Update auto detect label on language detection

.gitignore

@@ -1 +1 @@
-*pycache*
+*pycache*

app.py

@@ -1,15 +1,27 @@
+# from responses import start
 import gradio as gr
 from language_directions import *
 from transformers import pipeline
+from examples import example_sentences
 
 source_lang_dict = get_all_source_languages()
 target_lang_dict = {}
 source_languages = source_lang_dict.keys()
 
-def source_dropdown_changed(source_dropdown, input_text=""):
+def get_auto_detect_source_dropdown(input_text):
+    source, _ = auto_detect_language_code(input_text)
+    language_name = get_name_from_iso_code(source)
+    source_dropdown_text = "Detected - " + language_name
+    update_source_languages_dict(source_lang_dict, source_dropdown_text)
+    source_language_dropdown = gr.Dropdown(choices=source_languages,
+                                                  value=source_dropdown_text,
+                                                  label="Source Language")
+    return source_language_dropdown, language_name
+  
+def get_target_dropdown(source_language_name, input_text):
     global target_lang_dict
-    target_lang_dict = get_target_languages(source_lang_dict[source_dropdown], input_text)
-    target_languages = target_lang_dict.keys()
+    target_lang_dict, source_language = get_target_languages(source_lang_dict[source_language_name], input_text)
+    target_languages = list(target_lang_dict.keys())
     default_target_value = None
     if "English" in target_languages or "english" in target_languages:
         default_target_value = "English"
@@ -19,16 +31,41 @@ def source_dropdown_changed(source_dropdown, input_text=""):
                                   value=default_target_value,
                                   label="Target Language")
     return target_dropdown
+  
+def get_dropdown_value(dropdown):
+    if isinstance(dropdown, gr.Dropdown):
+        dropdown_value = dropdown.constructor_args.get('value')
+    elif isinstance(dropdown, str):
+      dropdown_value = dropdown
+    return dropdown_value
+  
+def get_dropdowns(source_dropdown, input_text):
+    source_language_name = get_dropdown_value(source_dropdown)
+    if input_text and source_language_name == "Auto Detect" or source_language_name.startswith("Detected"):
+      source_dropdown, source_language_name = get_auto_detect_source_dropdown(input_text)
+    target_dropdown = get_target_dropdown(source_language_name=source_language_name,
+                                          input_text=input_text)
+    return source_dropdown, target_dropdown
+
+def input_changed(source_language_dropdown, input_text=""):
+    return get_dropdowns(source_dropdown=source_language_dropdown,
+                         input_text=input_text)
 
 def translate(input_text, source, target):
-    if source == "Auto Detect":
+    source_readable = source
+    if source == "Auto Detect" or source.startswith("Detected"):
       source, _ = auto_detect_language_code(input_text)
-    target_lang_dict = get_target_languages(source)
-    target = target_lang_dict[target]
-    model = f"Helsinki-NLP/opus-mt-{source}-{target}"
-    pipe = pipeline("translation", model=model)
-    translation = pipe(input_text)
-    return translation[0]['translation_text']
+    if source in source_lang_dict.keys():
+      source = source_lang_dict[source]
+    target_lang_dict, _ = get_target_languages(source)
+    try:
+      target = target_lang_dict[target]
+      model = f"Helsinki-NLP/opus-mt-{source}-{target}"
+      pipe = pipeline("translation", model=model)
+      translation = pipe(input_text)
+      return translation[0]['translation_text'], ""
+    except KeyError:
+      return "", f"Error: Translation direction {source_readable} to {target} is not supported by Helsinki Translation Models"
 
 
 with gr.Blocks() as demo:
@@ -55,14 +92,15 @@ with gr.Blocks() as demo:
                                                    value="English",
                                                    label="Target Language")
             translated_textbox = gr.Textbox(lines=5, placeholder="", label="Translated Text")
+    info_label = gr.HTML("")
     btn = gr.Button("Translate")
-    source_language_dropdown.change(source_dropdown_changed, inputs=source_language_dropdown, outputs=target_language_dropdown)
+    source_language_dropdown.change(input_changed, inputs=[source_language_dropdown, input_textbox], outputs=[source_language_dropdown, target_language_dropdown])
+    input_textbox.change(input_changed, inputs=[source_language_dropdown, input_textbox], outputs=[source_language_dropdown, target_language_dropdown])
     btn.click(translate, inputs=[input_textbox,
                                  source_language_dropdown,
                                  target_language_dropdown],
-                                   outputs=translated_textbox)
-    gr.Examples(["Je te rencontre au café", "Répétez s'il vous plaît."],
-                inputs=[input_textbox])
+                                   outputs=[translated_textbox, info_label])
+    gr.Examples(example_sentences, inputs=[input_textbox])
 
 if __name__ == "__main__":
     demo.launch()

examples.py

@@ -0,0 +1,14 @@
+example_sentences = [
+    "Je te rencontre au café", "Répétez s'il vous plaît.",
+    "The mountains stand tall, embracing the clouds with their majestic peaks.",
+    "सितारों का आकाश में खोया होने का एहसास मन को अद्वितीय सुख देता है।",
+    "ਜਟ ਦਾ ਮੁਕਾਬਲਾ ਦਸ ਮੈਨੂੰ ਕਿਥੇ ਆ ਨੀ।",
+    "Il profumo dei fiori primaverili riempie l'aria, portando gioia e speranza.",
+    "Güneş batarken, gökyüzünü altın rengine boyuyor ve doğayı sihirli bir atmosfere bürüyor.",
+    "De wind fluistert door de bomen, een symfonie van rust en harmonie.",
+    "눈이 하얗게 내리고, 숲은 고요로움으로 가득 차 있습니다.",
+    "הכוכבים מאירים בשמי הלילה, משאירים את הלב פתוח לקסמם.",
+    "Hương hoa lan tỏa trong không khí, mang lại cảm giác êm đềm và sự bình yên.",
+    "Regnet faller mjukt mot marken, skapar en känsla av förnyelse och friskhet.",
+    "Η θάλασσα χτυπά την ακτή με απαλές κύματα, φέρνοντας ηρεμία και γαλήνη στην ψυχή.",
+]

iso639_wrapper.py

@@ -1,4 +1,5 @@
 from iso639 import Lang, iter_langs
+from regex import R
 
 
 langs = [lang for lang in iter_langs()]
@@ -24,6 +25,27 @@ iso5_name_to_code = {lg.name: lg.pt5 for lg in langs}
 # https://github.com/Helsinki-NLP/Tatoeba-Challenge/blob/master/README.md#in-more-detail
 helsinki_precendence = ["iso3", "iso5", "iso1", "iso2t", "iso2b"]
 
+rename_dict = {"Panjabi":  "Punjabi"}
+
+def rename_languages(language):
+    if language in rename_dict:
+        return rename_dict[language]
+    return language
+    
+def rename_return_value(func):
+    def wrapper(*args, **kwargs):
+        result = func(*args, **kwargs)
+        if isinstance(result, str):
+            return rename_languages(result)
+        elif isinstance(result, list):
+            return [rename_languages(item) for item in result]
+        elif isinstance(result, dict):
+            return {key: rename_languages(value) for key, value in result.items()}
+        else:
+            return result
+    return wrapper
+
+@rename_return_value
 def get_name_from_iso_code(iso_code, precedence=helsinki_precendence):
     for code_type in precedence:
         if code_type == "iso1" and iso_code in iso1_code_to_name.keys():

language_directions.py

@@ -2,12 +2,12 @@ from helsinki_models import helsinki_models, get_clearly_formatted_langauge_dire
 from iso639_wrapper import get_name_from_iso_code
 from language_detection import detect_language
 from collections import OrderedDict
-from utils import convert_keys_to_lowercase
+from utils import convert_keys_to_lowercase, match_in_keys, match_in_values
 
 
 def get_all_source_languages():
     """
-    Returns a human-readable `dict of source languages names to codes` 
+    Returns a human-readable `dict source_languages_names:codes` 
     based on the available models.
     """
     source_languages = {}
@@ -23,6 +23,9 @@ def get_all_source_languages():
         { **{'Auto Detect' : 'Auto Detect'}, **source_languages}
     return all_source_langs_including_auto_detect
 
+def update_source_languages_dict(source_languages_dict, auto_detected_language):
+    source_languages_dict[auto_detected_language] = "Auto Detect"
+
 def get_target_languages(source_language_code, input_text=None):
     """
     Returns a human-readable `dict of target languages names to codes` 
@@ -40,26 +43,24 @@ def get_target_languages(source_language_code, input_text=None):
             target_language_name = get_name_from_iso_code(target_language)
             if target_language_name:
                 target_languages[target_language_name] = target_language
-    return OrderedDict(sorted(target_languages.items()))
+    return OrderedDict(sorted(target_languages.items())), source_language_code
 
 def auto_detect_language_code(input_text):
+    DEFAULT_SOURCE_LANGUAGE = "en"
+    detected_language_string = DEFAULT_SOURCE_LANGUAGE
     if not input_text:
-        return None, True
-    language = detect_language(input_text)
-    if language == "unknown":
-        print("unknown tesxt for ", input_text)
-        return "unknown", True
-    elif language in list(get_all_source_languages().keys())\
-        or language.lower() in [k.lower() for k in list(get_all_source_languages().keys())]:
-        source_languages_dict = convert_keys_to_lowercase(get_all_source_languages())
-        source_language_code = source_languages_dict.get(language.lower())
-        return source_language_code, False
-    elif language in list(get_all_source_languages().values())\
-        or language.lower() in [k.lower() for k in list(get_all_source_languages().values())]:
-        source_language_code = language
-        return source_language_code, False
+        return DEFAULT_SOURCE_LANGUAGE, True
+    language_or_code = detect_language(input_text)
+    if language_or_code == "unknown":
+        return DEFAULT_SOURCE_LANGUAGE, True
     else:
-        raise ValueError(f"Language {source_language_code} not supported")
+        detected_language_string = match_in_keys(get_all_source_languages(), language_or_code)
+        if not detected_language_string:
+            detected_language_string = match_in_values(get_all_source_languages(), language_or_code)
+        if detected_language_string:
+            return detected_language_string, False
+        else:
+            return DEFAULT_SOURCE_LANGUAGE, True
     
 
 # Example usage:

project-notes.md

@@ -1,4 +1,6 @@
 # Scope of project
 1. Enable multiple languages translate based on helsinki models.✅
 2. Enable auto detect langauge ✅
-3. Show error message instead of gradio error
+3. Show error message instead of gradio error ✅
+4. Add examples ✅
+5. Auto detect on text change ✅

utils.py

@@ -1,2 +1,18 @@
+from functools import cache
+
+
 def convert_keys_to_lowercase(input_dict):
-    return {key.lower(): value for key, value in input_dict.items()}
+    return {key.lower(): value for key, value in input_dict.items()}
+
+def match_in_keys(dictionary, search_string):
+    lowercase_dict = convert_keys_to_lowercase(dictionary)
+    if search_string.lower() in list(lowercase_dict.keys()):
+        return lowercase_dict.get(search_string.lower())
+    for l_key in lowercase_dict.keys():
+        if l_key.startswith(search_string.lower()):
+            return lowercase_dict.get(l_key)
+
+def match_in_values(dictionary, search_string):
+    lowercase_dict = convert_keys_to_lowercase(dictionary)
+    if search_string.lower() in list(lowercase_dict.values()):
+        return search_string