"""Prompt templates and response schema for anomaly explanation LLM calls."""

SYSTEM_PROMPT = """You are a data-quality and macroeconomic diagnostics assistant that validates anomalies in time-series indicators such as those from the World Development Indicators (WDI).

GOAL:
Identify and validate anomalies in indicator time series, explain the most likely verifiable causes, and classify each anomaly window.

CONSTRAINTS:
- Use only historically documented, verifiable facts — including well-known global or national events that are publicly recognized and corroborated by multiple reputable sources (e.g., natural disasters, wars, global crises, major policy reforms).
- You may use your general world knowledge to recall such events if they are clearly established in history.
- Do not speculate or hypothesize beyond documented or widely recognized events.
- If you cannot recall a clear, well-documented event that explains the anomaly, classify it as "insufficient_data".
- Provide concise, factual explanations; avoid storytelling or uncertain reasoning.
- Cite specific events or sources whenever possible with approximate date ranges and event types.
- Never reveal internal reasoning — output only the structured JSON that matches the schema.
- Only mark an anomaly as valid (is_anomaly=true) when there is at least one well-documented event or data-related cause.
- Prefer "insufficient_data" over creative inference when unsure.
- When in doubt, lower confidence ≤ 0.5.
- Use "YYYY-MM" format in `date_range`.
- Do not explain imputed values."""

USER_PROMPT_TEMPLATE = """# TASK
Validate the anomalies in the time series below, explain their most likely verifiable causes, and classify each anomaly window.

# ANALYSIS RULES
1. Treat anomalies as windows ([start, end]), not individual points; merge contiguous anomalous years.
2. Confirm anomalies only if they align with a verifiable event or clear data-quality issue.
3. The time series includes imputed values indicated by the "Imputed" column. Do not attempt to explain these values.
4. You may use general, well-documented historical knowledge (e.g., wars, natural disasters, global crises, pandemics, major policy reforms, or statistical revisions) when such events are clearly established in history and widely recognized.
5. If there is no match with known history or documented statistical events set is_anomaly=false and classification="insufficient_data".
6. Use one of these primary classifications:
   - "data_error" — placeholder, rounding, rebasing artifact, template issue, ingestion error, logical computation impossibility.
   - "external_driver" — macroeconomic or geopolitical event, conflict, policy reform, disaster, pandemic, global cycle.
   - "measurement_system_update" — rebasing, SNA/PPP revision, new census benchmark, classification change.
   - "modeling_artifact" — anomaly detector or transformation artifact.
   - "insufficient_data" — no verifiable cause.
7. Assign "evidence_strength" as one of:
   - "strong_direct" — clearly linked, well-documented event or revision.
   - "moderate_contextual" — plausible contextual relationship but without direct documentation.
   - "weak_speculative" — weak or uncertain linkage, minimal documentation.
   - "no_evidence" — used when no evidence_source applies (e.g., data error or missing data).
   Only "strong_direct" counts as a valid confirmation for is_anomaly=true.
8. Include an "evidence_source" list only when you can name a specific, verifiable event or document.
   - Use concise factual titles (e.g., "COVID-19 pandemic", "2014 GDP rebasing, Nigeria").
   - Do NOT invent or generalize sources (e.g., "IMF", "World Bank") to fill the field.
   - If no such source exists, use "evidence_source": [] and, if appropriate, set "verifiability": "not_applicable".
   - For generic data or reporting errors (e.g., implausible magnitudes, placeholders, ingestion artifacts), an empty evidence_source is expected and correct.
9. Keep explanations concise and factual, citing only verifiable events or mechanisms; avoid speculation or uncertain reasoning.
10. Output ONLY valid JSON matching the provided schema—no markdown, comments, or extra text.

# CONTEXT
{{INPUT_SERIES_INFO}}"""


def get_anomaly_response_format() -> dict:
    """Return the JSON schema for structured anomaly explanation responses."""
    from ai4data.anomaly.explanation.schemas import AnomalyExplanation

    schema = AnomalyExplanation.model_json_schema()
    return {
        "type": "json_schema",
        "json_schema": {
            "name": "anomaly_explanation",
            "strict": True,
            "schema": schema,
        },
    }