change file

app.py

@@ -1,8 +1,6 @@
-import gradio as gr
 from openai import OpenAI
 import gradio as gr
 import requests
-import os
 from PIL import Image
 import numpy as np
 import ipadic
@@ -11,17 +9,18 @@ import difflib
 import io
 import os
 
-client = OpenAI(api_key=os.getenv('OPENAI_API_KEY'))
+client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
+
 
 def generate_image(text):
-    image_path = f"/content/images/{text}.png"
+    image_path = f"./{text}.png"
     if not os.path.exists(image_path):
         response = client.images.generate(
-          model="dall-e-3",
-          prompt=text,
-          size="1024x1024",
-          quality="standard",
-          n=1,
+            model="dall-e-3",
+            prompt=text,
+            size="1024x1024",
+            quality="standard",
+            n=1,
         )
         image_url = response.data[0].url
         image_data = requests.get(image_url).content
@@ -30,27 +29,28 @@ def generate_image(text):
         img.save(image_path)
     return image_path
 
+
 def calulate_similarity_score(ori_text, text):
     if ori_text != text:
         model_name = "text-embedding-3-small"
-        response = client.embeddings.create(input = [ori_text, text], model=model_name)
+        response = client.embeddings.create(input=[ori_text, text], model=model_name)
         score = cos_sim(response.data[0].embedding, response.data[1].embedding)
         score = int(round(score, 2) * 100)
         if score == 100:
             score = 99
     else:
-      score = 100
+        score = 100
     return score
 
+
 def cos_sim(v1, v2):
     return np.dot(v1, v2) / (np.linalg.norm(v1) * np.linalg.norm(v2))
 
+
 def tokenize_text(text):
     mecab = MeCab.Tagger(f"-Ochasen {ipadic.MECAB_ARGS}")
-    return [
-        t.split()[0]
-        for t in mecab.parse(text).splitlines()[:-1]
-    ]
+    return [t.split()[0] for t in mecab.parse(text).splitlines()[:-1]]
+
 
 def create_match_words(ori_text, text):
     ori_words = tokenize_text(ori_text)
@@ -58,74 +58,68 @@ def create_match_words(ori_text, text):
     match_words = [w for w in words if w in ori_words]
     return match_words
 
+
 def create_hint_text(ori_text, text):
     response = list(difflib.ndiff(list(text), list(ori_text)))
     output = ""
     for r in response:
-      if r[:2] == "- ":
-          continue
-      elif r[:2] == "+ ":
-          output += "^"
-      else:
-          output += r.strip()
+        if r[:2] == "- ":
+            continue
+        elif r[:2] == "+ ":
+            output += "^"
+        else:
+            output += r.strip()
     return output
 
-def update_question(selected_option):
-    if selected_option == "Q1":
-        return "/content/images/白い猫が木の上で休んでいる.png"
-    elif selected_option == "Q2":
-        return "/content/images/サメが海の中で暴れている.png"
-    elif selected_option == "Q3":
-        return "/content/images/東京スカイツリーの近くで花火大会が行われている.png"
-    elif selected_option == "Q4":
-        return "/content/images/イカとタイがいた都会.png"
-    elif selected_option == "Q5":
-        return "/content/images/赤いきつねと緑のたぬき.png"
-    if selected_option == "Q6":
-        return "/content/images/宇宙に向かってたい焼きが空を飛んでいる.png"
-    elif selected_option == "Q7":
-        return "/content/images/イケメンが海岸でクリームパンを眺めている.png"
-    elif selected_option == "Q8":
-        return "/content/images/生麦生米生卵生麦生米生卵生麦生米生卵.png"
-    elif selected_option == "Q9":
-        return "/content/images/サイバーエージェントで働く人たち.png"
-    elif selected_option == "Q10":
-        return "/content/images/柿くへば鐘が鳴るなり法隆寺.png"
-    elif selected_option == "Q11":
-        return "/content/images/鳴くよウグイス平安京.png"
-    else:
-        return "/content/images/abc.png"
+
+def update_question(option):
+    answer = os.getenv(option)
+    return f"./{answer}.png"
+
 
 def main(text, option):
-    ori_text = update_question(option).split("/")[-1].split(".png")[0]
+    ori_text = os.getenv(option)
     image_path = generate_image(text)
     score = calulate_similarity_score(ori_text, text)
 
     if score < 80:
         match_words = create_match_words(ori_text, text)
-        hint_text = "一致している単語リスト: "+" ".join(match_words)
+        hint_text = "一致している単語リスト: " + " ".join(match_words)
     elif 80 <= score < 100:
-        hint_text = "一致していない箇所: "+create_hint_text(ori_text, text)
+        hint_text = "一致していない箇所: " + create_hint_text(ori_text, text)
     else:
         hint_text = ""
     return image_path, f"{score}点", hint_text
 
+
 with gr.Blocks() as demo:
     with gr.Row():
         with gr.Column():
             gr.Markdown(
                 "# プロンプトを当てるゲーム \n これは表示されている画像のプロンプトを当てるゲームです。プロンプトを入���するとそれに対応した画像とスコアとヒントが表示されます。スコア100点を目指して頑張ってください！ \n\nヒントは80点未満の場合は当たっている単語、80点以上の場合は足りない文字を「^」で示した文字列を表示しています。",
             )
-            selected_option = gr.components.Radio(["Q1", "Q2", "Q3", "Q4", "Q5", "Q6", "Q7", "Q8", "Q9", "Q10", "Q11"], label="問題を選んでください！")
+            option = gr.components.Radio(
+                ["Q1", "Q2", "Q3"], label="問題を選んでください！"
+            )
             output_title_image = gr.components.Image(type="filepath", label="お題")
-            selected_option.change(update_question, inputs=[selected_option], outputs=[output_title_image])
+            option.change(
+                update_question, inputs=[option], outputs=[output_title_image]
+            )
 
-            input_text = gr.components.Textbox(lines=1, label="画像にマッチするテキストを入力して！")
+            input_text = gr.components.Textbox(
+                lines=1, label="画像にマッチするテキストを入力して！"
+            )
             submit_button = gr.Button("Submit")
         with gr.Column():
             output_image = gr.components.Image(type="filepath", label="生成画像")
             output_score = gr.components.Textbox(lines=1, label="スコア")
             output_hint_text = gr.components.Textbox(lines=1, label="ヒント")
+    with gr.Row():
+        gr.Dropdown()
 
-    submit_button.click(main, inputs=[input_text, selected_option], outputs=[output_image, output_score, output_hint_text])
-demo.launch()
+    submit_button.click(
+        main,
+        inputs=[input_text, option],
+        outputs=[output_image, output_score, output_hint_text],
+    )
+demo.launch()