from pydantic import BaseModel, Field
from typing import List, Optional

class PersonaStakeholder(BaseModel):
    id: str = Field(description="Unique persona identifier from persona_pedagogy.db")
    name: str = Field(description="Canonical persona name")
    role: str = Field(description="Specific domain role title")
    tag: str = Field(description="Stakeholder perspective tag (e.g. Working Constituent)")
    quote: str = Field(description="Grounded dialectical take on the legislative mandate")

class OfficeIntelligence(BaseModel):
    hiring_manager_name: str
    hiring_manager_title: str
    district_desc: str = Field(description="Geographic and demographic footprint")
    district_issue: str = Field(description="Primary policy or regulatory friction")
    staffer_hook: str = Field(description="Tailored opening value proposition")
    persona_panel: List[PersonaStakeholder]
    recommendations: List[str] = Field(min_length=3, max_length=3)

class CapitolApplicationManifest(BaseModel):
    target_id: str
    job_id: int
    chamber: str  # Assembly or Senate
    principal: str
    target_position: str
    salary_range: str
    contact_email: str
    office_intelligence: OfficeIntelligence

def build_executive_deliberation(job_manifest: dict, candidate_background: dict) -> str:
    """
    Fog Bank Studio multi-persona graph node. Pits 4 distinct personas against 
    the job mandate to derive robust, defensible recommendations.
    """
    personas = [
        "everyday_mechanic_frank",    # Constituent cost-of-living lens
        "gov_regulatory_inspector",    # CEQA & Administrative Law lens
        "econ_macro_strategist",       # General Fund deficit stress-test
        "rhetoric_speechwriter_orator" # 60-second floor speech clarity
    ]
    
    deliberation_context = {
        "mandate": job_manifest["target_position"],
        "principal": job_manifest["principal"],
        "district_issue": job_manifest["office_intelligence"]["district_issue"],
        "candidate_assets": candidate_background["prior_service"]
    }
    
    # Deterministic dialectic resolution without big-tech cloud routing
    return render_deliberation_brief(deliberation_context, personas)

def audit_cognitive_clarity(prose_text: str) -> dict:
    """
    Enforces Cognitive Clarity & Federal Plain Language standards:
    - Active voice >= 85%
    - Sentence length <= 20 words
    - Zombie noun density <= 5% (eliminates -tion, -ment, -ance bloat)
    """
    sentences = tokenize_sentences(prose_text)
    zombie_nouns = [w for w in prose_text.split() if w.endswith(('tion', 'ment', 'ance'))]
    nominalization_density = len(zombie_nouns) / max(1, len(prose_text.split()))
    
    assert nominalization_density <= 0.05, "Reject: bureaucratic padding detected"
    return {
        "flesch_reading_ease": calculate_flesch(prose_text),
        "zombie_density": nominalization_density,
        "plain_english_compliant": True
    }