Merge pull request #1377 from phoneee/fix/empty-string-guards

fix: guard nighit, check_sara, check_marttra against empty input

Author: bact

pythainlp/khavee/core.py

@@ -53,9 +53,15 @@ def check_sara(self, word: str) -> str:
         sara = []
         countoa = 0
 
+        if not word:
+            return ""
+
         # In case of การันย์
         if "์" in word[-1]:
             word = word[:-2]
+            # After removing the karun, the word may become empty (e.g. "ก์")
+            if not word:
+                return ""
 
         # In case of สระเดี่ยว
         for i in word:
@@ -251,6 +257,9 @@ def check_marttra(self, word: str) -> str:
         word = self.handle_karun_sound_silence(word)
         word = remove_tonemark(word)
 
+        if not word:
+            return ""
+
         # Check for ำ at the end (represents "am" sound, ends with m)
         if word[-1] == "ำ":
             return "กม"

pythainlp/morpheme/word_formation.py

@@ -35,14 +35,28 @@ def nighit(w1: str, w2: str) -> str:
         >>> nighit("สํ", "โยค")
         'สังโยค'
     """
+    if not isinstance(w1, str) or not isinstance(w2, str):
+        raise TypeError("Both w1 and w2 must be strings.")
+    w1 = w1.strip()
+    w2 = w2.strip()
+    if not w1:
+        return w2
+    if not w2:
+        return w1
     if not str(w1).endswith("ํ") and len(w1) != 2:
         raise NotImplementedError(f"The function doesn't support {w1}.")
     list_w1 = list(w1)
     list_w2 = list(w2)
     newword = []
     newword.append(list_w1[0])
     newword.append("ั")
-    consonant_start = [i for i in list_w2 if i in set(thai_consonants)][0]
+    _consonants = set(thai_consonants)
+    consonants_in_w2 = [i for i in list_w2 if i in _consonants]
+    if not consonants_in_w2:
+        raise ValueError(
+            f"w2 {w2!r} contains no Thai consonants."
+        )
+    consonant_start = consonants_in_w2[0]
     if consonant_start in ["ก", "ช", "ค", "ข", "ง"]:
         newword.append("ง")
     elif consonant_start in ["จ", "ฉ", "ช", "ฌ"]: