Create korean_process.py

korean_process.py

@@ -0,0 +1,115 @@
+import os
+import dotenv
+import openai
+import random
+import make_dataset
+import re
+from langchain_core.prompts import PromptTemplate
+from langchain_core.output_parsers import StrOutputParser
+from langchain_community.chat_models.friendli import ChatFriendli
+
+dotenv.load_dotenv()
+
+llm = ChatFriendli(
+    friendli_token = os.getenv("FRIENDLI_API_KEY"),
+    model = "meta-llama-3-70b-instruct",
+    streaming = False,
+    # temperature=0.5,
+    # max_tokens=100,
+    # top_p=0.9,
+    # frequency_penalty=0.0,
+    # stop=["\n"],
+)
+
+frame_prompt = """Suggest a single line of Korean lyric that matches with given syllables,lyrics, and title.
+Ensure to avoid repeating previous lyrics. Focus on creative and original expression.
+Match the length of the sentence to the syllables I provide as closely as possible.
+For example, if Syllables: 7 given, you should write a 6~8 letter korean sentence.
+Your answer should feel like soft, trendy K-pop lyrics without any profanity.
+Your answer should be short, and only composed with a single sentence.
+Answer with a single line of lyrics you created, and nothing else.
+
+Here,
+Title: {title}
+Mood and Symbol: {mood}
+Previous Lyrics: {lyric}
+Syllables: {syllables}
+
+Your korean lyric that should be added to the previous lyrics:
+"""
+
+client = openai.OpenAI(base_url="https://inference.friendli.ai/dedicated/v1", api_key=os.getenv("FRIENDLI_API_KEY"))
+
+def update_language(current_language):
+    if random.random() < 0.8:
+        return current_language
+    else:
+        return random.choice(make_dataset.languages)
+
+def extract_message(text, ending):
+    pattern = re.escape(ending) + r'([^"\n]+)'
+    match = re.search(pattern, text)
+    if match:
+        message = match.group(1).strip()
+        return message.strip('"')
+    return "" 
+
+def find_mood(lyric):
+    prompt = PromptTemplate.from_template("find mood and symbolic keyword of the given song. one sentence each. song: {lyric}. Mood and symbol: ")
+    formatted_prompt = prompt.format(lyric=lyric)
+
+    feedback_chain = (
+        llm
+        | StrOutputParser()
+    )
+
+    return feedback_chain.invoke(formatted_prompt)
+
+def generate_korean_lyrics(title, lyric):
+    """
+    prompt = make_dataset.frame_prompt.format(title=title, lyric=lyric, syllables=make_dataset.count_syllables(lyric))
+    response = client.chat.completions.create(
+                    model=os.getenv("KR_ENDPOINT"),
+                    messages=[
+                        {"role": "system", "content": f"You are an assistant of lyricist"},
+                      
+                        {"role": "user", "content": prompt},
+                    ],
+                )
+    return response.choices[0].message.content
+    """
+    lines = lyric.split('\n')
+    result = ""
+    mood = find_mood(lyric)
+    print(mood)
+    for line in lines:
+        if len(line) > 0:
+            candidate = ""
+            candidate_count = 7
+            syllables = make_dataset.count_syllables(line)
+            for i in range(1):
+                #print("Original:", line)
+                response = client.chat.completions.create(
+                    model=os.getenv("KR_ENDPOINT"),
+                    messages=[
+                        {"role": "system", "content": f"You are an assistant of lyricist"},
+                        {"role": "user", "content": frame_prompt.format(title=title, lyric=result, syllables=syllables, mood=mood)},
+                    ],
+                )
+                suggestion = response.choices[0].message.content
+                count = make_dataset.count_syllables(suggestion)
+                if count == syllables:
+                    #print(f"[Success!] Suggestion {i}:", suggestion)
+                    candidate = suggestion
+                    break
+                elif count < syllables:
+                    if syllables - count < candidate_count:
+                        candidate = suggestion
+                        candidate_count = syllables - count
+                else:
+                    if count - syllables < candidate_count:
+                        candidate = suggestion
+                        candidate_count = count - syllables
+                #print(f"[Fail!] Suggestion {i}:", suggestion)
+            result += candidate + '\n'
+    return result