fix(toEnglish): properly detect word boundaries (#177)

#Patch release Co-authored-by: Harjot Singh <harjot@harkul.com>
shabados · Nov 3, 2020 · 76925a6 · 76925a6
1 parent 4adbf3f
commit 76925a6
Show file tree

Hide file tree

Showing 2 changed files with 7 additions and 1 deletion.
diff --git a/lib/toEnglish.js b/lib/toEnglish.js
@@ -106,7 +106,7 @@ const replacements = [
   [ /mÚ/g, 'mhlw' ], // Mehla replcement
   [ /i(.)/g, '$1i' ], // Place sihari in correct position
   [ /(.)[i]([R®H§´ÍÏçœ˜†])/ug, '$1$2i' ], // Move sihari in front of pairin akhars
-  [ new RegExp( `(\\S[^ha])([iu])([\\s$${vishraams.join( '' )}])`, 'ug' ), '$1$3' ], // Remove trailing Aunkar (u) and Sihari (i) except when on Haha (h), Ooraa (a), or on a standalone akhar
+  [ new RegExp( `(\\S[^ha])[iu]([${vishraams.join( '' )}]|\\b)`, 'ug' ), '$1$2' ], // Remove trailing Aunkar (u) and Sihari (i) except when on Haha (h), Ooraa (a), or on a standalone akhar
   [ new RegExp( `(\\b\\S)h([^iIuUyYwWoONM§¨®´µÍÏçüœˆ˜†]\\b|[${vishraams.join( '' )}])`, 'ug' ), '$1yh$2' ], // Add y to three consonant letter words with haha in middle per issue #123
 ]
 

diff --git a/test/toEnglish.spec.js b/test/toEnglish.spec.js
@@ -24,6 +24,12 @@ const transliterations = [
   [ 'ਸਹਜ; ਸਸਹਜ ਅਨਹਦ ਰਹਤ ਕਹਤ ਪਹਰ, ਸਹਸ ਮਹਲ ਟਹਲ ਕਹਨਨ ਕਹਨ', 'sehaj; sasahaj anahad rehat kehat pehar, sehas mehal ttehal kahanan kehan' ],
   [ 'ਸਭ ਭਇਓ ਪਰਾਇਓ', 'sabh bheo paraaeo' ],
   [ 'ਆਸਾ ਮਹਲਾ ੫ ਪੰਚਪਦੇ₃ ॥', 'aasaa mahalaa 5 panchapade₃ |' ],
+  [ 'ਹਰਿ', 'har' ],
+  [ 'ਸਚੁ', 'sach' ],
+  [ 'ਰਾਹੁ', 'raahu' ],
+  [ 'ਭਾਉ', 'bhaau' ],
+  [ 'ਸਤਿਗੁਰੁ ਸਤਿਗੁਰੁ ਸਚੁ; ਸਚੁ ਹਰਿ ਹਰਿ ਹਿੰਙੁ', 'satigur satigur sach; sach har har hing' ],
+  [ 'ਸੁ ਉ ਜੁ', 'su u ju' ],
 ]
 
 describe( 'toEnglish()', () => {