Merge pull request #481 from PyThaiNLP/hotfix-bug-normalize

bact · web-flow · commit bdafa578ed2c · 2020-09-17T09:01:19.000+01:00
Fix remove_repeat_vowels() bug that remove spaces between vowel
diff --git a/pythainlp/util/normalize.py b/pythainlp/util/normalize.py
@@ -38,7 +38,7 @@
     f"{follow_v}{lead_v}{above_v}{below_v}\u0e3a\u0e4c\u0e4d\u0e4e"
 )
 _NOREPEAT_PAIRS = list(
-    zip([f"({ch}[ ]*)+" for ch in _NOREPEAT_CHARS], _NOREPEAT_CHARS)
+    zip([f"({ch}[ ]*)+{ch}" for ch in _NOREPEAT_CHARS], _NOREPEAT_CHARS)
 )
 
 _RE_TONEMARKS = re.compile(f"[{tonemarks}]+")

Original file line number	Diff line number	Diff line change
`@@ -38,7 +38,7 @@`
`38`	`38`	`f"{follow_v}{lead_v}{above_v}{below_v}\u0e3a\u0e4c\u0e4d\u0e4e"`
`39`	`39`	`)`
`40`	`40`	`_NOREPEAT_PAIRS = list(`
`41`		`- zip([f"({ch}[ ]*)+" for ch in _NOREPEAT_CHARS], _NOREPEAT_CHARS)`
	`41`	`+ zip([f"({ch}[ ]*)+{ch}" for ch in _NOREPEAT_CHARS], _NOREPEAT_CHARS)`
`42`	`42`	`)`
`43`	`43`
`44`	`44`	`_RE_TONEMARKS = re.compile(f"[{tonemarks}]+")`