vipranarayan14 / aksharas

An utility for analysing akṣaras and varṇas in an Devanagari text.

Aksharas

Aksharas is an utility for analysing akṣaras and varṇas in a Devanagari text.

Installation

npm i @vipran/aksharas

Usage

import Aksharas from "@vipran/aksharas";

// OR for CommonJS:
// const Aksharas = require("@vipran/aksharas").default;

const input = "सर्वे भवन्तु सुखिनः।"

const results = Aksharas.analyse(input);

const aksharas = results.aksharas.map(akshara => akshara.value);

console.log(aksharas); // "स", "र्वे", "भ", "व", "न्तु", "सु", "खि", "नः"

API

Aksharas.analyse()

Accepts a string input and returns a Results object.

const input: string = 'नमः';
const results: Results = Aksharas.analyse(input);

Aksharas.TokenType

It is an enum with the following values:

• TokenType.Akshara
• TokenType.Symbol
• TokenType.Whitespace
• TokenType.Invalid
• TokenType.Unrecognised

These can be used to filter the tokens in the Results object. Example:

import Aksharas from "@vipran/aksharas";
// OR import Aksharas, { TokenType } ...

const input = "हे! हरेऽत्र नागच्छ।";
const results = Aksharas.analyse(input);
const symbols = results.all
  .filter((token) => token.type === Aksharas.TokenType.Symbol)
  .map((token) => token.value);

console.log(symbols); // "ऽ", "।"

Aksharas.VarnaType

It is an enum with the following values:

• VarnaType.Svara
• VarnaType.Vyanjana

These can be used to filter the varnas in Results.varnas. Example:

import Aksharas from "@vipran/aksharas";
// OR import Aksharas, { VarnaType } ...

const input = "गुरुः";
const results = Aksharas.analyse(input);

const svaras = results.varnas
  .filter((varna) => varna.type === Aksharas.VarnaType.Svara)
  .map((varna) => varna.value);

console.log(svaras); // "उ", "उः"

Results

The Results object contains the following properties:

• all
  • type: Token[]
  • An array of Token objects containing all the tokens analysed from input string. It includes Devanagari akṣaras, Devanagari symbols (१, २, ।, ॥, etc.) and non-devangari characters (i.e. characters in other scripts, special characters, whitespace characters, etc.)
• aksharas
  • type: Token[]
  • Devanagari syllables like रा, सी, etc. Here, halanta consonants such as क्, च्, य्, etc. are also considered as aksharas when they are at the end of a word.
• varnas
  • type: Varna[]
  • Devanagari consonants and vowels in the input. (Only in v0.4.0 or above.)
• symbols
  • type: Token[]
  • Devanagari symbols such as १, २, ।, ॥, etc.
• whitespaces
  • type: Token[]
  • All whitespace characters: \s, \t, \n, etc.
• invalid
  • type: Token[]
  • All Devanagari characters whose occurance in the input string do not conform to the definition of an akṣara. For example, a virāma or a vowel mark which is not preceded by a consonant is invalid. ("अ्", "गोु", etc.)
• unrecognised
  • type: Token[]
  • Non-devangari characters (i.e. characters in other scripts and special characters such as @, #, etc.)
• chars
  • type: string[]
  • All Unicode characters in the input string. Same as String.prototype.split().

Token

Many of the properties in the Results object consists of an array of Token-s. A Token object has the following properties:

• type
  • type: TokenType
  • Type of the token. One of the values of Aksharas.TokenType.
• value
  • type: string
  • Conatins an analysed part of the input string.
• from
  • type: number
  • From index - representing the start position of the token in the input string.
• to
  • type: number
  • To index - representing the end position of the token in the input string.
• attributes
  • type: Record<string, any>
  • An optional key-value object which may contain other attributes of the token. It is currently used only in the Akshara tokens for storing the varnas in that akshara.

Varna

Results.varnas consists of an array of Varna objects. A Varna object has the following properties:

• type
  • type: VarnaType
  • Type of the token. One of the values of Aksharas.VarnaType.
• value
  • type: string
  • Conatins an analysed part of the input string.

License

MIT © Prasanna Venkatesh T S

aksharas's Issues

Add matra count info in the 'Akshara' tokens

add support for word count

Using aksharas to build a utility for comparing text

Hello. We are reaching out for building a utility that can compare texts for Indic languages, in particular Kannada. The functionality should approximate the working of this online tool https://countwordsfree.com/comparetexts. At this moment our understanding is that character counting as implemented in aksharas would be a prerequisite. In addition, we have been using aksharas-web for plain character counting in Kannada text by transliterating to Devanagari. We would like to build an interface for Kannada as well. Hoping to hear your thoughts on it, and on the GitHub workflow to take this forward.

akshara and varna freqency counts

recognise ardhavisarga ('ᳲ') as vowel mark

• Vedic ardhavisarga ('ᳲ') is used for indicating jihvamuliya and upadhmaniya. For example, "काकᳲकृष्णः".
• It should be recognised as a vowel mark.
• It is only valid if the following letter is k/kh/p/ph. Refer: कुप्वोः ≍क≍पौ च। Ast. 8/3/37

Q: Should the ardhavisarga be considered as part of the previous vowel or later consonant?
Q: Can ardhavisarga come only after 'a' or any vowel?

Consider "ॐ" as an akshara

Example

Input text: "ॐ"

Current behaviour

"ॐ" is considered a symbol (TokenType.Symbol).

Expected behaviour

It should be considered an akshara (TokenType.Akshara). It should be treated as equivalent to "ओम्". It should have varnasLength as 2 (and varnas as ओ and म्).

Add support for varnas in `analyse` results

Currently, the results returned by the analyse function contain only the aksharas from the input text. Add support for varnas also. Later, remove varnasLength from results since varnas itself can be used to find the length.

Add hrasva/dirgha/pluta info in the 'Akshara' tokens

Remove `varnasLength` from `analyse` results

Remove varnasLength from analyse results, since varnas itself can be used to find the length.