# From query_generation.py:17-21
class QueryGenerator:
    """LLM-powered query strategy generator using Google Gemini.

    Generates 8-12 diverse strategies across channels depending on search depth.
    """

# From search.py:14-24
def build_query(self, company_domain: str) -> str:
    # jobs_search doesn't use domain/company placeholders
    if self.channel == "jobs_search":
        return self.query_template
        
    # Other channels use domain/company substitution
    company_name = company_domain.split(".")[0]
    
    return self.query_template.format(
        DOMAIN=company_domain, COMPANY_NAME=company_name
    )

# From query_generation.py:115-120 (system instruction)
"""Score each strategy based on:
1. Specificity: More specific queries score higher than broad ones
2. Evidence Quality: Strategies likely to find direct evidence score higher
3. Source Reliability: Company websites and job postings often score higher than general web searches
4. Keyword Relevance: Strategies using exact research goal keywords score higher"""

# From query_generation.py:222-240
def _is_valid_strategy(self, strategy_data: dict, required_fields: set, supported_channels: set) -> bool:
    # Fast set membership checks
    if not required_fields.issubset(strategy_data.keys()):
        return False
    
    if strategy_data["channel"] not in supported_channels:
        return False
    
    # jobs_search doesn't need domain/company placeholders
    channel = strategy_data["channel"]
    template = strategy_data["query_template"]
    
    if channel == "jobs_search":
        return True  # Any template is valid for jobs_search
    else:
        # Other channels require domain/company placeholders
        return "{DOMAIN}" in template or "{COMPANY_NAME}" in template

# From pipeline.py:145-152
tasks: List[asyncio.Task[Tuple[str, SourceResult]]] = []
for domain in self.company_domains:
    for strategy in self.strategies:
        tasks.append(
            asyncio.create_task(
                self._execute_one(domain, strategy, self.search_depth)
            )
        )

# Quick: Focus on highest-yield sources
{
  "search_depth": "quick",
  # Generates: Company website, jobs, news (4-6 strategies)
}

# Standard: Balanced coverage
{
  "search_depth": "standard",
  # Generates: All sources, moderate depth (7-10 strategies)
}

# Comprehensive: Maximum evidence
{
  "search_depth": "comprehensive",
  # Generates: All sources, maximum depth (11-13 strategies)
}

# From routes.py:28-35
query_generator = QueryGenerator()
strategies = await query_generator.generate_strategies(
    research_goal=payload.research_goal,
    search_depth=payload.search_depth,
)
end_time = time.time()
print(f"Query generation time: {end_time - start_time} seconds")