Implement Story vs Narrative Modification POC

[leonvanbokhorst]

Summary by Sourcery

New Features:

• Introduce a proof of concept module for modifying stories using narrative modifiers such as tone, perspective, and purpose.

[sourcery-ai]

Reviewer's Guide by Sourcery

This PR implements a proof of concept for story modification using narrative elements. The implementation uses language models and embedding techniques to alter story presentations while maintaining core elements. The system processes story elements (events, characters, settings) and applies narrative modifiers (tone, perspective, purpose) through embedding combinations and text generation.

Class diagram for Story vs Narrative Modification POC

classDiagram
    class OllamaInterface {
        +generate_embedding(text: str) async
        +generate(prompt: str) async
        +cleanup() async
    }
    class EmbeddingCache {
        +get(key: str) -> List[float]
        +set(key: str, value: List[float])
        +clear()
    }
    class StoryNarrativeModifier {
        +embed_elements(elements_dict: Dict[str, Any]) async -> Dict[str, Any]
        +apply_narrative(story_embedding: List[float], narrative_embedding: List[float], weight: float) -> np.ndarray
        +generate_modified_text(original_text: str, modified_embedding: List[float], narrative_modifier: str) async -> str
        +main() async
    }
    StoryNarrativeModifier --> OllamaInterface
    StoryNarrativeModifier --> EmbeddingCache
    note for StoryNarrativeModifier "This class orchestrates the story modification process using narrative elements."

Loading

File-Level Changes

Change	Details	Files
Implementation of core story modification functionality	Created async function to generate and cache embeddings for story and narrative elements Implemented narrative application logic using weighted combinations of embeddings Added text generation function that applies narrative modifiers to original text Developed main orchestration function that coordinates the entire modification process	src/poc_story_vs_narrative.py
Data structure setup for story and narrative elements	Defined sample story elements including events, characters, and settings Created narrative modifier categories for tone, perspective, and purpose Structured the data as nested dictionaries for easy access and modification	src/poc_story_vs_narrative.py
Integration with external dependencies and utilities	Set up logging configuration and error handling Integrated with language model interface for text generation Implemented embedding cache for optimization Added cosine similarity calculations for comparing embeddings	src/poc_story_vs_narrative.py

---

Tips and commands

Interacting with Sourcery

• Trigger a new review: Comment @sourcery-ai review on the pull request.
• Continue discussions: Reply directly to Sourcery's review comments.
• Generate a GitHub issue from a review comment: Ask Sourcery to create an
  issue from a review comment by replying to it.
• Generate a pull request title: Write @sourcery-ai anywhere in the pull
  request title to generate a title at any time.
• Generate a pull request summary: Write @sourcery-ai summary anywhere in
  the pull request body to generate a PR summary at any time. You can also use
  this command to specify where the summary should be inserted.

Customizing Your Experience

Access your dashboard to:

• Enable or disable review features such as the Sourcery-generated pull request
  summary, the reviewer's guide, and others.
• Change the review language.
• Add, remove or edit custom review instructions.
• Adjust other review settings.

Getting Help

• Contact our support team for questions or feedback.
• Visit our documentation for detailed guides and information.
• Keep in touch with the Sourcery team by following us on X/Twitter, LinkedIn or GitHub.

[sourcery-ai]

Hey @leonvanbokhorst - I've reviewed your changes - here's some feedback:

Overall Comments:

• Consider moving the sample story elements and narrative modifiers to a configuration file for better maintainability and easier testing with different datasets.

Here's what I looked at during the review

• 🟡 General issues: 2 issues found
• 🟢 Security: all looks good
• 🟢 Testing: all looks good
• 🟡 Complexity: 1 issue found
• 🟢 Documentation: all looks good

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}

[sourcery-ai]

suggestion (performance): Consider using asyncio.gather() to parallelize embedding generation for better performance

The current implementation processes embeddings sequentially. Using asyncio.gather() would allow concurrent processing of multiple embeddings, significantly improving performance for larger datasets.

async def embed_elements(elements_dict: Dict[str, Any]) -> Dict[str, Any]:
    async with asyncio.TaskGroup() as tg:
        tasks = [tg.create_task(embed_element(k, v)) for k, v in elements_dict.items()]
    return dict(await asyncio.gather(*tasks))

[sourcery-ai]

suggestion: Add validation for the weight parameter to ensure it's between 0 and 1

Invalid weight values could lead to unexpected results. Consider adding a check like if not 0 <= weight <= 1: raise ValueError('Weight must be between 0 and 1')

Suggested change

	def apply_narrative(story_embedding: List[float], narrative_embedding: List[float], weight: float = 0.5) -> np.ndarray:
	def apply_narrative(story_embedding: List[float], narrative_embedding: List[float], weight: float = 0.5) -> np.ndarray:
	if not 0 <= weight <= 1:
	raise ValueError("Weight must be between 0 and 1")

[sourcery-ai]

issue (complexity): Consider extracting the nested event processing logic into a dedicated function to improve code organization.

The main() function's nested loop structure makes the code flow difficult to follow. Consider extracting the event modification logic into a separate function:

async def process_event(
    event_name: str,
    event_embedding: List[float],
    embedded_narratives: Dict
) -> Dict:
    """Process a single event with all narrative modifiers."""
    results = {}
    for category, modifiers in embedded_narratives.items():
        for modifier_name, modifier_embedding in modifiers.items():
            modified_embedding = apply_narrative(
                event_embedding, modifier_embedding, weight=0.6
            )
            similarity = cosine_similarity([event_embedding], [modified_embedding])[0][0]

            modified_text = await generate_modified_text(
                event_name, modified_embedding, f"{category}: {modifier_name}"
            )

            results[(category, modifier_name)] = {
                "modified_embedding": modified_embedding,
                "similarity": similarity,
                "modified_text": modified_text,
            }
    return results

async def main():
    # ... embedding generation code ...

    modified_story = {}
    for event_name, event_embedding in zip(
        story_elements["events"], embedded_story["events"]
    ):
        event_results = await process_event(
            event_name, event_embedding, embedded_narratives
        )
        modified_story.update({
            (event_name, cat, mod): data 
            for (cat, mod), data in event_results.items()
        })

This refactoring:

1. Reduces nesting depth from 3 to 2 levels
2. Makes the event processing logic easier to test and modify
3. Improves readability by separating concerns