feat(stripEndings): add stripEndings function to remove line endings

Closes #25
Closes #138

README.hbs

@@ -29,6 +29,7 @@ const {
   toShahmukhi,
   stripAccents,
   stripVishraams,
+  stripEndings
   isGurmukhi,
 } = require( 'gurmukhi-utils' )
 
@@ -41,7 +42,8 @@ toHindi('ਕੁਲ ਜਨ ਮਧੇ ਮਿਲੵੋਿ ਸਾਰਗ ਪਾਨ 
 toShahmukhi('ਹਰਿ ਹਰਿ ਹਰਿ ਗੁਨੀ') // => هر هر هر گُنی
 stripAccents('ਜ਼ਫ਼ੈਸ਼ਸ') // => ਜਫੈਸਸ
 stripVishraams('sbid mrY. so mir rhY; iPir.') // => sbid mrY so mir rhY iPir
-isGurmukhi('ਗੁਰਮੁਖੀ') // t=> true
+stripEndings('ਸੋ ਘਰੁ ਰਾਖੁ; ਵਡਾਈ ਤੋਇ ॥੧॥ ਰਹਾਉ ॥') // => ਸੋ ਘਰੁ ਰਾਖੁ; ਵਡਾਈ ਤੋਇ
+isGurmukhi('ਗੁਰਮੁਖੀ') // => true
 ```
 
 Additionally, the package is available for web use via [unpkg CDN](https://unpkg.com/gurmukhi-utils).

README.md

@@ -22,6 +22,7 @@ Want to speak with us? <p>[![Slack](https://slack.shabados.com/badge.svg)](https
   * [firstLetters(line, [stripNukta], [withVishraams]) ⇒ String](#firstlettersline-stripnukta-withvishraams-%E2%87%92-string)
   * [isGurmukhi(text, [exhaustive]) ⇒ boolean](#isgurmukhitext-exhaustive-%E2%87%92-boolean)
   * [stripAccents(text) ⇒ String](#stripaccentstext-%E2%87%92-string)
+  * [stripEndings(text) ⇒ String](#stripendingstext-%E2%87%92-string)
   * [stripVishraams(text, options) ⇒ String](#stripvishraamstext-options-%E2%87%92-string)
   * [toAscii(text) ⇒ String](#toasciitext-%E2%87%92-string)
   * [toEnglish(line) ⇒ String](#toenglishline-%E2%87%92-string)
@@ -45,6 +46,7 @@ const {
   toShahmukhi,
   stripAccents,
   stripVishraams,
+  stripEndings
   isGurmukhi,
 } = require( 'gurmukhi-utils' )
 
@@ -57,7 +59,8 @@ toHindi('ਕੁਲ ਜਨ ਮਧੇ ਮਿਲੵੋਿ ਸਾਰਗ ਪਾਨ 
 toShahmukhi('ਹਰਿ ਹਰਿ ਹਰਿ ਗੁਨੀ') // => هر هر هر گُنی
 stripAccents('ਜ਼ਫ਼ੈਸ਼ਸ') // => ਜਫੈਸਸ
 stripVishraams('sbid mrY. so mir rhY; iPir.') // => sbid mrY so mir rhY iPir
-isGurmukhi('ਗੁਰਮੁਖੀ') // t=> true
+stripEndings('ਸੋ ਘਰੁ ਰਾਖੁ; ਵਡਾਈ ਤੋਇ ॥੧॥ ਰਹਾਉ ॥') // => ਸੋ ਘਰੁ ਰਾਖੁ; ਵਡਾਈ ਤੋਇ
+isGurmukhi('ਗੁਰਮੁਖੀ') // => true
 ```
 
 Additionally, the package is available for web use via [unpkg CDN](https://unpkg.com/gurmukhi-utils).
@@ -138,6 +141,35 @@ Useful for generalising search queries.
 stripAccents('ਜ਼ਫ਼ੈਸ਼ਸਓ') // => ਜਫੈਸਸੳ
 stripAccents('Z^Svb') // => gKsvb
 ```
+### stripEndings(text) ⇒ <code>String</code>
+Strips line endings from any Gurmukhi or translation string.
+Accepts both Unicode and ASCII input.
+Useful for generating accurate first letters or modifying non-Gurbani for better display.
+*Not* designed for headings or Sirlekhs.
+
+**Returns**: <code>String</code> - A ending-less version of the text.  
+
+| Param | Type | Description |
+| --- | --- | --- |
+| text | <code>String</code> | The text to stip endings from. |
+
+**Example** *(Line ending phrases)*  
+```js
+stripEndings('ਸੋ ਘਰੁ ਰਾਖੁ; ਵਡਾਈ ਤੋਇ ॥੧॥ ਰਹਾਉ ॥') // => ਸੋ ਘਰੁ ਰਾਖੁ; ਵਡਾਈ ਤੋਇ
+stripEndings('ਹੁਕਮੁ ਪਛਾਣਿ; ਤਾ ਖਸਮੈ ਮਿਲਣਾ ॥੧॥ ਰਹਾਉ ਦੂਜਾ ॥') // => ਹੁਕਮੁ ਪਛਾਣਿ; ਤਾ ਖਸਮੈ ਮਿਲਣਾ
+stripEndings('ਜਨ ਨਾਨਕ. ਗੁਰਮੁਖਿ ਜਾਤਾ ਰਾਮ ॥੪॥੬॥ ਛਕਾ ੧ ॥') // => ਜਨ ਨਾਨਕ. ਗੁਰਮੁਖਿ ਜਾਤਾ ਰਾਮ
+```
+**Example** *(English Translations)*  
+```js
+stripEndings('O Nanak, Forever And Ever True. ||1||') // => O Nanak, Forever And Ever True.
+stripEndings('lush greenery. ||1||Pause||') // => lush greenery.
+stripEndings('always I live within the Khalsa. 519') // => always I live within the Khalsa.
+stripEndings('without your reminiscence.(1) (3)') // => without your reminiscence.
+```
+**Example** *(Spanish Translations)*  
+```js
+stripEndings('ofrece su ser en sacrificio a Ti. (4-2-9)') // => ofrece su ser en sacrificio a Ti.
+```
 ### stripVishraams(text, options) ⇒ <code>String</code>
 Removes the specified vishraams from a string.
 

docs/README.md

@@ -22,6 +22,7 @@ Want to speak with us? <p>[![Slack](https://slack.shabados.com/badge.svg)](https
   * [firstLetters(line, [stripNukta], [withVishraams]) ⇒ String](#firstlettersline-stripnukta-withvishraams-%E2%87%92-string)
   * [isGurmukhi(text, [exhaustive]) ⇒ boolean](#isgurmukhitext-exhaustive-%E2%87%92-boolean)
   * [stripAccents(text) ⇒ String](#stripaccentstext-%E2%87%92-string)
+  * [stripEndings(text) ⇒ String](#stripendingstext-%E2%87%92-string)
   * [stripVishraams(text, options) ⇒ String](#stripvishraamstext-options-%E2%87%92-string)
   * [toAscii(text) ⇒ String](#toasciitext-%E2%87%92-string)
   * [toEnglish(line) ⇒ String](#toenglishline-%E2%87%92-string)
@@ -45,6 +46,7 @@ const {
   toShahmukhi,
   stripAccents,
   stripVishraams,
+  stripEndings
   isGurmukhi,
 } = require( 'gurmukhi-utils' )
 
@@ -57,7 +59,8 @@ toHindi('ਕੁਲ ਜਨ ਮਧੇ ਮਿਲੵੋਿ ਸਾਰਗ ਪਾਨ 
 toShahmukhi('ਹਰਿ ਹਰਿ ਹਰਿ ਗੁਨੀ') // => هر هر هر گُنی
 stripAccents('ਜ਼ਫ਼ੈਸ਼ਸ') // => ਜਫੈਸਸ
 stripVishraams('sbid mrY. so mir rhY; iPir.') // => sbid mrY so mir rhY iPir
-isGurmukhi('ਗੁਰਮੁਖੀ') // t=> true
+stripEndings('ਸੋ ਘਰੁ ਰਾਖੁ; ਵਡਾਈ ਤੋਇ ॥੧॥ ਰਹਾਉ ॥') // => ਸੋ ਘਰੁ ਰਾਖੁ; ਵਡਾਈ ਤੋਇ
+isGurmukhi('ਗੁਰਮੁਖੀ') // => true
 ```
 
 Additionally, the package is available for web use via [unpkg CDN](https://unpkg.com/gurmukhi-utils).
@@ -138,6 +141,35 @@ Useful for generalising search queries.
 stripAccents('ਜ਼ਫ਼ੈਸ਼ਸਓ') // => ਜਫੈਸਸੳ
 stripAccents('Z^Svb') // => gKsvb
 ```
+### stripEndings(text) ⇒ <code>String</code>
+Strips line endings from any Gurmukhi or translation string.
+Accepts both Unicode and ASCII input.
+Useful for generating accurate first letters or modifying non-Gurbani for better display.
+*Not* designed for headings or Sirlekhs.
+
+**Returns**: <code>String</code> - A ending-less version of the text.  
+
+| Param | Type | Description |
+| --- | --- | --- |
+| text | <code>String</code> | The text to stip endings from. |
+
+**Example** *(Line ending phrases)*  
+```js
+stripEndings('ਸੋ ਘਰੁ ਰਾਖੁ; ਵਡਾਈ ਤੋਇ ॥੧॥ ਰਹਾਉ ॥') // => ਸੋ ਘਰੁ ਰਾਖੁ; ਵਡਾਈ ਤੋਇ
+stripEndings('ਹੁਕਮੁ ਪਛਾਣਿ; ਤਾ ਖਸਮੈ ਮਿਲਣਾ ॥੧॥ ਰਹਾਉ ਦੂਜਾ ॥') // => ਹੁਕਮੁ ਪਛਾਣਿ; ਤਾ ਖਸਮੈ ਮਿਲਣਾ
+stripEndings('ਜਨ ਨਾਨਕ. ਗੁਰਮੁਖਿ ਜਾਤਾ ਰਾਮ ॥੪॥੬॥ ਛਕਾ ੧ ॥') // => ਜਨ ਨਾਨਕ. ਗੁਰਮੁਖਿ ਜਾਤਾ ਰਾਮ
+```
+**Example** *(English Translations)*  
+```js
+stripEndings('O Nanak, Forever And Ever True. ||1||') // => O Nanak, Forever And Ever True.
+stripEndings('lush greenery. ||1||Pause||') // => lush greenery.
+stripEndings('always I live within the Khalsa. 519') // => always I live within the Khalsa.
+stripEndings('without your reminiscence.(1) (3)') // => without your reminiscence.
+```
+**Example** *(Spanish Translations)*  
+```js
+stripEndings('ofrece su ser en sacrificio a Ti. (4-2-9)') // => ofrece su ser en sacrificio a Ti.
+```
 ### stripVishraams(text, options) ⇒ <code>String</code>
 Removes the specified vishraams from a string.
 

example.js

@@ -7,7 +7,8 @@ const {
   toEnglish,
   stripAccents,
   stripVishraams,
-  isGurmukhi
+  stripEndings,
+  isGurmukhi,
 } = require( 'gurmukhi-utils' )
 
 console.log(toUnicode( 'Koj' ))
@@ -19,4 +20,5 @@ console.log(toShahmukhi( 'ਹਰਿ ਹਰਿ ਹਰਿ ਗੁਨੀ' ))
 console.log(toHindi( 'ਕੁਲ ਜਨ ਮਧੇ ਮਿਲੵੋਿ ਸਾਰਗ ਪਾਨ ਰੇ ॥' ))
 console.log(stripAccents('ਜ਼ਫ਼ੈਸ਼ਸ'))
 console.log(stripVishraams('sbid mrY. so mir rhY; iPir.'))
+console.log(stripEndings('ਸੋ ਘਰੁ ਰਾਖੁ; ਵਡਾਈ ਤੋਇ ॥੧॥ ਰਹਾਉ ॥'))
 console.log(isGurmukhi('ਗੁਰਮੁਖੀ'))

index.d.ts

@@ -13,10 +13,13 @@ export function firstLetters(text: string, stripNukta?: boolean = true, withVish
 export function isGurmukhi(text: string, exhaustive?: boolean): boolean
 
 export function stripAccents(text: string): string
+
 interface StripVishraamsOptions {
   heavy?: boolean;
   medium?: boolean;
   light?: boolean;
 }
 
 export function stripVishraams(text: string, options?: StripVishraamsOptions): string
+
+export function stripEndings(text: string): string

index.js

@@ -7,6 +7,7 @@ const toHindi = require( './lib/toHindi' )
 const isGurmukhi = require( './lib/isGurmukhi' )
 const stripAccents = require( './lib/stripAccents' )
 const stripVishraams = require( './lib/stripVishraams' )
+const stripEndings = require( './lib/stripEndings' )
 
 module.exports = {
   toAscii,
@@ -18,4 +19,5 @@ module.exports = {
   isGurmukhi,
   stripAccents,
   stripVishraams,
+  stripEndings,
 }

lib/stripEndings.js

@@ -0,0 +1,51 @@
+const toUnicode = require( './toUnicode' )
+const toAscii = require( './toAscii' )
+const { getRegexClass, getRegexGroup } = require( './regex-utils' )
+
+// Line endings in both ASCII, Unicode, and English
+const endingClass = getRegexClass( [ '।', '॥', ']', '[', '|' ] )
+// Sometimes translation line endings begin with these characters, before numbers
+const optionalEndingClass = getRegexClass( [ '(' ] )
+// Remove any broken endings
+const brokenEndingClass = getRegexGroup( [ '()' ] )
+
+// All numbers in ASCII, Unicode
+const numbers = [ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9 ].map( i => i.toString() )
+const numberClass = getRegexClass( [ ...numbers, ...numbers.map( toUnicode ) ] )
+
+// Rahao in English, ASCII, Unicode
+const pauseGroup = getRegexGroup( [ 'ਰਹਾਉ', toAscii( 'ਰਹਾਉ' ), 'Pause' ] )
+
+const matchers = [
+  // Endings followed by any number => match the rest of the line
+  ` ?(${endingClass}|${optionalEndingClass}?)${numberClass}.*`,
+  // || Rahao || style endings
+  ` ?${endingClass} ?${pauseGroup} ?${endingClass}`,
+  // Clean up any lingering ending characters
+  brokenEndingClass,
+  endingClass,
+].map( exp => new RegExp( exp, 'g' ) )
+
+
+/**
+ * Strips line endings from any Gurmukhi or translation string.
+ * Accepts both Unicode and ASCII input.
+ * Useful for generating accurate first letters or modifying non-Gurbani for better display.
+ * *Not* designed for headings or Sirlekhs.
+ * @param {String} text The text to stip endings from.
+ * @return {String} A ending-less version of the text.
+ * @example <caption>Line ending phrases</caption>
+ * stripEndings('ਸੋ ਘਰੁ ਰਾਖੁ; ਵਡਾਈ ਤੋਇ ॥੧॥ ਰਹਾਉ ॥') // => ਸੋ ਘਰੁ ਰਾਖੁ; ਵਡਾਈ ਤੋਇ
+ * stripEndings('ਹੁਕਮੁ ਪਛਾਣਿ; ਤਾ ਖਸਮੈ ਮਿਲਣਾ ॥੧॥ ਰਹਾਉ ਦੂਜਾ ॥') // => ਹੁਕਮੁ ਪਛਾਣਿ; ਤਾ ਖਸਮੈ ਮਿਲਣਾ
+ * stripEndings('ਜਨ ਨਾਨਕ. ਗੁਰਮੁਖਿ ਜਾਤਾ ਰਾਮ ॥੪॥੬॥ ਛਕਾ ੧ ॥') // => ਜਨ ਨਾਨਕ. ਗੁਰਮੁਖਿ ਜਾਤਾ ਰਾਮ
+ * @example <caption>English Translations</caption>
+ * stripEndings('O Nanak, Forever And Ever True. ||1||') // => O Nanak, Forever And Ever True.
+ * stripEndings('lush greenery. ||1||Pause||') // => lush greenery.
+ * stripEndings('always I live within the Khalsa. 519') // => always I live within the Khalsa.
+ * stripEndings('without your reminiscence.(1) (3)') // => without your reminiscence.
+ * @example <caption>Spanish Translations</caption>
+ * stripEndings('ofrece su ser en sacrificio a Ti. (4-2-9)') // => ofrece su ser en sacrificio a Ti.
+ */
+const stripEndings = text => matchers.reduce( ( text, exp ) => text.replace( exp, '' ), text ).trimRight()
+
+module.exports = stripEndings