # 0.0

Classes¶

`NepaliNormalizer`¶

Advanced normalizer with configurable options.

from nepalikit.normalizer import NepaliNormalizer

normalizer = NepaliNormalizer()
normalized = normalizer.normalize("म  नाम")
print(normalized)  # "म नाम"

script = normalizer.detect_script("मेरो नाम")
print(script)  # "devanagari"

# Check if text is purely Devanagari
normalizer.is_devanagari("नेपाल")  # True

# Get Devanagari fraction
normalizer.mixed_script_ratio("नेपाल Hello")  # 0.5

Methods¶

`NepaliNormalizer.normalize(text, form="NFC")`¶

Apply full normalization pipeline.

Parameters: - text (str): Input text - form (str): Unicode normalization form ('NFC' or 'NFKC')

Returns: Normalized text