agent_system/script_schema.py

"""
Defines structured JSON schema for video scripts.
"""
from typing import List, Dict, Any, Optional
from pydantic import BaseModel, Field

class VideoScriptSection(BaseModel):
    """
    Represents a section of a video script.
    """
    section_type: str = Field(..., description="Type of section: hook, key_points, explanation, conclusion, call_to_action")
    content: str = Field(..., description="Content text of this section")
    display_title: Optional[str] = Field(None, description="Display title for this section (for UI)")

class VideoScript(BaseModel):
    """
    Structured schema for video scripts.
    """
    title: str = Field(..., description="Title of the video")
    target_audience: Optional[str] = Field(None, description="Target audience for this video")
    tone: Optional[str] = Field(None, description="Tone of the video: professional, casual, etc.")
    sections: List[VideoScriptSection] = Field(..., description="Ordered sections of the script")
    call_to_action: Optional[str] = Field(None, description="Call to action for the video")
    key_summary_points: Optional[List[str]] = Field(default_factory=list, description="Key summary points")
    sources: Optional[List[str]] = Field(default_factory=list, description="Sources used for the script")
    fact_check_confirmation: Optional[str] = Field(None, description="Confirmation that facts have been checked")
    
    def get_full_script_text(self) -> str:
        """
        Generate the full script text suitable for narration.
        
        Returns:
            Combined script text from all sections
        """
        return "\n\n".join(section.content for section in self.sections)
    
    def get_formatted_display(self) -> str:
        """
        Generate formatted script for display in chat UI.
        
        Returns:
            Markdown formatted script for display
        """
        result = [f"## 🎬 {self.title}"]
        
        if self.target_audience:
            result.append(f"**Target Audience:** {self.target_audience}")
            
        if self.tone:
            result.append(f"**Tone:** {self.tone}")
        
        result.append("")  # Empty line
        
        # Add each section with its display title
        for section in self.sections:
            if section.display_title:
                emoji = self._get_emoji_for_section(section.section_type)
                result.append(f"### {emoji} {section.display_title.upper()}")
            
            # Add content, maintaining paragraph breaks
            result.append(section.content)
            result.append("")  # Empty line
            
        # Add summary points if available
        if self.key_summary_points and len(self.key_summary_points) > 0:
            result.append("### 📋 KEY SUMMARY POINTS")
            for point in self.key_summary_points:
                result.append(f"- {point}")
            result.append("")
            
        # Add sources if available
        if self.sources and len(self.sources) > 0:
            result.append("### 📚 SOURCES")
            for source in self.sources:
                result.append(f"- {source}")
        
        return "\n".join(result)
    
    def _get_emoji_for_section(self, section_type: str) -> str:
        """
        Get appropriate emoji for section type.
        
        Args:
            section_type: Type of section
            
        Returns:
            Emoji character
        """
        emoji_map = {
            "hook": "🎯",
            "key_points": "📊",
            "explanation": "💡",
            "detail": "📋",
            "context": "🔍",
            "conclusion": "🏁",
            "call_to_action": "⚡"
        }
        return emoji_map.get(section_type, "✨")
        
    def get_legacy_format(self) -> Dict[str, Any]:
        """
        Get script in the legacy format for backward compatibility.
        
        Returns:
            Dictionary in the original format used by the system
        """
        # Extract full script text
        script_text = self.get_full_script_text()
        
        # Find call_to_action text
        cta_text = self.call_to_action or ""
        for section in self.sections:
            if section.section_type == "call_to_action":
                cta_text = section.content
                break
        
        # Build legacy format
        return {
            "title": self.title,
            "script": script_text,
            "callToAction": cta_text,
            "factCheckConfirmation": self.fact_check_confirmation or "",
            "keySummaryPoints": self.key_summary_points or [],
            "sources": self.sources or []
        }
        
def convert_legacy_to_script(legacy_data: Dict[str, Any]) -> VideoScript:
    """
    Convert legacy script data to VideoScript format.
    
    Args:
        legacy_data: Legacy script data dictionary
        
    Returns:
        VideoScript object
    """
    sections = []
    
    # Extract script content and try to split into logical sections
    script_content = legacy_data.get("script", "")
    
    # Try to find a hook (first sentence or paragraph)
    hook = ""
    remaining_script = script_content
    if "." in script_content:
        hook_parts = script_content.split(".", 1)
        if len(hook_parts) > 0:
            hook = hook_parts[0].strip() + "."
            remaining_script = hook_parts[1].strip()
    
    # Add hook if found
    if hook:
        sections.append(
            VideoScriptSection(
                section_type="hook",
                content=hook,
                display_title="Hook"
            )
        )
    
    # Add main content
    if remaining_script:
        sections.append(
            VideoScriptSection(
                section_type="explanation",
                content=remaining_script,
                display_title="Main Content"
            )
        )
    
    # Add call to action if present
    call_to_action = legacy_data.get("callToAction", "")
    if call_to_action:
        sections.append(
            VideoScriptSection(
                section_type="call_to_action",
                content=call_to_action,
                display_title="Call To Action"
            )
        )
    
    # Create VideoScript object
    return VideoScript(
        title=legacy_data.get("title", "Video Script"),
        sections=sections,
        key_summary_points=legacy_data.get("keySummaryPoints", []),
        sources=legacy_data.get("sources", []),
        call_to_action=call_to_action,
        fact_check_confirmation=legacy_data.get("factCheckConfirmation", "")
    )