Merge pull request #116 from AnFreTh/main

Release v0.2.2

Author: AnFreTh

README.md

@@ -81,6 +81,35 @@ You can install STREAM directly from PyPI or from the GitHub repository:
     pip install git+https://github.com/AnFreTh/STREAM.git
     ```
 
+3. **Download necessary NLTK resources**:
+
+   To download all necessary NLTK resources required for some models, simply run:
+
+   ```python
+   import nltk
+
+   def ensure_nltk_resources():
+       resources = [
+           "stopwords",
+           "wordnet",
+           "punkt_tab",
+           "brown",
+           "averaged_perceptron_tagger"
+       ]
+       for resource in resources:
+           try:
+               nltk.data.find(resource)
+           except LookupError:
+               try:
+                   print(f"Downloading NLTK resource: {resource}")
+                   nltk.download(resource)
+               except Exception as e:
+                   print(f"Failed to download {resource}: {e}")
+
+   ensure_nltk_resources()
+   ```
+
+
 3. **Install requirements for add-ons**:
     To use STREAMS visualizations, simply run:
     ```bash

assets/movie_poster_topic1.png

@@ -54,6 +54,7 @@ def run(self):
     "plotting": ["dash", "plotly", "matplotlib", "wordcloud"],
     "bertopic": ["hdbscan"],
     "dcte": ["pyarrow", "setfit"],
+    "experimental": ["openai"],
 }
 
 

setup.py

@@ -1,4 +1,4 @@
 """Version information."""
 
 # The following line *must* be the last in the module, exactly as formatted:
-__version__ = "0.2.0"
+__version__ = "0.2.2"

stream_topic/__version__.py

@@ -0,0 +1,10 @@
+from .topic_poster import movie_poster
+from .topic_story import story_topic
+from .topic_summary import topic_summaries
+
+
+__all__ = [
+    "movie_poster",
+    "story_topic",
+    "topic_summaries",
+]

stream_topic/experimental/__init__.py

@@ -0,0 +1,88 @@
+from openai import OpenAI, OpenAIError
+from loguru import logger
+import os
+from IPython.display import Image, display
+
+# Define allowed models
+ALLOWED_MODELS = ["dall-e-3", "dall-e-2"]
+
+
+def movie_poster(
+    topic,
+    api_key,
+    model="dall-e-3",
+    quality="standard",
+    prompt="Create a movie poster that depicts that topic best. Note that the words are ordered in decreasing order of their importance.",
+    size="1024x1024",
+    return_style="url",
+):
+    """
+    Generate a movie-poster-style image based on a given topic using OpenAI's DALL-E model.
+
+    Parameters:
+    - topic: List of words/phrases or list of tuples (word, importance) representing the topic.
+    - api_key: API key for OpenAI.
+    - model: Model to use (e.g., 'dall-e').
+    - poster_style: Description of the style for the image, default is "Movie Poster".
+    - content: Initial system content.
+    - prompt: Prompt for the image generation.
+    - size: Size of the generated image, default is "1024x1024".
+
+    Returns:
+    - image_url: URL of the generated image.
+    """
+
+    # Load the API key from environment if not provided
+    if api_key is None:
+        api_key = os.getenv("OPENAI_API_KEY")
+
+    if api_key is None:
+        raise ValueError("API key is missing. Please provide an API key.")
+
+    assert return_style in [
+        "url",
+        "plot",
+    ], "Invalid return style. Please choose 'url' or 'plot'"
+
+    # Initialize the OpenAI client with your API key
+    client = OpenAI(api_key=api_key)
+
+    # Validate model
+    if model not in ALLOWED_MODELS:
+        raise ValueError(
+            f"Invalid model. Please choose a valid model from {ALLOWED_MODELS}."
+        )
+
+    # Create the prompt for the movie poster
+    if isinstance(topic[0], tuple):
+        # If the topic is a list of tuples with importance
+        topic_description = ", ".join(
+            [f"{word} (importance: {importance})" for word, importance in topic]
+        )
+    else:
+        # If the topic is a list of words in descending importance
+        topic_description = topic
+
+    image_prompt = f"Given the following topic: {topic_description}. {prompt}"
+
+    # Logging the operation
+    logger.info(f"--- Generating image with model: {model} ---")
+    response = client.images.generate(
+        model=model,
+        prompt=image_prompt,
+        size=size,
+        quality=quality,
+        n=1,
+    )
+
+    # Ensure the response is valid
+    if response:
+        image_url = response.data[0].url
+    else:
+        image_url = "No image generated. Please try again."
+
+    if return_style == "url":
+        return image_url
+
+    elif return_style == "plot":
+        display(Image(url=image_url))

stream_topic/experimental/topic_poster.py

@@ -0,0 +1,90 @@
+from openai import OpenAI, OpenAIError
+from loguru import logger
+import os
+
+
+# Define allowed models
+ALLOWED_MODELS = [
+    "gpt-3.5-turbo",
+    "gpt-3.5-turbo-16k",
+    "gpt-4",
+    "gpt-4o",
+    "gpt-4o-mini",
+    "gpt-4-turbo",
+]
+
+
+def story_topic(
+    topic,
+    api_key,
+    model="gpt-3.5-turbo-16k",
+    content="You are a creative writer.",
+    prompt="Create a creative story that includes the following words:",
+    max_tokens=250,
+    temperature=0.8,
+    top_p=1.0,
+    frequency_penalty=0.0,
+    presence_penalty=0.0,
+):
+    """
+    Generate a creative story using OpenAI's GPT model.
+
+    Parameters:
+    - topic: List of words or phrases to include in the story.
+    - api_key: API key for OpenAI.
+    - model: Model to use (e.g., 'gpt-3.5-turbo-16k').
+    - content: Initial system content.
+    - prompt: Prompt for the story generation.
+    - max_tokens: Maximum tokens for the response.
+    - temperature: Creativity level for the model.
+    - top_p: Nucleus sampling parameter.
+    - frequency_penalty: Penalty for word frequency.
+    - presence_penalty: Penalty for word presence.
+
+    Returns:
+    - story: Generated story as a string.
+    """
+
+    # Load the API key from environment if not provided
+    if api_key is None:
+        api_key = os.getenv("OPENAI_API_KEY")
+
+    # Initialize the OpenAI client with your API key
+    client = OpenAI(api_key=api_key)
+
+    # Validate model
+    if model not in ALLOWED_MODELS:
+        raise ValueError(
+            f"Invalid model. Please choose a valid model from {ALLOWED_MODELS}."
+        )
+
+    # Create the prompt
+    prompt = f"{prompt}: {', '.join(topic)}. Make it as short as {max_tokens} words."
+
+    # Logging the operation
+    logger.info(f"--- Generating story with model: {model} ---")
+
+    try:
+        response = client.chat.completions.create(
+            model=model,
+            messages=[
+                {"role": "system", "content": content},
+                {"role": "user", "content": prompt},
+            ],
+            max_tokens=max_tokens,
+            temperature=temperature,
+            top_p=top_p,
+            frequency_penalty=frequency_penalty,
+            presence_penalty=presence_penalty,
+        )
+
+        # Ensure the response is valid
+        if response and len(response.choices) > 0:
+            story = response.choices[0].message.content
+        else:
+            story = "No story generated. Please try again."
+
+        return story
+
+    except OpenAIError as e:
+        return f"An error occurred: {str(e)}"

stream_topic/experimental/topic_story.py

@@ -0,0 +1,93 @@
+from openai import OpenAI, OpenAIError
+from loguru import logger
+import os
+
+ALLOWED_MODELS = [
+    "gpt-3.5-turbo",
+    "gpt-3.5-turbo-16k",
+    "gpt-4",
+    "gpt-4o",
+    "gpt-4o-mini",
+    "gpt-4-turbo",
+]
+
+
+def topic_summaries(
+    topics,
+    api_key,
+    model="gpt-3.5-turbo-16k",
+    content="You are a creative writer.",
+    prompt="Provide a 1-2 sentence summary for the following topic:",
+    max_tokens=60,
+    temperature=0.7,
+    top_p=1.0,
+    frequency_penalty=0.0,
+    presence_penalty=0.0,
+):
+    """
+    Generate a 1-2 sentence summary for each topic using OpenAI's GPT model.
+
+    Parameters:
+    - topics: List of lists, where each sublist contains words/phrases representing a topic.
+    - api_key: API key for OpenAI.
+    - model: Model to use (e.g., 'gpt-3.5-turbo-16k').
+    - content: Initial system content.
+    - prompt: Prompt for the summary generation.
+    - max_tokens: Maximum tokens for each summary.
+    - temperature: Creativity level for the model.
+    - top_p: Nucleus sampling parameter.
+    - frequency_penalty: Penalty for word frequency.
+    - presence_penalty: Penalty for word presence.
+
+    Returns:
+    - summaries: List of summaries corresponding to each topic.
+    """
+
+    # Load the API key from environment if not provided
+    if api_key is None:
+        api_key = os.getenv("OPENAI_API_KEY")
+
+    # Initialize the OpenAI client with your API key
+    client = OpenAI(api_key=api_key)
+
+    # Validate model
+    if model not in ALLOWED_MODELS:
+        raise ValueError(
+            f"Invalid model. Please choose a valid model from {ALLOWED_MODELS}."
+        )
+
+    summaries = []
+
+    for idx, topic in enumerate(topics):
+        # Create the prompt for each topic
+        topic_prompt = f"{prompt} {', '.join(topic)}."
+
+        # Logging the operation
+        logger.info(f"--- Generating summary for topic {idx} with model: {model} ---")
+
+        try:
+            response = client.chat.completions.create(
+                model=model,
+                messages=[
+                    {"role": "system", "content": content},
+                    {"role": "user", "content": topic_prompt},
+                ],
+                max_tokens=max_tokens,
+                temperature=temperature,
+                top_p=top_p,
+                frequency_penalty=frequency_penalty,
+                presence_penalty=presence_penalty,
+            )
+
+            # Ensure the response is valid
+            if response and len(response.choices) > 0:
+                summary = response.choices[0].message.content
+            else:
+                summary = "No summary generated. Please try again."
+
+            summaries.append(summary)
+
+        except OpenAIError as e:
+            summaries.append(f"An error occurred: {str(e)}")
+
+    return summaries