Update app.py

app.py

@@ -1,32 +1,85 @@
 import streamlit as st
 import epitran
-
+import langcodes
+from langcodes import LanguageTagError
 # TODO: reverse transliterate? 
 
+
+def get_lang_description_from_mapping_name(string_to_check):  
+  if "generic-Latn" == string_to_check: 
+    return "Generic Latin Script"
+    
+  if len(string_to_check)<2:
+    return None
+  
+  try:
+    description = None
+    lang = langcodes.get(string_to_check)
+    if lang: 
+      items = []
+      for key, value in lang.describe().items():
+        items.append(f"{key}: {value}")
+        
+    
+      description = ", ".join(items))
+    return description
+
+  except LanguageTagError as e: 
+    if any(["out of place" in str(e), "must be followed by something" in str(e)]):  
+      # print("*****")
+      # print(e)
+      # LanguageTagError: This extlang subtag, 'red', is out of place. Expected territory, variant, extension, or end of string.
+      # LanguageTagError: This script subtag, 'east', is out of place. Expected territory, variant, extension, or end of string.
+      # LanguageTagError: The subtag 'p' must be followed by something
+      substrings = string_to_check.split("-")
+      substrings = substrings[:-1] # remove the last one
+      string_to_check = "-".join(substrings)
+      return get_lang_from_mapping_name(string_to_check)
+    else:
+      print("*****")
+      print(e)
+      return None
+
+
+def get_valid_epitran_mappings_list():
+  map_path = Path(epitran.__path__[0]) / "data"/"map"
+  map_files = list(map_path.glob("*.*"))
+  valid_mappings = [map_file.stem for map_file in map_files]
+  return valid_mappings
+
+
 if __name__ == "__main__":
-  iso_lang_code = st.text_input(
-    label="Three-letter ISO-639-3 (https://iso639-3.sil.org/) language code", 
-    value="swa"
-    )
-  st.write(f"iso code is {iso_lang_code}")
-  
-  iso_script_code = st.text_input(
-    label="ISO 15924 (https://unicode.org/iso15924/iso15924-codes.html) script code, e.g. 'Latn' for Latin script, 'Hans' for Chinese script, etc.", 
-    value="Latn"
-    )
-  st.write(f'iso code is {iso_script_code }')
+
+  valid_epitran_mappings = get_valid_epitran_mappings_list()
+  selected_mapping = st.selectbox("Which language/script pair would you like to use?", valid_epitran_mappings)
+  description = get_lang_description_from_mapping_name(selected_mapping)
+  st.write(f"You selected {selected_mapping}")
+  
+  
+#  iso_lang_code = st.text_input(
+#    label="Three-letter ISO-639-3 (https://iso639-3.sil.org/) language code", 
+#    value="swa"
+#    )
+#  st.write(f"iso code is {iso_lang_code}")
+#  
+#  iso_script_code = st.text_input(
+#    label="ISO 15924 (https://unicode.org/iso15924/iso15924-codes.html) script code, e.g. 'Latn' for Latin script, 'Hans' for Chinese script, etc.", 
+#    value="Latn"
+#    )
+#  st.write(f'iso code is {iso_script_code}')
   
   input_text = st.text_area(label="Whatever you type here will be transliterated!", value="Gari langu linaloangama limejaa na mikunga")
   
-  combined_code = "-".join([iso_lang_code, iso_script_code])
-  st.write(f"Combined code: {combined_code}")
+#  combined_code = "-".join([iso_lang_code, iso_script_code])
+#  st.write(f"Combined code: {combined_code}")
   st.info("attempting to instantiate epitran transliterator for your language/script")
-  epi = epitran.Epitran(combined_code)
+  epi = epitran.Epitran(selected_mapping )
 
   st.info(f"transliterating `{input_text}`\n\tusing {epi}...")  
   transliteration = epi.transliterate(input_text)
   
   st.success(transliteration)
+